a stabilized master laser system for differential absorption lidar · 2015-10-29 · a stabilized...
TRANSCRIPT
A Stabilized Master Laser System forDifferential Absorption LIDAR
by
Alex Dinovitser
Supervised by: Murray Hamilton and Robert Vincent
Thesis submitted for the degree of
Doctor of Philosophy
at
the University of Adelaide
Department of Physics
October 11, 2012
Contents
Abstract v
Declaration vii
Acknowledgements ix
List Of Figures xiii
List Of Tables xiv
List Of Acronyms xv
1 Atmospheric Remote Sensing of Water Vapour 1
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Water vapour sensing techniques . . . . . . . . . . . . . . . . . . . . . . 3
1.3 DIfferential Absorption Lidar (DIAL) . . . . . . . . . . . . . . . . . . . 10
1.4 Overview of modern water-vapour DIALs . . . . . . . . . . . . . . . . . 12
1.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2 Atmospheric Scattering and Spectroscopy 19
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.2 Propagation and scattering . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.3 Atmospheric spectroscopy . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.4 Line selection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5 Pressure shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.6 Spectral line and dither . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
i
ii CONTENTS
2.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3 DIAL Transmitter and System Design 37
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2 DIAL system design consideration . . . . . . . . . . . . . . . . . . . . . 37
3.3 Overview of this DIAL . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.4 On-line master laser control system . . . . . . . . . . . . . . . . . . . . 49
3.5 Off-line master laser control system . . . . . . . . . . . . . . . . . . . . 51
3.6 System timing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
3.7 Control system performance . . . . . . . . . . . . . . . . . . . . . . . . 54
3.8 Laser amplifier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
3.9 Proposed applications of laser control system . . . . . . . . . . . . . . . 57
3.10 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
4 Characterization, Calibration and Application 63
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.2 Humidity sensor calibration experiment . . . . . . . . . . . . . . . . . . 68
4.3 Spectral calibration experiment . . . . . . . . . . . . . . . . . . . . . . . 76
4.4 Atmospheric DIAL observation experiment . . . . . . . . . . . . . . . . 84
4.5 Extended observation experiment . . . . . . . . . . . . . . . . . . . . . . 89
4.6 Amplified Spontaneous Emission (ASE) measurement experiment . . . . 93
4.7 Optical amplifier power, ASE and alignment experiment . . . . . . . . . 96
4.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
5 Conclusion 101
A Electronic Systems 103
A.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
A.2 Digital and timing system . . . . . . . . . . . . . . . . . . . . . . . . . . 106
A.3 Analog wavelength control system . . . . . . . . . . . . . . . . . . . . . 108
A.4 Spectroscopic ratiometric detection systems . . . . . . . . . . . . . . . . 113
A.5 Power electronic pulse driver for laser amplifier . . . . . . . . . . . . . . 115
CONTENTS iii
A.6 Acousto-Optic Modulator (AOM) Radio Frequency (RF) driver . . . . . . 120
A.7 RF Beat Detector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
B Control System Model 125
B.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
B.2 Modeling the lock-in amplifier . . . . . . . . . . . . . . . . . . . . . . . 125
B.3 Simulation experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
C Master Laser Characterization Experiments 129
C.1 Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
C.2 Methods and results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
C.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
D On-line Extinction Measurement 141
E Master Laser Diode Wavelength Pair Matching 143
E.1 Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
E.2 Manual method using a slow ramp . . . . . . . . . . . . . . . . . . . . . 143
E.3 Laser characterization using a wavemeter . . . . . . . . . . . . . . . . . 145
F Dither Induced Offset Experiment 149
F.1 Aim . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
F.2 Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
F.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
G Matlab (Octave) Functions and Code Listing 153
H Publications 167
H.1 Stabilized master laser system for differential absorption lidar . . . . . . . 167
H.2 Transmitter design for differential absorption water vapour LIDAR . . . . 177
H.3 Towards low-cost water-vapour differential absorption lidar . . . . . . . . 183
H.4 Towards low-cost water-vapour differential absorption lidar . . . . . . . . 189
Bibliography 199
Short Abstract
This thesis presents a Differential Absorption Lidar (DIAL) based on a simple yet accu-
rate and robust dual master laser stabilization system built using optic fiber components.
A water-vapor absorption cell stabilizes the on-line wavelength, while the off-line wave-
length is beat-frequency stabilized using a 16 GHz bandpass filter. The Master Oscillator
Power Amplifier (MOPA) uses a Tapered laser to form the transmitted pulse. Calibration
and atmospheric measurements are demonstrated. The control system is built at the elec-
tronic component level, with schematics and code listings provided. The system can be
expanded for stabilization of multiple lasers.
iv
Abstract
In this thesis, we present a prototype water vapour DIfferential Absorption Lidar (DIAL)
instrument with accurate and precise wavelength control of master diode lasers. This
stabilization system design has a number of novel elements that work towards a robust
and low-cost autonomous DIAL observatory. With two continuous wave optical wave-
lengths stabilized, a pulse is formed using an Acousto-Optic Modulator (AOM) to switch
light out of each control system to form the transmitted pulse. The control systems em-
ploy synchronous reference signal detection that suppresses system perturbations due to
the optical switching, facilitating the use of deep dither modulation that aids in accurate
stabilization to weak absorption lines. Furthermore, ratiometric detection in the control
loop suppresses interference caused by back reflections in optical fiber components, as
well as amplitude modulation of the laser diode due to injection current. In our system,
the first laser is stabilized to an absorption line of a water vapour cell, while the second
is beat-frequency stabilized relative to the first using a passive 16 GHz bandpass filter.
This technique can be expanded to stabilize any number of reference lasers with respect
to each other and to an absolute optical standard. The prototype DIAL uses a Tapered
optical Amplifier (TA) to form 1 µs 500 mW optical pulses with a repetition rate of >3
kHz for atmospheric transmission. Fourteen observation experiments were conducted
over two years, with water vapour measurements obtained using a calibrated humidity
sensor, using three saturated salt solutions as humidity references. The measured pulse
extinction was used to calculate the effective absorption cross-section of the transmitter,
and therefore used to calculate quantitative water vapour measurements from the DIAL
observation data. It is hoped that this work will be useful to the further development and
commercialization of this unique and powerful remote sensing technique.
v
vi ABSTRACT
Declaration
I certify that this work contains no material which has been accepted for the award of any
other degree or diploma in any university or other tertiary institution and, to the best of
my knowledge and belief, contains no material previously published or written by another
person, except where due reference has been made in the text. In addition, I certify that
no part of this work will, in the future, be used in a submission for any other degree or
diploma in any university or other tertiary institution without the prior approval of the
University of Adelaide. I give consent to this copy of my thesis when deposited in the
University Library, being made available for loan and photocopying.
I also give permission for the digital version of my thesis to be made available on the web,
via the Universitys digital research repository, the Library catalogue and also through web
search engines.
This work is also licensed under a Creative Commons Attribution-NonCommercial-ShareAlike
3.0 Unported License.
SIGNED: ....................... DATE: .......................
vii
viii DECLARATION
Acknowledgements
My main gratitude is, of course, to the supervisors Murray Hamilton and Robert (Bob)
Vincent. When I embarked on this project in 2007, some of the components such as the
entire receiver system, components for the vapour cell and some of the RF electronics for
the passive stabilization system, as well as the laser components were provided by Murray.
Murray also contributed to the development of the ratiometric detection system that made
it possible to assemble a large part of this instrument with fiber optics. I also thank
them for their patience, their tolerance, their guidance, and for sharing the excitement
for this project; Murray, for his strict rigor and encouragement to work harder, AND for
never saying no to an idea no matter how unlikely he may have thought it would work.
Special thanks to Bob Vincent, for his encouragement and for sharing his passion for the
atmospheric sciences that motivated this whole project, as well as for the insight into the
physics and atmospheric measurement techniques.
Many thanks to all my colleagues, past and present in the optics and atmospheric
groups, with particular thanks to Tom Rutten who was the first to show me the ropes,
David Ottaway and Nikita Simakov, with whom we had many productive and interest-
ing discussions on lasers, physics, among other subjects. Thanks to Trevor Waterhouse,
for trusting me with the workshop and letting me make stuff for this project (before the
lawyers took over). Blair Middlemiss, for his technical help and equipment, Max Lohe
and Sergey Cherkis, for a sounding board on the beat spectra problem and Marthinus van
der Westhuizen for proofreading and wise advice. Thanks also to Peter Veitch for lending
me the Pellicle beamsplitter.
Special thanks to Nick Chang, David Hosken, Matthew Heintze, Keiron Boyd, Miftar
Ganija and others for making me welcome at the start and at the end of this journey.
ix
x ACKNOWLEDGEMENTS
Thanks also to all the school of chemistry and physics staff, facilities and the helpful,
productive environment that was provided.
Finally, a very big thanks to my partner Lyndia, for her love and support, her encour-
agement and tolerance, and her very positive outlook, without which none of this would
have happened. Thanks to my Parents Anna and Emmanuel, for their support, their pa-
tience and their encouragement for me to finish this thesis.
I have a terrible amnesia for people, as Bob would know, and have been known to
forget to acknowledge the most worthy. Consequently, I feel it is likely I have forgotten
someone. If this has occurred, I hope I can be forgiven.
List of software used in this work
Credit must also be given to the tools used to do and create this work.
• Labview
• Matlab, Octave and Mathematica
• Latex, Okular, Texstudio and Texmaker
• Inkscape, Xfig and Gimp
• Kicad and GEDA
• Linux, Ubuntu and KDE
List of Figures
1.1 LIDAR configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.2 Illustration of LIDAR and DIAL . . . . . . . . . . . . . . . . . . . . . . 13
1.3 Spectral attenuation in DIAL . . . . . . . . . . . . . . . . . . . . . . . . 13
2.1 Water vapour resonance line . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2 Lorentzian and Voigt line shape comparison at STP . . . . . . . . . . . . 28
2.3 Lorentzian and Voigt line shape comparison at 16 km . . . . . . . . . . . 28
2.4 Water vapour absorption spectra 0.5-2.0 µm . . . . . . . . . . . . . . . . 29
2.5 Temperature sensitivity model results . . . . . . . . . . . . . . . . . . . 30
2.6 Pressure shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
2.10 Laser power modulation induced error . . . . . . . . . . . . . . . . . . . 35
3.1 Single laser dual wavelength system . . . . . . . . . . . . . . . . . . . . 41
3.2 Dual laser dual wavelength system . . . . . . . . . . . . . . . . . . . . . 41
3.3 Tapered optical Amplifier (TA) Diagram . . . . . . . . . . . . . . . . . . 45
3.4 DIAL control system diagram . . . . . . . . . . . . . . . . . . . . . . . 47
3.5 DIAL instrument photograph 1 . . . . . . . . . . . . . . . . . . . . . . . 48
3.6 DIAL instrument photograph 2 . . . . . . . . . . . . . . . . . . . . . . . 48
3.7 DIAL timing diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.8 Synchronous noise rejection . . . . . . . . . . . . . . . . . . . . . . . . 54
3.9 Control system model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.10 Wavelength perturbation results . . . . . . . . . . . . . . . . . . . . . . 55
3.11 Master wavelength stability results . . . . . . . . . . . . . . . . . . . . . 56
3.12 Wavelength stability histogram . . . . . . . . . . . . . . . . . . . . . . . 56
xi
xii LIST OF FIGURES
3.13 Optical amplifier photograph . . . . . . . . . . . . . . . . . . . . . . . . 58
3.14 Locking to an arbitrary beat frequency . . . . . . . . . . . . . . . . . . . 59
3.15 Trace gas detection using Continuous Wave (CW) THz . . . . . . . . . . 60
3.16 Locking multiple off-line wavelengths . . . . . . . . . . . . . . . . . . . 61
4.1 Master laser optical spectrum . . . . . . . . . . . . . . . . . . . . . . . . 66
4.2 Calibration experiment setup . . . . . . . . . . . . . . . . . . . . . . . . 71
4.3 Calibration experiment photograph . . . . . . . . . . . . . . . . . . . . . 72
4.4 Mounted humidity sensor photograph . . . . . . . . . . . . . . . . . . . 73
4.5 Calibration result: least-squares plot . . . . . . . . . . . . . . . . . . . . 74
4.6 Sensor stabilization time . . . . . . . . . . . . . . . . . . . . . . . . . . 76
4.7 Spectral calibration layout . . . . . . . . . . . . . . . . . . . . . . . . . 79
4.8 Measured absorption spectra . . . . . . . . . . . . . . . . . . . . . . . . 81
4.10 Atmospheric observation experiment setup . . . . . . . . . . . . . . . . . 85
4.11 DIAL Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
4.14 DIAL Extended observation results . . . . . . . . . . . . . . . . . . . . . 91
4.15 Optical amplifier alignment fringes . . . . . . . . . . . . . . . . . . . . . 94
4.16 ASE measurement results . . . . . . . . . . . . . . . . . . . . . . . . . . 95
4.17 Relative on-line transmission . . . . . . . . . . . . . . . . . . . . . . . . 95
4.18 Output power and spectral purity results . . . . . . . . . . . . . . . . . . 99
A.2 Main electronics panel photo . . . . . . . . . . . . . . . . . . . . . . . . 104
A.3 System overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
A.4 Timing schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
A.5 Synchronous transient suppression illustration . . . . . . . . . . . . . . . 109
A.6 Analog control system schematic . . . . . . . . . . . . . . . . . . . . . . 112
A.7 Ratiometric detection system . . . . . . . . . . . . . . . . . . . . . . . . 114
A.8 Tapered laser amplifier photograph . . . . . . . . . . . . . . . . . . . . . 118
A.9 Photovoltaic TA alignment . . . . . . . . . . . . . . . . . . . . . . . . . 118
A.10 TA driver schematic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
A.11 Pulse driver performance results . . . . . . . . . . . . . . . . . . . . . . 120
LIST OF FIGURES xiii
A.12 AOM RF driver schematic . . . . . . . . . . . . . . . . . . . . . . . . . 122
A.13 Beat frequency detector schematic . . . . . . . . . . . . . . . . . . . . . 123
B.1 Control system model and experimental results . . . . . . . . . . . . . . 127
B.2 DIAL control system model diagram . . . . . . . . . . . . . . . . . . . . 128
E.1 Laser diode matching . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
List of Tables
4.1 %RH of saturated salt solutions . . . . . . . . . . . . . . . . . . . . . . . 69
4.2 Sensor calibration results . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.3 HITRAN parameter update table . . . . . . . . . . . . . . . . . . . . . . 77
4.4 Measured absorption spectral results . . . . . . . . . . . . . . . . . . . . 82
4.5 Humidity measurement results . . . . . . . . . . . . . . . . . . . . . . . 82
4.6 Voigt model comparison . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.7 DIAL observation results summary . . . . . . . . . . . . . . . . . . . . . 88
4.8 DIAL-radiosonde comparison . . . . . . . . . . . . . . . . . . . . . . . 88
4.9 Measured absorption spectral results . . . . . . . . . . . . . . . . . . . . 91
4.10 Humidity measurement results . . . . . . . . . . . . . . . . . . . . . . . 91
4.13 DIAL-Radiosonde Comparison . . . . . . . . . . . . . . . . . . . . . . . 92
xiv
Acronyms
AERI Atmospheric Emitted Radiance Interferometer
AIRS Atmospheric Infra-Red Sounder
AMSU Advanced Microwave Sounding Unit
AOM Acousto-Optic Modulator
APD Avalanche Photo Diode
AR Anti Reflection
ARM Atmospheric Radiation Measurement facility
ASE Amplified Spontaneous Emission
ATMS Advanced Technology Microwave Sounder
AVHRR Advanced Very-High Resolution Radiometer
BOM Bureau Of Meteorology
BP Band-Pass filter
CHAMP CHAllenging Minisatellite Payload
CODI COmpact water vapour DIal
CW Continuous Wave
DAQ Data AQuisition
DFB Distributed Feedback diode laser
xv
xvi ACRONYMS
DIAL DIfferential Absorption Lidar
DLR Deutsches zentrum fur Luft- und Raumfahrt
DOE Department Of Energy
ECDL External Cavity Diode Laser
ESR Equivalent Series Resistance
FIR Far Infra-Red
FM Frequency Modulation
FOV Field Of View
FP Fabry-Perot
FSR Free Spectral Range
FTIR Fourier Transform Infra-Red (spectrometer)
FWHM Full Width at Half Maximum
GDPFS Global Data processing and Forecasting System
GHG Green-House Gas
GIFTS Geosynchronous Imaging Fourier Transform Spectrometer
GNSS Global Navigation Satellite System
GOES Geostationary Operational Environmental Satellite(s)
GPS Global Positioning System
GRACE Gravity Recovery And Climate Experiment
GRAS GNSS Receiver for Atmospheric Sounding
HIRS High-Resolution Infrared Radiation Sounder
HITRAN HIgh resolution TRANsmission
ACRONYMS xvii
HWHM Half Width at Half Maximum
IASI Infrared Atmospheric Sounding Interferometer
IHOP International H2O Project
IR Infra-Red
LASE Laser Atmospheric Sensing Experiment
lidar LIDAR
LIDAR LIght Detection And Ranging
LITE Lidar In space Technology Experiment
LPF Low Pass Filter
MODIS Moderate-Resolution Imaging Spectroradiometer
MOPA Master Oscillator Power Amplifier
MSLP Mean Sea Level Pressure
NASA National Aeronautics and Space Administration
NOAA National Oceanic and Atmospheric Administration
NPOESS National Polar-orbiting Operational Environmental Satellite System
NPP NPOESS Preparatory Project
NWP Numerical Weather Prediction
OPO Optical Parametric Oscillator
OSA Optical Spectrum Analyzer
PID Proportionl Integral Differential
PIN P-Intrinsic-N doped semiconductor
PMT Photo Multiplier Tube
xviii ACRONYMS
QPF Quantitative Precipitation Forecasting
RAM Residual Amplitude Modulation
RF Radio Frequency
RH Relative Humidity %
SG Savitzky-Golay
SSM/I Special Sensor Microwave Imager
STP Standard Temperature and Pressure
TA Tapered optical Amplifier
TPW Total Precipitable Water
TRMM Tropical Rainfall Measuring Mission
VCO Voltage Controlled Oscillator
VCSEL Vertical Cavity Surface Emitting Laser diode
VHF Very High Frequency
WMO World Meteorological Organization
WM Wavelength Modulation
Chapter 1
Atmospheric Remote Sensing of Water
Vapour
1.1 Introduction
The Earth’s climate depends on the radiative output of the sun and the balance between
short wave radiation absorbed from the sun and the long-wave radiation emitted into space
from the surface, clouds, and atmosphere. Our current understanding of the Earth’s at-
mosphere comes from observations of the atmospheric temperature, ocean surface and
land surface temperature, as well as humidity, clouds, albedo, and distribution of green-
house gases. Water vapour is responsible for 60% of the so-called greenhouse effect,
followed by carbon dioxide (26%), ozone (8%) and methane and others (6%) (Kiehl
and Trenberth, 1997). Quantitative measurement of water vapour in the atmosphere is
therefore of considerable interest for climate research as well as for weather forecast-
ing. Much uncertainty in atmospheric models arises from the way water vapour changes
phase, rapidly releasing and absorbing latent heat while driving the convection in the tro-
posphere. This phase change simultaneously alters atmospheric radiant emissivity and so-
lar albedo, which dominates the net radiative cooling of the troposphere (Sherwood et al.,
2010). With the strong temperature dependence of the saturation pressure and evapoura-
tion rates, water vapour produces a strong positive feedback on climate changes driven by
other greenhouse gases such as anthropogenic CO2 (Hansen, 2008). Indeed, tropospheric-
1
2 CHAPTER 1. ATMOSPHERIC REMOTE SENSING OF WATER VAPOUR
averaged column water vapour has increased by 4% since 1970, in line with increasing
sea surface temperatures (Trenberth et al., 2007). These phenomena are responsible for
much of the complexity in modeling of the hydrological cycle, aerosol-cloud interactions,
the energy budget and its interaction with the other greenhouse gases. Water vapour is
therefore considered the most important trace gas in the atmosphere (Schneider et al.,
2010).
Even though there are several techniques for the remote sensing of water vapour, pre-
cise and extensive measurement of atmospheric state and composition with good hori-
zontal and vertical resolution is an ongoing challenge and a significant obstacle to further
advancement of forecasting models. The lack of coverage, accuracy and resolution in the
current observing systems makes accurate Quantitative Precipitation Forecasting (QPF)
a more challenging objective due to the large data gaps in the initial state (Wulfmeyer
et al., 2006). For this reason, the skill of short- and medium range QPF is not sufficient
to serve many user communities. These deficiencies result in a lack of accuracy and cer-
tainty in the prediction of extreme weather events where larger amounts of precipitation
are involved, such as the recent floods in Australia. Furthermore, these data gaps hamper
the understanding of the atmospheric processes and conditions by which precipitation is
initiated (Girolamo et al., 2008).
The standard weather modeling approach is to use explicit equations of motion trun-
cated at some prescribed scale, with empirical parametrization models for the smaller
scales. Despite great progress over the past 50 years, weather models are still far from
being a perfect representation of reality (Sherwood et al., 2010). It is believed that param-
eterizations of atmospheric processes are the main sources of error, but it is not known
whether the right parameterizations are yet to be formulated or perhaps the current mod-
els simply need to have the parameters tweaked. Perhaps the current models need to be
run at a higher resolution with the parametrization of smaller scales, or perhaps a totally
different methodology is needed. By acquisition of atmospheric data of higher accuracy
and resolution, over larger spatial and temporal domains, these questions can begin to
be addressed. There are currently several major developments on this front. For ex-
ample, the Australian Bureau Of Meteorology (BOM) is currently installing nine new
1.2. WATER VAPOUR SENSING TECHNIQUES 3
Very High Frequency (VHF) ground based radars around the country, and at the same
time rolling out new products such as the next-generation forecasting system (Bureau Of
Meteorology (BOM), 2011).
DIfferential Absorption Lidar (DIAL) is a promising type of LIght Detection And
Ranging (LIDAR) remote sensing technique that utilizes the elastic scattering cross-
section of aerosols and molecules, and a molecular resonance frequency specific to a
particular molecule. These characteristics result in a number of unique advantages over
other technologies, including high absolute accuracy, sensitivity and range resolution
(Wulfmeyer et al., 2005). Compared to other laser remote sensing techniques that use
inelastic scattering, DIAL is potentially a more suitable choice where power supply, cost
or instrument size and weight are important. DIAL may therefore be well suited to both
low-cost ground-based installation, where a large number of instruments are desired to
cover a geographic area, as well as for space-based applications (Browell et al., 1979)
where power consumption and size are of prime concern (Browell et al., 1998). Such
applications could improve the accuracy of forecasts with the current models and lead to
further improvements in the weather and climate models themselves.
However, despite many decades of DIAL development, there is no continuously op-
erational instrument in existence due at least in part, to the complexity of achieving the
required laser power and spectral characteristics. This project addressed some of these
difficulties with the design, prototyping and utilization of a DIAL instrument for atmo-
spheric observations of water vapour.
In the following chapter we survey remote sensing techniques before focusing on
DIAL, while the following chapters describe our work including the dual wavelength
stabilized control system, that is a step towards a low-cost, reliable, self-contained stand-
alone operational instrument.
1.2 Water vapour sensing techniques
Broadly speaking, water vapour can be detected remotely by either passive or active
means. Passive techniques involve the interaction of ambient radiation with the vapour,
4 CHAPTER 1. ATMOSPHERIC REMOTE SENSING OF WATER VAPOUR
while active sensing by radar or LIDAR involves the transmission and reception of radi-
ation for interaction with atmospheric constituents. Passive techniques employ relatively
simple instruments with no transmitter, however, there are severe limitations on accuracy
and range resolution that can be achieved. The use of ever higher spectral resolutions and
more sophisticated numerical data analysis techniques over the past three decades has pro-
duced only modest improvements (W. Smith et al., 2008). Active profiling techniques, by
comparison achieve high accuracy and resolution while presenting fewer computational
challenges for data inversion, however, these also require more complex and expensive
instruments to acquire the raw data.
1.2.1 Radiometry
The passive measurement of emission spectra of the atmosphere in the mid-IR or the
absorption spectra of solar radiation at shorter wavelengths are currently among the tech-
niques used for acquiring an atmospheric constituent profile. The most widely used satel-
lite based technique known as multi-spectral or hyper-spectral imaging involves the mea-
surements of thermal emission lines of a particular species at various parts of the electro-
magnetic spectrum. These diverse data channels are combined to improve accuracy par-
ticularly where cloud attenuation is present (Baron et al., 2008) (Goldberg et al., 2003).
Data analysis uses advanced algorithms to calculate the total precipitable water profile.
However, the physical basis for all radiometric techniques is the inversion of the radiative
transfer equation at different wavelengths, which is the main source of its difficulty.
Even under the best circumstances, with independent calibration, the accuracy of
these techniques is limited to no more than 10% with a resolution of no better than 1 km
above the boundary layer (Feltz et al., 1998). The state of the art Atmospheric Infra-Red
Sounder (AIRS)+Advanced Microwave Sounding Unit (AMSU) instrument currently in
service on the Aqua satellite achieves 15% accuracy with a 2 km resolution in the tropo-
sphere above the boundary layer (Divakarla et al., 2006) (Aumann et al., 2003). The latest
operational instrument, the Advanced Technology Microwave Sounder (ATMS) (Muth
et al., 2004) launched as part of the NPOESS Preparatory Project (NPP), is yet to de-
liver results. An incremental refinement in space-borne instrumentation of this type has
1.2. WATER VAPOUR SENSING TECHNIQUES 5
been ongoing since the first specialized weather satellites of the early 1970s (Pierce et al.,
2006), (JPL, 2002), however, with an improvement in the optical spectral resolution of
two orders of magnitude since the 1980’s, the improvement in range resolution and ac-
curacy has been of the order of two or three (W. Smith et al., 2008). Despite significant
limitations in accuracy and resolution, the assimilation of data from satellite-borne in-
struments such as AIRS into numerical weather forecast models over the past decade has
been largely responsible for the great range and reliability improvements in weather and
climate forecasting. Furthermore, there are other significant gaps that radiometry cannot
address. These include the understanding of the physics of cloud formation and the role
that aerosols play.
1.2.2 Global Positioning System (GPS)
The GPS signal deployed for global positioning has precision time and phase coding
that makes it useful for measuring atmospheric refractive index by measuring the total
delay of the GPS signals form the constellation of satellites currently in orbit. It could
be considered a quasi-passive technique since it uses an artificial signal deployed for
another purpose. There are actually two GPS receiver techniques employed for atmo-
spheric water vapour and temperature probing; ground-based and space-based, which
have received considerable attention over recent years. Ground-based GPS employed
by the National Oceanic and Atmospheric Administration (NOAA) (NOAA, 2000) mea-
sures the total precipitable water vapour vertical column at a number of sites across the
USA, and by research facilities across the world. The widespread distribution of sites
provides an additional input to atmospheric models for weather forecasting. The space-
based satellite GPS radio occultation is a relatively new water vapour and temperature
profiling technique offering global coverage and long-term stability. This technique uses
orbiting receivers such as CHAllenging Minisatellite Payload (CHAMP) (Heise et al.,
2006), Gravity Recovery And Climate Experiment (GRACE) (Wickert et al., 2005) and
GNSS Receiver for Atmospheric Sounding (GRAS) (Luntama et al., 2008) to measure
the vertical refraction gradient as a function of the change of the measurement geometry
during a radio occultation event between one of the current operational GPS transmitting
6 CHAPTER 1. ATMOSPHERIC REMOTE SENSING OF WATER VAPOUR
satellites and the earth’s horizon. Since the refractivity of the atmosphere is a function of
temperature and water vapour concentration, with some assumption about the temperature
profile, the water vapour profile can be calculated. However, the ambiguity between tem-
perature and humidity in the troposphere results in some uncertainties (Kursinski et al.,
2001). Furthermore, water vapour never contributes more than about one third of the
total refractivity, even in the wet tropics, therefore measurement noise becomes signifi-
cant for dryer atmospheres. Errors in deriving the near-surface refractivity, together with
the dependence of accuracy and vertical resolution on atmospheric conditions, results in
considerable measurement uncertainties under most conditions (Kursinski et al., 1997).
However, the global coverage, combined with very good absolute accuracy of refraction
provides for very good long-term stability, making this a useful tool for long-term moni-
toring of climate change (Kursinski et al., 1997).
1.2.3 In-situ sampling
In-situ sampling involves direct sensor contact with the air, and can in principle provide
the most accurate measurements by which all other techniques can be calibrated. How-
ever, for routine measurements, such accuracy would be cost-prohibitive. The ubiquitous
expendable radiosonde is still the mainstay of weather forecasting, however, even with
a low, $300 cost instrument payload, this still represents a significant expense for me-
teorological services. With a 100-year legacy, radiosondes remain a vital input to all
mesoscale (∼100km) circulation models for weather forecasting (Edwards, 2001), as no
other currently deployed technology provides the required accuracy and resolution. Ra-
diosondes also provide the longest historical record of direct atmospheric measurement,
even though this data is punctuated by technological improvements in sensor technolo-
gies. Prior to 1990, the early humidity sensors were prone to icing which made some
results uncertain, while a lack of a GPS receiver resulted in uncertainty of the location
from where the data was acquired. Later instruments have been refined with new types
of sensors and telemetry, and the current Vaisala RS92 includes radio and GPS location
tracking.
Despite these improvements in the sensors and telemetry, the initial calibration of each
1.2. WATER VAPOUR SENSING TECHNIQUES 7
sonde remains a major issue. Pressure-independent random errors, with sonde-to-sonde
variability as high as 25% have been reported (Turner et al., 2004), where calibration
against a column average measurement by a microwave radiometer was used. A recent
study of radiosonde accuracy has found water vapour mixing ratios with dry biases of up
to 20% in the middle troposphere (Miloshevich et al., 2009) under dry conditions.
The radiosonde is not the only in-situ sampling technique available. One of earli-
est atmospheric sounding techniques was the dropsonde, where a hollow metal ball was
dropped from a great height. Radar tracking provided a measurement of its terminal ve-
locity from which atmospheric density and temperature could be calculated. In recent
years, commercial aircraft have been fitted with sensors that provide regular atmospheric
profiles near airports as planes take off and land.
The common factor in all types of in-situ measurement techniques, is that cost limits
the spatial and geographical distribution, as well as the accuracy of data that can be practi-
cally obtained. For this reason, other techniques have, and will continue to be developed.
1.2.4 LIDAR
LIght Detection And Ranging (LIDAR) is an active profiling technique that employs the
propagation of a pulse of radiation, and the detection of the backscattered signal, where
the time of flight gives the range, while the instantaneous return power provides a mea-
surement of scattering and attenuation at that range. The range resolution ∆l is given by
the speed of light c and laser pulse duration ∆l = 12c∆t. This means that range resolution
can be almost arbitrarily short.
Three basic arrangements are applicable to all lidar and radar active remote sensing
systems, as illustrated in figure 1.1. The idea of using light to probe the atmosphere
was first proposed by E. H. Synge in 1930 (Johnson et al., 1939) (Synge, 1930), who
proposed “A method for investigating the higher atmosphere ”up to a height of 50 km,
using a permanent assemblage of several hundred searchlights, since ’..the belligerent na-
tions in the late war have many thousands of searchlights on their hands,...’ Although his
proposal was never implemented, the idea of remotely measuring gas density as a func-
8 CHAPTER 1. ATMOSPHERIC REMOTE SENSING OF WATER VAPOUR
(a) Pseudo monostatic and Monostatic (b) Bistatic with one transmitter, or Multistaticwith two or more transmitters
Figure 1.1: Lidar configurations illustrating monostatic systems with infinite overlap be-tween receiver and transmitter, or complete overlap beyond some fixed range, and multi-static systems where the overlap between the transmitter and receiver has a finite extent
tion of wavelength and scatter, by utilizing a modulated light source, a telescope and a
detector was born (Clemesha et al., 1966). Through the 1950’s and at its peak around
the mid 1960’s, the searchlight technique competed with the emerging lidar, (known as
laser radar at the time). In 1966, the structure of aerosol scatter up to a height of 70 km
was profiled over a 6-hour period (Elterman, 1966) using searchlights. The first mea-
surement of the atmospheric scattering of laser radiation was reported in 1963 by Fiocco
and Smullin (Fiocco and Smullin, 1963), shortly after they successfully detected laser
echoes from the surface of the moon (Smullin and Fiocco, 1962). G. Elford began mak-
ing atmospheric observations in Adelaide in 1969. This instrument was based on a ruby
laser, and was used for regular observations until 1976. The results from these obser-
vations were used to calculate the tropospheric extinction due to aerosols (Young and
Elford, 1979). By 1994, LIDAR (lidar) had been deployed in the Lidar In space Technol-
ogy Experiment (LITE) (Winker et al., 1996), and in 2008, the first lidar experiment was
deployed on another planet, Mars, (Whiteway et al., 2008). These instruments had the
ability to measure backscatter, but unlike DIAL, had no capability for chemical species
sensing.
Raman LIDAR
The first detection of an atmospheric Raman return signal was made by Melfi et.al. in
1969 (Melfi et al., 1969), and with ongoing development, it has now become the most
1.2. WATER VAPOUR SENSING TECHNIQUES 9
widely used species profiling LIDAR technique. In Raman lidar, the detected signals
are due to inelastic interactions that produce characteristic energy (or wavelength) shifts
for each species. These emitted wavelengths can be discriminated from scattering by
other species using optical filters for the desired wavelength. The main advantage of
this technique is that unlike DIAL, there is no need for highly accurate laser wavelength
stabilization, so long as the detected wavelength falls inside the optical receiver’s filter
bandwidth. By detecting the elastic scatter, as well as the Raman scatter at two or more
spectral channels with a bistatic or multistatic system, it becomes possible to profile some
aerosol characteristics with a single transmitted wavelength (Philbrick et al., 2010). For
the profiling of molecular density of a variable species such as water vapour, the Raman
spectral returns from both water and nitrogen are acquired, and the vapour mixing ra-
tio profile is calculated from the ratio of the two received spectral channels, with some
corrections for the wavelength dependence of the optics and the atmosphere.
The main disadvantage of Raman lidar is the very low inelastic scattering cross sec-
tion, three orders of magnitude smaller than for Rayleigh scattering (Wandinger, 2005) at
a given wavelength. However, non-resonant Raman scattering occurs over a wide spectral
range. This allows for the use of highly efficient laser materials such as Nd:YAG to make
more powerful laser transmitters. Furthermore, since molecular scattering scales with
wavelength as λ−4, shorter wavelengths can be used, compared with DIAL where longer
wavelengths are required to reach adequately strong absorption bands. Nevertheless, the
scattering efficiency limits the application of the Raman technique to atmospheric species
with a relatively high concentration, as well as requiring a proportionally larger product
of laser power and receiver aperture, for a given measurement range. Furthermore, the
high laser power and large aperture requirements of Raman lidar make daytime operation
more difficult, as well as posing a problem for applications where space and power supply
is a premium, such as on space and air-borne platforms. Furthermore, the relatively large
wavelength shifts inherent in the inelastic scattering process, requires calibration for the
wavelength dependence in the efficiency of the optical receiving systems, as well as of the
transmission and attenuation of the atmosphere. In this sense, Raman lidar cannot be de-
scribed as self calibrating, unlike DIAL. The weakness of the Raman scattering, however,
10 CHAPTER 1. ATMOSPHERIC REMOTE SENSING OF WATER VAPOUR
is also one of its strengths. The first Stokes Raman returns for multiple species can be ac-
quired, in addition to the elastic scattered signal using multiple receiver channels. Since
the distributions of species like Nitrogen and Oxygen are well known, the lidar equation
for atmospheric attenuation and backscatter can be solved.
Raman lidar has been used in several field campaigns in recent years (Eichinger
et al., 1999) (Whiteman et al., 2006), including the International H2O Project (IHOP)
(UCAR, 2002) and Atmospheric Radiation Measurement facility (ARM) (DOE, 2011)
deployments. Furthermore, a water Raman instrument has been in regular operation
at MeteoSwiss since 2008 (Simeonov et al., 2010) with results assimilated into current
mesoscale forecast models (Calpini et al., 2011). These results are being used to fur-
ther advance these numerical models to improve the skill of short-range weather forecasts
(Grzeschik et al., 2008), especially with respect to cloud formation and precipitation.
1.3 DIfferential Absorption Lidar (DIAL)
A typical DIAL system transmits at least two wavelengths, one that coincides with a
molecular resonance, and another off-resonance wavelength. The return signal is due to
Rayleigh and Mie elastic scattering, while the dipole resonance results in attenuation of
the wavelength close to the center of the absorption line. Due to the narrow absorption
spectrum, the transmitted on-line laser wavelength needs to be precisely regulated with
good spectral purity. However, this also means that the off-line wavelength can be very
close, less than 20 GHz from the resonance peak. This virtually excludes the possibility
that any other phenomena can produce a difference in return signal at the two wavelengths.
For this reason the DIAL technique has been considered ’self calibrating’.
The earliest DIAL systems were built just a few years after the very first lidar experi-
ments, by R. M. Schotland who pioneered this technique (Schotland, 1964). By 1966 he
had utilized a temperature-tuned 3-level ruby laser for atmospheric observations (Schot-
land, 1966). Due to laser wavelength instabilities of these early systems, the H2O absorp-
tion coefficient was not known (Browell et al., 1979), and the vapour profile results could
therefore be considered only relative. Schotland improved the spectral characteristics of
1.3. DIfferential Absorption Lidar (DIAL) 11
this laser by using an intra-cavity etalon for wavelength selection, and an external Fabry-
Perot for monitoring output wavelength (Browell et al., 1979). The main difficulty with
using a ruby gain medium for water vapour was that the gain bandwidth centered at 694
nm only encompassed weak water absorption lines.
Schotland’s DIAL was an early example of a switched wavelength system, operating
a single laser for some considerable length of time at one wavelength, before switching
over to another wavelength. The advantage of this simple technique is that it requires a
single master laser with no need for rapid retuning. However, due to atmospheric changes,
the time delay between on-line and off-line acquisition introduces errors that are difficult
to quantify, even though they are reduced by averaging. In recent years, at least two
DIAL systems have employed the same technique using a diode laser (Machol et al.,
2004) (Nehrir et al., 2009b), with wavelength change-over times as short as 10 seconds.
A modern example of this technique uses a single laser with stable tuning characteristics
such that the wavelength can be switched in less than 40 ms (Koch et al., 2004), which
allows for consecutive pulses of alternating wavelengths to be transmitted.
Our system transmits consecutive pulses of alternating wavelength, however, we use
two continuously stabilized Fabry-Perot (FP) master lasers. This is the conventional
DIAL technique employed by most other groups involved in this field, where two or more
wavelengths are transmitted and received sequentially, and consecutive time-slots are al-
located to alternate wavelengths, as illustrated in figure 1.2. This time division multiplex-
ing of the wavelength does away with any need for separate wavelength-specific detector
channels, and can be implemented using a receiver consisting of a single detector with an
optical bandpass filter to reduce background. This is a significant advantage over the other
DIAL techniques since it facilitates the use of arbitrarily small optical wavelength offsets
without any trade offs involving channel separation and crosstalk. By allowing a sufficient
time interval between consecutive pulses, channel crosstalk is eliminated, while a particu-
larly small difference between the on-line and off-line wavelengths, as illustrated in figure
1.3, reduces errors due to instrumental and atmospheric transmission characteristics.
Another DIAL technique transmits multiple wavelengths simultaneously and uses
chromatic filters for the separate receiver channels. An early example of this technique
12 CHAPTER 1. ATMOSPHERIC REMOTE SENSING OF WATER VAPOUR
used a ruby laser as both a pump for another laser (on-line), as well as serving as the off-
line transmitter (Browell et al., 1979). However, a large wavelength difference between
the channels degraded the self-calibrating property of DIAL, while a small wavelength
difference introduces channel crosstalk that degrades DIAL accuracy.
The Raman DIAL is another example of this simultaneous wavelength - multiple
channel technique. However, in this case, one wavelength is transmitted through the
atmosphere, and two or more wavelengths are simultaneously generated by inelastic scat-
tering (Browell et al., 1979). Unfortunately, this only works for ozone because it of its
broad absorption line in the ultraviolet that accommodates the multiple Raman shifted
wavelengths inside the absorption feature (Lazzarotto et al., 1999). This actually is a
Raman technique, and therefore subject to the small inelastic scattering cross-section.
Another interesting modern example of the simultaneous wavelength - multiple chan-
nel technique, uses different dither frequencies of the two laser wavelengths, with elec-
tronic demodulation at the receiver to recover the relative strengths of the two separate
wavelength channels (Kameyama et al., 2008). However, this is only applicable to CW
DIAL, which only works for the measurement of a species column average.
Simultaneous wavelength - multiple channel DIAL techniques are therefore less ap-
plicable where sensitivity, precision as well as range resolved measurements are required.
1.4 Overview of modern water-vapour DIALs
In this section, we briefly summarize some of the published DIAL systems employing
contemporary technologies with aspects relevant to our requirements. This is far from a
comprehensive review, the aim here is merely to sample the scope of work being done in
this field. Some CO2 systems are also included where there is spectral proximity to water
absorption bands.
One of the first DIAL instruments to provide useful atmospheric results was con-
structed by Browell in 1979 (Browell et al., 1979). This used a ruby laser to pump a dye
laser, tunable for water lines within the 715-740 nm range. The ruby laser served two
1.4. OVERVIEW OF MODERN WATER-VAPOUR DIALS 13
Figure 1.2: This system sequentially transmits and acquires the returns of wavelengthsthat coincide with on-line and off-line absorption of a specific species. The differencein return provides a measurement of the absorbing species, while the time provides ameasurement of range.
822.85 822.9 822.95 8230
0.5
1
1.5x 10
−22
Wavelength nm
Mole
cula
r absorp
tion c
ross−
section c
m2
16 GHz
Laser #1Laser #2
Figure 1.3: Illustration of on-line and off-line attenuation as the principle of DIAL spec-troscopy. In this application, the off-line absorption cross-section at 16 GHz is 3% that ofthe on-line, at sea level.
14 CHAPTER 1. ATMOSPHERIC REMOTE SENSING OF WATER VAPOUR
functions, firstly, it pumped the on-line master laser, and secondly, it served as the off-line
wavelength tuned well away from any strong water absorption lines. The on-line dye laser
was tuned to the 724.372 nm water absorption line, and the absorption cross-section was
measured on a pulse-by-pulse basis using a multi-pass absorption cell with a 300 m path
length. The transmitted output energy was 250 mJ off-line and 165 mJ on-line with a 1
Hz repetition rate and a linewidth of 5 GHz. The on-line and off-line wavelengths were
transmitted simultaneously and coaxially, and the backscattered light received using a 0.5
m Newtonian telescope. The 30 nm difference made it possible to dichroically separate
the two wavelengths in the detector package. Results up to a 3 km altitude were consistent
with radiosonde data, that was estimated to have an accuracy of 9 %.
In 1998, the Max Plank institute reported development of their DIAL system based on
an Alexandrite ring laser with a Ti:sapphire master pumped with an Nd:YVO laser. This
system produced 50 mJ pulses at 15 Hz near 720 nm with a 99.99% spectral purity. In or-
der to achieve the required stability to prevent mode hops, and for single mode operation,
the master laser had to be carefully designed and operated in a temperature controlled
environment of ±0.1 K. The wavelength is stabilized to the edge of a fringe produced by
a computer stabilized interferometer, set with a wavemeter that was previously calibrated
to the absorption line using a slow scan of the absorption spectrum with a photoacous-
tic cell. The off-line wavelength is produced by the same master laser, using a Pockel
cell to produce a longitudinal mode hop (Wulfmeyer, 1998). The detector consists of a
14 cm telescope coupled to a variable optical attenuator and a single Avalanche Photo
Diode (APD) (Wulfmeyer and Bosenberg, 1998).
NASA had been actively developing water vapour DIAL systems throughout the 1990s.
By 2000, they had the Laser Atmospheric Sensing Experiment (LASE) instrument (Brow-
ell et al., 2000) as an airborne DIAL with an injection seeded Ti:Sapphire laser using an
unstable resonator, pumped by frequency doubled, flash lamp pumped Nd:YAG laser. It
operated near 815 nm with a spectral purity of 99%. The seed laser was a single mode
diode, frequency modulated and stabilized using a 200 m absorption cell. The absorp-
tion cell was also used to measure spectral purity. The control system locked to a strong
absorption line, and used electronic de-tuning away from line center for the desired cross-
1.4. OVERVIEW OF MODERN WATER-VAPOUR DIALS 15
section, depending on water concentration in the atmosphere, with an estimated accuracy
was 6% in the troposphere (Moore et al., 1996). The 40 cm receiver was coupled to two
APD channels for high and low gain. The APDs were EG&G model C30955E with a 1.5
mm diameter (Refaat, 2000).
In 2001, Bruneau et al published details of their LEANDRE II DIAL instrument. The
transmitter used a double-pulse (50 mJ x2, 10 Hz) Alexandrite laser with the spectral
separation between the pulses (0.44 nm) and a 50 µ m interval. Wavelength stabilization
of the self-seeded laser was by intra-cavity electro-optics (Bruneau et al., 2001a) (Bruneau
et al., 2001b) however, LEANDRE is no longer under active development.
A tapered optical amplifier was first used in the COmpact water vapour DIal (CODI)
instrument in 2004. This instrument tuned to three different wavelengths around 823 nm
using a Distributed Feedback diode laser (DFB) master laser manufactured by Sarnoff
Corporation. The single tapered amplifier with 0.5 W output and a 8 kHz repetition rate
produced ∼600 ns pulses. For wavelength control, the master laser was stabilized to the
edge of the resonance of an etalon made from ultra-low expansion glass. The entire instru-
ment was operated inside a temperature controlled enclosure stabilized with an accuracy
of better than 1°C (Machol et al., 2004). This etalon was automatically re-calibrated to
the required water absorption line every 30 minutes, to maintain a wavelength stabiliza-
tion accuracy of ±80 MHz. The system operated for 45 seconds on the on-line wavelength
and 10 seconds on the offline, taking 3 seconds to switch wavelengths. The spectral purity
was 99.9%. The receiver consisted of a 35 cm telescope coupled to a EG&G APD.
In 2007, Obland et al published DIAL results. This system consisted of a single
custom designed External Cavity Diode Laser (ECDL) at 830 nm with 17 nm tuning
range. Two tapered optical amplifiers were cascaded for the 400 mW output. The receiver
included a multimode fiber coupled APD detector. The pulse rate was 20 kHz with the
use of a ASRC Aerospace AMCS-USB card (Obland, 2007). The wavelength control is
essentially open-loop, using an optical spectrum analyzer as a reference.
Also in 2007, Deutsches Zentrum fur Luft- und Raumfahrt published their airborne
DIAL system that used four DFB diode seed lasers near 935 nm, for three on-line and
one off-line wavelength (Schwarzer et al., 2007). The multiple on-line wavelengths were
16 CHAPTER 1. ATMOSPHERIC REMOTE SENSING OF WATER VAPOUR
tuned to the centers of different strength lines, facilitating greater accuracy over a greater
range of water concentrations from 0 to 25 gkg . This facilitated water profile retrievals from
a greater range, almost approaching the stratosphere when operated from the ground. The
transmitter output stage is based on two injection seeded Optical Parametric Oscillator
(OPO)s pumped by a doubled Nd:YAG, Q-switched at 100 Hz. Measurement resolutions
of 1 km up to a 10 km range, and resolution of 2 km up to 16 km range were achieved.
Frequency stability of ± 30 MHz is reported, while the seed laser diodes themselves
produce a 30 MHz linewidth with a side-mode suppression of 50 dB. The first master
laser stabilization used a 36-m sealed vapour cell, a 60-kHz dither rate and a lock-in
amplifier. The other three seed lasers were locked to the master using wavemeters. In
2009, this system was rebuilt and upgraded for space qualification. Optical switching
with polarizing fiber components was performed synchronously with zero-crossing of the
dither (Wirth et al., 2009a) (Wirth et al., 2009b) like the system described in this thesis,
however unlike our system that requires no compensation, a constant offset was required
to compensate the stabilization error due to dither power modulation. Like many other
DIAL control systems, this used a computer and a wavemeter (HighFinesse-Angstrom
WS/7) to stabilize off-line wavelengths. With an output up to 12 W, this is one of the
most powerful DIAL systems to date.
In 2008, a DIAL instrument was combined with a Raman and Rayleigh lidar with a
80 cm receiver, making it one of the most sensitive. The DIAL system at the University
of Hohenheim, was operated in the 815-820 nm range with >4 W average power. This
resulted in the highest range resolved remote water vapour sensing worldwide (Pal et al.,
2008). The laser transmitter consisted of a Ti:Sapphire ring-resonator pumped by a dou-
bled Nd:YAG.(Behrendt et al., 2005) (Wagner et al., 2004). The detector consisted of a
dichroic beamsplitter with both a Photo Multiplier Tube (PMT) for short wavelengths and
a APD for longer wavelengths. The aerosol properties acquired with the Raman channels,
were also used to correct for the Rayleigh Doppler errors in the DIAL results. By 2008,
NASA developed a coherent CO2 DIAL (Koch et al., 2008), as a further development of
their previous work of 2004 (Koch et al., 2004). The 2 µm coherent CO2 DIAL system
uses a precisely controlled side-line wavelength to optimize DIAL sensitivity and preci-
1.5. SUMMARY 17
sion by adjusting the optical depth. This system used a total of three lasers, one reference
to the absorption line center, the side line laser stabilized by beat frequency method at a set
frequency between 0.1 and 2.9 GHz, and an un-stabilized off-line laser that drifts around
14 GHz from the line center. A side-line wavelength results in less on-line attenuation at
higher altitudes, resulting in a better SNR from a space-based platform. Furthermore, the
side-line wavelength did not need to be modulated to lock to the on-line laser, which is
itself locked to an absorption line.
Most recently, Montana State have published details of their first-generation system
using cascaded tapered amplifiers with peak optical power exceeding 1 W. A Littman-
Metcalf ECDL is stabilized using a wavemeter to the 828.187 nm line, a ∼3 GHz side line,
and ∼44 GHz off-line, operated at each wavelength for 10 seconds. The reported stability
is estimated at ±88 MHz, based on the resolution of the optical spectrometer, a Burleigh
wavemeter (WA-1500) used for the stabilization. A spectral purity of 99.5% was reported,
similar to the CODI instrument. These used the same type of tapered amplifier as our
instrument. A Geiger-mode APD is fiber coupled to a 28 cm telescope for a narrow Field
Of View (FOV). Daytime observation is reported, with good night-time observational
agreement with radiosondes up to 2.5 km (Nehrir et al., 2009b) (Nehrir et al., 2009a).
1.5 Summary
Accurate and well range-resolved measurement of water vapour is of great interest for
both meteorology and climatology. While the most common application of lidar is to ac-
quire a backscatter profile, DIAL, and Raman lidar techniques are available for species
profiling. The DIAL technique offers significant advances in this field, as well as for mea-
surement of other trace gases. However, despite decades of technological developments,
there are still no DIAL instruments in production or in permanent installation anywhere
in the world. One of the main reasons for this is the cost and difficulty of achieving the
necessary wavelength stability and spectral purity of the laser power transmitter.
In this thesis, we firstly delve into the physics of atmospheric spectroscopy responsible
for these stringent requirements, and then describe our transmitter with its calibration and
18 CHAPTER 1. ATMOSPHERIC REMOTE SENSING OF WATER VAPOUR
application to atmospheric observation.
Chapter 2
Atmospheric Scattering and
Spectroscopy
2.1 Introduction
This chapter describes a methodology for calculation of an absorption spectrum of an
individual line, as required in subsequent chapters for its measurement. Atmospheric
scattering is responsible for the lidar signal, while DIAL measures its wavelength specific
attenuation. In this chapter we discuss how lidar signals are scattered and attenuated,
and how that relates to DIAL design considerations and measurement accuracy. The last
section describes how a wavelength stabilization system can be subject to an error that
is a function of the spectral shape and laser modulation, and later sections 3.4 and A.4
describes how this error is suppressed in our design.
2.2 Propagation and scattering
All scattering phenomena are caused by the interaction between the propagating electro-
magnetic radiation and electric charges present in all matter. Scattered radiation can be
re-radiated in a random direction, re-radiated at a different wavelength, or in the case of
particulates, be absorbed and converted to heat. The phase function is a description of the
pattern of the scattered radiation, and is generally not isotropic. The larger the molecule
19
20 CHAPTER 2. ATMOSPHERIC SCATTERING AND SPECTROSCOPY
or aerosol, the more power is scattered near 0° and 180° to the propagating radiation.
For this reason, most of the measured lidar signal in our case is due to Mie-like scattering
by aerosols.
In the context of DIAL, attenuation of radiation is due to absorption as well as due to
the scattering away from the direction or wavelength of the receiver. A detailed discussion
is beyond the scope of this chapter, but is well summarized in (Bohren and Huffman, 1998)
and (Young, 1982).
The volume absorption coefficient of a substance αext is expressed in terms of the
absorption σabs and scattering σsca cross sections, where N is the volume number density
of the scattering species,
αext = N(σabs + σsca), (2.1)
and the absorption cross section, σabs is a function of the complex refractivity of parti-
cles, that depends on many factors, including their composition. The volume backscatter
coefficient at an angle of 180 °, βπ, is responsible for some of the signal returning to the
location from where it originated,
βπ =Nσπ
F(π), (2.2)
where F(θ) is the phase function that describes the angular distribution of backscatter,
and depends on the shape and size of the particles. With these two parameters, we can
use the lidar equation 2.3 to calculate the received signal P(R) from range R, given range
dependent system constants KG(R), and the pulse length ∆h as follows
P(R) =KG(R)βπ(R)∆h
8πR2 exp(−2
∫ R
0αext(r)dr
). (2.3)
The ratio of extinction to backscatter, also known as lidar ratio, is rather difficult to mea-
sure with elastic lidar. Furthermore, the aerosol shape and orientation introduces depolar-
ization of the scattered radiation.
Molecular scattering cross-section increases near a molecular resonance, and in DIAL
this results in increased extinction because nearly all of the measured backscatter is due
to aerosol Mie scattering, and not due to molecular scattering. On the other hand, this
2.2. PROPAGATION AND SCATTERING 21
Figure 2.1: Water vapour resonance line acquired with our instrument illustrating thereduction in absorption cross section at 15 GHz. We used 16 GHz for our lidar
molecular resonance scattering is relevant to Raman lidar, since the Raman signal would
be greatly increased at or near a resonance. Indeed there have been proposals to use
such a technique (Rosen et al., 1975), however, just like in DIAL, this would require
precise laser wavelength stabilization. DIfferential Absorption Lidar (DIAL) utilizing
elastic scattering measures the extinction at two slightly different wavelengths with one
wavelength stabilized relative to the center of an absorption line, is illustrated in figure
2.1. The relative attenuation at the two wavelengths therefore provides a measurement of
the absolute concentration of the absorbing species.
The DIAL inversion approximation equation 2.4 was originally derived by (Schotland,
1974) by inversion of the lidar equation 2.3. The concentration of the absorbing species
na(r) can be approximated in terms of the absorption cross-section at the two wavelengths
σon and σo f f and the online Non and offline No f f photocounts by
na(r) =1
2∆r(σon − σo f f )ln
(Non(r1)No f f (r2)Non(r2)No f f (r1)
). (2.4)
This is sometimes called the Schotland approximation. The derivation of this equation as-
sumes that there is no need to calibrate the receiving equipment for dispersion or spectral
sensitivity, due to the very close spectral proximity of the two wavelengths. It also ne-
glects errors due to the Doppler broadened Rayleigh scattering in the atmosphere (Braun,
1985). The conservation of momentum between aerosols and molecules means that the
22 CHAPTER 2. ATMOSPHERIC SCATTERING AND SPECTROSCOPY
molecular scattered part of the return signal exhibits a much greater Doppler broadening
profile than for the Mie scattered part. The broadened line shape is convolved with the
Voigt line shape of the absorbing species. However, these effects are relatively small,
and (2.4) presents good accuracy where there are no steep gradients in aerosol density or
character (Ansmann, 1985), for example, due to fog or cloud. It is possible, however, to
correct for some errors due Doppler broadened Rayleigh scattering from the off-line re-
turn signal (Girolamo et al., 2008). Due to our observing conditions, this correction was
not necessary.
2.3 Atmospheric spectroscopy
Since DIAL is based on a measurement of wavelength specific extinction with atmo-
spheric propagation, quantitative DIAL measurements depend on the spectroscopic pa-
rameters of the selected absorption line. These parameters include line position, intensity
and shape, as well as the sensitivity to pressure, temperature and mixing ratio. The HIgh
resolution TRANsmission (HITRAN) spectroscopic database (Rothman et al., 2009) pro-
vides line parameters suitable for modeling of molecular transmission and radiance prop-
erties of rotational and rotational-vibrational transitions. However, as changes in the most
recent updates would suggest, there is some residual uncertainty regarding the accuracy
of its dataset, even for strong absorption lines. For example, the strong line in the 830 nm
band, at 822.922 nm, increased in intensity by 14% from HITRAN-2006 to HITRAN-
2008. The following sections, therefore, are not a replacement for laboratory calibrations.
However, the HITRAN database is still very useful as a tool for selecting candidate lines,
as well as for understanding the effects of pressure and temperature on DIAL errors.
2.3.1 Spectral line modeling
The full theory of pressure broadening and shift of spectral lines is far is beyond the scope
of this work. However, It is possible to define some simplified and empirical models
which are sufficiently accurate for the purpose of atmospheric profiling. We will restrict
the discussion to modeling physical phenomena that would contribute more than a 1 %
2.3. ATMOSPHERIC SPECTROSCOPY 23
of the DIAL signal, however the expected accuracy of these models should be an order
of magnitude better than this under most of the conditions of temperature and pressure
encountered in the troposphere.
The models, discussion and diagrams that follow, have been implemented in, and
produced by Matlab code listed in Appendix G.
2.3.2 Molecular absorption
Molecular absorption of radiation follows the Lambert relation
I = I0 exp(−αextL), (2.5)
where II0
is the attenuation through a slab of atmosphere thickness L with extinction
coefficient αext. For the purposes of this modeling, extinction due to both scattering and
attenuation is combined, and for DIAL is primarily a function of optical frequency ν that
produces the differential signal. The effects of temperature, T , and pressure P are also
modeled using available HITRAN data.
αext(T, P, ν)/N = S (T )g(T,P,ν) = σ(T,P,ν) (2.6)
Here S (T ) is the line intensity that is modeled as a function of temperature T , and g(T,P,ν)
is the line shape that is also a function of pressure P. N is the volume number density of
the absorbing species, and σ(T,P,ν) is the molecular absorption cross-section, as discussed
in equation 2.1.
The following sections delve into the spectral models based on HITRAN parameters
that are used in the Matlab code set out in appendix G to evaluate equation 2.6.
Line intensity model
HITRAN provides line intensity S and lower state of the transition EL parameters. These
parameters, together with the thermodynamic partition function Q, determine the temper-
24 CHAPTER 2. ATMOSPHERIC SCATTERING AND SPECTROSCOPY
ature dependence of a particular absorption line (Nagali et al., 1997)
S (T )S (T0)
=Q(T0)Q(T )
(T0
T
)exp
(hcEL
k
(1T0−
1T
))(2.7)
where h is the Planck constant, k is the Boltzmann constant and T0 is conventionally
defined as 296 K. The partition function Q(T ) term is an empirical polynomial that repre-
sents the summation of states over all thermal energy levels. Since the partition function
can only be expressed in closed-form for non-interacting particles (Moore, 1984), the par-
tition function for water vapour cannot be expressed in closed form because of the strong
interaction between water molecules, as indicated by the fact that the self-broadening
coefficient in the HITRAN data has a significantly different value to the air-broadening
coefficient. For this reason, an empirical approach is required here too. For water vapour,
the polynomial and its coefficients are given by
Q(T ) = a + bT + cT 2 + dT 3,
where
a = −4.44405; b = 0.27678; c = 1.2536 × 10−3; d = −4.8938 × 10−7;
(2.8)
for temperatures between 70 and 405 K (Zhou et al., 2005). The coefficient in equation
2.7 is often approximated as (ToT )
32 , as in Browell et al. (1991).
Line shape model
Atmospheric molecular absorption is a convolution of Lorentzian pressure broadened and
Gaussian Doppler broadened line shapes, resulting in a Voigt profile. The Lorentz line
shape emerges from statistical time distribution between collisions that perturb the dipole
oscillation induced by the incident radiation, resulting in a linear pressure dependence.
The line intensity S , coefficient of atmospheric collision broadening γa, self broadening γs
and coefficient of temperature dependence of line width n, are provided by the HITRAN
database, with the total pressure broadened line width given by
γL =
(T0
T
)n
(Psγs + (Pa − Ps)γa), (2.9)
2.3. ATMOSPHERIC SPECTROSCOPY 25
where Pa and Ps are the partial pressures of air and water vapour respectively. With (2.9),
the peak absorption cross-section is given by
σ0 =S T
πγL. (2.10)
The Lorentzian line shape can now be expressed as a function of wavelength displacement
ν − ν0 from line center ν0,
gL(ν − ν0) =γL
π((ν − ν0)2 + γL2). (2.11)
To a first approximation, the absorption cross section σL is then given by
σL = S gL. (2.12)
The other broadening mechanism is due the Doppler effect with the Gaussian statistical
velocity distribution of molecular thermal energy,
The Doppler broadening Half Width at Half Maximum (HWHM) is given by
γD = ν0
√2ln2· kT
mc2 (2.13)
with a Gaussian shape function
gD(ν − ν0) =
√1πγ2
D
exp(−ln2(ν − ν0)2
γ2D
). (2.14)
The Doppler halfwidth for a water molecule at standard conditions at 820 nm, at STP, is
about 0.02 cm−1, while the Lorentz halfwidth for most lines in this range is close to 0.1
cm−1. The Voigt model for the line shape arises from the largely valid assumption that the
two broadening mechanisms are independent, except under the most extreme conditions
(Peach, 1981). The Voigt distribution has a peak absorption cross section that is about
3% lower than the Lorentzian alone at Standard Temperature and Pressure (STP), and
therefore it is necessary to model it for the desired DIAL accuracy of <1%.
26 CHAPTER 2. ATMOSPHERIC SCATTERING AND SPECTROSCOPY
The Voigt convolution gv can be expressed in closed form;
gV(x, y) =S x
π√γD
∫ ∞
−∞
e−t2
(x − t)2 + y2 dt (2.15)
where S is the line intensity and x and y are defined in terms of the Gaussian, γD, and
Lorentzian, γL line widths as follows;
x =ν − ν0
γD, (2.16)
y =γL
γD. (2.17)
Unfortunately, however, there are no analytical closed-form solutions to the error-function
contained in the integral in equation 2.15, and numerous authors have investigated the best
ways to evaluate the Voigt profile using various numerical techniques (Milman, 1978;
Abrarov et al., 2010; Schreier, 2009; Drayson, 1976; Bykov et al., 2008) with varying
trade-offs between calculation speed and numerical accuracy. One of the simplest tech-
niques that provides adequate accuracy for our purpose is described by Whiting (Whiting,
1968), combined with the parameter modification by Olivero (Olivero and Longbothum,
1977). This method seems particularly suitable for our application because it has a min-
imum error where the Lorentz broadening is significantly greater than Doppler, which is
the case for tropospheric spectroscopy. Under these conditions, the worst-case error will
be less than 0.2 %. This method is also very computationally efficient using no iteration.
The Whiting-Olivero technique uses a Voigt width model ΥV as a function of Doppler
ΥD = 2γD and Lorentz ΥL = 2γL full-widths,
ΥV = 0.5346ΥL +
√0.2166ΥL
2 + ΥD2, (2.18)
a peak absorption, Igv , as a function of ΥV and ΥL
Igv =1
ΥV(1.065 + 0.447 ΥLΥV
+ 0.058( ΥLΥV
)2, (2.19)
and the Voigt shape model as a function of equations 2.19, 2.18 and the optical wavelength
2.4. LINE SELECTION 27
difference, D = (ν − ν0).
gV/Igv =
(1 −
ΥL
ΥV
)exp
−2.772(
DΥV
)2 +
ΥLΥV
1 + 4( DΥV
)2
+ 0.016(1 −
ΥL
ΥV
) (ΥL
ΥV
) exp
−0.4(
DΥV
)2.25 − 1010 + ( D
ΥV)2.25
(2.20)
2.4 Line selection
Figure 2.4a illustrates the water vapour absorption spectrum across the visible and near
Infra-Red (IR). For DIAL applications, we are primarily concerned with the strength of
the line, and how the absorption cross-section changes with temperature. As the figure
illustrates, stronger lines are available at the longer wavelengths. The shorter wavelengths
only offer weaker lines that result in a smaller differential signal, however, the elastic
Rayleigh and Mie scattering cross-section due to aerosols is larger, which increases the
power in the return signal.
Each of the thousands of absorption lines in each of the bands has unique spectral
characteristics, and the choice of a particular line depends on the intention of the mea-
surement.
Temperature sensitivity of both broadening and intensity effects the measured absorp-
tion cross-section, and a line with high temperature sensitivity can be selected as a way
of profiling temperature (Bosenberg, 1998). For measuring water vapour independently
from temperature, a line with a near zero temperature sensitivity over the prevailing tem-
perature range, can be selected.
Re-writing equation 2.7 in a simplified form (Browell et al., 1991), substituting into
2.9 and 2.10 and differentiating, it can be shown that the temperature at which temperature
sensitivity of the absorption cross-section is zero, Tnull, can be written as equation 2.21.
This facilitates line selection of a particular line for the lower state energy, EL, so that
it has near zero temperature dependence over the temperature range that is likely to be
28 CHAPTER 2. ATMOSPHERIC SCATTERING AND SPECTROSCOPY
822.89 822.9 822.91 822.92 822.93 822.94 822.950
0.5
1
1.5x 10
−22
Wavelength (nm)
Mole
cula
r absorp
tion
cro
ss−
section a
t S
TP
(cm
2)
Lorentzian
Voigt
Figure 2.2: Lorentzian and Voigt spectral models of the 822.922 nm line at STP, includingthe (negligible) effects of adjacent lines. This model was also used to calculate the peaksand troughs for more than 30,000 thousand adjacent lines in HITRAN between 500 nmand 2 µ m, illustrated in figure 2.4.
822.89 822.9 822.91 822.92 822.93 822.94 822.950
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x 10−21
Wavelength (nm)
Mole
cula
r absorp
tion
cro
ss−
section a
t 0.1
atm
(cm
2)
Lorentzian
Voigt
Figure 2.3: Lorentzian and Voigt spectral models of the 822.922 nm line at 0.1 atm and296 K. This atmospheric pressure corresponds to an altitude of 16 km, where Dopplerbroadening becomes significant.
2.4. LINE SELECTION 29
0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2
10−26
10−24
10−22
10−20
Wavelength (µm)
Absorp
tion c
ross−
section (
cm
2)
822.922 nm
(a) Water vapour absorption cross-section spectrum 0.5-2.0 µm
0.81 0.812 0.814 0.816 0.818 0.82 0.822 0.824 0.826 0.828
10−26
10−24
10−22
10−20
Wavelength (µm)
Absorp
tion c
ross−
section (
cm
2) 822.922 nm
(b) Water vapour absorption cross-section Spectrum around 820 nm and the selected line with the Voigtcross-section σv = 1.46 × 10−22cm2
Figure 2.4: Water vapour line spectra at sea level, showing cross-sections at line peaks andmidway between each consecutive HITRAN line. The peaks and troughs show absorptionmaxima and minima respectively. This figure was generated with Matlab using code inAppendix G
30 CHAPTER 2. ATMOSPHERIC SCATTERING AND SPECTROSCOPY
100 150 200 250 300 350 400−4
−3
−2
−1
0
1
2x 10
−3
Tempratue (K)
Sensitiv
ity o
f absorp
tion c
ross−
section
Sensitivity of absorption cross−section to temperature with lower−state energy
0
50
100
150
200
n=0.7
E’’=
Figure 2.5: Temperature sensitivities of the absorption cross-section (dimensionless) ofvarious hypothetical absorption lines with lower state energies in the range of 0-200 cm−1,and a broadening coefficient of around 0.7. This figure is intended as a guide for selectionof possible candidate lines. The lower state energy EL, and coefficient of temperaturedependence of line width n data for each line is supplied in HITRAN.
encountered in the atmosphere during DIAL operation.
Tnull '
(hck
) ( EL
1.5 − n
)(2.21)
Due to the relationship between molecular number density, local temperature and pres-
sure, the temperature of a zero sensitivity mass mixing ratio measurement, follows a
slightly different relation compared to the absorption cross-section null equation 2.21
(Cahen et al., 1982).
Tnull '
(hck
) ( EL
2.5 − n
)(2.22)
The choice of lower state energy therefore also depends on what type of measurement
is desired. To optimize the accuracy of a number density measurement, the lower state
energy should be selected according to equation 2.21, while to optimize the accuracy of a
mixing ratio measurement, the lower state energy should be selected according to equation
2.22. As an illustration of these considerations, figure 2.5 presents a sample calculation
of temperature sensitivity of the absorption cross-section for various lower state energy
values with a fixed n of 0.7.
2.5. PRESSURE SHIFT 31
2.5 Pressure shift
The HITRAN database also provides the spectral line air-pressure shift parameter. This
figure, which is in the range of 0 to 0.03 cm−1/atm at 296 K for most lines in the 820 nm
water absorption band, measures the shift of the line center as a function of air pressure,
which can be related to altitude using equation 2.23,
P = P0 exp(−
zH
). (2.23)
For example, the 822.92 line has a shift parameter of -0.016 cm−1/atm, which means
that we can expect a pressure induced wavelength shift of 0.008 cm−1 at an altitude of
5.5 km relative to STP. This corresponds to a frequency shift of 240 MHz at that altitude
relative to the shift at sea level. Substituting into the Voigt model at half atmospheric
pressure, we get an error of 2.1% in the absorption cross-section measurement with the
laser stabilized to the line at STP, as illustrated in figure 2.7. However if we shift the on-
line laser frequency by 240 MHz, we will cancel the error at 5.5 km, while introducing
a smaller error of 0.7% at STP, as illustrated in figure 2.8. This difference is due to the
shape change of the pressure broadened absorption profile with altitude, as illustrated in
figure 2.6 . An investigation of the subject (Zuev et al., 1985) recommended tuning the
on-line laser to the center of the Doppler profile, which in this case corresponds to a 480
MHz offset, such that the error approaches zero near the top of the atmosphere, as illus-
trated in figure 2.9. However, the results presented in these figures, suggest that it is better
to tune the laser to scale height. Furthermore, since most DIAL systems are really only
intended for tropospheric applications below the boundary layer, such a large optical fre-
quency offset is not justifiable here. Where precise wavelength control is available, a shift
corresponding to that at the maximum altitude from which DIAL results are obtainable,
would provide the best trade off.
32 CHAPTER 2. ATMOSPHERIC SCATTERING AND SPECTROSCOPY
822.89 822.9 822.91 822.92 822.93 822.94 822.950
0.5
1
1.5
2
2.5
3
3.5
4x 10
−5
Wavelength nm
Extinction C
oeffic
ient (c
m−
1)
Extinction Coefficient vs Pressure for 1 %vol water vapour
1.0 atm − sea level
0.5 atm − 5.5 km
0.1 atm − 16 km
Figure 2.6: Voigt model at various altitudes, including pressure shift. As the altitude in-creases, pressure broadening and shift decrease. Doppler broadening dominates at highaltitudes. At 16 km, the peak absorption cross-section is reduced by a factor of ∼2, butthe pressure is reduced by a factor of 10, therefore, the peak extinction is still very signif-icant despite the sparse atmosphere. Furthermore, for a given shift in wavelength (dottedvertical lines) the reduction in absorption cross-section is much greater at 0.1 atm (red)than at STP (blue).
0 2000 4000 6000 8000 10000 12000 14000 16000−35
−30
−25
−20
−15
−10
−5
0
Altitude (m)
Cro
ss−
section m
easure
ment err
or
(%)
Pressure shift induced error with laser wavelength centered on the STP profile
Figure 2.7: This figure illustrates the measurement error of the absolute water numberdensity using the 822.92 nm line, with a laser tuned to the absorption line center at STP(sea level)
2.5. PRESSURE SHIFT 33
0 2000 4000 6000 8000 10000 12000 14000 16000−7
−6
−5
−4
−3
−2
−1
0
Altitude (m)
Cro
ss−
section m
easure
ment err
or
(%)
Pressure shift induced error with laser wavelength centered to profile at 5.5 km (0.50 atm)
Figure 2.8: Measurement error of the absolute water vapour number density using the822.92 nm line, with a laser tuned to line center at scale height, 5.5 km.
0 2000 4000 6000 8000 10000 12000 14000 16000−3
−2.5
−2
−1.5
−1
−0.5
0
Altitude (m)
Cro
ss−
section m
easure
ment err
or
(%)
Pressure shift induced error with laser wavelength centered on the Doppler profile
Figure 2.9: Measurement error of the absolute water vapour number density using the822.92 nm line, with a laser tuned to the Doppler broadened line.
34 CHAPTER 2. ATMOSPHERIC SCATTERING AND SPECTROSCOPY
2.6 Spectral line and dither
Many systems utilize modulated current injection as a method for generating the error
signal from the derivative of the absorption feature (Demtrder, 2003). However, the re-
sulting amplitude modulation causes a shift in the locked optical frequency (Hollberg
et al., 1996) (Taubman and Hall, 2000). We utilize the line spectral model to calculate
the magnitude of the error resulting from the amplitude modulation of the laser, as ex-
plained in Appendix F. These results are not applicable to this DIAL system because we
utilize a ratiometric detection system described in Appendix A which removes the optical
power modulation from the control loop. Even with lowest-cost components, our system
provides a multiplicative common-mode rejection of approximately 40 dB which reduces
this error by two orders of magnitude.
With an optical frequency modulation of 500 MHz, we measure a power modulation
of 0.6 % with the Hitachi HL8325 laser diode at 95 mA, using a method described in
Appendix C.
This simulation uses the Voigt model described above, combined with a linear model
for laser power, to calculate the result in figure 2.10 using Matlab code dithermodel.m.
This simulation is based on a linear time-invariant model of the lock-in amplifier, which
is also utilized in the control system description in Appendix B. The simulated error was
found to be 580 MHz with atmospheric water mixing ratio of 1%, 33 m optical path
length, and measured laser diode characteristics from experiment C.2.1.
2.7. SUMMARY 35
822.905 822.91 822.915 822.92 822.925 822.93 822.935 822.94−0.01
−0.005
0
0.005
0.01
0.015
0.02
Wavelength (nm)
Err
or
sgnal −
norm
aliz
ed
Error signal simulation with Laser Current Dither
580 MHz
Figure 2.10: Simulation of the error signal with power and wavelength modulation byinjection current, using the spectral model at 822.92 nm and the characterized laser diode.The effect of the power modulation is a shift of the zero-crossing of the error signal.Without ratiometric detection, the control system will stabilize with a 580 MHz offsetfrom the true absorption peak. Furthermore, there will be another zero crossing with avery large offset, where the control system could also stabilize
2.7 Summary
This chapter provided the background to spectroscopy and the models necessary for an
informed selection of specific absorption lines, model the effects of laser wavelength ac-
curacy, and the effects of pressure broadening and shift. The line model, together with
measured laser device data was used to calculate wavelength stabilization error with a
lock-in amplifier. The following chapter describes how ratiometric detection at the vapour
cell is used to suppress this error, as well as providing a description of the other elements
of this DIAL instrument.
36 CHAPTER 2. ATMOSPHERIC SCATTERING AND SPECTROSCOPY
Chapter 3
DIAL Transmitter and System Design
3.1 Introduction
In Chapter 2 we discussed water vapour spectroscopy and the significance of wavelength
accuracy and stability, which are key requirements and challenges. In this chapter we
review the requirements for DIAL with a survey of past and current components and
designs, followed by a description of the system developed by the author, with additional
information, schematics and other data provided in the appendices.
3.2 DIAL system design consideration
The following section summarizes the specific DIAL system requirements, design and
performance considerations. This is followed by a summary of how these have been
achieved by various research groups over recent decades, albeit on a laboratory scale.
Wavelength and line selection
The operating wavelength of a DIAL transmitter is primarily determined by the availabil-
ity of laser at a wavelength with suitable absorption cross-section. Molecules exhibit a
number of spectral bands that are of the order of 10 nm in width, each containing tens of
thousands of vibrational and vibrational-rotational lines. Resonance lines are generally
stronger at longer wavelengths, and from HITRAN data, the apparent fundamental dipole
37
38 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
resonance for water vapour near 70 µm (4 THz) has an absorption cross section some 5
orders of magnitude larger than the strongest lines near 820 nm. In a DIAL application, a
strong line produces a larger differential signal from each range cell, but suffers increased
attenuation which limits the measurement range. Therefore, a suitable line strength is a
trade-off between range and precision. From a study of DIAL error sources, the signal
to noise ratio is optimized with an optical extinction between 0.03 and 0.1 in each range
cell (Wulfmeyer and Walther, 2001) as determined either by the data averaging, or by the
optical pulse length.
To maximize scattering, we would select the shortest wavelength with suitably strong
absorption lines, however, this is not always possible due to availability and cost of the
laser and system components. For example, the shortest wavelength band at which water
has suitably strong absorption features is near 720 nm, which also corresponds to good
receiver quantum efficiency, as well as the availability of good large receivers such as
astronomical optical telescopes. However, there are currently no low-cost lasers around
this wavelength. Consequently, water absorption bands near 820 nm and 940 nm are more
often utilized. We selected the 820 nm band for our instrument due to the availability of
laser diodes, adequate and low cost PMT detectors, as well as the usability of a standard
Schmidt-Cassegrain telescope.
Once a suitable band is identified, a suitable absorption line is selected. The line
intensity is selected based on range and prevailing atmospheric conditions, while other
considerations rest on other spectroscopic parameters including lower state energy, coef-
ficient of temperature dependence of air-broadened halfwidth, and air-broadened pressure
shift of transition as explained in Chapter 2.
Wavelength accuracy and stability
The absolute species concentration is calculated from the inversion of the differential sig-
nal from at least one pair of wavelengths. The differential extinction depends on the
species concentration, as well as the effective absorption cross-section. The effective
absorption cross section is a convolution of the laser signal with the actual physical ab-
sorption profile. Therefore, the precision of the on-line wavelength, as well as its spectral
3.2. DIAL SYSTEM DESIGN CONSIDERATION 39
purity, determines the effective absorption cross-section. The pressure-broadened Voigt
profile is of the order of 3 GHz HWHM for most water lines near 820 nm at STP. This
broadening effect is proportional to pressure, and absorption lines are sharper and nar-
rower at higher altitudes. For a given change in wavelength, the change in the on-line
absorption cross-section is greater at higher altitudes than at sea level. This was illus-
trated in figure 2.6 in Chapter 2.
Linewidth and spectral purity
The effective absorption cross-section comes from a convolution of the transmitted laser
spectrum with the Voigt shape of the absorption feature. If a significant portion of the
optical energy is spread into the wings of the spectral line, and is within the passband
of the receiving detector, the effective absorption cross-section will be reduced by this
convolution. These issues can be addressed by the design of the laser power transmitter
at some additional cost. If a simpler, lower cost power laser amplifier stage is employed,
the effective absorption cross-section can be calibrated against a reference sample of the
absorbing species, providing that this measurement remains stable over time. For ex-
ample, the attenuation in an air sample of known humidity can be used to calculate the
effective absorption cross-section for a particular absorption line. These measurements
and techniques are further explained in Chapter 4.
Master laser wavelength stabilization
Wavelength stabilization for DIAL has been implemented in a number of ways includ-
ing wavemeters, calibrated etalons, and vapour cells. Etalons and wavemeters provide
extremely good short-term stability because they provide a large error signal to the wave-
length control loop, but require calibration and temperature stabilization. Multipass molec-
ular absorption cells come in two varieties; Herriott (Herriott et al., 1964) and White
(White, 1976) cells. These generally provide a weaker and hence a noisier and slower
error signal to the wavelength control system, but require no calibration or temperature
stabilization, and hence offer much better long-term optical frequency stability (Fox et al.,
1993). Some DIAL systems had extensive and difficult procedures for wavelength control
40 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
requiring either continuous human intervention, or wavemeter calibrations against a refer-
ence laser and absorption lines, as a separate procedure prior to any observations (Bruneau
et al., 2001a). Our aim was to design a system with the most appropriate techniques for
a robust, low-cost, fully autonomous system that aims for <5% precision and stability of
the effective cross-section and humidity measurements, as recommended for meteorol-
ogy (World Meteorological Organization (WMO), 2010). For this reason we employed a
modified Herriott cell for our prototype.
Laser stabilization often employs some type of modified Pound-Drever-Hall technique
(Black, 2001). The modulation of the laser wavelength can be at a relatively low fre-
quency where the resulting optical sidebands are inside the absorption feature (Wang
et al., 1989), or at a high frequency such that the optical sidebands are spaced wider than
the optical spectral width of the absorption feature. Both cases can be used for wave-
length stabilization, as well as to suppress the effects of amplitude noise on the wave-
length stabilization. The advantage of a high frequency and a faster control loop, is a fast
wavelength stabilization that compensates for laser frequency fluctuations over a wider
frequency range. However, most of the laser amplitude and frequency flicker noise is at
the lowest frequencies due to the 1f characteristic (McManus et al., 1995). For this reason,
the costly high speed modulation, detection and control is not always justifiable, since a
well behaved laser can perform adequately well with a slow control loop (Silver, 1992).
Since our target is for low cost and simplicity, we selected the low-frequency wavelength
modulation. This also facilitates the synchronization of dither and system timing, that has
other advantages to be discussed later.
3.2.1 Survey of diode lasers
In systems employing low power tunable master oscillators, several different types of
devices can be employed. The following section outlines their salient characteristics.
Distributed Feedback diode laser (DFB) These provide excellent tuning character-
istics over a small tuning range with good spectral purity (Wieman and Hollberg, 1991)
3.2. DIAL SYSTEM DESIGN CONSIDERATION 41
Master Laser
Wavelength ControlSystem
Wavelength MeasurementSystem
LaserAmplifier
+
Figure 3.1: Block diagram illustrating a stabilized dual wavelength reference built witha single master laser where the wavelength is square-wave modulated between on-lineand off-line wavelengths (Koch et al., 2004). Using an absorption cell as a reference, thecontrol system finds a quiescent point such that the difference signal between the on-lineand off-line transmission through the cell is maximized, as the wavelength alternates be-tween two states. The amplitude of the square wave then sets the wavelength separation.This type of system requires a master laser that has a sufficiently fast, linear and stabletuning characteristic for the desired pulse repetition rate. Furthermore, the binary oper-ation means that it might not be well suited to applications requiring multiple off-linewavelengths.
Master Laser 1
Wavelength ControlSystem
Wavelength MeasurementSystem
LaserAmplifier
Master Laser 2
Wavelength ControlSystem
Wavelength MeasurementSystem
OpticalSwitching
Figure 3.2: DIAL block diagram illustrating a dual stabilized master laser system, and asimplified representation of our system. This has the advantage that both CW lasers arecontinuously stabilized making it easier to achieve a higher degree of wavelength accuracyand stability. The cost of this is the requirement for multiple lasers and optical switching,resulting in optical power loss and greater system complexity. This arrangement has beenproposed for applications where multiple on-line or side-line wavelengths are required(Koch et al., 2008) for space-based DIAL. For nadir applications from space, the on-linelaser is stabilized to the side of an absorption line with a small fixed offset away fromline center to avoid saturated absorption at low pressure, where Doppler broadening isdominant in the upper atmosphere.
42 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
(Posthumus et al., 2005), however, availability of desirable DIAL wavelengths is rather
limited. Furthermore, at a custom wavelength, the cost of these devices is of the order of
US$10k. The output power is only moderate (∼40 mW). However, the tuning and spectral
characteristics make these devices extremely attractive for DIAL application.
Fabry Perot diode lasers This is the most common and lowest cost type of laser diode
available today, with common applications in bar code scanners, CD players and laser
pointers. Some varieties of this device lase in only one longitudinal mode (or the integer
number of standing waves in the laser cavity) at a time under most conditions, which
makes them a potentially suitable candidate for spectroscopy. However, the optical length
of the resonant cavity and the spectral gain profile, means that it can lase in one of several
possible longitudinal modes under most conditions. This makes it potentially difficult to
coax the device into the desired mode for the required wavelength. However, experience
has shown that these devices can maintain the mode well if kept continuously powered
with temperature control and 60 db optical isolation. Furthermore, these devices can have
reasonably good spectral purity. Single mode devices up to several hundred milliwatts
are available at a very low cost, however, the exact wavelength is poorly determined and
lasers need to be spectrally selected and matched for the required DIAL wavelength. Cost
considerations mandated their use for this DIAL application.
External Cavity Diode Laser (ECDL) These devices are commercially available at a
moderately high cost. They exhibit very good spectral purity and wavelength tuneability.
One of the main drawbacks of these devices is the high sensitivity to vibrations, which
limits their use to stationary or laboratory conditions, and renders them unsuitable for a
robust system. Furthermore, these devices seem to require regular re-alignment [private
correspondence with Toptica].
Vertical Cavity Surface Emitting Laser diode (VCSEL) These devices have similar
structure to a FP laser, except with a very short cavity that can only support one longitudi-
nal mode. The tuning and spectral characteristics are similar to the DFB laser. However,
output power per device is very low at around 1 mW. These devices could be manufac-
3.2. DIAL SYSTEM DESIGN CONSIDERATION 43
tured for a wide range of wavelengths, and in large quantities, at very low cost. This
makes them potentially interesting for low-cost DIAL, where multi-stage or high-gain
optical amplifiers are available, as well as for other spectroscopy applications.
3.2.2 Receiver and detector
The received signal is proportional to the area and efficiency of the receiving optics and
detector, as well as the power of the transmitter, as expressed in equation 2.3. For most
DIAL applications, we are operating in the photon counting regime where low-noise (low
dark count) detectors are required. This rules out conventional PIN diodes despite their
high quantum efficiency. Avalanche type detectors including APDs and PMTs have suit-
ably low noise, and for wavelengths shorter than about 830 nm, photomultipliers provide
adequate quantum efficiency with very low dark count (∼ 1s−1) and a wide field of view.
However, the PMT’s quantum efficiency decreases rapidly at longer wavelengths, leav-
ing only APDs as a suitable candidate. The main difference between these two types of
detectors is the size of the detection area. PMTs have a large detection area that results
in a wide FOV if the full device aperture is utilized. This eases alignment, but also in-
creases the angle of background sky light impinging on the detector. In order to reduce
background, an aperture can be placed in front of a PMT to select the desired FOV.
Silicon APDs are a viable alternative to the PMT because of their high quantum ef-
ficiency even up to wavelengths of 1100 nm. They can be operated over a range of re-
verse bias voltages that trades-off quantum efficiency against dark count, and are generally
thermoelectrically cooled to improve overall performance in the photon-counting regime.
They exhibit a dark count of ∼ 50s−1, at a quantum efficiency of ∼50 % above 800 nm
where PMT performance is poor. The size of the detection area of a typical APD is often
measured in square microns, orders of magnitude smaller than a typical PMT. This is
advantageous to reduce background, as well as facilitating the use of fiber optics in the
receiving systems, however, it makes alignment more difficult, and mechanical stability
more critical. For these reasons, PMTs are still widely utilized at wavelengths shorter
than 820 nm.
44 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
3.2.3 Transmitter
The power amplifier can be a gain-switched or a Q-switched injection seeded output stage,
or a parametric amplifier (eg: OPO (Amediek et al., 2008) (Weibring et al., 2003)). Other
optical gain mechanisms such as Raman and SBS have not been reported in lidar applica-
tions to date. Seeded Q-switched lasers exhibit very high gain and a consequent spectral
narrowing which suppresses ASE as well as reducing the optical linewidth at the output
of the transmitter while providing high output power. However, these are expensive and
generally produce pulses too short for spectral purity required for DIAL. Furthermore,
the cavity mode of these types of amplifiers needs to be precisely resonant with the wave-
length of the seed laser at the time the Q-switch is fired, to form the output pulse. This
complicates the already difficult timing and wavelength control requirements.
On the other hand, devices such as tapered semiconductor optical laser amplifiers,
are relatively cheap and easy to implement. However, they have limited output power,
relatively low gain, and spectral purity issues. Since the precise wavelength control as
well as high power output is most easily implemented using some type of Master Oscil-
lator Power Amplifier (MOPA), the TA is ideal for a prototype instrument. Futhermore,
We found that the tapered optical amplifier can be suitable low-power DIAL transmitter
candidate when operated at a high pulse repetition rate, which partly compensates for its
relatively low output power. Furthermore, we have shown how to calibrate the effective
absorption cross-section due to ASE, as illustrated in Appendix 4.3. Finally, the availabil-
ity of relatively high power, low cost FP and DFB diode lasers, can provide adequate seed
power to saturate the TA input.
The Tapered optical Amplifier (TA)
The Tapered optical Amplifier (TA), also known as the Semiconductor Optical Amplifier
(SOA) was developed in 1992 for tunable frequency doubling and free-space communi-
cations (Mehuys et al., 1992). These devices use ion-implantation to produce a graded
refractive index that guides the light along the device’s optical structure. This optical
structure consists of a ridge waveguide followed by a tapered section (Yeh et al., 1993) as
3.3. OVERVIEW OF THIS DIAL 45
InputFacet
OutputFacet
RidgeSection
Taper Section
Figure 3.3: Semiconductor Tapered Amplifier Illustration
illustrated in Figure 3.3.
The ridge waveguide is much longer than the Rayleigh range of the input signal, and
forms a near single-mode spatial filter that provides a near diffraction-limited wavefront
for amplification at the start of the taper section. The ridge waveguide also acts as a sat-
urable preamplifier to suppress ASE before entering the tapered section. On entering the
taper, the beam expands freely to achieve the large small-signal-gain. The angle of the
tapered section is designed to diverge at a slightly wider angle than the freely diffracted
angle to avoid edge diffraction effects from the taper itself. The taper suppresses parasitic
gain loss due to ASE while also providing the increasing gain volume to maintain a con-
stant power density and gain saturation along the length of the taper. Both ends of the TA
are Anti Reflection (AR) coated to the best extent possible to prevent cavity modes.
TAs are manufactured by several companies including Toptica, Eagleyard, m2k-Laser
GmbH and Sacher Lasertechnik at various wavelengths from 650 nm to 1200 nm.
The availability and ease of application of the TA made it a sensible choice for this
DIAL prototype.
3.3 Overview of this DIAL
In this section, we describe our laser transmitter system for DIAL in the context of water
vapour, and how we work towards meeting these challenges using low cost and rudi-
mentary components. By providing an overview of this system, its design and operation,
we also illustrate techniques applicable to other laser technologies and wavelengths suit-
able for detection of other trace gas species. Unlike other designs that use spectrometers
(Bruneau et al., 2001a) (Yu et al., 1997), or etalons (Ponsardin et al., 1994) (Machol et al.,
46 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
2004) with a computer, to control laser wavelengths, this on-line laser is stabilized to wa-
ter vapour absorption while the off-line laser is beat-frequency stabilized to a passive 16
GHz bandpass filter, without the need for a microwave oscillator or RF mixer. The last
section of this chapter deals with extension of these techniques for stabilization of multi-
ple off-line wavelengths, as well as stabilization of the beat frequency using any type of
passive frequency reference.
Figure 3.4 provides the overview diagram of the entire DIAL system, while more de-
tailed and specific descriptions, including schematics of our system, are supplied in the
Appendices. Our dual master laser system used two CW 40 mW Hitachi HL8325G laser
diodes, operating near 822.9 nm. A pair of Faraday isolators (Newport ISO-04, EOT and
Conoptic) were used to provide 60 dB of optical isolation for each of the lasers to achieve
reasonably reproducible tuning characteristics and longitudinal mode stability. Pulses for
amplification and transmission to the atmosphere were formed by two AOMs (Isomet
1205-603 and Crystal technology 3080-120), with the remaining light (that is comple-
mentary to the pulsed beams) used for wavelength stabilization of the master lasers. One
laser was servo locked to the wavelength of the peak of a water absorption line. The
second was maintained at a fixed wavelength offset from the first by combining the two
laser beams and stabilizing the frequency of the beat signal to the peak transmission of
a microwave RF bandpass filter (Reactel 4C1-16G-500-S11) at 16 GHz. After passing
through each AOM, each undeflected beam is coupled into a single-mode optical fiber
which takes it to the stabilization electronics.
The pulsed beam that is Bragg scattered by the acoustic wave is directed to a beam
splitter used as a combiner and then to the optical amplifier (TEC-400-830-500). Using
the Bragg scattered wave, rather than the zeroth-order straight-through beam, as the basis
for the pulse transmitted to the atmosphere ensures that there is essentially no optical
power leakage from the master when the AOM is not energized. An added benefit of using
the Bragg scattered beam in this way is that the AOM then contributes to the isolation of
the master lasers from back reflections from the optical amplifier.
The acoustic wave in each AOM is repetitively pulsed on for 1 µs with a period of
3.3. OVERVIEW OF THIS DIAL 47
Σ
On-
Line I T I T
−LP
F
On-
Line
Lase
r2
BP
16 G
Hz
Det
ecto
r
Off-
Line
∫
∫
Off-
Line
split
ter
Fibe
r
TOA
ISO
ISO
Lase
r1
Opt
ical
Rec
eive
r
Opt
ical
Pul
se O
utpu
t
Tim
ing
Syst
em
RF
Driv
e
Dith
er
AOM
AOM
∫ ∫
GaA
s PI
NLP
F
TOA
AOM
LPF
Bea
msp
litte
r
Mir
ror
Tap
ered
Op
tica
l Am
pli
fier
Acc
ou
sto
-Op
tic
Mo
du
lato
r
Fre
e-sp
ace
fib
er c
ou
ple
r
Lo
w P
asss
Filt
er
An
alo
g M
ult
iplie
r
An
alo
g D
ivid
er
BPB
and
pas
s F
ilter
RF
ISO
Far
aday
Iso
lato
r 60
db
Op
tica
l Fib
er
Fre
e-sp
ace
alig
nm
ent
Ele
ctri
cal s
ign
al
IL
aser
Cu
rren
t C
on
tro
l
TL
aser
Tem
per
atu
re C
on
tro
l
Figure 3.4: DIAL control system. Also see simplified diagram provided in figure 3.2.
48 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
Figure 3.5: DIAL instrument showing receiver, lasers, optical switching, electronics andassociated components
Figure 3.6: DIAL instrument showing multi-path absorption cell, electronics and associ-ated components
3.4. ON-LINE MASTER LASER CONTROL SYSTEM 49
667 µs. The pulse length, which determines the transmitted pulse energy, is chosen as
a trade-off between signal-to-noise and range resolution in the lidar return. This pulse
width and period correspond to a range resolution of 150 m and a maximum range of
100 km. The effective maximum vertical range is very much less than this because of
the relatively low transmitted pulse energy (∼500 nJ). Indeed the data acquisition system
only records for 50 µs after the pulse is transmitted, corresponding to a maximum range
of about 7 km. Vertical resolution could be improved by reducing the pulse width, but
this would reduce pulse energy and signal-to-noise ratio as well as compromising the
maximum vertical range because of background light. The pulse repetition rate could be
considerably higher, up to ∼10kHz or more, without introducing range ambiguities, but
in our case this is limited by the LabView software and the Licel data acquisition system.
Each laser operates in a single longitudinal mode and provides a mode-hop-free tun-
ing range of more than 200 GHz. These laser diodes have been found to operate in several
discrete regions over a wavelength band from 820 to 834 nm. Device selection is neces-
sary to have two lasers that tune to the same wavelength range, as discussed in Appendix
E. The optical tapered optical amplifier (TEC-400-830-500 from Sacher Lasertechnik)
injection current is pulsed synchronously with the arrival of the master laser pulses of
alternating wavelength. The maximum rated output power is 500 mW, however typical
output varied, and was consistently below 400 mW, typically ∼ 300 mW.
More detail of the system is presented in Chapter A, while system performance and results
are presented in Chapter 4.
3.4 On-line master laser control system
As illustrated in figure 3.4, The zeroth-order on-line laser beam after the AOM is coupled
into a single-mode optical fiber, and then to a 50: 50 fiber coupler, which serves as both
a splitter and a combiner..
Our multi pass cell is a home-made variant of a Herriott cell using near confocal
spherical gold-coated mirrors. The light is coupled into the space between the mirrors via
a small periscope, and after 66 traversals of the space between the mirrors and a 33 m
50 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
path length, the beam encounters the periscope again and is coupled out of the cell.
Before light enters the absorption cell, it encounters a pellicle beamsplitter. This type
of beamsplitter has negligible thickness, and therefore does not produce multi-path inter-
ference by itself. A ratiometric detector measures the reflection from the pellicle, as well
as the light transmitted through the cell. The ratio of these two optical intensities is what
provides the feedback signal to the control system.
The phase sensitive detection uses a dither modulation of ∼1.5 kHz via the on-line
laser current control, so that the laser has a peak-to-peak optical frequency modulation of
500 MHz. This optical frequency modulation through the absorption profile, results in an
amplitude modulation at the ratiometric output, with a magnitude and phase that depends
on the position of the laser wavelength relative to the center of the absorption peak. As
the laser is tuned through the resonance, phase-sensitive detection of this signal produces
a derivative of the absorption profile, as illustrated in figure 2.10. The control signal can
then lock to the zero crossing of this error signal in order to keep the laser locked to the
resonance peak.
However, there are unwanted sources of amplitude modulation with changing laser
wavelength. First, the dither of laser injection current also modulates the laser power
which shifts the error signal as discussed in section 2.6. Second, the multiple beam inter-
ference fringes due to reflections within the fiber splitter and other optical fiber compo-
nents, have a greater contrast than many water absorption lines. Fortunately, both of these
interferences are greatly reduced by the ratiometric detection illustrated in figure 3.4. Un-
fortunately, however, due to electronic non-linearities in our low-cost design, these errors
were not completely eliminated in our system. Furthermore, ratiometric detection could
not reduce the effect of the fringes due to multi-path interference in the cell itself, because
there is a small amount of overlap of the laser spots on the mirrors. Some light at the
mirrors is scattered into paths that effectively shortcut one or more of the passes in the
cell, and this is manifested as fringes at the output of the ratiometric detector, with a free
spectral range corresponding to the mirror separation of the absorption cell. The use of
the single-mode fiber provided a good transverse mode that was less susceptible to multi-
path interference in the absorption cell, as well as simplifying the assembly of the system.
3.5. OFF-LINE MASTER LASER CONTROL SYSTEM 51
In our case, these fringes were more than an order of magnitude weaker than the absorp-
tion line at 822.9 nm, and caused only a small random offset error in the stabilization to
the center of the water line. Unfortunately, this effect was dependent on alignment and
changed with ambient conditions.
The AOM also cause an error, albeit a much smaller one. The pulses transmitted to
the sky are shifted in frequency from the absorption line center by the 80 MHz acoustic
frequency of the AOM. In the lower troposphere, where the water absorption lines are at
least 2 GHz HWHM, this introduces negligible systematic error of<0.3%.
To minimize the cost and component count of the optical system, we use one fiber
splitter, such that the light from both the off-line and on-line lasers pass through the ab-
sorption cell. Since the off-line laser wavelength is not modulated, it does not affect the
on-line control system, however, there is also a 16 GHz beat signal at the water-vapour
cell. This has no effect, however, since the photodiodes here are relatively slow (∼100
MHz).
3.5 Off-line master laser control system
Just like the on-line laser, the off-line master is Faraday isolated, passed through an AOM,
and coupled into the single-mode fiber. The second output from the fiber coupler goes to a
GaAs PIN photodiode (New Focus Model 1481-S) which has a frequency response from
DC to 25 GHz which detects the beat frequency of the two lasers.
Since the on-line master laser already includes a dither, the beat signal around 16 GHz
also has a frequency modulation depth of 500 MHz at a rate of 1.5 kHz. The bandpass
filter, with a center frequency of 16 GHz and a 3 dB bandwidth of 500 MHz, converts
this frequency modulation to an amplitude modulation whose phase depends on which
side of the filters transfer function the beat is tuned to. The microwave power is detected
by a tunnel diode (Herotek DT2018), which has an output bandwidth much greater than
1.5 kHz, and the phase sensitive amplification of this dither component of the microwave
power is used to generate an error signal. This is integrated and fed back to the off-line
laser injection current controller. Thus, the off-line control system locks the frequency
52 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
difference of the lasers to the zero crossing of the derivative of the bandpass filters transfer
function, in a similar way to the stabilization used for the on-line laser.
There are two lock points for the off-line laser, one on each side of the absorption
line, and these are easily distinguishable when setting up the system by observing the
injection current or the temperature of the master laser. This contrasts with the active filter
technique (Schilt et al., 2008), that uses a microwave mixer and oscillator. This produces
four such points of ambiguity with a zero crossing error signal, and makes initial setup
more difficult. Furthermore, because we dispense with the active microwave electronics,
we reduce cost and complexity, and enable offset locking at any frequency where a passive
bandpass filter is available.
The main disadvantage of the passive locking technique is having a fixed offset fre-
quency, that is more difficult to change than simply varying an oscillator. Obtaining a
different frequency offset (from 16 GHz) is a matter of picking a bandpass filter centered
at the desired offset frequency.
When choosing the RF bandpass, the ideal bandwidth Full Width at Half Maximum
(FWHM) of the filter should be equal to the magnitude of the optical dither, which in
our case is 500 MHz. Furthermore, the filter’s transfer function should not have a ripple
or a wide flat region in the passband. Also, the photodiode that detects the beat signal
needs to have a sufficiently large bandwidth, and photodiodes up to 100 GHz are currently
available, while other types of microwave components can be used at higher frequencies,
as discussed in section 3.9.
3.6 System timing
For water vapour in the lower troposphere, the linewidth of the molecular resonance at
822.9 nm is dominated by pressure broadening and ranges from about 5 GHz at sea level
to about 3 GHz at the highest altitude at which we can hope to measure water vapour con-
centrations with a transmitted pulse energy of 500 nJ (about 4 km). For a 1% accuracy, the
wavelength accuracy of the on-line master laser needs to be less than ∼100 MHz. How-
ever, the optical frequency deviation due to the on-line master laser dither in our system is
3.6. SYSTEM TIMING 53
÷2
Synchronous Dither
Data Acquisition Ch1
AOM1 RF Drive
Slave Pulse Output
Data Acquisition Ch2
AOM2 RF Drive
∫Master Oscillator ∫
Td
Td
Td
Td
÷2
Q Q
∫
÷2
TdAnalog Integrator
Digital Divider
Variable Time Delay
50 ohm cable driver
Figure 3.7: DIAL timing diagram illustrating how the dither signal is synchronous withall other operations.
500 MHz. It is advantageous to have this large dither to obtain a strong error signal from
the phase-sensitive amplifier for the vapour cell, however, the wavelength deviation this
produces results in unacceptably high variability in the on-line optical frequency. To get
around this difficulty, we synchronize the extraction of the optical pulses by the AOMs
with the zero crossings of the dither signal. This is illustrated in figure 3.7, and figure 3.8a
shows the synchronicity between the dither sinusoid and the optical pulses. This ensures
that the laser frequency within each pulse is consistent from pulse to pulse, however, as
the optical frequency is modulated by the dither, there is an optical frequency chirp dur-
ing each pulse. In our system, the pulse length is 1 µs and the dither period is 1333 µs,
resulting in an optical frequency chirp of less than 1 MHz. figure 3.8b illustrates this with
a temporal zoom of figure 3.8a.
The output pulse can be set to any phase of the dither signal, however there is an ad-
vantage to keeping it close to the zero crossing, as a method of suppressing the switching
transient from the control loop. This consists of a synchronous amplifier which is im-
plemented using an analog multiplier. Switching transients that occur close to the zero
crossing of the reference are therefore effectively suppressed by the analog multiplier,
and therefore are rejected from the error signal. This is further illustrated and discussed
in appendix A.3.
54 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
(a) Optical pulse near zero crossing of dither (b) Zoom of pulse interval showing little dither sweep
Figure 3.8: Timing dither synchronization illustrating (a) the optical pulse timing with thesinusoidal dither, and (b) a zoom of (a) illustrating the negligible dither chirp during theoptical pulse.
3.7 Control system performance
A simplified schematic of the control system is illustrated in figure 3.9, and a more de-
tailed analysis of the control model and system characterization is presented in Appendix
B. The on-line wavelength controller consists of a conventional lock-in amplifier set up to
measure the water vapour resonance. The error signal for the off-line laser is generated by
measuring the transmission of the beat frequency through a passive bandpass filter. With
the feedback loops, Sw1 and Sw2 closed, a step perturbation of the on-line wavelength
alone will result in transient waveforms at both test points TP1 and TP2, as illustrated in
figure 3.10. However, a perturbation to the off-line wavelength, will perturb the off-line
control system (TP2) alone, with no effect on the on-line wavelength (not shown).
The step response illustrated in figure 3.10 was used to calibrate the optical frequency
response as a function of step voltage. This was then be used to characterize the wave-
length stability performance of our control system by measuring the closed loop noise at
both TP1 and TP2 in figure 3.9. A small sample of the acquired data is illustrated in figure
3.11. Since the off-line system includes noise contributions from more sources, we can
see a greater noise amplitude at TP2 compared to TP1. A much longer sample of noise
at TP1 was acquired, with a histogram of the data shown in figure 3.12. Note that these
3.7. CONTROL SYSTEM PERFORMANCE 55
Off-line outputλ2
Pλ1
Fiber coupler - photodiode -
I λLaser Diode 2
Test Point TP2
Bandpass filter
On-line laser output
λ1
P λ
Ratiometric Vapor cell
Test Point TP1
+
Perturbation
Sw2
Sw1
λ2
∫On-line
I λLaser Diode
λ = λ0 + gl(I − 40)V I
Laser Diode Controller
I = gc1(V )
V
Detector with Lock-in
V(t)=ga1(P(t-k))
V ILaser Diode Controller 2
I = FC2(V ) λ = λ0 + gl(I − 40)
∫Off-line
P = gv(λ)
V(t)=ga2(P(t-k))
Detector with Lock-in
P = gb(| λ1− λ2 |)PV
P
Figure 3.9: Simplified control system model block diagram
0 0.5 1 1.5 2 2.5 3
−0.04
−0.02
0
0.02
0.04
0.06
time (s)
lock−
in a
mp.
out
puts
(V
) off−line
on−line
Figure 3.10: An online control system perturbation effects both the on- and off-line wave-lengths, but they both stabilize within a few seconds. Figure courtesy Dr. Hamilton.
56 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
Figure 3.11: Laser wavelength variability measured over a period of 10 seconds from theerror signals at TP1 and TP2, with the feedback loops closed. TP1 shows the fluctuationsin the on-line laser frequency, and TP2 shows those for the beat frequency. The signalsshown are the voltages measured at the respective test points, scaled so that the verticalaxis is in frequency units. The vertical separation between the traces is an arbitrary DCshift introduced for clarity. Figure courtesy Dr. Hamilton.
0.5 1 1.5 2 2.5 3 3.5 4 4.5 50
0.05
0.1
0.15
0.2
0.25
0.3
Optical Frequency Deviation (MHz)
Pro
babili
ty
Figure 3.12: Histogram showing the probability density of optical frequency deviationof the on-line master laser. With a sampling rate of 100 Hz and a measurement time ofalmost 10 minutes, we are measuring the laser frequency deviations on a time scale from500 to 0.02 seconds.
3.8. LASER AMPLIFIER 57
results don’t include DC offset errors and other sources of error. These will be considered
in further detail in Appendix A.
3.8 Laser amplifier
The tapered SOA or TA, consists of a bare mounted diode manufactured by Sacher
Lasertechnik (type TEC-400-830-500), with the optical and pulse electronic systems de-
signed around the characteristics of this device. A fast pulse driver having a slew rate
of 500 A/µs with no overshoot, was designed and built specifically for this purpose as
described in Appendix A.5. This involved a significant design effort as there were no
suitable and commercially available current drivers at the time. Appendix A.5 also dis-
cusses these design considerations including a schematic and layout.
The input free-space coupling to the TA used a Thorlabs A390TM-B aspheric lens,
while output coupling consisted of a Thorlabs C330TM-B aspheric followed by a plano-
cylindrical lens with a focal length of approximately 10 cm to reduce the transverse astig-
matism. The flat face of the cylindrical lens was turned at fixed angle, to prevent feedback
to the laser amplifier, and this reflection was utilized to sample the transmitted pulse. In
order to maintain input alignment to the laser amplifier during experiments and observa-
tions, we kept tabs on the output pulse, and tweaked the input occasionally to maintain
output pulse power as explained in Appendix 4.7. Circuits and further details are provided
in Appendix A.5, with the assembly of optics and drive electronics shown in figure3.13
and A.8.
3.9 Proposed applications of laser control system
3.9.1 Introduction
During the development of this DIAL instrument, two interesting questions naturally
arose. Firstly, How can the 100 GHz frequency ceiling of RF electronics for the offset
locking be overcome? Also, how can more than one off-line laser be stabilized with re-
58 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
Figure 3.13: Optical amplifier with electronics mounted as close as possible to the laserchip
spect to the on-line? In this section, the author proposes some novel ideas that illustrate
how to do just that.
The first idea removes the 100 GHz beat frequency limitation imposed by microwave
electronics by substituting Far Infra-Red (FIR) optics. This should enable beat stabi-
lization beyond 3 Terahertz using currently available components Toptica Photonics Inc.
(2012). Furthermore, this technique would allow the beat frequency stabilization to an
absolute reference such as a molecular resonance line. This could be of great interest to
high-sensitivity trace gas detection as polar molecules have a fundamental dipole reso-
nance, and hence the largest absorption cross-section than in any other part of the electro-
magnetic spectrum.
In order to stabilize multiple off-line lasers, the author proposes an extension to the
system topology presented thus far, that would allow any number of lasers to be stabilized
to each other. Furthermore, using a combination of these two techniques, it should be
possible to lock an arbitrary number of lasers to an on-line laser using molecular beat-
3.9. PROPOSED APPLICATIONS OF LASER CONTROL SYSTEM 59
THz Detector
Reference-2
+-
Photomixer
Gas Cell
BP
16 GHzDetector GaAs PIN
Optical Input
Optical Input
Electronic Output
Electronic Output
Figure 3.14: This diagram shows the beat frequency reference used for DIAL (top), andthe proposed THz beat frequency reference (bottom).The photomixer is essentially an an-tenna dipole separated by an intrinsic semiconductor substrate. When a DC electric fieldis applied across the substrate, the optical beat signal produces a corresponding currentflow. This is guided through a frequency matched dipole THz antenna. Various antennadesigns are currently being developed for broadband operation as well as for higher con-version efficiency (Gregory et al., 2007). This device essentially performs the same func-tion as the GaAs PIN photodiode, converting an optical beat signal into a correspondingRF field. By using a photomixer and replacing the coaxial waveguide with a free-spacepropagation, the limitations on the magnitude of beat frequencies that may be employedare removed.
frequency references, with the on-line laser itself locked to another absolute molecular
reference. To the best of my knowledge, this has not yet been done either.
3.9.2 A wideband dual-frequency locked laser system
The utilization of a passive stabilization reference offers unique advantages beyond DIAL
for stabilization of the beat frequency to an absolute standard, such as a vapour cell, as
well as for very high beat frequencies in the Terahertz range.
The beat frequency generated by linear wave mixing of two lasers has long been rec-
ognized as a viable Terahertz RF frequency source (Evenson et al., 1984) for such ap-
plications as gas phase spectroscopy (Hindle et al., 2008) and imaging (Mueller, 2003).
However, the ability to precisely tune a CW Terahertz signal to an absorption line as a
detection technique as illustrated in figure 3.15, does not seem to have received attention
from published research.
The system illustrated in figure 3.14 is a modification from section 3, where the band-
pass filter is replaced by a second gas reference cell, thereby removing most restrictions
on the magnitude of the beat frequency, while stabilizing it with absolute accuracy. The
60 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
Σ
Detector
∫Reference cell Dither
splitterFiber
∫∫
LPF
+-
Photomixer
Gas Cell
IT
IT
Laser2
ISO
ISO
Laser1
KTim
ingSystem
∫∫
PhotomixerDetector +
-
Gas Detection Cell
∫∫
LPF Data Aquisition SystemFIR THz signalElectric signalOptical Fiber
Figure 3.15: This diagram shows a how a stabilized continuous wave FIR source can beused for trace gas detection. The beat frequency is modulated by the dither signal thatsimultaneously performs the stabilization using the reference cell, while at the same timeproviding the synchronous measurement at the gas detection cell. By selecting a strongsharp resonance line with deep frequency modulation from the dither, a long detectionpath length and a low-noise THz detector, very high sensitivity can be achieved.
THz detector measures the transmitted radiation through the gas cell where an absorp-
tion line serves as a reference. Where there are very strong absorption lines, a short path
length can suffice. In this application, a square-law detector is required after the gas cell
to destroy phase sensitivity, and provide the same functionality as a RF Tunnel detector
at RF frequencies, thereby measuring only the frequency specific attenuation of atomic
resonance. There are some low-cost room-temperature devices available at various fre-
quencies, including Pyroelectric detectors, Schottky diodes up to 5 THz and MIM diodes
(Cowell et al., 2011) (Berland, 2003) up to tens of THz. There are also a variety of room-
temperature photoconductive intrinsic small band-gap semiconductor detectors that cover
much of the THz band, such as GeBe at 7-9 THz (Odashima et al., 1999).
3.9. PROPOSED APPLICATIONS OF LASER CONTROL SYSTEM 61
Optical Receiver
Σ
On-Line
IT
IT
− LPF
On-Line
Laser2
∫
SOA
ISO
ISO
Laser1
Optical Pulse Output
TimingSystem
RF Drive
Dither
AOM
AOM
∫∫
splittersFiber
BP
RF-1Detector∫ LPF
BP
RF-2Detector∫ LPF
IT
Laser3 ISO AOM
Further Stages, etc.
Further Stages, etc.
Figure 3.16: Proposed system for locking any number of side-line and off-line lasers. Anadvantage of this topology is that none of the off-line or side-line channels have any dithermodulation of the optical frequency.
3.9.3 Locking multiple off-line or side-line channels
In this section we illustrate an extension for stabilizing multiple off-line lasers to the on-
line master, using a separate beat frequency standard for each off-line channel. Off-line
channels with an offset of a few GHz, also known as side-line channels, are important
for space-based DIAL due to the opacity of the upper atmosphere to an optical frequency
stabilized to the center of an absorption line, as illustrated in figure 2.6. A highly stable
side-line optical frequency therefore serves as the on-line channel in such applications.
Such an instrument will therefore require three or more stabilized lasers. Furthermore,
the data acquired with the additional off-line channels can provide some temperature data
(Bosenberg, 1998), as well improved signal over a wider range of humidity conditions.
The principle of operation of much of the system illustrated in Figure 3.16 is largely ex-
plained in previous chapters. The extension to the number of channels is possible because
only one laser, the on-line channel, is subject to dither modulation. None of the off-line
62 CHAPTER 3. DIAL TRANSMITTER AND SYSTEM DESIGN
optical frequencies are modulated, which eliminates any possibility of beat signal ambi-
guity in the locking scheme. The main restriction on the numbers of off-line lasers, is
optical losses in the optical free-space and fiber splitters. However, there are alternative
optical system designs that mitigate these difficulties, which are beyond the scope of this
work.
3.10 Summary
This chapter provided an overview of this DIAL design in the context of its require-
ments, and past and present work in this field. The coupled control systems were modeled
and characterized, with measurements illustrating the stability of the master laser wave-
lengths. Timing, synchronization and optical amplification were also discussed. Finally,
two novel extensions to this design were described, one applicable to DIAL, while the
other applicable to trace gas detection. The next chapter discusses experiments that were
performed with this DIAL, including calibration and atmospheric results.
Chapter 4
System Characterization, Calibration
and Application
4.1 Introduction
The DIAL system described in the previous chapters was studied and subjected to exper-
iments to understand its characteristics and error sources. In the process of characterizing
and calibrating the system, 14 atmospheric H2O measurements were performed between
September 2008 and April 2010 during which time there were virtually no changes to
the system hardware. The actual observations were timed to occur close to the sched-
uled radiosonde launches around 12:00 UTC from Adelaide Airport, 8 km SW from our
location, which enabled us to compare this data with our observations.
The significance of the following error sources were considered.
1. Master laser wavelength accuracy
2. Master laser spectrum
3. Laser amplifier ASE
4. DIAL receiver acquisition precision and noise
5. Alignment and lidar overlap
63
64 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
Although DIAL measurements are in principle self calibrating, not all systematic and
random error sources are necessarily eliminated in a practical instrument. Self-calibrating
DIAL requires the transmitted wavelength to be well centered on the spectral feature, and
the spectral power needs to be confined well within the absorption line width, for the
duration of an observation. Although we were able to achieve very good wavelength
accuracy, it was not always possible to maintain spectral purity due to mechanical optical
misalignments, as described later.
Master laser wavelength accuracy
The absolute accuracy of our master laser was obtained from the measurement of the
absorption line itself and could achieve stability of ∼5 MHz as illustrated in figure 3.12.
However, our hand-made absorption cell had the input and output beams grazing the tops
of the coupling mirrors causing alignment dependent scattering and diffraction. This is the
probable reason why it sometimes exhibited strong multi-path interference. Furthermore,
there were occasional other, unexplained sources of Fabry-Perot type interference that
overwhelmed the ratiometric detection which had a limited linearity over a finite dynamic
range.
During previous laboratory experiments, it was observed that the fringe intensity var-
ied with the cell alignment as well as time and temperature. Before and after the at-
mospheric observation experiment, spectral calibration experiments were performed as
described in section 4.3. The results illustrate the incipient nature of this phenomenon,
showing very little fringe contrast before the experiment, and 30% interference after the
experiment, relative to the peak extinction.
While we don’t know the exact cause of this, it appears to be due to the random nature
of alignment drift. There was no way of detecting this problem during an experiment, as
the fringes were only made visible by scanning the laser wavelength, which could not be
done during the observation while the laser wavelength was stabilized.
4.1. INTRODUCTION 65
Master laser spectrum
To a first approximation, our master laser diode consists of a low-finesse Fabry-Perot
cavity with a temperature dependent length, and a gain medium with a peak gain wave-
length determined by the injection current (Dechiaro and Chemelli, 2003). During lasing,
the cavity will contain one dominant standing wave with an integer number of periods,
known as the longitudinal mode of the cavity. However, there will also be some lasing on
adjacent longitudinal modes.
Since the injection current also changes the carrier density and hence the refractive in-
dex in the cavity, changing both current and temperature makes it possible to produce the
same dominant lasing wavelength from a different longitudinal mode. However, because
the peak gain of the medium relative to the wavelength will be different, the amount of
lasing energy on adjacent modes will be different as well.
Our lasers exhibited a high degree of excited state depletion by the dominant lasing
mode, resulting in continuous mode-hop free tuning over a span approaching 200 GHz.
When the laser mode was forced to hop, the jump in optical wavelength was more than 1
nm. Therefore, the longitudinal mode hysteresis was very strong, with quite good spectral
purity over a significant portion of the continuous tuning range. Under some conditions,
however, adjacent modes could be significant, and easily visible on an optical spectrum
analyzer as illustrated in figure 4.1. Since our master lasers could produce significant
sidebands, it was desirable to determine the operating state prior to each experiment, in
order to make sure that we return the lasers to the same longitudinal mode as well as the
same wavelength each time.
The longitudinal mode state of the laser can be unambiguously determined by the
injection current, temperature and wavelength, and a technique to control the state of
each laser was developed. This was done by setting the initial temperature and current to
maximize the probability of landing the laser in the desired longitudinal mode. The laser
could then be powered up with current and temperature adjusted to get to the required
wavelength. If the laser failed to be at the required wavelength, the laser was powered
down, and the procedure repeated. With a longitudinal mode spacing of 0.14 nm (62
GHz), the laser could easily reach the same wavelength from a different longitudinal
66 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
818 819 820 821 822 823 824 825 826 827−50
−40
−30
−20
−10
0
10
Wavelength nm
Opt
ical
Pow
er d
bm
0.14 nm Longitudinal Mode Spacing
Figure 4.1: Optical spectrum of master laser operating at full power, illustrating lasing onadjacent longitudinal modes. Acquired with a Yokogawa AQ-6315 OSA
mode, but this would require a different injection current and temperature. Returning
the laser to the same wavelength, current and temperature for each experiment, therefore
ensured the same longitudinal mode and hence the same spectrum.
Laser amplifier ASE
The laser amplifier produced significant ASE that depended on input alignment which
will be discussed in Section 4.7. During observation experiments, this apparently drifted
with time and temperature. Furthermore, our device exhibited some random behavior and
comparing the different experiments on different days, we still see a ∼ 5% discrepancy in
the effective absorption cross section immediately after alignment. One possible explana-
tion was that this tapered amplifier had been damaged during a previous project, judging
from the shape of the transverse mode produced from its input port.
These devices are sensitive to damage by feedback to the output, especially when there
is no optical input. Stimulated emission works in both directions, so the highly pumped
medium can produce power in excess of the optical damage threshold at the input of the
taper. This type of damage may have resulted in partial occultation of the gain medium,
increasing input alignment sensitivity, and possibly explaining some of the random errors
in the results that follow.
4.1. INTRODUCTION 67
DIAL receiver precision and noise
Noise calculations based on a detailed study of the propagation of independent error in
the simplified DIAL approximation, with complete overlap between laser transmitter and
receiver, have been published (Wulfmeyer and Walther, 2001). The result from this study
is a set of equations that consider shot noise, background photon count, amplifier noise
and speckle.
Due to the averaging of many pulses, we can discount speckle noise in our experi-
ments. Furthermore, since we are dealing with photon counting in the digital domain, we
can consider amplifier gain to be infinite which eliminates amplifier noise. Furthermore,
the relatively high repetition rate and the averaging of a large number of photocounts,
reduces the magnitude of the uncertainty due to shot noise despite the very low transmit-
ted power and low backscatter, except where the signal is almost indistinguishable from
background. The main source of error in this DIAL system is therefore due to background
(Hamilton et al., 2008). The sky background in Adelaide city is quite significant due to
aerosols from cars that scatter light from our excessively well lit streets with broad emis-
sion spectra. This instrument could therefore be expected to perform much better when
located away from metropolitan areas.
The simplified equation for variance σ2n from the random uncertainty due to back-
ground alone from (Wulfmeyer and Walther, 2001) and also from the Schotland approxi-
mation
σ2n =
(1
2k∆Rσon
)2 [Non(r1)+Nb
(Non(r1)−Nb)2+
No f f (r2)+Nb
(No f f (r2)−Nb)2+
Non(r2)+Nb
(Non(r2)−Nb)2+
No f f (r1)+Nb
(No f f (r1)−Nb)2
](4.1)
where the background photocount is Nb while Non and No f f are the on-line and off-line
photocounts respectively during each 1 µs interval, ∆R = r2 − r1 is the range interval, k
is the number of laser pulses used in the averaging, while σon = 1.46 × 10−22 cm−2 is
the on-line absorption cross-section, with the off-line absorption cross-section taken to be
zero.
68 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
Alignment and lidar overlap
The overlap function G(R) is one of the coefficients in the lidar equation 2.3. It is a
measure of how much of the transmitted beam’s scatter is in the field of view of the
receiver at range R. The low pulse power available from the gain-switched Tapered optical
Amplifier (TA) made lidar alignment very challenging. This resulted in uncertainties in
G(R), especially at close range, as discussed in section 4.4. Furthermore, it was found that
changes in the input alignment changed the shape of the transverse mode at its output. In
other words, different alignments of the on-line and off-line lasers relative to the TA input,
result in different shape and propagation direction of the transmitted pulse. This was most
likely a fault of the TA chip itself.
4.2 Humidity sensor calibration experiment
4.2.1 Introduction
A calibrated Relative Humidity % (RH) measurement, together with the measurement
of peak attenuation of a laser wavelength scanned across a water resonance line, pro-
vides a measurement of the effective absorption cross-section and a calibration of DIAL
observations that will be discussed in sections 4.3 and 4.5. This calibration, together
with temperature, therefore performs an independent atmospheric water number density
measurements without reliance on HITRAN parameters using calculations discussed in
Chapter 2 and an ideal laser.
Although the saturated salt solution is not a primary standard, it has long been con-
sidered a simple and accurate calibration technique (Greenspan, 1977). In this section,
we describe the calibration of a thin-film capacitive RH sensor using three saturated salt
solutions.
This calibration was performed after most of the observations, and its results were
used to subsequently analyze the observation data to calculate water number density, and
to estimate the effective absorption cross-section and spectral purity of our on-line laser.
4.2. HUMIDITY SENSOR CALIBRATION EXPERIMENT 69
4.2.2 Aim
The aim of this experiment was to obtain a calibration quadratic polynomial correction
for our humidity sensor chip, a Honeywell HIH4000, using three saturated salt solutions.
4.2.3 Method
We selected three saturated salt solutions to provide the three humidity references as
shown in Table 4.1. In order to maintain the salt solution at 25°C, we constructed the
Salt RHS @ 25°CCalcium Chloride CaCl2 29.0%
Magnesium Nitrate Mg(NO3)2 52.9%Sodium Chloride NaCl 75.3%
Table 4.1: Equilibrium %RH of saturated salt solutions, selected to span the humidityrange encountered in Adelaide.
experimental apparatus shown in figure 4.2 illustrating the overall system setup including
the electronic schematic. A PID controller (Shinko JCS-33) provided a 4-20 mA current
output that was converted to a roughly proportional power dissipation of up to 5 W with
a MOSFET transistor. A fan circulated the air in the glass jar to keep the temperature
homogeneous, while an opaque shield was placed around the jar (not shown) to prevent
radiant heat from entering the enclosure and disturbing the thermal equilibrium.
A salt vessel was constructed of a smaller glass jar with its lid machined to snugly
accommodate a connector shaft. With the application of some silicone grease, the con-
nector stem formed an airtight seal that suspended the sensor inside the sealed salt vessel,
without any contamination of the salt or the sensor. A separate vessel was used for each
salt solution. When a new salt solution container was placed into the apparatus, the sen-
sor connector stem was pushed through a freshly cleaned machined lid. In this way, no
part of sensor or assembly came into any contact with salt. Without air circulation (inside
the vessel), the height of the container should not exceed the smallest dimension of the
free surface of the solution (Organisation Internationale de Metrologie Legale (OIML),
1996). The rate at which equilibrium is attained increases with the surface area of the salt
solution, and decreases with the volume of gas in the sealed vessel. The dimensions of
70 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
this vessel came close to meeting this recommendation.
The sensor placed in each salt vessel in turn, that itself was placed in the temperature
controlled enclosure illustrated in figure 4.3 for a sufficiently long period of time. The
HIH4000 is a ratiometric sensor where the relative humidity is related to the ratio of input
and output voltages. Therefore, both the supply and output voltages were monitored until
the ratio reached equilibrium, or when it was observed to waver around the final point. It
was observed to take several days to reach equilibrium, probably due to the dimensions
of the vessel and the slow heat transfer with the air bath. For each salt solution in turn,
the apparatus was left for at least a week before the final readings were taken.
The apparatus consisted of the following components;
• Regulated 5 V power supply for the sensor.
• 5 v ± 1 V power supply for the heating assembly.
• Heating assembly consisting of a fan and a controlled heat source.
• Forced air enclosure, glass, with radiant shield.
• Sealed glass vessels containing saturated salt solutions.
• HIH4000 humidity sensor and voltmeter.
• 3-wire Platinum RTD temp. sensor and PID controller (Shinko JCS-33).
Sensor self-heating
With a sensor current of 400 µA and a supply of 5 V, the power dissipation is 2 mW.
The temperature rise depends primarily on the still-air (RCA) case-ambient thermal resis-
tance (K/W) which was not specified by the manufacturer (Honeywell, 2010). Using the
specifications for another temperature measuring device (the AD590MF) with a similar
package, we have RCA = 650 K/W. This thermal resistance would result in a tempera-
ture rise of about 1 K, resulting in an underestimate of relative humidity by about 5%.
In order to reduce this uncertainty, the thermal resistance RCA was reduced by attaching
a copper strip to the back of the sensor, which was then attached to the connector stem
with Alumina packed Epoxy to minimize thermal resistance. This is illustrated in figure
4.4. However, it was later recognized that this reduced the thermal resistance between the
4.2. HUMIDITY SENSOR CALIBRATION EXPERIMENT 71
25.025.0
PID Temp.Control
'C
Fan
Salt
Thermal Enclosure
Radiant Shield
Hydrostatic
HIH4000
DMM5. VSensorSupply
SensorPt RTD
5V
HeaterSupply
4-20mA
Heater
Water Vapor
Vessel
Figure 4.2: Experimental Setup
72 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
Figure 4.3: Experimental apparatus with the radiant shield removed
sensor and the ambient room air temperature as well, since the tip of the connector was
outside the thermal enclosure. The entire experiment therefore had to be repeated with the
sensor connector and stem completely inside the thermal enclosure, however, the results
did not differ significantly.
Sensor stabilization
In this experiment, the sensor was in the stabilized environment for an extended period
of time, and the sensor was left turned off overnight. The sensor was turned on and the
output was recorded as shown in figure 4.6. The sensor appears to stabilize in about two
hours. Like most electronic devices, the sensor exhibits 1f noise which implies that the
peak-to-peak drift increases with the observing time. This probably explains the slow
drift in the output after two hours.
4.2. HUMIDITY SENSOR CALIBRATION EXPERIMENT 73
(a) RH Sensoron connectorstem without Cuheatsink.
(b) Sensor connector assembly with sensor behind Cuheatsink.
Figure 4.4: Humidity sensor with and without Cu heatsink.
4.2.4 Results
Table 4.2 shows the experimental results, with the calculated least squares polynomial 4.2
illustrated in figure 4.5.
Salt RH (%) Sensor Voltage (V)Calcium Chloride 29.0% 1.94
Magnesium Nitrate 52.9% 2.77Sodium Chloride 75.3% 3.57
Table 4.2: Normalized sensor voltage for the three fixed RH experiments.
RH% = −0.4878V2 + 31.0929V − 29.4842. (4.2)
4.2.5 Discussion
The RH sensor has a temperature sensitivity of ∼0.2%RH °C−1 which is unrelated to self-
heating that would actually change the RH at the surface of the sensor chip. This is the
74 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
1 1.5 2 2.5 3 3.5 40
20
40
60
80
100
Sensor Voltage output V
Rel
ativ
e H
umid
ty %
HIH4000 RH Sensor Calibration
Measured Data PointsQuadratic fitSensor Specification
Figure 4.5: Calibration result shown alongside the sensor manufacturer’s typical specifi-cation. This result is just barely within the specified accuracy tolerance for this device.
response of the sensor itself to temperature. In other words, this error is present with
the sensor at the same temperature as the air, if the temperature is not 25 °C. Although
the calibration was done at 25°C, this sensor was actually used at different temperatures.
Fortunately, the manufacturer has provided a correction equation with a null effect at
25°C (Honeywell, 2010). After calibrating RH at 25°C using Equation 4.2, Equation 4.3
RHC = RH(1.054 − 0.00216T ) (4.3)
was used to apply the temperature correction as part of the calibrated RH measurement.
Most of the atmospheric observation were made within 10 °of 25 °C and its probably safe
to assume that equation 4.3 provides the exact correction. However, there was no way to
test this.
The only other known possible source of error in this experiment was the uncertainty
in the figures in table 4.1.
The following potential errors were considered, and discounted, as follows.
1. Controlled temperature accuracy
The temperature accuracy depends on the accuracy of the sensor and the instrument,
as well as self heating of the temperature sensor. The sensor current was measured
at 165 µ A, with a resulting power dissipation I2R of less than 3µ W, which would
result in negligible self heating. The combined accuracy of the ’A’ class sensor
4.2. HUMIDITY SENSOR CALIBRATION EXPERIMENT 75
and control instrument is specified as ± 0.1 °C. The RH of the saturated salts is
temperature dependent. The sensitivities of the RH with temperature were obtained
or derived from (Wiederhold, 1997) and (Leon-Hidalgo et al., 2009) and are shown
below.
Salt ∆ RH(%)/°C
Calcium Chloride 0.1
Magnesium Nitrate 0.3
Sodium Chloride 0.04
With the worse case sensitivity of just 0.3 %/°C, a 0.1 °C change would result in a
0.03% RH error.
2. Temperature stability and uniformity
Temperature stability was ensured with the stabilized PID linear control loop. The
integrator constant was increased to ensure control system stability. Temperature
uniformity was ensured with the active forced airflow, with a high power fan for the
enclosure size.
3. Temperature difference between the RH sensor and the salt solution
The self heating of the sensor was reduced with a heat sink, with the RH connector
and stem completely inside the thermal enclosure, any temperature gradient would
be negligible.
4. Purity of the salt and added water
The purity of the reagent grade salt was better than 99.9%. Deionized water was
used for the solution. The error contribution from these sources is probably negli-
gible.
5. Sealing and pressure
The total pressure in the vessel at 25°C would be slightly higher than ambient,
since it was sealed at ∼ 20 °C. However, relative humidity of saturated salts is not
at all sensitive to pressure (World Meteorological Organization (WMO), 2008).
76 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
0 50 100 150 200 250 30097.5
98
98.5
99
99.5
100
Time (minutes)
Perc
enta
ge o
f final valu
e (
%)
HIH4000 Sensor stailization after power up at 75% RH
Figure 4.6: Sensor stabilization after power-on. The sensor was in the stabilized environ-ment with the power turned off. The power was then turned on and the output recorded.This indicates that an initial error of up to 2% can be expected, with an eventual stabilityof around ±0.5%
4.2.6 Conclusion
The calibration polynomial was found and the humidity sensor was found to produce an
almost linear relationship with relative humidity that was barely within the manufacturer’s
specifications. The absolute accuracy of the sensor was just within the ± 8% tolerance
at 75% RH specified by the device manufacturer. This calibration corrected for a sensor
error of around 5% at 50% RH. The sensor was found to stabilize to better than 1% of the
final value, two hours after power was applied.
4.3 Spectral calibration experiment
4.3.1 Introduction
The DIAL technique has the potential to perform self-calibrating measurements of vapour
pressure if the actual absorption cross section is known precisely and if the spectral pu-
rity is known. Although HITRAN is the most comprehensive and highly cited source for
spectroscopic parameters, the accuracy of the data in this reference is not well established.
One of the HITRAN parameters ’ierr’ is the uncertainty index for wavenumber, intensity,
4.3. SPECTRAL CALIBRATION EXPERIMENT 77
and air-broadened half-width Rothman et al. (1998). Even though the ’ierr’ parameter
for our line at 822.922 nm indicates the highest degree of certainty, the actual spectro-
scopic parameters for this line have changed significantly with the 2008 release. With the
Voigt model, the absorption cross-section of the line used for observations has changed
by almost 16%, with the parameters shown in table 4.3.
822.922 nm S T × 10−23cm γa(cm−1)HITRAN-2006 3.848 0.0933HITRAN-2008 4.470 0.0915
Table 4.3: Table showing changes to the intensity S T and halfwidth γa parameters for the822.922 nm line for two consecutive HITRAN releases.
Another source of uncertainty in the effective absorption cross-section, is due to its
dependence on the spectral purity. The measurement of the effective absorption cross-
section against fixed-point humidity references, serves as a calibration for the laser spec-
tral purity, as well as an independent measurement of molecular absorption cross-section,
without any reference to HITRAN parameters.
4.3.2 Aim:
Since there is uncertainty as to what the correct HITRAN value is, as well as the spectral
purity of our laser, the aim of this experiment was to measure the effective absorption cross
section at 822.92 nm around the same time as our atmospheric observations on 23/9/2009
using the same laser pulses and system configuration as used for the observation.
4.3.3 Method part 1. Spectrum acquisition
This method acquired the spectrum using the transmission pulses with the vapour cell.
The optical spectra were acquired before and after the atmospheric observation, with the
system quickly re-arranged between that shown in figure 3.4, and figure 4.7. The proce-
dure is detailed in Appendix D. In summary, the system was aligned and the ∼500 mW
optical pulses were re-directed into a single-mode optical fiber leading to the multi-path
cell that was normally used for wavelength stabilization. The wavelength of the on-line
78 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
master laser was then scanned across the absorption line, and the transmission through the
cell was measured at the center of the absorption feature and on the wings. This was done
by acquiring two datasets, one from each photodiode, using a high speed data acquisition
system consisting of a computer, a GageScope GS1205 and the adgmr.m code described
in Appendix G. The system layout is illustrated in figure 4.7. Each dataset consists of
7000 pulse acquisitions, where each acquisition contains 256, 12-bit samples, acquired at
100× 106s−1. For each scan calc.m code was used in post process to calculate the ratio of
pulse energy exiting the cell by that entering it, in the same way as the analog ratiometric
calculation during laser stabilization. From this data, the relative transmission of each
pulse was measured so as to obtain an optical spectrum due to the absorption cell. The
resulting data is plotted in figures 4.8 and 4.9. These figures, together with the Savitzky-
Golay (SG) fitted curves, served as a graphical aid to finding the maximum (on-line) and
minimum (off-line) absorption values, from which the effective absorption cross-sections
were calculated. A total of 8 separate scans were performed, 4 before and 4 after the
atmospheric observation experiment.
4.3.4 Method part 2. Number density measurement
This method was used to obtain the water molecular number density from the Relative
Humidity % (RH), U using the previously calibrated humidity sensor described in exper-
iment 4.2, and temperature T obtained using a precision mercury thermometer. From the
temperature measurement the saturation vapour pressure, es is given by
es = exp(−2991.2729T−2 − 6017.0128T−1 + 18.87643854 − 2.8354721 × 10−2T
+ 1.7838301×10−5T 2−8.4150417×10−10T 3 + 4.4412543×10−13T 4 + 2.858487 ln(T ))
(4.4)
This empirical equation actually provides far greater accuracy than required (Buck, 1981)
(Wexler, 1977). With the relative humidity measurement U, and the saturation pressure
es, the partial pressure of water vapour e′ is given by (World Meteorological Organization
4.3. SPECTRAL CALIBRATION EXPERIMENT 79
Fib
erC
ou
ple
r
Sys
tem
Tim
ing
Sp
ectr
al C
alib
rati
on
an
d M
easu
rem
ent
Sy
stem
To
OS
A
Sw
itch
Sh
utt
er
Fre
e S
pac
e P
rop
agat
ion
Sin
gle
Mo
de
Op
tic
Fib
erE
lect
ron
ic S
ign
al
Bea
msp
litte
r
Mir
ror
Osc
illo
sco
pe
ND
filt
er
Co
mp
ute
r W
ith
C
S14
105
I TL
aser
#1
Far
aday
Iso
lato
rsA
OM
TA
Fab
ry-P
ero
t
RF
Dri
veT
P
Figure 4.7: System layout illustrating a typical spectral calibration setup. This illustrationshows two possible configurations, one configuration is used to capture the spectrum us-ing the tapered amplifier, while the other captures the spectrum using the master laser. Inmaster laser mode, the shutter is open and the power laser switch is off. Inverted pulsesfrom the master laser are acquired from both sides of the vapour cell, as the laser tem-perature is scanned with a period of ∼ 5 seconds. In power laser mode, the TA is turnedon, and the shutter is closed. Positive pulses are acquired in the same fashion as before.The FP is used to find the TA input alignment, while the oscilloscope measures the outputpulse power. An OSA can be used to check the wavelength.
80 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
(WMO), 2008)
e′ =U.es
100. (4.5)
From the ideal gas law, the molar concentration n of any gas is
n =p.XvRT
=e′
RT. (4.6)
From this the absolute humidity, or number density per cubic centimeter N in the room
and hence inside the open vapour cell is calculated
N = N.n. × 10−6, (4.7)
where N = 6.022 × 1023 and R = 8.31.
A surprising result from the previous equations is that N is calculated without using
barometric pressure. However, to relate these measurements to Radiosonde data, the mass
mixing ratio r is calculated
r = 621.98e′
p − e′, (4.8)
with the barometric pressure, p. The results are summarized in Table 4.5.
4.3.5 Results
From sections 4.3.3 and 4.3.4 we have two independent sets of results that are combined to
calculate the effective absorption cross-section. From the acquired spectra measurements
summarized in Table 4.4, the relative on-line transmission T = 0.885. From the measured
humidity and temperature, Table 4.5 gives N = 2.47× 1017 cm−3. We can see that there is
no discernible change between the early and the late measurements for both sets, during
the course of the experiment. With the vapour cell path-length x = 3300 cm, we can
evaluate equation 4.9 to find the effective absorption cross-section σe f f
σe f f =−1Nx
ln(T) = 1.49 × 10−22 cm2. (4.9)
4.3. SPECTRAL CALIBRATION EXPERIMENT 81
0 1000 2000 3000 4000 5000 6000 70000.32
0.33
0.34
0.35
0.36
0.37
0.38
Wavelength scan (data points)
Tra
nsm
issio
n (
arb
itra
ry u
nits)
(a) Spectrum at 7.54pm
Figure 4.8: Absorption spectrum example before observation on 23/09/2009
0 1000 2000 3000 4000 5000 6000 70000.31
0.32
0.33
0.34
0.35
0.36
0.37
0.38
Wavelength scan (data points)
Tra
nsm
issio
n (
arb
itra
ry u
nits)
(a) Spectrum at 11.13pm
Figure 4.9: Absorption spectrum example after observation on 23/09/2009
82 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
Time Off-line On-line Transmission (T*100) %19.54 .367 .325 88.619.52 .376 .333 88.619.40 .495 .428 88.519.25 .495 .438 88.523.13 .358 .318 88.823.11 .360 .318 88.423.09 .360 .318 88.423.07 .358 .318 88.8
Table 4.4: Measured off-line and on-line optical power and transmission results
Time (approx) Sensor Vout T °C RH % r gkg N cm−3
19.30 2.48 19.4 44.2 6.24 247×1015
21.00 2.64 17.8 48.4 6.19 246×1015
23.30 2.47 19.5 44.0 6.24 247×1015
Table 4.5: Number density results from calibrated RH sensor and thermometer, with mix-ing ratio using BOM data for 23/9/2009 with MSLP = 1016 hPa.
ν0(cm−1) S T (cm) γa(cm−1) γs(cm−1) n σv(cm2)12151.8236 4.47×1023 0.0915 0.432 0.7 1.46×10−22
Table 4.6: Voigt model at 822.922 nm to calculate absorption cross section, σv fromHITRAN parameters ν0, S T , γa, γs and n
4.3.6 Discussion
Table 4.4 shows the transmission results measured from the graphical data presented in
figures 4.8 and 4.9, while table 4.5 presents the results of the water vapour number density
measured using the calibrated humidity sensor. Putting both of these results into equation
4.9, we get our measured effective absorption cross-section.
It is interesting to compare the result in 4.9 with the cross-section calculated from
HITRAN in Table 4.6. This uses the full spectral model described in Chapter 2 with
Equation 2.9 to calculate the total pressure broadened linewidth, and the Doppler width
from Equation 2.13, while equations 2.18 and 2.19 were used for the Voigt model.
Our cross-section result is very close to, but slightly higher than the HITRAN Voigt
model. This would be impossible if the experimental results and HITRAN data were to
4.3. SPECTRAL CALIBRATION EXPERIMENT 83
be perfectly accurate, since a finite laser linewidth would always result in a reduction
of the effective absorption cross-section. If the laser spectral purity on this occasion was
particularly good, and the HITRAN parameters are accurate, then this result sits well with
our estimated precision of ±5% for the extinction measurements. However, on another
occasion a reduction in absorption cross-section due to TA of some 13% was found, as
described in Experiment 4.6. However, due to difference in amplifier alignment described
in Experiment 4.7, as well as a lack of error bounds in the HITRAN parameters, it is not
possible to resolve this ambiguity.
A possible major source of error is clearly visible in figure 4.8a that shows apparent
laser amplifier instability of the transverse mode. This could not have been due to mas-
ter laser mode hopping, since this type of mode-hop would have been irreversible due to
excited state depletion by the lasing mode. The tapered laser chip was used in a previous
project where it was partly damaged, and the transverse mode of this device has a com-
plex structure, and random fluctuations have been observed during previous experiments.
By observation of the actual pulse train near sample number 6000, (possible transverse-
mode) bi-stability is clearly visible (not shown). Although this had no significant effect
on the measured absorption cross-section results from data, this type of noise could have
degraded the DIAL observation.
In order to correct for the alignment drift, the input alignment of the tapered amplifier
was adjusted during the observation experiment, as will be discussed in detail in the next
section. This may have been partly responsible for the fringes in figure 4.9 that are not
present in figure 4.8. The presence of this signal on both channels suggests an origin
prior to the absorption cell. The contrast of these fringes in each channel was so strong
(about 70%), that the linear range of one of the amplifiers was exceeded, causing their
appearance in the ratiometric results. Although the ratiometric fringe contrast was less
than 4% of the full scale range, they amount to ∼30% relative to the magnitude of the
absorption peak. This type of behavior would have a significant impact on the accuracy
of the on-line laser locking system, as a random variation in an interference fringe relative
to the actual water absorption peak, would modulate the on-line absorption cross-section
by the same fraction.
84 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
4.4 Atmospheric DIAL observation experiment
4.4.1 Aim
The aim of this experiment was to conduct an atmospheric observation with a calibrated
DIAL instrument and compare the results with radiosonde and local humidity data, with
a view to identifying remaining issues with our instrument as part of the development
process towards a field-deployable, low-cost instrument.
4.4.2 Method
This experiment was conducted concurrently with calibration experiment described in
section 4.3 that used results from calibration described in section 4.2 with traceability to
saturated salt solutions. The system schematic is illustrated in figure A.1 and the overall
setup is shown in figures 3.5 and 4.10 showing the external mirror and periscope. The
initial part of the setup method was described in section 4.3, while other setup details are
provided in section 3.8.
Alignment
The alignment of the transmitted beam with the FOV of the receiver was particularly
difficult due to the weak pulse and significant background. Alignment was found by
a trial and error iteration by scanning the X-Y direction of the transmitted beam while
observing the return signal.
Receiver data acquisition
The receiver consisted of a Hamamatsu R7400U-20 PMT with enhanced IR response
(Hamamatsu, 2001) operating at 1000 V, in a shielded enclosure near the focal plane of
a 400 mm diameter SchmidtCassegrain telescope. In order to reduce the large inner city
background signal, the FOV of the receiver was reduced with an aperture in front of the
PMT. This reduced the effective diameter of the PMT from 10 mm to 4.43±0.02 mm. The
Licel (LICEL GmbH, 2002) data acquisition recorder consisted of a 2 channel counter
4.4. ATMOSPHERIC DIAL OBSERVATION EXPERIMENT 85
Optical Receiver
Optical Pulse Output
Wall
WindowExternal mirror
SOA
A390TM-B C330TM-B
CylindricalLens
Lab Interior Exterior
TEC-400-830-500
Periscope
Figure 4.10: Observation arrangement for atmospheric transmission. In order to transmitthe pulse vertically through the atmosphere, the astigmatically corrected and collimatedoutput pulse was aligned using a periscope to a large mirror, mounted at 45°outside thelaboratory’s window. This same mirror was also used to collect scattered light by aligningit collinear to the receiving telescope.
with a maximum rate of 250 MHz and a 4094 acquisition memory with a time resolution
of 50 ns. When this memory is full, the data is transferred to a computer running Labview
that controls the Licel recorder.
4.4.3 Results
The data acquisition commenced at 8.42 pm on 23/09/09 and continued for 40 minutes.
Figure 4.11 shows differential signal up to 1.2 km. The figure shows the shape of the
transmitted pulse scattered from the external mirror, with a slight rise after this pulse.
This rise is likely due to the increasing overlap between the transmitted beam and the
field of view of the receiver, which is complete beyond some finite range.
The on-line signal drops to essentially the background level beyond ∼600 m beyond
which it is not possible to calculate water vapour number density above that altitude,
however, some cloud scatter is clearly visible in the off-line channel at a range of ∼2.4
km.
86 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
0 500 1000 1500 2000 2500 300010
5
106
107
Range (m)
Photo
n C
ount
DIAL Results 23−Sep−2009
Offline
Online
Figure 4.11: Photon count per 50 ns interval, summed over the duration of the observation.Cloud return is visible in the offline channel at 2.4 km.
Figure 4.12 provides a zoom of the data up to 1 km, and clearly shows the differential
return. Since the scatter in the 0-150 m range is due to the deflecting mirror, as well as
the atmosphere, it cannot be used to calculate atmospheric returns. The different levels of
scatter in this range is due to different laser power levels at the two wavelengths, as well
as slightly different beam alignments, resulting from different transverse beam profiles.
This would also result in a different overlap function at the two wavelengths that could
also to be a source of error at close range.
This data was stored as 614 files, each corresponding to a 4 second interval, during
which approximately 3 seconds worth of actual data was acquired. The Licel recorder ac-
quires the 4094 shots, and then transfers the data, during which acquisition is apparently
paused. The averaged acquisition for each channel consisted of 1023 temporally consec-
utive data points, each acquired at a rate of 50 ns. This results in a maximum acquisition
range of about 7 km, however, there was no discernible return above about 3 km during
this experiment. Earlier experiments showed that high cloud at 6 km is easily detectable.
Some patchy cloud was noticeable on the night, and figure 4.11 shows the cloud return in
the offline channel at around 2.4 km. The absence of any signal in the online channel at
that range provides confirmation of DIAL operation as expected.
The background plus dark count was 160×103 per µs, summed over the duration of
the experiment, differing by less than 1% between the online and offline channels. This
4.4. ATMOSPHERIC DIAL OBSERVATION EXPERIMENT 87
0 100 200 300 400 500 600 700 800 900 100010
5
106
107
Range (m)
Photo
n C
ount
DIAL Results 23−Sep−2009
Offline
Online
Figure 4.12: Same data as in figure 4.11, truncated to 1 km range
figure was obtained by averaging the data acquired immediately before the laser pulses,
which was the first 100 samples immediately prior to each laser shot, and consists of the
sum of all sources including sky background that gets through our 1 nm optical bandpass
filter, scattered laser radiation from the bench that may find its way into the telescope,
as well as the dark count characteristic of the photomultiplier tube operated at its abso-
lute maximum voltage of 1000 V. The number of photocounts due to backscatter were
then estimated by subtracting the dark count from the total count. Table 4.7 shows the
backscatter count at each of the range bins corresponding to our resolution of 150 m, with
the background subtracted. The final column shows the resulting calculated mixing ratio
using the effective absorption cross-section measured in section 4.3.5.
88 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
Range (m) Online count Offline count Number density N cm−3 Mixing ratio gkg
0-150 2118152 1171808 n/a n/a150-300 953584 1299632 1.93E+017 5.03300-450 234664 752056 1.83E+017 4.85450-600 31608 251416 1.94E+017 5.25
Table 4.7: DIAL observation results
Range (m) Radiosonde mixing ratio ( gkg ) % difference
0-150 7.6 n/a150-300 7.4 32300-450 6.7 28450-600 5.9 11
Table 4.8: Comparison of DIAL results from Adelaide University and radiosonde datafrom Adelaide airport
4.4.4 Discussion
Tables 4.7 and 4.8 compare our experimental results with that of the radiosonde launched
at around the same time, about 8 km south-west of out position. Our results systematically
measure a significantly lower mixing ratio. It is significant to note that our calibrated
humidity sensor also measured a lower water mixing ratio of 6.2 gkg (see table 4.5) than
the radiosonde’s 7.6 gkg . Such differences could be due to radiosonde calibration, as well
as the local conditions at the airport that is mostly open ground, while the University
campus ground is mostly sealed. Another interesting observation, is that the discrepancy
seems to decline with altitude, as would be expected with lidar overlap approaching unity,
and more homogeneous atmospheric composition due to lesser ground effects. Since our
lidar is effectively blinded in the first 150 m by mirror scatter, we were not able to directly
compare the DIAL result with the humidity sensor, furthermore, it was not possible to
make a horizontal measurement by removing the external mirror as our lab window looks
out directly towards another building less than 80 m away.
As discussed in section 4.3.5, some optical interference commenced at some time
during the observation experiment, that could have produced a significant random error,
however, while this could not be discounted, it was not evident in these results.
4.5. EXTENDED OBSERVATION EXPERIMENT 89
DIAL random error analysis
Random errors due to the various noise sources are another possible consideration, but
not significant one in this case, as previously evaluated in (Hamilton et al., 2008). This
indicates an error due to background less than 1% up to an altitude of 1 km.
4.4.5 Conclusion
In these experiments we calibrated a humidity sensor and measured the effective absorp-
tion cross-section of the transmitted on-line laser pulses, followed by an atmospheric
observation experiment to measure atmospheric water content up to 600 m with a 150 m
resolution. On comparison between our humidity sensor and radiosonde data at ground
level, we found that the radiosonde’s humidity sensor gave a significantly higher reading
than ours, which could be due to ground effects, as well as due to the different locations
and radiosonde calibration. At higher altitudes, our DIAL results converge with that of
the radiosonde, and in the 450-600 m range, the discrepancy was less than 12%, which
was close to the discrepancy between the respective humidity sensors at ground level.
Possible random errors due to optical interference among others, made it impossible to
further resolve error sources.
4.5 Extended observation experiment
4.5.1 Aim:
The aim of this experiment was twofold. First, we selected a different wavelength with
a slightly smaller absorption cross-section. Second, to conduct an extended continuous
observation over many hours to test the operation and stability of our system in order
to identify remaining issues and considerations for the further development towards a
compact and robust field-deployable instrument.
90 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
4.5.2 Method
This experiment was conducted on 19-20 April 2010, and the methods used here were
largely the same as described in section 4.4 and 4.3, except that we tuned the on-line
master laser to an adjacent, weaker absorption line at 823.2 nm. A spectrum was acquired
using the absorption cell, prior to the observation. The effective absorption cross-section
was calculated from the water vapour number density, which was measured using the cal-
ibrated humidity sensor and thermometer. As in the previous experiment, this result was
used to perform the DIAL inversion with the acquired on-line and off-line data, after sub-
tracting background and dark count. Our results were also compared against radiosonde
data obtained from the Bureau of Meteorology.
4.5.3 Results
The calibrated humidity sensor measurements are presented in Table 4.10, where the cal-
culated water number density is also shown. figure 4.13 illustrates the measured absorp-
tion spectrum, and table 4.9 shows the measured transmission at the line center. From
these two results, we have the measured effective absorption cross-section in equation
4.10. We compare this result to the calculated absorption cross-section for the HITRAN
data shown in table 4.11.
Hourly DIAL results are provided in table 4.12 and illustrated in figure 4.14. Ra-
diosonde data is provided in table 4.13.
σe f f =−1Nx
ln(T) = 7.5 × 10−23 cm2. (4.10)
This result compares with the HITRAN result using the model described in Chapter 2
and summarized in Table 4.11, giving a cross-section figure of 5.6 ×10−23 cm2.
4.5. EXTENDED OBSERVATION EXPERIMENT 91
Time (approx) Off-line On-line Transmission (T*100) %20:00 0.687 0.642 93.5
Table 4.9: Measured off-line and on-line optical power and transmission
Time (approx) Sensor Vout T °C RH % r gkg N cm−3
23.30 2.34 21.7 40.3 6.47 257×1015
03.30 2.66 19.5 49.2 6.89 276×1015
Table 4.10: Mixing ratio and number density results with calibrated RH sensor
ν0(cm−1) S T (cm) γa(cm−1) γs(cm−1) n σv(cm2)12147.6898 1.56×1023 0.0842 0.328 0.67 5.6×10−23
Table 4.11: Voigt model at 823.202 nm to calculate absorption cross section, σv fromHITRAN parameters ν0, S T , γa, γs and n
0 1000 2000 3000 4000 5000 6000 70000.63
0.64
0.65
0.66
0.67
0.68
0.69
0.7
Wavelength scan (arbitrary unit)
Absorp
tion facto
r
Absorption Spectrum through vapor cell
Acquired data points
SG fit
Figure 4.13: Acquired laser spectrum before observation
0 100 200 300 400 500 600 700 800 900 100010
5
106
107
Range (m)
Photo
n C
ount
Dial Results 20−April−2010 00:00−01:00 hr
Online
Offline
Figure 4.14: DIAL Results in the first hour of extended observation, starting at 00:00 hron 20-April-2010
92 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
Range (m) Online count Offline count Number density cm−3
0-150 81334 203311 n/a150-300 16586 61179 2.1981e+17300-450 2132 12399 2.5715e+17450-600 781 3462 -error
(a) Dial observation results 00:00-01:00 hr
Range (m) Online count Offline count Number density cm−3
0-150 83119 201169 n/a150-300 16404 59934 2.3267e+17300-450 2594 11717 1.1979e+17450-600 509 3484 2.3457e+17
(b) DIAL observation results 01:00-02:00 hr
Range (m) Online count Offline count Number density cm−3
0-150 71615 165705 n/a150-300 13856 50094 2.5212e+17300-450 2108 9853 1.4503e+17450-600 707 3400 1.5887e+16
(c) DIAL observation results 02:00-03:00 hr
Table 4.12: Hourly DIAL results
Range (m) RH % P hPa T Mixing ratio r gkg Number density N cm−3
0-150 76 1010 288 10.6 4.3e+17150-300 62 993 285 9.1 3.6e+17300-450 59 975 285 8.8 3.4e+17450-600 55 958 283 8.2 3.2e+17
Table 4.13: Radiosonde data from Adelaide airport on 19/04/2010 at 12.00 hr UTC
4.6. ASE MEASUREMENT EXPERIMENT 93
4.5.4 Discussion
The absorption cross-section obtained from the transmission and humidity measurements
at this wavelength is about 20% higher than HITRAN Voigt model results shown in table
4.11. The transmission experiment for spectral calibration was not repeated after the
atmospheric observation.
With this absorption line being one-third the strength of that used in experiment in sec-
tion 4.4, any errors would be more significant. This suggests that there is some systematic
error in our system that has not been accounted for.
However, in this experiment, the radiosonde data seems to inconsistent with these
results, indicating a 67 % higher number density compared to the sensor result shown
in table 4.10 for 23:30 hours, which was some four hours after the balloon launch. The
four hour time difference may have resulted in a change in atmospheric conditions, as a
possible explanation for this discrepancy.
4.5.5 Conclusion
This experiment illustrated the consistency of operation of our DIAL instrument, with
constant operation for over 3 hours. A possible systematic error in the absorption cross-
section measurement has been confirmed.
4.6 ASE measurement experiment
4.6.1 Aim:
In section 4.3, we measured the effective absorption cross section using the high power
pulsed output from the laser amplifier. The aim of this experiment was to measure the
relative difference in the absorption cross-section between the master laser, and the laser
amplifier. By measuring the relative absorption using the master laser, as well as the
amplified laser pulses, this will provide some indication of the contribution of ASE of the
laser amplifier.
94 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
(a) Pure ASE with no seed input (b) Fringe contrast after alignment
Figure 4.15: This figure illustrates the optimal TA alignment by maximizing the visiblefringe contrast from an etalon.
4.6.2 Method
This experiment is done in two parts, using the method described in section 4.2 and il-
lustrated in figure 4.7, to measure the effective absorption cross section from the tapered
amplified, and then repeating the same measurements using the on-line master laser. As
before, we use fast pulse acquisition with post-processing to obtain both sets of results.
The post-processing to extract the master laser pulse data is different since we are acquir-
ing negative-going pulses as the AOM switches optical power out of its zeroth order. The
laser amplifier was optimally aligned as described in experiment 4.3.3 by maximizing
Fabry-Perot fringe contrast as illustrated in figure 4.15.
The laser amplifier input was aligned as in other experiments, by maximizing the
fringe contrast observed through an etalon.
4.6.3 Results
The results obtained on July 6 2009 are shown in figure 4.17.
4.6. ASE MEASUREMENT EXPERIMENT 95
0 1000 2000 3000 4000 5000 6000 7000 8000 90000.88
0.9
0.92
0.94
0.96
0.98
1
Wavelength Scan
Pul
se p
ower
rat
io
Master and Transmitted pulse absorption spectra
Amplified pulse power ratioAmplified SG−fitMaster Laser pulse power ratioMaster SG−fit
822.92 nm absorption line
Figure 4.16: Scanned absorption line results from both master laser and optical amplifierMaster laser 0.893Laser amplifier 0.907
Figure 4.17: Relative on-line transmission
4.6.4 Discussion
As indicated in the results above, the peak attenuation of the amplified output was lower
than the master laser by some 13%, which was most likely due to the ASE introduced
by the optical amplifier. As discussed in Section 4.3 and 4.7, the ASE depends on input
alignment, as well as any random behavior of this TA device.
However, this also means that results from experiment 4.3 disagree with HITRAN by
around ∼14%. On the other hand, it’s quite possible that the device was not aligned in the
same way on this occasion, and produced a different ASE spectrum as will be discussed
in the next Section 4.7,.
4.6.5 Conclusion
In this experiment, the ASE of the laser amplifier reduced the effective absorption of water
vapour by 13%.
96 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
4.7 Optical amplifier power, ASE and alignment experi-
ment
4.7.1 Introduction
In section 4.6, we measured the reduction of σe f f with the amplifiers’ optical input opti-
mally aligned, using the same technique as for observation and calibration experiments.
In this experiment however, we deliberately mis-align the TA input and observe the cor-
relation between any relative change in absorption cross-section and output pulse power.
During experimental work we found that the mechanical input alignment of the optical
tapered amplifier drifts during the course of an observation. In this experiment we delib-
erately mis-align the TA, and see what additional effect it has on the effective absorption
cross-section, as well as the effect it has on output power.
4.7.2 Aim
The aim of this experiment was to characterize the amplifier optical gain in relationship
ASE as measured by variations in relative absorption cross-section, and provide a handle
on the output spectral purity as a function of output pulse power, which can be easily
measured.
4.7.3 Method
In this experiment we measure the effective absorption cross section and output pulse en-
ergy due to horizontal and vertical mis-alignment. The effect of input alignment drift was
simulated by deliberately mis-aligning the optical amplifier input in the horizontal and
vertical orientations and measuring the on-line and off-line attenuation vs output power,
as a function of the horizontal and vertical mis-alignment. We acquired multiple sets of
pulse absorption data as the optical wavelength was scanned through the center of the
absorption line, and out to ∼ 10 GHz either side, at various horizontal and vertical input
alignments. From the acquired pulse spectra, curve-fitting was used with each acquired
spectrum to find the relative online to offline absorption ratios by comparing the peak rel-
4.7. OPTICAL AMPLIFIER POWER, ASE AND ALIGNMENT EXPERIMENT 97
ative extinction to that in the wings. For every input alignment, the spectrum acquired in
the previous step, we also acquired some pulse waveforms using an oscilloscope to find
the average relative output pulse energy during each scan.
Figure 4.7 illustrates the experimental setup. A signal generator (Agilent 33120A)
slowly scans the laser temperature controller with a period of 100 mHz, producing an
approximately triangular waveform of the diode’s temperature and resulting optical fre-
quency. This causes the laser wavelength to be scanned repeatedly across the absorption
line, with a total excursion of approximately 20 GHz p-p. At some point in time, the
user triggers the data acquisition system consisting of the high-speed card (GageScope
GS14105) controlled by a real-time data acquisition program (see adgmr.m). After the
system is activated, the actual pulse acquisition is triggered by a transition in the optical
signal, to capture 256, 12-bit samples before and after an optical transition. This acqui-
sition is repeated to capture 7000 consecutive ∼ 1µs pulses, for both channels simultane-
ously, which takes approximately 5 seconds. The same code works for both the positive
pulses produced by the laser amplifier, as well as the negative-going pulses produced from
the zeroth-order output of the AOM.
This data is acquired directly from the photodiode amplifiers. These 2-stage photodi-
ode amplifiers, described in Appendix A.4, were designed for both high speed (100 MHz)
and low noise 3 nV/√
Hz, and work well for this high-speed measurement, as well as for
normal DIAL laser on-line stabilization requiring low-noise signal detection.
Since we are traversing the absorption peak every 5 seconds, a 5 second acquisition
usually captures both wings of the absorption spectrum. However, since the user actuated
trigger is not synchronized to the temperature cycle, the absorption peak is in a random
position relative to the acquisition frame.
The experimental setup illustrated in figure 4.7 was found to be susceptible to electri-
cal noise, probably due to the computer powering the acquisition card. This problem was
minimized by reducing ground loops, and by using the maximum analog gain before the
data acquisition hardware (500 mV inputs), taking care to avoid clipping or saturation of
the digitized data.
98 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
Master laser waveform
For this acquisition, the laser amplifier is disabled by powering down the pulse driver. The
zeroth order from the AOM is coupled into the absorption cell using the single-mode opti-
cal fiber as illustrated in figure 4.7, and the wavelength is scanned as previously described.
The data is processed using a script (calcm.m) that searches for each instance of a nega-
tive pulse in the data and integrates for a set number of subsequent acquisitions, for each
channel. The pulse-by-pulse ratio of the two channels is the result, which is subsequently
smoothed by a Savitskiy-Golay filter. The filtered result provides for a convenient way
to measure the online attenuation relative to the wing attenuation at 10 GHz. This mea-
surement was required in this experiment to calculate the relative absorption cross-section
degradation due to the laser amplifier.
Amplified laser waveform
For this acquisition, the zeroth order AOM output from the master laser is blocked, and
the laser amplifier was energized. The output of the laser amplifier was focused into the
single-mode optic fiber and coupled into the absorption cell as illustrated in figure 4.7.
We found that the output waveform of the laser amplifier was sensitive to the alignment
of the optic fiber coupler. This was most likely due to feedback between the (flat) fiber
face and the output facet of the optical amplifier itself. This alignment sensitivity was
almost completely eliminated by inserting a ND1.6 filter as shown. The alignments of
other optical components could be angled so as to avoid this type of feedback.
The center alignment of the laser amplifier was found by using both an etalon, and by
monitoring the output power using a photodiode. It was found that an alignment that max-
imizes the output pulse power, also maximizes the fringe contrast produced by the etalon.
The optical power output of the laser amplifier therefore served as a convenient proxy
for optical alignment. Starting with the optimal alignment each time, the alignment was
scanned either vertically or horizontally, while the absorption spectra for each alignment
was acquired. For each alignment, we also captured at least three optical pulse power
waveforms using an oscilloscope (Tektronix TDS1002). The average of the area under
each of these pulses, for each of the alignments, was calculated (Matlab script calcpall.m)
4.7. OPTICAL AMPLIFIER POWER, ASE AND ALIGNMENT EXPERIMENT 99
0.05 0.1 0.15 0.2 0.25 0.3 0.35
0.08
0.09
0.1
0.11
0.12
0.13
0.14
0.15
0.16
Average Optical Power per Pulse (W)
Pe
ak O
n−
line
Ab
so
rptio
n
Horizontal misalignment
Vertical misalignment
Figure 4.18: Optical pulse power vs. On-line absorption
as an estimate of the average energy per pulse. As before, the absorption spectrum was
measured from the observed pulse train as the wavelength was scanned across the spec-
tral peak. Matlab scripts (see calc.m and calcm.m in Appendix G) were used to find the
positive-going edge and integrate for each pulse, with Savitskiy-Golay filtering to calcu-
late the ratio of peak vs off-peak absorption. Therefore, for every alignment we had a
measurement of pulse energy and the ratio between peak absorption and absorption in the
spectral wing.
4.7.4 Results
Figure 4.18 summarizes the results from this experiment. It is noteworthy that the maxi-
mum power was less than 350 mW, despite being driven at full current. The photodiode
was calibrated against a Thorlabs DT120 power meter. If this is correct, we are seeing
some degradation in the TA over time, as well as an example of its unpredictability.
4.7.5 Discussion
The method for estimating the output pulse power was prone to some random error as we
sampled the pulses at random intervals during each scan, as the power decreased by about
100 CHAPTER 4. CHARACTERIZATION, CALIBRATION AND APPLICATION
20% as the temperature increased from minimum to maximum. Since we didn’t ensure a
representative sample of output power over the 5 s measurement interval, we inevitably
have some random uncertainty in the pulse power measurement. This sweep explains
most of the scatter in the result. It was possible to use the fast photodiode channel for this
measurement to reduce this random error, however, the Fabry-Perot interference in the
splitter and other fiber components, would have also resulted in some errors. Since this is
not a calibration experiment, but merely a means to set some bounds on spectral purity, a
higher level of precision was not required.
A much more significant factor for consideration is that we only changed the posi-
tion of the waist of the beam at the optical amplifier, not its angle of incidence on the
input port, nor the position of the waist along its optic axis, nor its polarization axis. In
reality, therefore, this experiment only considered one of a multiple possible kinds of
mis-alignment.
4.7.6 Conclusion
The spectral purity of the amplified laser output depends on the optical alignment into the
TA. The results show a rapid decline in spectral purity as well as a decrease of optical
power decreases by more than 25% from maximum, for either horizontal or vertical mis-
alignment of the beam waist relative to the TA gain region . The effect of incidence angle
on spectral purity and power were not measured.
4.8 Summary
In this chapter we calibrated a humidity sensor from which the water number density
was calculated and used to measure the effective absorption cross-section due to the spec-
tral purity of the transmitted laser pulses. This measurement was then used to obtain
quantitative DIAL measurements on two different spectral lines. Random variability and
instability, as well as free-space optical alignment drift were some of the issues found
with this prototype. However, the master lasers, the wavelength control and associated
systems worked well, paving the way for future DIAL development.
Chapter 5
Conclusion
During the course of this project, a low-cost Differential Absorption Lidar system was
designed, constructed, characterized and used for atmospheric observations. The design
included a number of elements that were developed specifically for the low-cost and ro-
bust implementation. These included optical fiber interconnects, ratiometric detection
for on-line stabilization, synchronous dither and timing and a novel off-line stabilization
technique. Although the construction of this instrument left much to be desired with
mechanical, alignment and other unresolved issues, the observation results showed the
expected differential signal, cloud return in the off-line channel and the expected decrease
in humidity with altitude. As a step towards an operational observatory, this work also
developed a novel calibration technique based on saturated salt solutions from which the
effective absorption cross-section was calculated, with the result in agreement with the
current HITRAN data at 822.92 nm..
The concepts and techniques that were developed as a part of this project are now
ready for further refinement and a higher quality level of implementation. The develop-
ment of a fully autonomous, portable, stand-alone DIAL instrument based on our pro-
totype, would also benefit from further development of the reference cell, as well as the
use of a different type of master laser diode with more predictable tuning characteristics.
In particular, the system can benefit greatly by a re-design of the way the laser-amplifier
input is aligned into the rest of the optical system, since changes in this alignment made it
difficult to calibrate for ASE, while continuous re-alignment of this device was required
101
102 CHAPTER 5. CONCLUSION
during the course of an observation. The need for a higher optical power from a precisely
controlled master laser remains an unresolved issue. The tapered optical amplifiers are
unfortunately limited to about ∼1 W, and the possibility of using an optically pumped
gain medium or an OPO device therefore warrants further investigation.
Appendix A
Electronic Systems
A.1 Introduction
This chapter describes the electronic design of the various sub-systems that comprise this
DIAL instrument.
The objective of these designs was to produce the simplest fully functional circuits
with the minimal total part count, as well as utilizing components that were most readily
available at the time. The electronics was not a limitation on system performance which
suffered mostly from optical phenomena such as multi-path interference and mechanical
alignment, as described in section 4.9.
This chapter describes the electronic component level, with the schematics and the
operation of the systems outlined in the green squares in figure A.1. with a block diagram
in figure A.3 illustrating how the electronics are interconnected. In the following pages,
the design and function of these blocks will be described in the context of the system as a
whole.
103
104 APPENDIX A. ELECTRONIC SYSTEMS
Σ
LPF
I
T
I
T
− LPF
On−Line
Laser2
Bandpass
16 GHz
Off−Line
∫
∫
Fiber
ISO
ISO
Laser1
Optical Pulse Output
System
Off−Line Wavelength Measurement
On−Line Wavelength Measurement
RF DriveTiming
splitter
RF Det.Dither
GaAs PIN
SOA
AO
AO
Multipass cell
A4
A5A2
A6
A3
A2
A3
A7
To DAQ
Figure A.1: DIAL system diagram illustrating the functional blocks in green.
Figure A.2: Main electronics panel with timing A2, analog control A3, and the beatfrequency electronics A7. Figures 3.5, 3.6 and 3.13 show the other electronics.
A.1. INTRODUCTION 105
DIA
L d
ua
l la
se
r
wa
ve
len
gth
co
ntr
ol syste
m
AO
M2
AO
M1
Dith
er
tim
ing La
ser
Am
plif
ier
DA
Q1
DA
Q2
DIA
L t
imin
g s
ou
rce
Nu
me
rato
r
De
no
min
ato
rH
igh
Sp
ee
d A
mp
lifie
r
An
alo
g
Div
ide
r
AO
M2
AO
M1
Lic
el C
om
pu
ter
Da
ta A
qu
isitio
n S
yste
m
On
−lin
e la
se
r syste
m
Off
−lin
e la
se
r syste
m
AO
M−
1
AO
M−
2
Wa
ter
Va
po
r C
ell
16
GH
z B
ea
t D
ete
cto
r
AO
M R
F D
rive
r
La
se
r A
mp
lifie
r a
nd
Pu
lse
Drive
r
On−
line C
on
tro
l
Off
−L
ine C
on
tro
l
Vap
or
cell
RF
dete
cto
rH
igh
Sp
ee
d A
mp
lifie
r
Tra
nsm
itte
r
Co
mp
ute
r
Op
tica
l
Ele
ctr
ica
l
Figure A.3: System overview
106 APPENDIX A. ELECTRONIC SYSTEMS
Digital and timing system Section A.2
Analog wavelength control system Section A.3
Spectroscopic ratiometric detection systems Section A.4
Pulse driver for Laser amplifier Section A.5
AOM RF driver Section A.6
RF beat detector Section A.7
A.2 Digital and timing system
The digital system provides for the synchronous operation of the dither, optical pulse
timing, phase synchronous measurement and external data capture acquisition timing as
described in section 3.6. In this way, the timing system is the coordination center of the
DIAL instrument as a whole. The schematic of the timing system is illustrated in figure
A.4.
The digital system timing performs the following functions.
AOM timing Section A.2.1
Laser amplifier timing Section A.2.2
Data AQuisition (DAQ) system timing Section A.2.3
Dither signal timing Section A.2.4
A.2.1 AOM timing
The system clock is obtained from a quartz crystal module, U22, and divided down by a
ripple counter U21, a 74HC4060 which provides two outputs at 3 kHz and 1.5 kHz respec-
tively, to the delay logic that follows, as shown in figure A.4. The timing and pulse widths
for all the subsequent systems are generated using a number of HC4538 monostable mul-
tivibrators. The first two multivibrator strings, U16 and U17 provide the AOM timing
where the first monostable provides the adjustable delay, while the second provides the
adjustable pulse-width for each device. By utilizing the complementary triggering inputs,
the timing phase of the two multivibrators, U16 and U17, is at 180° relative to the output
at pin-3 of U21. This provides the required timing for the on-line and off-line lasers and
A.2. DIGITAL AND TIMING SYSTEM 107
1 J 3Q
4 K 2Q
13 CLR
12 CLK
74107
74HC107
U20
8 J 5Q
11 K 6Q
10 CLR
9 CLK
74107
74HC107
U20
+5V
2 RC
4 I1
5 I0
3 CLR
6Q
7Q
4538
1 C74HC4538
U15
14 RC
12 I1
11 I0
13 CLR
10Q
9Q
4538
15 C
U15
+5V
9 C
11 RC
10 R
12 Reset
15Q9
13Q8
14Q7
6Q6
4Q5
5Q4
7Q3
3Q13
2Q12
1Q11
4060 U21
74HC4060
+5V
3OUT
4Vcc
OSC 25.175MHz
U22
1
2
3
S1 150
10k
1nF+5V
4701nF
+5V10k
+5V
+5V
5
6
8
3
4
1
2 7
TSC428
NC NC
GND VDD
U23
TSC428
14 RC
12 I1
11 I0
13 CLR
10Q
9Q
4538
15 C
U16
2 RC
4 I1
5 I0
3 CLR
6Q
7Q
4538
1 C74HC4538
U16
470p
3k9
150p
+5V2k
+5V
11k
+5V1k
+5V+5V
5
6
8
3
4
1
2 7
TSC428
NC NC
GND VDD
U24
TSC428
14 RC
12 I1
11 I0
13 CLR
10Q
9Q
4538
15 C
U17
2 RC
4 I1
5 I0
3 CLR
6Q
7Q
4538
1 C74HC4538
U17
470p
3k9
150p
+5V2k
11k
+5V1k
+5V+5V
51
512
1
2
1
51
512
1
2
1
14 RC
12 I1
11 I0
13 CLR
10Q
9Q
4538
15 C
U18
2 RC
4 I1
5 I0
3 CLR
6Q
7Q
4538
1 C74HC4538
U18
470p
3k9
150p
+5V2k
12k
+5V2k
+5V+5V
475k 5
6
8
3
4
1
2 7
TSC428
NC NC
GND VDD
U25
TSC428
51
2
1
14 RC
12 I1
11 I0
13 CLR
10Q
9Q
4538
15 C
U19
2 RC
4 I1
5 I0
3 CLR
6Q
7Q
4538
1 C74HC4538
U19
150p
3k9
150p
+5V2k
5k6
+5V
+5V+5V
+5V
5
6
8
3
4
1
2 7
TSC428
NC NC
GND VDD
U26
TSC428 51
512
1
2
1
150150
AOM2
AOM1
Dither timing
Test signal
BothOfflineOnline
SOA driver
DAQ2
DAQ1
DIAL timing system
Figure A.4: DIAL timing schematic
108 APPENDIX A. ELECTRONIC SYSTEMS
allows advance timing adjustment for acoustic delays in the AOM crystal, as well as the
pulse overlap with the laser amplifier.
A.2.2 Laser amplifier timing
The laser amplifier is energized with a delay and pulse-width set at multivibrator U18. In
order to facilitate alignment and testing, this timing is switchable at S1 so that it can be
energized only for on-line, off-line, or for both pulses, using a 74HC107 JK flip-flop and
logic.
A.2.3 Data AQuisition (DAQ) system timing
The data acquisition system is triggered ∼4 µs in advance of everything else, so that a
background signal level can be acquired prior to the transmitter firing, for use in sub-
sequent DIAL inversion calculations. This is done by a delay of all the other systems,
while providing a non-delayed pulse to the DAQ. The two halves of U19 are used to form
suitable pulses on complimentary transitions of the master timing signal.
A.2.4 Dither clock
The dither signal is obtained directly the master clock signal. This facilitates the removal
of the optical switching transient from the wavelength controller by synchronizing zero-
crossing of the dither signal with the switching transient at an analog multiplier, that forms
a lock-in amplifier, as described in section 3.6. In order to perfectly synchronize dither
zero crossing with the pulse timing, a variable delay was added between the master clock
and the clock output driver U23 formed by U15 and U20.
A.3 Analog wavelength control system
The analog control circuit implements the system described in Chapter B and is illus-
trated in figures A.1 and A.6 below. The system consists of a dither generator, an on-line
controller and a off-line controller as follows.
A.3. ANALOG WAVELENGTH CONTROL SYSTEM 109
(a) (b)
Figure A.5: Synchronous transient suppression. Unfiltered mixer output at TP3 fromfigure A.6 shown on bottom trace, illustrating the optical switching transients A.5a andtheir suppression with correct dither bias and timing A.5b. This prevents the opticalswitching from significantly perturbing the control system.
A.3.1 Dither sinusoid generator
The sinusoidal dither signal is obtained synchronously from the timing clock using analog
double-integration. When power is switched out of the CW beam by the AOM to form the
output pulse, a transient appears on the output of the ratiometric detector as illustrated in
figure A.5a, which although short in duration, perturbs the feedback control loop resulting
in a systematic error at the output of the integrator. However, if this transient is timed
to coincide with the zero crossing of the dither signal as illustrated in figure A.5b, the
transient will be attenuated because to a first approximation it will have zero amplitude
at the analog multiplier. In other words, the impulse due to optical switching will be
attenuated by the lock-in, as it coincides with an instantaneous zero reference input. This
is implemented by utilizing the fact that the double integrator places the sinusoidal dither
180° with respect to the clock signal, with the zero-crossings coinciding with the timing
transitions of the original square wave.
The digital timing signal (square wave 0-12 V) input at J1 in figure A.6 is obtained
from a gate driver chip, that provides a ∼12V square wave with a 50 Ω impedance through
a coaxial cable terminated at the driver end to absorb reflections from R5. The voltage at
J1 is converted to a current through Z3, resulting in a regulated 0-5 V square wave at
Z3 and R6. The Zener regulation provides power supply immunity for the dither signal
amplitude, while R6 adjusts the amplitude at the outputs of U3 and U4. To maximize
110 APPENDIX A. ELECTRONIC SYSTEMS
system signal to noise ratio, R6 was set so that the outputs at pins 1 and 7 of U3 and U4
are large, but not clipping. If we ignore C2, R8 and R10, the circuit including U3 and U4
together form a second-order integrator. Each integrator produces a phase shift of near
90, so the total phase shift of 180 places the zero-crossings of the resulting sinusoid
close to the square wave transitions at J1.
A.3.2 On-line controller
The on-line controller is connected to the ratiometric circuit, which provides a high-level
signal at J2 proportional to the ratio of two photodiode currents that is a measure of
absorption due to the vapour cell alone. Diodes Z1 and Z2 together with R1 provide input
protection for U1, while C1 removes the DC offset of the analog divider. R1 together
with R3 and R4 set the adjustable gain of this stage. This gain was set at a high level
without clipping at pin-6 of U1, so as to maximize the system signal-to-noise ratio, which
is limited by the performance of the analog multipliers. The on-line controller consists of
a lock-in amplifier made with U11 (AD633 analog multiplier), and R20 and C14 forming
the low-pass output filter. U9 and associated circuitry provides the unfiltered signal for
diagnostics, without perturbing the control loop. To complete the control system, U5
and associated components provides the integrator to form a stable controller. This can be
turned off by closing S 1 which is required for alignment, diagnostics and testing. The time
constants were selected for system stability, rather than for fast response. If it was required
to optimize performance, the low-pass filter after U11 would be replaced by a higher-
order filter, while this integrator would be supplemented by proportional and derivative
amplifiers to form a Proportionl Integral Differential (PID) system.
A.3.3 Off-line controller
The input of the controller, J3, is connected directly to a low-noise and low impedance mi-
crowave Tunnel detector diode, Herotek DT-2018, which produces a signal proportional
to the power transmitted through the 16 GHz bandpass filter. The AD797 was selected for
U2 as the first-stage preamplifier with a gain-bandwidth product adequate to implement
A.3. ANALOG WAVELENGTH CONTROL SYSTEM 111
the required signal conditioning in one stage, as well as a voltage and current noise den-
sity of 1 nV/√
Hz, and 2 pA/√
Hz respectively. This means that a source impedance of
less than ∼500 Ω is required, making it a very good match for the Tunnel diode detector
with an output impedance of 125 Ω (Herotek, 2008). D1 and D2 provide protection for
both the detector diode, as well as for U2, while R11 provides 50 Ω cable termination as
well as setting the gain with R12.
This system only uses the AC component due to the amplitude modulation produced
by the bandpass filter with a dithered beat frequency for synchronous detection. This
makes it necessary to remove the DC offset with C11. The value of this capacitor was
selected for a minimal phase delay of 1.2 at the dither frequency of 1.5 kHz and a source
resistance of 125+51 Ω.
The offline controller consists of a lock-in amplifier consisting of analog multiplier,
U10 and a low-pass filter, that function in the same way as U11 described in section A.3.2.
The output filter for the lock-in comprises of R22 and C8 that are the same value and func-
tion as R20 and C14. The integrator circuit with U7, R23, C12, perform the same function
for the off-line control system as U5, etc, described above. Similarly, S 2 and R31 provide
for disabling the control loop while keeping the rest of the system operational, which is
useful for testing purposes.
The schematic shows dither injection for both the off-line and on-line systems, how-
ever, the off-line dither was turned off at the potentiometer R15. This has the advantage of
reducing the required number of expensive optical fiber splitters to one. Only one on-line
laser needs to carry an optical dither to produce a phase synchronous signal at the vapour
cell. This means that the two lasers can be combined at one splitter, and the resulting
combined optical signal, available at both outputs of the splitter, carry both the 500 MHz
wavelength modulation that is stabilized to the vapour cell, as well as the 16 GHz beat
note with the same 500 MHz FM dither modulation depth, that is stabilized to a bandpass
filter, as illustrated in the system figure A.1.
112 APPENDIX A. ELECTRONIC SYSTEMS
OP
37
23
4
6
7U
1 AD
79
72 3
4
6
7
U2
Z1
16
vZ
2
16
v
C1
2u
F
R1
1k1
R2
1k1
2
1J2
R3
5k6
+1
2V
R4
50
k
5 6
7
48
TL
07
2
U3
3 2
1T
L0
72U
4R
51k
2
1T
imin
gJ1
Z3
5v1
R6
5k
C2 33
uF
R7
1k5
C3
15
0n
F
R8
22
0k
R9
39
0
C4
15
0n
F
R1
02
20
k
2
1J3
C1
1
33
uF
R1
15
1
R1
25
0k
R1
35
1
R1
4
10
k
R1
5
50
kC
6
82
0p
F
R1
69
10
k
R1
710
k
R1
8
50
kC
782
0p
F
R1
99
10
k
D1
1N
91
4D
2
R2
0
10
0k
R2
1
1M
R2
2
10
0k
C8
1u
FR
23
1M
C9
1u
F21
S1
R2
415
0
C1
21
uF
3 2
1
4 11
TL
07
4
U5
R2
5
10
0k
10
9
7T
L0
74U
6
5 6
14
TL
07
4U7
R2
61
0k
C1
0
33
0p
F
R2
7
51
2
1J4
R2
81
0k
C1
33
30
pF
R2
9 51
2
1J5
10
0k
R3
0
12
13
8T
L0
74U8
−1
2V
+1
2V
+1
2V
+1
2V+1
2V
27
0n
FC
5
C1
4
1u
F
R3
11
50
21
S2
R3
2
475k
R3
3
475k
3 2
1T
L0
72
U9
5 6
7
48
TL
07
2
10
0u
F
10
0u
F
−1
2V
10
0u
F
10
0u
F−
12
V
10
0u
F
+1
2V 10
0u
F−
12
V
10
0u
F
10
0u
F−
12
V
10
0u
F
10
0u
F
−1
2V
10
0u
F
+1
2V
−1
2V
−1
2V
+1
2V
Z4
16
vZ
5
16
v
467
8
1
510
−V
S
+V
S
W
Z
X Y
+
U1
0
2 3
AD
63
3
467
8
1
510
−V
S
+V
S
W
Z
X Y
+
U1
1
2 3
AD
63
3
D4
D3
TP
3T
P4
DIA
L laser
wavele
ngth
contr
ol syste
m
Onlin
e
Contr
ol
Offlin
e
Contr
ol
Vapor−
Cell
RF
−
Dete
cto
r
Figure A.6: DIAL analog control system schematic
A.4. SPECTROSCOPIC RATIOMETRIC DETECTION SYSTEMS 113
A.4 Spectroscopic ratiometric detection systems
This system detects the absorption line for the on-line control system. This consists of
three components
• Absorption cell with Pellicle beamsplitter
• Pair of high-speed photodiode amplifiers
• Analog divider
The optical detection system includes high-speed photodiode preamplifiers around the
vapour cell, as well as the analog divider, as illustrated in figures A.1 and A.7. The reason
for the high speed design was two fold. Firstly, we needed fast recovery from optical
switching, where transition times are about 10 ns. Secondly, these amplifiers were also
required for pulse waveform acquisition at 100 MS/s for spectral calibration experiments
described in Chapter 4.
A.4.1 Photodiode amplifiers
The photodiode amplifier schematic is presented as a part of figure A.7 The SHF203 is an
extremely low-cost Siemens PIN photodiode with low capacitance and a detection area of
1 mm2 in a clear package that eases alignment. The capacitance of the device was reduced
to 3 pf by operating it at a relatively high reverse voltage of 12 V, and adding a protection
resistor, R41 and R46, to limit maximum input current below the 30 mA limit for the
OPA657, in case of excessive laser power.
A.4.2 Analog divider
The analog divider was built using available components, and is probably the most prob-
lematic electronic component in the prototype due to oscillations under certain conditions.
This is essentially a variable gain amplifier block, and under certain conditions of high-
gain, when the denominator signal is particularly low, results in instability. The TL072
was not the correct choice for U13. The circuit comprising U13 and U14 should be re-
placed by an AD734, or similar, to perform the analog division function.
114 APPENDIX A. ELECTRONIC SYSTEMS
23
4
6
7
OP
A6
57 U
923
4
6
LM
63
65 U1
07
+5
V
−5
V
+1
2V
−1
2V
SF
H2
03
PD
1
27
0
R4
3
10
kR
42
+1
2V
27
0
R4
1
7k5
R4
5
27
0
R4
4
−1
2V
+1
2V
+5
V
−5
V
23
4
6
7
OP
A6
57 U
11
23
4
6
LM
63
65 U1
27
+5
V
−5
V
+1
2V
−1
2V
SF
H2
03
PD
1
27
0
R4
8
10
kR
47
+1
2V
27
0
R4
6
7k5
R5
0
27
0
R4
9
TP
5
TP
6
467
8
1
510
−V
S
+V
S
W
Z
X Y
+
U1
4
2 3
AD
63
3
3 2
1
48
TL
07
2
U1
3
5 6
7
48
TL
07
2U
13
+1
2V
−1
2V
+1
2V
−1
2V
10
k
R5
2
10
k
R5
1
51
R5
4
27
0
R5
3
+1
2V
+1
2V
−1
2V
−1
2V
TP
7
Ou
tpu
t
Ou
tpu
t
Nu
me
rato
r
De
no
min
ato
r
Hig
h S
pe
ed
Am
plif
iers
An
alo
g d
ivid
er
5v6
5v6
16
v
16
v
+1
2V
−1
2V
Ou
tpu
t
Figure A.7: Ratiometric detection system consisting of high-speed transimpedance am-plifiers with analog divider
A.5. POWER ELECTRONIC PULSE DRIVER FOR LASER AMPLIFIER 115
A.5 Power electronic pulse driver for laser amplifier
Introduction
The tapered semiconductor optical amplifier (TA) is an injection current pumped laser
with input and output facets that are anti-reflection coated to achieve an optical gain in
excess of 30 dB over a bandwidth of several nanometers, currently available in the near
infrared and visible parts of the spectrum. Conventionally, these devices are specified
for CW operation with a constant current driver, however without a seeding input, they
produce a broad incoherent optical spectrum due to ASE. Many application, including
DIAL require pulsed operation without ASE, and requires injection current switching syn-
chronized with the seeding laser pulse. Despite these requirements, there was no known
suitable controller commercially available. In our application, a tapered laser amplifier
achieved rapid switching with a ∼3 A drive current and a symmetrical turn-on and turn-
off time of 5 nanoseconds. This was adequate to produce sharp 1 µS pulses with a duty
cycle of less than 1:100, and was ideal for this application. In this section we present a
laser pulse driver design with an innovative application for a MOSFET gate driver chip,
that also facilitates optical input alignment of the device. This circuit has been in routine
use as part of our lidar for over 5 years with many hundreds of operating hours, with no
detectable performance degradation.
Component selection
Gate drivers are commonly used in many types of power electronic devices to rapidly
charge and discharge the gate capacitance inherent in high power MOSFET transistors. In
these applications, the high transient currents maximize energy conversion efficiency by
reducing switching time, thereby minimizing switching losses. These low-cost ubiquitous
devices are therefore designed to rapidly switch high transient currents with a small duty
cycle, which is a very similar requirement to this application. We used the EL7158 due
to free sample availability courtesy of Intersil, and there were several other suitable chips
available from other manufacturers including Maxim and Texas Instruments.
116 APPENDIX A. ELECTRONIC SYSTEMS
Design
With a switching time of less than 10 ns, a stand-alone driver would require some type
of transmission line interface to the laser. However, the extreme sensitivity of this device
to reverse transients, made this approach problematic. In order to avoid these difficulties,
the driver was placed in very close proximity to the laser so as to minimize any parasitic
inductance. This could only be achieved by making the circuit as small and compact as
possible, as shown in figure A.8. For this reason, the circuit was built without the use of a
circuit board, and assembled simply by soldering most of the components to and around
the surface-mount IC.
The datasheet for EL7158 doesn’t specify a maximum pulse energy or maximum drive
capacitance, however, it should be safe to assume that the thermal time constant of the chip
would be much greater than the 1 µs pulse duration used in this application. Therefore, it
was reasonable to use average power dissipation calculations for safe operation.
PD = VsIs + CintV2s f +
I2dRi fτ
(A.1)
where Vs is the supply voltage, 11 v, Is is the quiescent current, 3 mA, Cint is the internal
capacitance, 100 pF, Id is the maximum drive current of 3A, Ri is the maximum internal
resistance of 1 Ω, f is pulse frequency, 3 kHz and τ is the pulse width of 1 µs. Evaluating
equation A.1 gives a power dissipation of just 60 mW. With a thermal resistance of around
200 K/W, results in a temperature rise of just 12 °.
Circuit operation
The schematic in figure A.10 illustrates the electronic design. The current through the
laser is determined by the internal resistance of the driver chip, 0.5 Ω, the ballast resistor
R3, 2.7 Ω, the power supply voltage to the chip which is limited to 11 V by a zener
diode D6, and the voltage across the laser device itself. The pulse current was adjusted
by varying the power supply voltage using a dedicated power supply. In order to prevent
inadvertent operation at very low duty cycle that could destroy the driver chip, a 56 nF
capacitor C1 was placed before the 51 Ω termination resistor R1, which limited the output
A.5. POWER ELECTRONIC PULSE DRIVER FOR LASER AMPLIFIER 117
pulse duration to several microseconds. Resistor R2 limited the input current to the IC
due to the charge stored in C1 and the forward voltage drop of the Schottky diodes D1
and D2. The combination of components consisting of C1, R1, D1 and D2 also serves as a
level shifter required by the positive-ground nature of the TA, and the more conventional
negative-ground of the other electronics used in our lidar. Furthermore, a 100 mA power
supply fuse limits the total power dissipation, and provides a degree of protection from
damage due to a very high duty cycle with an excessively high repetition rate. The other
components shown in figure A.10 include Zener diodes, Schottky diodes and resistors to
protect the laser device and the driver chip from inadvertent damage due to static and
other transients at the input and at the two test points.
Furthermore, the non-linear voltage-dependent capacitance of the Schottky diodes D3
and D4 at the output of the EL7158 served to prevent ringing and overshoot, particularly
for the positive current transition that takes the output to the negative power supply. The
relatively large diode capacitance of D4 when the voltage across the diode approaches
zero, is responsible for the slowing rise time seen in Figure A.11b.
A total of three different types of capacitors were used to decouple the power supply
and store the energy to provide a near-constant current during each pulse. Using different
types of capacitors reduces the possibility of high frequency resonance by providing a
more distributed characteristic impedance. Ten low-ESR Tantalum capacitors totaling ∼
750 µ F with Equivalent Series Resistance (ESR) ' 50 mΩ, as well as two 1 µF ceramic
capacitors, were placed near the IC, with the positive terminals well grounded to the TA.
Alignment and operation
The addition of test points T P1 and T P2 at both ends of the ballast resistor R3, enables
measurement of current waveforms as shown in figure A.11, as well as the voltage across
the laser device. With the power supply turned off, test point T P2 was used to observe
the photovoltaic effect in the active region inside the TA with the optical seed pulses
illustrated in figure A.9. This served as a very useful coarse-alignment aid for the optical
waist coupling to the input port of the TA.
118 APPENDIX A. ELECTRONIC SYSTEMS
Figure A.8: Tapered laser secured on Lambda mounts, with pulse driver circuit attachedand oscilloscope leads connected for coarse alignment using the photovoltaic signal.
Figure A.9: With the TA power turned off, optical pulses produce photovoltaic signals atT P2 when aligned with the active region of the device. In this case, one of the two sourcesexhibit a better alignment, producing the stronger signal.
A.5. POWER ELECTRONIC PULSE DRIVER FOR LASER AMPLIFIER 119
Figure A.10: TA driver schematic
120 APPENDIX A. ELECTRONIC SYSTEMS
(a) 1 µs current pulse waveform (b) Rise time (c) Fall time
Figure A.11: Pulse driver performance results showing TA current waveforms. The 1 µs2.5 A pulse has a 5 ns rise (b) and 5 ns fall (c) time to 80% of final value.
Driver performance
The driver provided a well controlled, high speed current pulse for the TA timed to co-
incide with the input optical pulse. Figure A.11 illustrates the voltage waveform across
the 2.7 Ω ballast resistor R3 to measure the supplied current. The ballast resistor was
built with a series-parallel combination of surface-mount metal-film resistors to ensure
low inductance with good linearity.
A.6 AOM RF driver
The purpose of this driver was to provide a switchable 80 MHz RF signal at a ∼1 W,
to energize the AOMs to effect the optical switching. This block was assembled mostly
from commercially available components made by Mini-Circuits, and was interfaced to
the rest of the DIAL control system using customized electronics, as illustrated in figure
A.12. The variable voltage reference consisting of a LM369 and a potentiometer, pro-
vides a constant voltage close to 10 V, as required by the Mini-Circuits ZOS-100 Voltage
Controlled Oscillator (VCO) to produce an 80 MHz signal required by the AOMs. The
stability of this frequency is critically important to the optical alignment since the first-
order diffracted beam from the AOM is a function of the acoustic frequency generated
by this VCO, and is aligned with the input to the optical amplifier which is a very small
optical waveguide. It is quite possible that reference voltage drift, as well as drift due
to the VCO was one of the causes of alignment problems that we encountered with our
A.7. RF BEAT DETECTOR 121
prototype system. The RF output from the VCO is then switched by Mini-Circuits ZSWA-
4-30 four-way GaAs absorptive switch. The Mini-Circuits ZHL-1A were used as the RF
power output stage for each of the AOMs. The output power can be calculated as the sum
of VCO output, switch attenuation and amplifier gain. This resulted in approximately 25
dBm of RF output, which was close to the maximum allowed by the Piezoelectric trans-
ducers on the AOMs. Due to less than perfect impedance matching, however, the actual
RF power output was lower than calculated.
A.7 RF Beat Detector
The function of this block was to convert an optical modulation near 16 GHz to an electri-
cal signal, amplify it, and measure the RF power transmitted through a 16 GHz bandpass
filter. The output here being a DC or low-frequency modulated signal that is a measure of
the transmitted RF power. Due to the high RF frequencies employed here, this block used
custom components and special RF patch cables. It was not possible to obtain detailed
data for many of the components. Figure A.13 illustrates the arrangement.
122 APPENDIX A. ELECTRONIC SYSTEMS
1
2VC
O
ZO
S−
10
0
5 6
21 43
−5 +5
control
1
2 3
4
5 ZS
WA
−4
−3
0
1
2ZH
A−
1A
1
2ZH
A−
1A
LM
36
9+
15
V
+1
5 V
+5
V
−5
V
+2
4 V
1k
10
k
76
40
49
32
910
54
1
8+5
V
51
51
AO
M1
AO
M2
AO
M1
AO
M2
AO
M R
F D
rive
r
80
MH
z
+2
4 V
Inp
ut
Ou
tpu
tU
H6
UH
7
UH
9
UH
8
UH6 VCO Mini Circuits ZOS-100UH7 RF switch Mini Circuits ZSWA-4-30
UH8, UH9 RF Amplifier Mini Circuits ZHA-1A
Figure A.12: AOM driver schematic
A.7. RF BEAT DETECTOR 123
IN OUT
7812
GND
1
2
3
UH1UH3
UH2
+12 V
UH5UH4
+15 V
−15 V
+15 V +12 VOptical Beat Detector
To analog
control system
Optical
Fiber
UH1 GaAs PIN Photodiode New Focus 1481-SUH2 RF amplifier ADI Advanced Systems QCJ-12182530UH3 RF splitter MCLI PS3-14UH4 16 GHz Bandpass Filter REACTEL inc. 4C1-16G-500-S11UH5 Detector diode Herotek DT1218
Figure A.13: 16 GHz beat frequency detector schematic
124 APPENDIX A. ELECTRONIC SYSTEMS
Appendix B
Control System Model
B.1 Introduction
Figure 3.9, and sections 3.7 and A.3 provided a general description of the operation of
the wavelength control system. In this section, a more rigorous control system descrip-
tion with measured system parameters from experimental results, is input to Matlab, with
results compared with actual system performance. This simulation model provides an
independent verification of operation in that both wavelengths can be independently con-
trolled in this dual coupled system. Furthermore, a modeling approach helps to understand
a system, and can be used to easily observe its characteristics as various parts are altered.
This can facilitate the development of a new design where system variables can be opti-
mized for desired characteristics. For example, wavelength stability can be improved by
increasing the speed of the control loop with a PID controller, or a multiple-wavelength
control system such as that described in section 3.16 can be modeled and optimized before
it is actually constructed.
B.2 Modeling the lock-in amplifier
The modeling of most components was straightforwards. The model for the lock-in ampli-
fiers uses a time-invariant approximation. This works by taking two points corresponding
to the peak-to-peak dither modulation around the unperturbed laser wavelength, with the
125
126 APPENDIX B. CONTROL SYSTEM MODEL
sum of the spectral profile function calculated at these two points. This sum is propor-
tional to the averaged output of the lock-in amplifier and the derivative of the spectral
profile which is also the control system feedback error signal.
B.3 Simulation experiment
The off-line model is a close analogue of the on-line system and we assume that the elec-
tronic bandpass filter, like the vapour cell is a damped resonator modeled with a Cauchy
distribution. Any optical frequency perturbation in the on-line laser are transferred di-
rectly to the off-line control system as two-laser beat frequency. We know that the average
optical power on the high-speed detector is about 3 mW, hence we can estimate the RF
power at the beat frequency as produced by the detector. There are, however, uncertainties
as to the RF gain of the preamplifier (which is only specified in terms of minimum gain of
30 dB), as well as other losses in the splitter and cables. We can, however, make a reason-
able estimate of the efficiency of the Tunnel diode based on specifications. In the end, we
selected a gain of 3000 for the RF preamplifier, which matched the device specification,
and produced the observed response. Figure B.2 presents the control system model for
the DIAL system as a whole.
For this modeling experiment, we applied a step perturbation to the online laser diode
controller equivalent to an optical frequency change of 50 MHz, and observed the re-
sponse of the system at 3 points as illustrated in figure B.2. This is equivalent to an
experiment that was performed for the publication (Dinovitser et al., 2010). The modeled
and experimental results are presented in figure B.1, illustrating good agreement.
B.3. SIMULATION EXPERIMENT 127
(a) Matlab model results showing online (top) and offline (center) voltage step responses. The thirdcurve is the offline wavelength transient
0 0.5 1 1.5
−40
−30
−20
−10
0
Time (s)
On
−L
ine
Lo
ck−
in O
up
ut
(mV
)
(b) Observed On-line step response
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8−0.03
−0.02
−0.01
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
Time (s)
Off
−L
ine
Lo
ck−
in A
mp
lifie
r O
utp
ut
(v)
(c) Observed Off-line step response
Figure B.1: Comparison of control system model and experimental results.
128 APPENDIX B. CONTROL SYSTEM MODEL
0.0
5A
/V1
TH
z/A
1/H
z
10V
/W
VA
Hz
WV
V
3000
0.3
0.0
02
110
0
0.0
2A
/V1
TH
z/A
cm
^−1
Gd=
0.0
176
Gl1
=0
.094
9
L=
300
0
N=
2.6
87e
19
P1=
1
S=
4.4
7e
−23
1e−
4*4
0 m
W1/H
z
0.6
A/W
1e
4 V
/A
V
Acm
^−1
WA
V
transim
pe
da
nce
−K
−
pre
am
p1
−K
−
pre
am
p
−K
−
multip
lier
−K
−
inpu
t*m
ultip
lier
−K
−
d(V
oig
t)/d
f
MA
TLA
B
Fu
nctio
n
Vapor
Ce
ll
−K
−
Unit c
onvers
ion
−K
−S
tep
1
Scope1
RF
split
ter
−K
−
RF
filt
er
MA
TL
AB
Fun
ction
RF
dete
cto
r
−K
−
RF
am
plif
ier
−K
−
Pho
todio
de1
−K
−
Photo
dio
de
10
Outp
ut
Buff
er1
−R
26
/R25
1
Ou
tput
Bu
ffer
−R
30
/R28
1
Low
Pa
ss F
ilte
r1
1
den
(s)
Low
Pass F
ilte
r
1
R22
*C8
.s+
1
Lase
r2
MA
TLA
B
Fu
nctio
n
Lase
r1
MA
TLA
B
Fu
nctio
n
Lase
r D
rive
r 2
−K
−
La
ser
Drive
r 1
−K
−
Inte
gra
tor1
1
R21
*C9.s
Inte
gra
tor
1
R2
3*C
12.s
Con
sta
nt1
4
Co
nsta
nt
1.6
Attenuation
−K
−
Figure B.2: Detailed DIAL control system model
Appendix C
Master Laser Characterization
Experiments
C.1 Aim
The aim of these experiment was to measure the approximate system parameters relevant
to the design of a control system for master laser tuning in a DIAL system, as well as
for system modeling. For example, based on these experimental results, the prototype
design selected laser injection current for dither modulation and for wavelength control.
The relevant laser system parameters included:
Laser wavelength-power coefficient
Laser wavelength-temperature coefficient
Laser power modulation input sensitivity and bandwidth (laser module)
Laser power control input sensitivity and bandwidth (current control module)
Laser thermal control input sensitivity
Laser thermal slew rate (non-linear system)
Laser thermal time constant (linear system)
129
130 APPENDIX C. MASTER LASER CHARACTERIZATION EXPERIMENTS
C.2 Methods and results
C.2.1 Method part 1
Master laser wavelength -current coefficient: The experimental setup is shown in figure
C.1.
A triangular dither modulation of the laser wavelength by approximately 500 MHz p-
p, also triggered an oscilloscope sweep. The quiescent laser current is adjusted using the
current controller to align a Fabry-Perot peak to the center of the oscilloscope screen. The
DC current is increased repeatedly using the current controller to find consecutive peaks.
Current readings from the controller, and power are recorded every 7.5 GHz. Temperature
held constant with a thermistor reading of 13.05 kΩ which corresponds to 19.2 °C.
Results part 1
Table
Diode Current (mA) Optical Power (mW) Relative optical frequency (GHz)
52.8 3.81 0
58.1 5.03 -7.5
63.3 6.26 -15
68.6 7.51 -22.5
73.9 8.76 -30
79.1 10.0 -37.5
84.3 11.23 -45
89.5 12.48 -52.5
94.6 13.72 -60
99.6 14.95 -67.5Laser diode optical frequency current coefficient = 0.7 mA/GHz
Discussion part 1
These results provide the laser optical power and wavelength characteristics for the mod-
eling experiment described in Chapter B. One source of error in these results is the Free
C.2. METHODS AND RESULTS 131
Oscilloscope
Power Meter
S120
Isolator
Fabry Perot Cavity7.5 GHz
IsolatorLaserSync.
T
I Current
Temp.PID
control
Dither signal Current Adjust
Temperature = 19.2’C
60 dB
(a) Experimental setup for laser diode current and wavelength characterization
30 40 50 60 70 80 90 1000
5
10
15
Injection Current (mA)
Optical P
ow
er
(mW
)
(b) Laser power vs injection current
0 5 10 15−80
−60
−40
−20
0
20
40
Optical Power (mW)
Optical F
requencey C
hange (
GH
z)
(c) Laser frequency vs output power
Figure C.1
Spectral Range (FSR) of the Fabry-Perot interferometer. This was measured to be very
close to the nominal 7.5 GHz, but this varies with temperature, time, etc. This error is
systematic for every consecutive measurement, and is cumulative from the first to the last
measurement presented in table C.1.
The laser power measurements are also relative since we are using isolators, optical fibers
132 APPENDIX C. MASTER LASER CHARACTERIZATION EXPERIMENTS
and a fiber splitters that exhibit loss that we don’t measure here. There is also a Fabry-
Perot interference effect inside the fiber splitter that causes the output power to vary with
optical frequency with an amplitude of about 5 % at an FSR of about 3.4 GHz, which is
likely to be the main source of error in this experiment.
The optical frequency factor takes into consideration the fact that both the power and op-
tical frequency measurements are relative. This is basically a measurement of how the
optical frequency would change over a full-scale change in optical power, by a straight-
line extrapolation from the stated value. This figure is utilized for small signal or dither
sensitivity to output power for operation near the stated quiescent current.
Since these results are for modeling purposes, the errors have little significance.
C.2.2 Method part 2
This experiment is identical to part 1 except instead of adjusting the quiescent laser cur-
rent, the temperature is adjusted with a constant current set at 80 mA. The experimental
layout is shown in figure C.2. The results are presented in terms of the thermistor readings.
The thermistor temperature was calculated from the 3rd order Steinhart-Hart equation co-
efficients provided by Thorlabs (Thorlabs, 2011) shown in C.2.
From these, we get a frequency/Temperature coefficient of ∼25 GHz/°C near 19°C.
C.2.3 Method part 3
In this part we characterize the biased-T input of the laser diode housing module Thorlabs
TCLDL9. The quiescent conditions for the laser are 80 mA and 13.00 kΩ. A 1 V p-p or
200 mV p-p sinusoidal signal is applied to the biased T, and the AC voltage from a tran-
simpedance amplifier which is proportional to the laser power modulation, is measured.
The phase between the signal generator and photodiode output is measured as well to
indicate the nature of the impedance between the input and the laser. Reading are taken
at different frequencies from 5 kHz to 1 MHz and are shown in C.3.
C.2. METHODS AND RESULTS 133
Oscilloscope
Isolator
Fabry Perot Cavity7.5 GHz
IsolatorLaserSync.
T
I Current
Temp.PID
control
Dither signal
Temperature adjust
60 dBCurrent = 80 mA
(a) Method Part 2
Thermistor R (kΩ) Temperature °C ∆ Optical frequency (GHz)12.380 20.142 0.012.550 19.837 7.512.724 19.529 7 15.012.900 19.223 22.513.082 18.912 30.013.264 18.606 37.5
18.6 18.8 19 19.2 19.4 19.6 19.8 20 20.2 20.40
5
10
15
20
25
30
35
40
Temperature ’C
Optical F
requency c
hange G
Hz
(b) Laser temperature frequency characteristic
1T
= a + b ln( R10k
)+ c ln
( R10k
)2
+ d ln( R10k
)3
a = 0.003354017; b = 0.00025617244; c = 2.1400943 × 10−6; d = −7.2405219 × 10−8
Figure C.2: Results part 2
C.2.4 Discussion part 3
Figure C.3 illustrates the dominant Zero near 75 kHz, with phase approaching zero, and
gain approaching maximum above this frequency. This input clearly is not particularly
useful for our applications, however, due to the high-frequency access to the diode, it
could be used in future where suppression of laser noise over a wide bandwidth is de-
sirable. For example, high speed laser frequency noise reduction servo has been used
134 APPENDIX C. MASTER LASER CHARACTERIZATION EXPERIMENTS
IsolatorIsolatorLaserT
I Current
Temp.PID
control
Function GeneratorOscilloscope
60 dB(a) Experimental setup for laser diode housing biased-T characteriza-tion
1 V p-p modulationFrequency (kHz) Phase () Voltage (mV)5 -90 2.510 -90 520 -80 1030 -65 1540 -60 2050 -55 23
200 mV p-p modulationFrequency (kHz) Phase () Voltage (mV)50 -55 575 -45 7.5100 -30 7.5200 -20 7.5500 0 91000 0 10
0 10 20 30 40 50 60 70 80 90 1000
10
20
30
40
Modulation Frequency (kHz)
Lin
ear
Gain
(arb
itra
ry u
nits)
0 10 20 30 40 50 60 70 80 90 100
−100
−80
−60
−40
−20
Phase °
(b) Gain and Phase results for laser housing biased-T input
Figure C.3: Results part 3
to achieve ultra-low noise laser sources (Ottaway et al., 2000). However, as we were
only interested in operation in the DC and low-frequency regime, we didn’t measure the
high-frequency pole of this input.
C.2. METHODS AND RESULTS 135
C.2.5 Method part 4
This experiment measured the optical wavelength modulation as a function of the voltage
input to the current control module. This was done in two parts. First, the Fabry-Perot
was used to calibrate the output voltage measured by the photodiode in terms of optical
frequency modulation. Assuming the modulation was a linear function of output power up
to 7.5 GHz, which was a good approximation based on results from part 1, the amplitude
of the function generator was adjusted so that the crest of two Fabry-Perot interference
peaks were visible. This provided a conversion factor between the photodiode signal
and optical wavelength. Second, the modulation voltage to the control input of the laser
current driver was measured. The temperature was kept constant with thermistor at 14
kΩ and quiescent current at 80 mA. The frequency response of the control input was also
measured by increasing the frequency of the input signal until the output amplitude halved
and the phase lag increased to about 45 degrees. The experimental setup is illustrated in
figure C.4.
Oscilloscope
Oscilloscope
Function Generator
LaserT
I Current
Temp.PID
control
Fabry Perot Cavity7.5 GHz
60 dB
Isolator
I= 80 mARt= 14.0 k
Figure C.4: Experimental setup for laser diode current controller and wavelength charac-terization
Results part 4
System current controller wavelength coefficient = 50 mV/GHz (no input attenuator).
Measured -3 db Frequency response = DC - 200 kHz (± 50 kHz)
136 APPENDIX C. MASTER LASER CHARACTERIZATION EXPERIMENTS
Discussion part 4
This experiment illustrated in C.4 was performed in the small-signal regime since at that
stage we weren’t assuming that the relationship between the laser injection current and
wavelength was linear. It was found to be very linear, over the entire laser output power
range, as indicated in other results C.1. This experiment was later repeated with a larger
signal, and found to give a similar result of 28 GHz/V, which was subsequently used in
the models.
The 200 kHz AC response, as well as DC response, together with sensible sensitivity,
made this a good candidate for wavelength control and dither modulation.
C.2.6 Method part 5
In this part we characterize the laser wavelength from the laser temperature controller
input. A signal generator is connected to the current controller to modulate optical fre-
quency by about 500 MHz so that a Fabry-Perot fringe can be visible on a synchronously
triggered oscilloscope. A variable 0-1 V DC supply is connected to the temperature con-
troller input. With the input voltage set to zero, the initial temperature was adjusted using
the control module so that a Fabry-Perot fringe was visible in the center of the oscillo-
scope screen. The input voltage was increased until the next fringe was centered on the
oscilloscope screen, and the voltage at each consecutive fringe was recorded. This was
repeated up to a 0.5 V input.
Discussion part 5
From the results in C.5, the system temperature controller wavelength coefficient is 11.5
mV/GHz. This is quite sensitive. Furthermore, we can expect the response speed of this
input to be slow compared to C.4.
C.2. METHODS AND RESULTS 137
Oscilloscope
LaserT
I Current
Temp.PID
control
Fabry Perot Cavity7.5 GHz
60 dB
Isolator
I= 80 mA
Function Generator
0−1. V DC
(a) Experimental setup for laser diode temperature controllerand wavelength characterization
Vin (mV) Relative optical frequency (GHz)0 0
80 7.5163 15249 22.5337 30426 37.5515 45
0 100 200 300 400 500 6000
10
20
30
40
50
Temperature Controller Input (mV)
Optical F
requency C
hange (
GH
z)
(b) Laser frequency change vs temperature controller voltage input
Figure C.5: Results part 5
C.2.7 Method part 6
In this part we measure the slew rate of the thermal response of laser wavelength to a
voltage step input to the temperature control module with the Peltier current set to 1 A.
Before the step voltage is applied, the initial temperature is set for a Fabry-Perot fringe
to be centered on the oscilloscope screen. A 1.1 V step, sufficient to saturate the Peltier
current driver at 1 A, is then applied, and the laser temperature drops at a constant rate.
A stopwatch is used to count the first 10 Fabry-Perot fringes, corresponding to a 75 GHz
138 APPENDIX C. MASTER LASER CHARACTERIZATION EXPERIMENTS
wavelength scan. The fringes are counted as they traverse across the oscilloscope screen.
This is repeated for a -1.1V step for heating of the laser diode. Both experiments are
repeated 3 times, with the setup illustrated in figure C.6.
Results part 6
Oscilloscope −1
1
0
0b)
a)
LaserT
I Current
Temp.PID
control
Fabry Perot Cavity7.5 GHz
60 dB
Isolator
I= 80 mA
Function Generator
(a) Experimental setup for laser temperature controller wave-length slew rate characterization
Cooling time (s) Heating time (s)13.8 9.32
13.85 10.0014.05 9.27
Figure C.6: laser temperature controller: 75 GHz optical slew time results
Thermistor resistance range with 1.1 V input = 11.7 kΩ −→ 13.7 kΩ
Heating rate = 7.9 GHz/sec. Cooling rate = 5.4 GHz/sec.
C.2.8 Discussion part 6
The wavelength slew rate would be proportional to the heat pumping rate that depends
on the Peltier current as well as the efficiency of the heat pump. Since the laser is near
room temperature, the difference in the two results is due mainly to the inefficiency of the
Peltier device. The rate of change of the optical frequency was observed to be reasonably
constant over the 75 GHz scan which permitted a relatively long duration for the scan.
C.2.9 Method part 7
In this part, the thermal small-signal time constant is measured. The main difference with
part 6 is that the Peltier is always operated at less than the limiting current. The temper-
C.2. METHODS AND RESULTS 139
ature is adjusted so that a Fabry-Perot fringe is centered on an oscilloscope screen and
a very small, low frequency modulation voltage is applied to the temperature controller.
The optical frequency excursion is measured by the motion of the peak of the Fabry-Perot
interference fringe on the oscilloscope screen. Since we want to keep the peak derivative
value of the modulation constant, the signal amplitude is scaled against the frequency.
The ratio of the peak to peak frequency excursion versus the peak to peak modulation
voltage provides the characteristic thermal response. The setup is illustrated in figure C.7.
Oscilloscope
LaserT
I Current
Temp.PID
control
Fabry Perot Cavity7.5 GHz
60 dB
Isolator
I= 80 mA
Function Generator
Function Generator
(a) Experimental setup for laser thermal time constant charac-terization
C.2.10 Results part 7
The results are presented in figure C.7.
C.2.11 Discussion part 7
The results in part 7 are fairly approximate due to the nature of the wavelength modula-
tion measurement described in the method. Nevertheless, it is clearly evident from the
response ratio that there is a pole in the thermal response somewhere near 350 mHz (±50
mHz). This corresponds to a time constant of about 0.5 sec ±0.1 s.
These experiments illustrate the low speed of the controller, which renders it unsuitable
for dither modulation. It would be suitable, however, for the stabilization loop. However,
as discussed in other chapters, we decided not to use this input for any control function.
140 APPENDIX C. MASTER LASER CHARACTERIZATION EXPERIMENTS
Frequency (Hz) Input Modulation (mV) Optical Freq.Modulation (GHz) Gain0.2 10 1.4 0.14
0.25 8 1.0 0.130.3 6.7 0.7 0.10
0.35 5.7 0.45 0.080.4 5 0.3 0.06
0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34 0.36 0.38 0.40.06
0.08
0.1
0.12
0.14
0.16
Frequency (Hz)
Contr
olle
r G
ain
(G
Hz/m
V)
(a) Temperature controller input voltage frequency response
Figure C.7
C.3 Conclusion
In this set of experiments we measured the laser system parameters required to make a
decision on how best to control the laser wavelength, as well as the system parameters re-
quired to implement a control system. It would have been attractive to implement a wave-
length controller system using temperature control since there was virtually no change in
laser power with wavelength. However, the difference between the heating and cooling
rate, as well as the generally slow response, made this unsuitable for a good control sys-
tem. On the other hand, the biased-T on the laser housing had excellent high-frequency
response, but was useless for DC stabilization. For these reasons, the DC current con-
troller was used for all modulation and control functions.
Appendix D
Experimental Procedure for On-line
Extinction Measurement
The master lasers were tuned to the 822.92 line, taking extra care that we were on the cor-
rect line by observing nearby absorption features. Furthermore, the master laser was on
the same longitudinal mode as during previous experiments, established from the wave-
length, as well as the voltage and current to the diode as described in Section 4.1. Only
the on-line master laser was used for the calibration.
1. The first master laser was tuned to the desired water line (at 822.92nm) operating at
95 mA. This was done by connecting the dither signal to the injection current driver,
and following a previously determined procedure involving setting the laser diode
temperature and current. The temperature is then scanned over a reasonable range
(∼ 1kΩ), and the pattern of absorption line positions and intensities is observed
from the output of the ratiometric detector at the absorption cell (see Fig A.1). This
facilitates identification of the desired water line without a calibrated OSA. Dither
signal is then turned off for the rest of the experiment.
2. The system was re-arranged according to figure 4.7. The control input of the tem-
perature controller was connected to a function generator to repeatedly rapidly scan
over the selected range every 5 seconds. The wavelength scan range around the
absorption peak was setup by adjusting the current controller while observing the
optical wavelength shift with an OSA, or by observing the modulation of the beat
141
142 APPENDIX D. ON-LINE EXTINCTION MEASUREMENT
frequency using an RF spectrum analyzer, with the second laser held at a fixed op-
erating state. The scan was as fast as possible with the Peltier current limit at 2
A, in order to fit the acquisition into the memory of the data acquisition card, with
the 100 MS/s sampling rate required to capture the actual shape of each evolving
optical pulse.
3. Optical attenuators for the high speed detectors were adjusted for a linear output.
The output stages of the LM6563 were found to exhibit soft-clipping nonlinearity
due to the 50 Ω cable termination loads. This was avoided by keeping the output
current below 20 mA, with an output voltage below 4 V p-p.
4. The direct beam from the master laser to the water vapour cell via splitter was
blocked.
5. The laser amplifier was turned on and aligned. This was done by observing the
fringe intensity with a Fabry Perot interferometer, and by maximizing the pulse
output amplitude.
6. The ratiometric circuit was disconnected, and the outputs from the high speed pho-
todiode amplifiers were connected by coaxial cables to the CS14105 high speed
data acquisition system.
7. A Matlab function see adgmr.m was run (See Appendix G).
8. The two sets of photodiode output data were then analyzed using calc.m. This
program firstly performs a time shift to compensate for the delay through the 33
m multi-pass cell. It then detects the start of each pulse, and sums 85 subsequent
acquisitions, which integrates most of the pulse with a 100 Ms/s sampling rate. The
resulting pair of data sets therefore provide a measurements of the relative energy
in each pulse, before and after the absorption cell. This ratio provides a relative
measurement of the extinction in the cell at each wavelength data point. Measuring
the ratio in this extinction at the absorption line center, and on the wings of the
spectrum, provides an absolute measurement of the extinction at the center of the
absorption line.
Appendix E
Master Laser Diode Wavelength Pair
Matching
E.1 Aim
A dual laser DIAL requires two laser diodes that both share a dominant wavelength in
a stable longitudinal mode, that also corresponds to the wavelength of a desirable water
absorption line. The aim of this experiment was to select a master laser diode pair such
that their wavelengths overlap while operating near full current and near room tempera-
ture, and the overlapping wavelength range includes suitable water absorption lines. The
first part describes a manual experimental procedure without using a wavemeter, which
was actually used to find the diode pair used to construct our instrument. The second part
describes the much more rapid procedure utilizing a wavemeter (HighFinesse-Angstrom
WS/7) when it was available.
E.2 Manual method using a slow ramp
Figure E.1 illustrates the experimental setup. A slow ramp is applied to the temperature
controller that slowly scans the laser wavelength over approximately a 300 GHz range
while a WM dither signal with a depth of about 500 MHz p-p is applied using the injection
current input of the diode controller with the quiescent current set at 80 mA. The laser is
143
144 APPENDIX E. MASTER LASER DIODE WAVELENGTH PAIR MATCHING
Dither signal
Diode Wavelength Measurement Experiment
I
TLaser DUT Isolator
Ramp signal
Vapor Cell
SR530
Lock−inAmplifier
Absorption Spectrum OutputCH1
CH2
Fabry Perot CavityFabry−Perot Output
7.5 GHz
12−bit data aquisition
Computer with GS1205
Figure E.1: Manual laser diode frequency identification experiment
fiber coupled and split two ways. The relative optical frequency sweep is measured with a
Fabry-Perot interferometer with a FSR of 7.5 GHz that provides sharp interference fringes
at that spacing that are acquired with one channel of a data acquisition card (GageScope
GS1205).
The second fiber goes to the water vapour absorption cell where the derivative spec-
trum is acquired with a photodiode and a lock-in amplifier, the second channel of the data
acquisition card. The temperature is swept over a period of about 50 seconds. The exper-
iment was repeated twice for each diode, once with the dither turned on, and again with
the dither turned off. This was necessary since the dither signal enabled the detection of
the absorption lines, but blurred the Fabry-Perot fringes.
The experiment was repeated for a total of 7 diodes before a matched pair was found.
The best match was at 823 nm, which also coincided with the strongest water absorption
line.
E.2.1 Wavelength identification from HITRAN
Once the absorption spectrum, and the fringes were acquired, the HITRAN database was
utilized to find the wavelength of the laser. This was a combination of a manual process of
measuring the intensity ratios and the spacings of the lines using the derivative spectrum
and the 7.5 GHz fringes respectively. This data was then incorporated into Matlab routines
’findl.m’ that scanned through the HITRAN database of water lines in the 820-830 nm
E.3. LASER CHARACTERIZATION USING A WAVEMETER 145
range, to find candidate matches within selected tolerance bounds. If the tolerance bounds
were set too strictly, there would be no matches found due to the manual measurement
errors and approximations. If the tolerance were too loose, then there would be too many
matches to consider. The tolerances were set by trial and error to return a small number of
matches that could then be manually compared to the HITRAN data. Not all diodes could
be characterized correctly using this method within given reasonable time constraints,
mostly because it was difficult to find longitudinal modes where a significant number of
absorption lines were visible.
E.2.2 Sample results
The experiment was repeated for a total of 7 diodes before a matched pair at 823 nm
was found. This technique was not as reliable as the wavemeter. Some of the successful
results are presented in figure E.2
E.3 Laser characterization using a wavemeter
An Angstrom WS/7 super precision wavemeter was made available for one day to char-
acterize our diodes. This was a far simpler and more reliable method. The experimental
apparatus was set up as shown in figure E.3 with a constant laser diode current of 90
mA when turned on. The diode under test was initially cooled to about 10 °C (20 kΩ),
laser current was turned on, and the wavelength measured. The temperature was then
gradually increased and measured by reading the resistance of the thermistor inside the
laser housing. At every temperature interval corresponding to a 2 kΩ temperature rise,
the laser wavelength was recorded. The measurements for each diode were recorded up
to a temperature of about 40 °C (6 kΩ). A total of 10 Hitachi diodes, HL8325, and two
Sanyo diodes, DL8032 (DUT No 15 and 16) were characterized.
146 APPENDIX E. MASTER LASER DIODE WAVELENGTH PAIR MATCHING
(a)
(b)
Figure E.2: Manual laser diode matching using acquired water absorption spectra, aFabry-Perot (FP) spectrum, HITRAN data and some MATLAB code. The FP spacingis 7.5 GHz, and the modulation of the FP fringes is due to internal reflections in the fiberoptic splitter.
E.3. LASER CHARACTERIZATION USING A WAVEMETER 147
I
TLaser DUT Isolator
Ramp signal
AngstromWS/7 Wavemeter
Temperature I=90 mA
Diode Wavelength Measurement Experiment
Figure E.3
E.3.1 Results
Table of laser diode wavelength measurements for pair selection, with wavelength given
in nanometers.Resistance DUT No
kΩ 2 3 4 5 6 7 11 12 13 14 15 16
20 825.7 825.4 828.4 825.8 826.0 822.0 825.7 827.0 820.4 823.9 820.5 827.8
18 826.4 826.5 829.3 826.1 826.4 821.3 826.4 827.1 821.5 824.3 820.5 828.5
16 827.3 827.4 829.3 826.5 827.3 822.1 827.1 828.3 821.6 825.0 821.3 829.3
14 827.6 827.9 830.0 827.7 828.1 823.9 827.9 829.0 822.4 826.1 822.1 829.4
12 828.9 828.6 831.8 828.7 828.8 824.8 828.5 829.8 823.3 826.9 822.8 830.1
10 829.6 829.7 832.3 829.6 829.9 824.3 830.1 830.7 824.4 827.8 823.6 831.6
8 831.1 831.4 833.8 830.9 831.1 825.9 830.8 831.9 825.9 828.7 825.3 832.9
6 833.6 833.2 835.6 833.9 833.2 829.0 832.5 833.7 826.9 830.4 827.2 834.0
E.3.2 Discussion
The above results obtained using the wavemeter, facilitated the selection of diode pairs
for water vapour spectroscopy. These results also shed some insight into the longitudinal
modal stability and behavior of the diodes. Following the results for a particular device, it
is evident how the wavelength changes monotonically with longitudinal mode-hops as the
temperature increases. These results are only intended for diode wavelength pair matching
selection purposes, since the temperature measurements are only very approximate, and
there is no characterization of the longitudinal mode stability at specific temperatures.
E.3.3 Conclusion
A wavemeter has been used successfully to characterize and wavelength match diode
pairs for differential absorption spectroscopy applications. When a wavemeter was not
available, it was often possible to determine laser diode wavelength from the water ab-
sorption spectrum by pattern-matching the measured spectrum to HITRAN data using a
simple Matlab script.
148 APPENDIX E. MASTER LASER DIODE WAVELENGTH PAIR MATCHING
Appendix F
Dither Induced Offset Experiment
Introduction
The injection current modulation used to provide the optical frequency dither affects both
the optical power and frequency of the master laser. An analytical approach to evaluate
this effect indicated that it would cause an offset error, however, it was too difficult to
evaluate its magnitude accurately due to the Voigt line shape. This experiment presents
an entirely numerical approach to do just that, utilizing a numerical approximation of
the line shape described in Chapter 2.3, and the measured laser characteristics described
below as well as in Chapter C. The phase sensitive lock-in amplifier produces an offset
error that is measured as an optical frequency offset between the actual center of the
absorption line, and the zero-crossing of the error-signal from the lock-in amplifier. The
model uses both the characteristic power and wavelength modulation in our DIAL laser
stabilization setup, as well as the characteristic 822.92nm absorption line as measured
through our open vapour cell.
F.1 Aim
The aim of this computer simulation experiment is to model the effects of the amplitude
modulation of the master laser on the stabilized optical frequency. This error would be
present if our system did not have ratiometric detection as described in Chapter A. In our
149
150 APPENDIX F. DITHER INDUCED OFFSET EXPERIMENT
system, this error is attenuated by the linearity of the analog divider, which is about two
orders of magnitude.
F.2 Method
The first part of this experiment involved measuring the optical frequency and power
characteristics of our laser diode. The laser was energized such that it was in the desired
longitudinal mode, and its wavelength was close to the nominal operating wavelength of
823 nm. The laser temperature was adjusted slightly so that a Fabry Perot fringe coincided
with the maximum injection current close to 100 mA, and the stabilized temperature was
kept constant for the duration of the experiment. The current was then reduced and the
optical power was measured at each consecutive Fabry-Perot fringe, which corresponds
to a 7.5 GHz spacing. The power was measured after passing through a pair of Faraday
isolators, however, since we are only interested in the relative power/frequency character-
istic, the optical attenuation is irrelevant. The results are presented in table C.1 and figure
F.1. From this data, we can calculate the coefficient to be 83 GHz at 95 mA, and the laser
power dither of 0.6% with an optical frequency dither of 500 MHz. In other words, due to
the inverse relationship between the injection current and optical frequency, as the optical
power increases by 0.6%, the optical frequency drops by 500 MHz and vice versa.
The second part of this experiment used these results, together with the Voigt model
described in section 2.3.2, to calculate the error signal shift from line center at STP. The
model (dithermodel.m) in section G simulates a square-wave dither with the resulting
vapour cell transmission and laser power modulation, as the optical frequency is scanned
across the absorption line, just like a lock-in amplifier. This provides a measurement of
the displacement of the zero-crossing from the absorption line center.
F.3 Results
F.3. RESULTS 151
0 10 20 30 40 50 60 70 80 90 100 1100
2
4
6
8
10
12
14
16Laser Power vs Injection Current measured at 7.5 GHz (optical) intervals
Injection Current (mA)
Opt
ical
Pow
er (
mW
atte
nuat
ed)
Figure F.1: This figure illustrates the measured laser power as measured at consecutive7.5 GHz Fabry-Perot fringes, by adjusting the injection current.
822.905 822.91 822.915 822.92 822.925 822.93 822.935 822.94−0.01
−0.005
0
0.005
0.01
0.015
0.02
Wavelength (nm)
Err
or
sgnal −
norm
aliz
ed
Error signal simulation with Laser Current Dither
580 MHz
Figure F.2: This figure illustrates the result of the simulation model. The error signal isdisplaced vertically and there is a corresponding shift in the zero-crossing
152 APPENDIX F. DITHER INDUCED OFFSET EXPERIMENT
Appendix G
Matlab (Octave) Functions and CodeListing
=====================================The adgmr.m routine performs the triggered frame capture as described in the text
%adgmr.m
% This is based on GageMultipleRecord.m sample program and uses setup.m
% The system’s acquisition, channel and trigger parameters are set. The data is
% captured and retrieved and the data for each channel and multiple segment
% is saved to a separate file.
clear;
systems = CsMl_Initialize;
CsMl_ErrorHandler(systems);
[ret, handle] = CsMl_GetSystem;
CsMl_ErrorHandler(ret);
[ret, sysinfo] = CsMl_GetSystemInfo(handle);
s = sprintf(’-----Board name: %s\n’, sysinfo.BoardName);
disp(s);
Setup(handle);
CsMl_ResetTimeStamp(handle);
ret = CsMl_Commit(handle);
CsMl_ErrorHandler(ret, 1, handle);
[ret, acqInfo] = CsMl_QueryAcquisition(handle);
ret = CsMl_Capture(handle);
CsMl_ErrorHandler(ret, 1, handle);
status = CsMl_QueryStatus(handle);
while status ˜= 0
status = CsMl_QueryStatus(handle);
end
% Get timestamp information
transfer.Channel = 1;
transfer.Mode = CsMl_Translate(’TimeStamp’, ’TxMode’);
transfer.Length = acqInfo.SegmentCount;
transfer.Segment = 1;
[ret, tsdata, tickfr] = CsMl_Transfer(handle, transfer);
153
154 APPENDIX G. MATLAB (OCTAVE) FUNCTIONS AND CODE LISTING
transfer.Mode = CsMl_Translate(’Default’, ’TxMode’);
transfer.Start = -acqInfo.TriggerHoldoff;
transfer.Length = acqInfo.SegmentSize;
% Regardless of the Acquisition mode, numbers are assigned to channels in a
% CompuScope system as if they all are in use.
% For example an 8 channel system channels are numbered 1, 2, 3, 4, .. 8.
% All modes make use of channel 1. The rest of the channels indices are evenly
% spaced throughout the CompuScope system. To calculate the index increment,
% user must determine the number of channels on one CompuScope board and then
% divide this number by the number of channels currently in use on one board.
% The latter number is lower 12 bits of acquisition mode.
MaskedMode = bitand(acqInfo.Mode, 15);
ChannelsPerBoard = sysinfo.ChannelCount / sysinfo.BoardCount;
ChannelSkip = ChannelsPerBoard / MaskedMode;
% Format a string with the number of segments and channels so all filenames
% have the same number of characters.
format_string = sprintf(’%d’, acqInfo.SegmentCount);
MaxSegmentNumber = length(format_string);
format_string = sprintf(’%d’, sysinfo.ChannelCount);
MaxChannelNumber = length(format_string);
format_string = sprintf(’%%s_CH%%0%dd-%%0%dd.dat’, MaxChannelNumber, MaxSegmentNumber);
d(1:acqInfo.SegmentSize*acqInfo.SegmentCount,1:2)=0;
for channel = 1:ChannelSkip:sysinfo.ChannelCount
transfer.Channel = channel;
for i = 1:acqInfo.SegmentCount
transfer.Segment = i;
%data0=data;
[ret, data, actual] = CsMl_Transfer(handle, transfer);
CsMl_ErrorHandler(ret, 1, handle);
% Note: to optimize the transfer loop, everything from
% this point on in the loop could be moved out and done
% after all the channels are transferred.
% Adjust the size so only the actual length of data is saved to the
% file
length = size(data, 2);
if length > actual.ActualLength
data(actual.ActualLength:end) = [];
length = size(data, 2);
end;
% Get channel info for file header
[ret, chanInfo] = CsMl_QueryChannel(handle, channel);
CsMl_ErrorHandler(ret, 1, handle);
% Get information for ASCII file header
info.Start = actual.ActualStart;
info.Length = actual.ActualLength;
info.SampleSize = acqInfo.SampleSize;
info.SampleRes = acqInfo.SampleResolution;
info.SampleOffset = acqInfo.SampleOffset;
info.InputRange = chanInfo.InputRange;
info.DcOffset = chanInfo.DcOffset;
info.SegmentCount = acqInfo.SegmentCount;
info.SegmentNumber = i;
info.TimeStamp = tsdata(i);
155
filename = sprintf(format_string, ’MulRecResult’, transfer.Channel, i);
%CsMl_SaveFile(filename, data, info);
d(1+(i-1)*acqInfo.SegmentSize:i*acqInfo.SegmentSize,transfer.Channel)=data;
end;
dat(channel,1:256)=data;
end;
%A=(dat(2,11:256)-dat(1,1:246));%.*(1+(1:246)*;
%plot(A)
%figure
plot(d)
ret = CsMl_FreeSystem(handle);
==============================
The setup.m code sets up the parameters for the frame size and acquisition rate re-quired for adgmr.m
function [ret] = Setup(handle)
%setup.m
% Set the acquisition, channel and trigger parameters for the system and
% calls ConfigureAcquisition, ConfigureChannel and ConfigureTrigger.
[ret, sysinfo] = CsMl_GetSystemInfo(handle);
CsMl_ErrorHandler(ret, 1, handle);
acqInfo.SampleRate = 100000000;
acqInfo.ExtClock = 0;
acqInfo.Mode = CsMl_Translate(’Dual’, ’Mode’);
acqInfo.SegmentCount = 7000;
acqInfo.Depth = 128;
acqInfo.SegmentSize = 256;
acqInfo.TriggerTimeout = 1000;
acqInfo.TriggerDelay = 0;
acqInfo.TriggerHoldoff = 128;
acqInfo.TimeStampConfig = 0;
[ret] = CsMl_ConfigureAcquisition(handle, acqInfo);
CsMl_ErrorHandler(ret, 1, handle);
% Set up all the channels even though
% they might not all be used. For example
% in a 2 board master / slave system, in single channel
% mode only channels 1 and 3 are used.
for i = 1:sysinfo.ChannelCount
chan(i).Channel = i;
chan(i).Coupling = CsMl_Translate(’DC’, ’Coupling’);
chan(i).DiffInput = 0;
chan(i).InputRange = 2000;
chan(i).Impedance = 50;
chan(i).DcOffset = 0;
chan(i).DirectAdc = 0;
chan(i).Filter = 0;
end;
[ret] = CsMl_ConfigureChannel(handle, chan);
CsMl_ErrorHandler(ret, 1, handle);
trig.Trigger = 1;
trig.Slope = CsMl_Translate(’Positive’, ’Slope’);
trig.Level = 15;
trig.Source = 1;
156 APPENDIX G. MATLAB (OCTAVE) FUNCTIONS AND CODE LISTING
trig.ExtCoupling = CsMl_Translate(’DC’, ’ExtCoupling’);
trig.ExtRange = 2000;
[ret] = CsMl_ConfigureTrigger(handle, trig);
CsMl_ErrorHandler(ret, 1, handle);
ret = 1;
==============================
The calc.m code detects the start of a positive pulse in the data frame, and calculatesthe sum of samples, storing the result as a single vector. An array of all these vectorsproduces the absorption curve described in the text.
a=0;
b=0;
c=1;
s=size(d); %d is the raw data
s=s(1,1)-100; %size of raw array
data=0;
data(1:s,1)=d(1:s,1); %before vapour cell
data(1:s,2)=d(11:s+10,2);
%after vapour cell with temporal adjustment to compensate for 33 m delay.
%See adgmr for sampling rate, etc.
while c<(s-200); %normalizing routine if needed; otherwise comment out.
if (data(c,1)>0.1) %look for start of pulse because of trigger jitter
a=a+sum(data(c:c+84,1));
%average out the pulses to calculate the normalizing factor a/b if used.
b=b+sum(data(c:c+84,2));
c=c+110; %advance counter to next sample
end
c=c+1;
end %normalizing routine
datan(1:s,2)=data(1:s,2);%*a/b;
datan(1:s,1)=data(1:s,1);
n=0;
c=1;
vps1=0;
vps2=0;
a/b
while c<(s-200); %start calculation routine
if (datan(c,1)>0.1), %look for start of pulse because of trigger jitter
n=n+1; %increment pulse counter
vp1(n)=sum(datan(c:c+84,1));%average out pulses
vp2(n)=sum(datan(c:c+84,2));%for both acquisitions
if size(vps1)<2 %if this is the first pulse do this
vps1=datan(c:c+84,1);
else
vps1=vertcat(vps1,datan(c:c+84,1)); %else concatenate subsequent acquisitions
end
if size(vps2)<2 %if this is the first pulse do this
157
vps2=datan(c:c+84,2);
else
vps2=vertcat(vps2,datan(c:c+84,2)); %else concatenate subsequent acquisitions
end
c=c+110; %advance counter to next sample
end
c=c+1;
end
n
r=vp2./vp1;
plot(r)
p=sgolayfilt(r,4,201);
y=r;
y(2,:)=p;
plot(1:n,y)
==============================
The calcm.m code detects the start of a negative pulse in the data frame, and calculatesthe sum of samples, storing the result as a single vector. An array of all these vectorsproduces the absorption curve described in the text. calcm.m is used for the master laseracquisition, whereas calc.m is for the laser amplified pulses.
a=0;
b=0;
c=1;
s=size(d);
s=s(1,1)-100;
data=0;
data(1:s,1)=d(1:s,1);
data(1:s,2)=d(11:s+10,2);
while c<(s-200);
if (data(c,1)<-0.1) %look for start of pulse because of trigger jitter
a=a+sum(data(c:c+84,1));
b=b+sum(data(c:c+84,2));
c=c+110; %advance counter to next sample
end
c=c+1;
end
datan(1:s,2)=data(1:s,2)*a/b;
datan(1:s,1)=data(1:s,1);
n=0;
c=1;
vps1=0;
vps2=0;
%a/b
while c<(s-200);
if (datan(c,1)<-0.1), %look for start of pulse because of trigger jitter
n=n+1; %increment pulse counter
vp1(n)=sum(datan(c:c+84,1));
vp2(n)=sum(datan(c:c+84,2));
158 APPENDIX G. MATLAB (OCTAVE) FUNCTIONS AND CODE LISTING
if size(vps1)<2 %if this is the first pulse do this
vps1=datan(c:c+84,1);
else
vps1=vertcat(vps1,datan(c:c+84,1)); %else concatenate subsequent acquisitions
end
if size(vps2)<2 %ditto
vps2=datan(c:c+84,2);
else
vps2=vertcat(vps2,datan(c:c+84,2));
end
c=c+110; %advance counter to next sample
end
c=c+1;
end
n
%r=vp2./vp1;
plot(r)
figure; plot(vps1(1:1000))
figure; plot(vps2(1:1000))
==============================
The calcpall.m code reads selected consecutive arrays of oscilloscope acquired datato estimate optical pulse power from consecutive measurements made at the same timeas pulse spectra were acquired. Note that only a few puses were acquired for powermeasurement, whereas 7000 pulses were used to reconstruct a single spectrum.
%calcpall.m
N=’67:69’,’70:72’,’73:75’,’79:80’,’84:87’,’88:90’;
q=size(N,2);
P(1:q,1:2500)=zeros;
S(1:q)=zeros; %initialize
result=’ok’;
for l=1:q
for m=str2num(char(N(l))); %eg; 67:69
fi=fopen([char(’//home/ad/Desktop/Experiment100328/csv/F00’) num2str(m) char(’CH1.CSV’)])
fseek(fi,0,’bof’); %set pointer to start of file
c=textscan(fi,’%s%s%s%n%n’,’delimiter’, ’,’);
c=c5; %we are only interested in the last column
z=size(c,1); %=2500
y=1
while c(y)<0.005 %find start of pulse at y
y=y+1;
end
S(l)=S(l)+sum(c(y-1:y+451)); %the pulse is 450 samples long
if c(y+452)> 0.005 result=’error’; end %detect error
P(l,1:z)=P(l,1:z)+c’;
end %m
S(l)=S(l)/(size(str2num(char(N(l))),2)*450);
P(l,1:z)=P(l,1:z)/size(str2num(char(N(l))),2);
figure; plot(P(l,:))
159
end %l
==============================
The rhitran.m code reads the required HITRAN data from a file called hitran out.txt,which is a subset of the 01 hit08.par file, supplied as a part of the HITRAN database.
%rhitran.m
fid=fopen(’hitran_out.txt’,’r’);
a=cell2mat(textscan(fid,
’%*d %*d %f %f %*f %f %f %f %f %f %*s %*s %*s %*s %*d %*s %*s %*f %*f’, ’delimiter’, ’,’));
fclose(fid)
% We have the following read from the HITRAN output file;
%.1 Wavenumber in cm-1
%.2 Intensity in cm/molecule
%.3 Air broadened halfwidth HWHM in cm-1/atm
%.4 Self broadened halfwidth HWHM in cm-1/atm
%.5 Lower state energy in cm-1
%.6 Coefficient of temperature dependence of air-broadened halfwidth
%.7 Air-broadened pressure shift of line transition in cmˆ-1/atm @ 296K
==============================
The avg1m.m code reads the set of files produced by the LICEL and LABVIEWacquisition system used for lidar observations, and produces an ensemble average forDIAL inversion.
%avg1m.m
%This file contains code for both Octave and Matlab ...for disk access.
%cd /home/ad/Desktop/Thesis/observation090923/data090923_ASC_2P/
%addpath /home/ad/Desktop/Thesis/observation090923/data090923_ASC_2P/
%nfiles=size((ls(’//home/ad/Desktop/Thesis/observation090923/data090923_ASC_2P’)),1);
%nfiles=size((ls(’//home/ad/Desktop/Thesis/observation090923/data090923_ASC_2P’)),2);
files=dir(’/home/ad/Desktop/Thesis/observation090923/data090923_ASC_2P’);
nfiles=size(files,1)-2;
n=1024
online(1:n)=zeros;
offline(1:n)=zeros;
%for c=1:nfiles;
%cfname=files(c,1:20);
for c=3:nfiles
%cfname=files(c:c+20)
cfname=files(c).name;
cf=load(’-ascii’,[’/home/ad/Desktop/Thesis/observation090923/data090923_ASC_2P/’ cfname]);
for j=1:1023;
online(j)=online(j)+cf((j),1);
offline(j)=offline(j)+cf((j),3);
end
end
save(’-ascii’,’/home/ad/Desktop/Thesis/observation090923/offline1’,’offline’)
save(’-ascii’,’/home/ad/Desktop/Thesis/observation090923/online1’,’online’)
%load(’-ascii’,’/home/ad/Desktop/Thesis/observation090923/offline1’,’offline1’);
160 APPENDIX G. MATLAB (OCTAVE) FUNCTIONS AND CODE LISTING
%load(’-ascii’,’/home/ad/Desktop/Thesis/observation090923/online1’,’online1’);
%online1=online;
%offline1=offline;
%offline=offline*max(online)/max(offline);
%semilogy((0:7.5:1800),offline(101:341),’-;offline;’,(0:7.5:1800),online(101:341)
,’-;online;’);
%semilogy((-30:7.5:1800),offline(101:345),(-30:7.5:1800),online(101:345));
%figure;semilogy((-60:7.5:3000),offline(98:506),’r’,(-60:7.5:3000),online(98:506),’b’);
figure;semilogy((-60:7.5:1005),offline(98:240),’r’,(-60:7.5:1005),online(98:240),’b’);
% plot((0:7.5:1500),offline(140:340),"-;offline;",(0:7.5:1500),online(140:340)
,"-;online;");
%print -dsvg plot1.svg
==============================
Script to model the effect of power dither on displacement of error signal from linecenter
%dithermodel
%Script to model the effect of power dither on displacement of error signal
%from line center.
%Input line index in hitran output array a(,,,,,,)
%rhitran
v=23257; %the 822.92nm line in this subset of the HITRAN dataset
T=296;
mr=0.01;%mixing ratio = 1%
M=18.0153/1000; %molar mass in Kg
c100=2.998e8*100; %100c for wavenumber conversion calculations
P1=1; %STP
N=2.687e19*mr; % Loschmidt constant for water molecules per cmˆ3
L=3000;%length of vapour cell in cm
dr1=250e6;% Hz dither frequency modulation depth amplitude (500 MHz p-p)
lpf1=83000e6;% Hz laser power coefficient at 95mA
Gd=a(v,1)*Dopplerw(T,M); %calculate the Doppler halfwidth
Gl1=(296/T)ˆa(v,6)*P1*(a(v,4)*mr+a(v,3)*(1-mr));
%calculate the pressure broadened halfwidth at STP
Fo=c100*a(v,1); %center wavenumber in Hz
dnu1=(Fo-dr1)/c100-a(v,1);% negative wavenumber dither modulation amplitude
dnu2=(Fo+dr1)/c100-a(v,1);% positive wavenumber dither modulation amplitude
p(1:401,1:7)=zeros;
p(1:401,1)=(-1:.005:1)/1; %wavenumber offset cm-1
for x= 1:401
p(x,2)=0.01*1e9/(p(x,1)+a(v,1)); %convert center wavenumber cmˆ-1 to wavelength in nm
p(x,3)= exp(-L*N*P1*a(v,2)*Voigt(p(x,1)+dnu1,Gl1,Gd))*(1+dr1/lpf1);
%negative wavenumber dither model- as optical frequency decreases, normalized power increases
p(x,4)= exp(-L*N*P1*a(v,2)*Voigt(p(x,1)+dnu2,Gl1,Gd))*(1-dr1/lpf1);
161
%positive wavenumber dither model- as optical frequency increases, normalized power decreases
p(x,5)= p(x,3)-p(x,4);
%simple discrete model of a lock-in amplifier, we take the two signals,
%multiply one by dither state (+ve in one case, and -ve in the other), and sum (integrate).
p(x,6)= (p(x,3)+p(x,4))/200;
%model of integrator for the absorption signal to remove the dither (no lock in amplifier).
p(x,7)= exp(-L*N*P1*a(v,2)*Voigt(p(x,1)+dnu1,Gl1,Gd));
end
%plot(p(1:401,2),p(1:401,5),p(1:401,2),p(1:401,6))
plot(p(1:401,2),p(1:401,5),p(1:401,2),p(1:401,5))
==============================
Function to calculate Doppler halfwidth coefficient for standard water vapour as afunction of temperature
%Dopplerw.m
%Function to calculate Doppler halfwidth coefficient for standard water
%vapour as a function of temperature
function Gd = Dopplerw(T,M)
M=18.0153/1000; %water molar mass in Kg !!
c=2.998e8;
R=8.3145;
Gd=sqrt(2*log(2)*R*T/M)/c;
==============================
Empirical Partition function calculation for water vapour from 70 K to 405 K.
%Partfn.m
%Partition function calculation for water vapour 70’K to 405’K
%- more accurate replacement for (To/T)ˆ1.5
function QT=Partfn(T)
a=-4.4405; b=0.27678; c=1.2536e-3; d=-4.8938e-7;
QT=a+b*T+c*Tˆ2+d*Tˆ3;
==============================
Function to calculate line strength coefficient using partition function
%Linest.m
%Functrion to calculate line strength coefficient using partition function
%with lower state energy and Temperature.
function st=Linest(E, T)
To=296;
h=6.626069e-34;
c=2.998e8;
k=1.38065e-23;
hck=100*h*c/k; %convert hc/k to centimeters cm !!
%st=(Partfn(To)/Partfn(T))*exp(hck*E*(1/To-1/T));
st=((To/T)ˆ1.5)*exp(hck*E*(1/To-1/T)); %alternative method in Browell
%ST=S*(Partfn(To)/Partfn(T))*(1-exp(-hck*V/T))/(1-exp(-hck*V/To))
%*exp(hck*E*(1/To-1/T));
%More general formula including line frequency. Not needed here.
162 APPENDIX G. MATLAB (OCTAVE) FUNCTIONS AND CODE LISTING
%ST=S*(To/T)ˆ(1.5)*(1-exp(-hck*V/T))/(1-exp(-hck*V/To))*exp(hck*E*(1/To-1/T
%)); %Simplified formula mentioned in text. Not used for calculation.
==============================
Function to calculate Lorentz line.%Lorentz.m
function csf = Lorentz(D,Gl)
csf=(Gl)/(pi*(Dˆ2+(Gl)ˆ2));
==============================
Function to calculate Lorentz line width.%Lorenzw.m
%Function to calculate the pressure and temperature corrected Lorentzian
%width or halfwidth assume 1% water partial pressure. ga=air-broadened
%, gs=self broadened , n=Coefficient of temperature dependence of width.
%This is an approximation for low water mixing ratios, since n(self) is not
%available
function Gl=Lorenzw(ga,gs,n,P,T)
Po=1;
To=296;
%mr=0.01; %input typical approximate water mixing ratio.
mr=0; %not considering self broadening
%Gl=(P/Po)*(To/T)ˆn simplified model- not used for calculation.
Gl=(gs*mr+ga*(1-mr)) * (P/Po) * ((To/T)ˆn);
==============================
Function to calculate Voigt line halfwidth.%Gvoigt.m
%Function to calculate Voigt line halfwidth from Gaussian Gd and Lorentzian
%Gl halfwidths. Ref Olivero78
function Gv=Gvoigt(Gl,Gd)
Gl2=Gl*2;%convert to FWHM for the Olivero formula
Gd2=Gd*2;
Gv=(0.5346*Gl2+sqrt(0.216597*Gl2ˆ2+Gd2ˆ2))/2;
==============================
Function to calculate the peak value of a Voigt.%Pvoigt.m
%Function to calculate the peak value of a Voigt profile from the
%Lorentzian halfwidth Gl, and the Doppler halfwidth Gd. Ref Whiting.
function csfvp = Pvoigt(Gl,Gd)
Gv2=2*Gvoigt(Gl,Gd); %Calculate Voigt line width FWHM
Gl2=Gl*2; %Lorentz FWHM for Whiting method
csfvp=1/(Gv2*(1.065+0.447*(Gl2/Gv2)+0.058*(Gl2/Gv2)ˆ2));
==============================
Empirical function to calculate the Voigt from Whiting Olivero.%Voigt.m
%function to calculate the Voigt function based on the Whiting method.
%Input wavelength Difference (cm-1), Lorentzian halfwidth Gl and Doppler
%halfwidth Gd.
function csfv = Voigt(D,Gl,Gd)
Gv2=2*Gvoigt(Gl,Gd); %Calculate Voigt line width FWHM
Gl2=Gl*2;
csfv = Pvoigt(Gl,Gd)*(((1-Gl2/Gv2)*exp(-2.772*(D/Gv2)ˆ2))+((Gl2/Gv2)/(1+4*(D/Gv2)ˆ2))
+0.016*(1-Gl2/Gv2)*(Gl2/Gv2)*(exp(-0.4*(D/Gv2)ˆ2.25)-10/(10+(D/Gv2)ˆ2.25)));
163
==============================
Script to plot peak line absorption cross-section vs temperature.
%plines.m
%script to plot peak line absorption cross-section vs temperature.
v=23257; %choose line from file 822.92nm
r(1:296,1:6)=zeros;
V=a(v,1)
S=a(v,2);
ga=a(v,3);
gs=a(v,4);
n=a(v,6)
mr=0.01 %mixing ratio approx.
To=296;
P=1; %atm
for e=1:5;
for T=101:396
r(T-100,1)=T;
zc=S*Linest(e*50-50,To)*Pvoigt(Lorenzw(ga,gs,n,P,To),V*Dopplerw(To));
zv=S*Linest(e*50-50,T)*Pvoigt(Lorenzw(ga,gs,n,P,T),V*Dopplerw(T));
r(T-100,e+1)=(2/(T-To))*(zv-zc)/(zv+zc);
end;
end;
plot(r(:,1),r(:,2),r(:,1),r(:,3),r(:,1),r(:,4),r(:,1),r(:,5),r(:,1),r(:,6))
==============================
Script to compare Lorentz and Voigt line shapes.
%plinev.m
%Script to plot line shapes to compare Lorentz and Voigt.
%Input line index in hitran output array a(,,,,,)
%rhitran
v=23257;
T=296;
mr=0.01;
%c=2.998e8;
M=18.0153/1000; %molar mass in Kg !!
Gd=a(v,1)*Dopplerw(T,M);
Gl=(296/T)ˆa(v,6)*(a(v,4)*mr+a(v,3)*(1-mr));
p(1:201,1:4)=zeros;
p(1:201,1)=(-1:.01:1); %wavenumber offset cm-1
ll=a(v,1)-a(v-1,1);
lr=a(v+1,1)-a(v,1);
for x= 1:201
p(x,2) = a(v,2)*Lorentz(p(x,1),Gl)+a(v-1,2)*Lorentz(ll+p(x,1)
,a(v,3))+a(v+1,2)*Lorentz(lr-p(x,1),a(v,3));
%Include the effects of adjacent lines to see if there is any effect.
p(x,3)= a(v,2)*Voigt(p(x,1),Gl,Gd);
%Voigt line assuming no effects from adjacent lines
p(x,4)= 1e7/(p(x,1)+a(v,1)); %wavelength nm
end
plot(p(1:201,4),p(1:201,2),p(1:201,4),p(1:201,3))
164 APPENDIX G. MATLAB (OCTAVE) FUNCTIONS AND CODE LISTING
==============================
Script to plot line shape and shift vs pressure.
%pressure_shift_profiles.m
%Script to plot line shape and shift vs pressure.
%Input line index in hitran output array a(,,,,,,)
%rhitran
v=23257; %the 822.92nm line in the HITRAN dataset
T=296;
mr=0.01;% volumetric mixing ratio of water vapour
%c=2.998e8;
M=18.0153/1000; %molar mass in Kg !!
P1=1;
P2=0.5;
P3=0.1;
S1=-1e7*a(v,7)*P1/(a(v,1)*(a(v,1)+a(v,7)*P1));
%convert shift in wavenumber (cm) at P1 to shift in wavelength (nm)
S2=-1e7*a(v,7)*P2/(a(v,1)*(a(v,1)+a(v,7)*P2));
S3=-1e7*a(v,7)*P3/(a(v,1)*(a(v,1)+a(v,7)*P3));
N=2.687e19*mr; % Loschmidt constant for water molecules per cmˆ3
Gd=a(v,1)*Dopplerw(T,M);
%Gd=0
Gl1=(296/T)ˆa(v,6)*P1*(a(v,4)*mr+a(v,3)*(1-mr)); %calculate the widths at the 3 pressures
Gl2=(296/T)ˆa(v,6)*P2*(a(v,4)*mr+a(v,3)*(1-mr));
Gl3=(296/T)ˆa(v,6)*P3*(a(v,4)*mr+a(v,3)*(1-mr));
p(1:401,1:5)=zeros;
p(1:401,1)=(-1:.005:1)/2; %wavenumber offset cm-1
for x= 1:401
p(x,2)= 1e7/(p(x,1)+a(v,1)); %wavelength nm
p(x,3)= N*P1*a(v,2)*Voigt(p(x,1),Gl1,Gd);
p(x,4)= N*P2*a(v,2)*Voigt(p(x,1),Gl2,Gd);
p(x,5)= N*P3*a(v,2)*Voigt(p(x,1),Gl3,Gd);
%Voigt line assuming no effects from adjacent lines
end
plot(p(1:401,2)+S1,p(1:401,3),p(1:401,2)+S2,p(1:401,4),p(1:401,2)+S3,p(1:401,5))
==============================
Script to plot error with altitude for fixed wavelength laser.
%p_alt_err0.m
%Script to plot error with altitude for fixed wavelength laser.
%Input line index in hitran output array a(,,,,,)
%rhitran
v=23257;
T=296;
mr=0.01;
%c=2.998e8;
M=18.0153/1000; %molar mass in Kg !!
H=7640; %scale height m
Po=1;
165
Gd=a(v,1)*Dopplerw(T,M);
p(1:444,1:5)=zeros;
p(1:444,1)=(0:40:17720); %waltitude m
for x= 1:444
p(x,2)=Po*exp(-p(x,1)/H); %calculate pressure from height
p(x,3)=(296/T)ˆa(v,6)*p(x,2)*(a(v,4)*mr+a(v,3)*(1-mr));
%Calculate the Lorentzian halfwidth from pressure
p(x,4)= a(v,7)*p(x,2); %Calculate the pressure shift in cm -1
l=Voigt(p(x,4),p(x,3),Gd); %Calculate the Voigt function at the shift
m=Voigt(0,p(x,3),Gd); %calculate the line center as measured at sea level
p(x,5)= 100*(l-m)/l; %calculate percentage error
end
plot(p(1:444,1),p(1:444,5)) %plot shift vs altitude
==============================
Script to calculate absorption cross sections at line centers and midway between linesusing Lorentzian approximation.
%acsv.m
%Script to calculate absorption cross sections at line centers and midway between
%lines using Lorentzian approximation. Requires lorentz.m function and
%rhitran.m script.
%rhitran
M=18.0153/1000;
mr=0.01 %mixing ratio
T=296
c=2.998e8;
gd=sqrt(2*log(2)*8.3145*T/M)/c;
n=size(a)
l(1:n(1)*2-2,1:2)=zeros;
for c=1:n(1)-1;
%c=23257
Gd=gd*a(c,1);
l(2*c-1,1)=0.01/a(c,1); %convert wavenumber to wavelength
m=(a(c,1)+a(c+1,1))/2; %current interpolated wavenumber
l(2*c,1)=0.01/m; %interpolate wavelength between points in cm-1
Yl1=(296/T)ˆa(c,6)*(a(c,4)*mr+a(c,3)*(1-mr));
Yl2=(296/T)ˆa(c+1,6)*(a(c+1,4)*mr+a(c+1,3)*(1-mr));
l(2*c-1,2) = a(c,2)*Voigt(0,Yl1,Gd); %calculate line center absorption cross-section
l(2*c,2) = a(c,2)*Voigt(a(c,1)-m,Yl1,Gd) + a(c+1,2)*Voigt(a(c+1,1)-m,Yl2,Gd);
%calculate interpolated absorption cross-section between lines (minimums)
end;
==============================
Script to plot line shape. Input line index in HITRAN output array.
166 APPENDIX G. MATLAB (OCTAVE) FUNCTIONS AND CODE LISTING
%chk.m
%Script to plot line shape. Input line index in hitran output array a(,,,,,,)
%rhitran
v=23257; %index
T=296;
mr=0.01;
%mr=0;
%c=2.998e8;
M=18.0153/1000; %molar mass in Kg !!
P1=1;
P2=0.5;
P3=0.1;
Gd=a(v,1)*Dopplerw(T,M);
%Gd=0
Gl1=(296/T)ˆa(v,6)*P1*(a(v,4)*mr+a(v,3)*(1-mr));
Gl2=(296/T)ˆa(v,6)*P2*(a(v,4)*mr+a(v,3)*(1-mr));
Gl3=(296/T)ˆa(v,6)*P3*(a(v,4)*mr+a(v,3)*(1-mr));
p(1:401,1:5)=zeros;
p(1:401,1)=(-1:.005:1)/2; %wavenumber offset cm-1
ll=a(v,1)-a(v-1,1);
lr=a(v+1,1)-a(v,1);
for x= 1:401
p(x,2)= P1*Voigt(p(x,1),Gl1,Gd);
p(x,3)= P2*Voigt(p(x,1),Gl2,Gd); %Voigt line assuming no effects from adjacent lines
p(x,4)= P3*Voigt(p(x,1),Gl3,Gd);
p(x,5)= 1e7/(p(x,1)+a(v,1)); %wavelength nm
end
plot(p(1:401,5)+0.001,p(1:401,2),p(1:401,5)+.0006,p(1:401,3),p(1:401,5),p(1:401,4))
==============================
Appendix H
Publications
H.1 Stabilized master laser system for differential absorp-tion lidar
• Dinovitser, A., Hamilton, M. W., and Vincent, R. A. (2010). Stabilized master laser systemfor differential absorption lidar. Applied Optics, 49:3274–+. http://adsabs.harvard.edu/abs/2010ApOpt..49.3274D
Stabilized master laser system for differentialabsorption lidar
Alex Dinovitser,* Murray W. Hamilton, and Robert A. VincentThe University of Adelaide, Adelaide, SA 5005, Australia
*Corresponding author: [email protected]
Received 2 March 2010; revised 11 May 2010; accepted 14 May 2010;posted 18 May 2010 (Doc. ID 124863); published 3 June 2010
Wavelength accuracy and stability are key requirements for differential absorption lidar (DIAL). We pre-sent a control and timing design for the dual-stabilized cw master lasers in a pulsed master-oscillatorpower-amplifier configuration, which forms a robust low-cost water-vapor DIAL transmitter system. Thisdesign operates at 823 nm for water-vapor spectroscopy using Fabry–Perot-type laser diodes. However,the techniques described could be applied to other laser technologies at other wavelengths. The systemcan be extended with additional off-line or side-line wavelengths. The on-linemaster laser is locked to thecenter of a water absorption line, while the beat frequency between the on-line and the off-line is locked to16 GHz using only a bandpass microwave filter and low-frequency electronics. Optical frequency stabi-lities of the order of 1 MHz are achieved. © 2010 Optical Society of AmericaOCIS codes: 140.3425, 280.1910.
1. Introduction
Differential absorption lidar (DIAL) is a spectro-scopic technique for measuring the distribution ofspecific trace gas species in the atmosphere. Watervapor is a key trace gas since a knowledge of the at-mospheric water-vapor concentration is vital formodeling meteorological phenomena and climate.It is now well known that water vapor plays a crucialrole in both radiative and convective energy transferthrough the atmosphere [1,2]. From the viewpoint ofgaining sufficient data for weather and climate mod-eling, the main difficulty with water measurementsin the atmosphere is the extreme and rapid variabil-ity of water concentrations compared to other gases[3]. Radiosondes are still the primary means of mea-suring atmospheric water vapor even though therecurrent costs are high, which limits the spatialand temporal extent of the data obtained. Despitemajor advances, satellite- and ground-based mea-surements based on spectral radiometry [4] and oc-cultation [5] techniques still offer limited verticalresolutions.
Differential absorption lidar has good accuracyand vertical resolution, and has the potential for de-velopment as a low-cost lidar, which would alleviatethe problems of horizontal and temporal resolution.Here we describe a laser transmitter system forDIAL, in the context of water vapor, but which is ap-plicable to the detection of many other species. InDIAL, laser pulses at two wavelengths, tuned so thatthey encounter different absorption cross sections forthe species being detected, are transmitted to the at-mosphere. The relative intensities of backscatteredlight at the two wavelengths depend on the concen-tration of the absorbing species. Typically, the on-linelaser is tuned to the center of a resonance, while theoff-line wavelength is tuned away from a resonancewhere the absorption cross section is small. If the ab-sorption cross sections are known, the concentrationbetween these two ranges can be deduced after mak-ing certain assumptions about extinction and scat-tering [6]. Lidar detection of trace gases usingRaman scattered light is an alternative method ofprofiling gas species in the atmosphere. However,DIAL can achieve a similar accuracy and detectionlimit with a power-aperture product that is morethan an order of magnitude smaller than Raman
0003-6935/10/173274-08$15.00/0© 2010 Optical Society of America
3274 APPLIED OPTICS / Vol. 49, No. 17 / 10 June 2010
[7]. This means that DIAL is more suitable for day-time applications, for power-critical space-basedapplications, and for cost-sensitive applications withlower power transmitters and smaller receivers. Onthe other hand, Raman lidar does offer the possibilityof simultaneous temperature measurement [8], andthe spectral purity and wavelength precision re-quirements on the laser are much less stringent thanis the case for DIAL.
A number of DIAL systems have been developed forthe profiling of water vapor [3,9–14]. Most of thesesystems rely on just one master laser, which isswitched between the two wavelengths. However, itis difficult to switch the laser wavelength accuratelybetween two wavelengths on a timescale of the orderof 1 ms. The alternative with a single laser is to runthe laser for several thousand pulses at one wave-length and then switch to the other wavelength,although this can lead to increased uncertainty ifthe concentration of the species being measured ischanging rapidly. A different approach is to use twolasers that are tuned to the relevant wavelengths[14–17]. This greatly simplifies the problem of main-taining the lasers at the required wavelengths.
The problem of maintaining one laser at the ab-sorption wavelength of the species of interest can besolved by either stabilizing the laser directly to theabsorption line using a reference cell [14,15,18], orby using a wavemeter [10,17]. Next, the second laserwavelength needs to be maintained at a fixed wave-length or frequency difference from the first. Infrequency terms, this difference should be at least10 GHz for tropospheric water vapor measurements.Again, some have opted to achieve this using a wave-meter [10,17]. However, it is also straightforward tocombine the two lasers and stabilize their beat fre-quency. In applications using extremely stable la-sers, the beat frequency can be phase locked to amicrowave reference oscillator, often an atomic clock[19] or a radio frequency oscillator [20]. In our appli-cation, extreme stability and phase coherence are notrequired. Methods of stabilizing the beat frequency tothe reference oscillator have been reported. Thesehave utilized a waveguide Mach–Zehnder interfe-rometer so that a fixed frequency difference betweenthe reference and the beat is achieved by locking toan interferometer fringe [21], or a microwave oscilla-tor and a mixer, to downshift the beat frequency, andthen using a low-pass electronic filter and powerdetector to generate a signal to stabilize the down-shifted beat signal to zero frequency [18].
Our system uses two cw semiconductor master la-sers, each continuously stabilized at their respectiveon-line and off-line wavelengths. Because this workis aimed at low-cost lidar, we have opted to stabilizethe on-line wavelength to a water absorption cell,rather than use a wavemeter, and we use a novelmethod of stabilizing the beat frequency betweenthe two lasers that does not require a microwave os-cillator or a mixer, giving a significant reduction incost. Because the lasers are operated in cw mode,
acousto-optic switching is used to produce the laserpulses.
Semiconductor laser technology is also suited forthe optical amplifier in a low-cost low-power system.The disadvantages of relatively low output power canbe partly offset by having quite large pulse repetitionrates of up to ∼10 kHz. When the absorption line istoo strong, a laser tuned to the side of the line can beemployed [15]. This side-line wavelength can be sta-bilized at a precise offset from the center of the linewith a straightforward extension of the techniquesdescribed in this paper. A simple calculation assum-ing a Lorentzian absorption line with width 1 GHz(HWHM) shows that the fractional change in absorp-tion cross section is about 0.1% if the laser fluctuatesby 3 MHz when tuned to the steepest part of the line.In both [15] and this work, a stability of better than3 MHz is achieved.
2. DIAL Transmitter
Our DIAL master laser system uses two cw 40 mWHitachi HL8325G laser diodes, operating near823 nm. Pulses for amplification and transmissionto the atmosphere are formed by acousto-optic mod-ulators (AOMs). The remaining light (that is comple-mentary to the pulsed beams) is used for wavelengthstabilization of the master lasers. One laser is servo-locked to the wavelength of the peak of a water ab-sorption line. The second is maintained at a fixedwavelength offset from the first by combining thetwo laser beams and stabilizing the frequency ofthe beat signal to the peak transmission of a micro-wave (RF) bandpass filter.
Figure 1 provides a simplified diagram of the entirelidar transmitter system. After passing through theAOM, each undeflected beam is coupled into a sin-gle-mode optical fiber. The pulsed beam that is Braggscattered by the acoustic wave is directed to a beamsplitter used as a combiner and then to the optical am-plifier. Using the Bragg scattered wave, rather thanthe zeroth-order “straight-through” beam, as the ba-sis for the pulse transmitted to the atmosphere en-sures that the pulse from the master returns tozero. An added benefit of using the Bragg scatteredbeam in this way is that the AOM then contributesto the isolation of the master lasers from backreflec-tions from the optical amplifier. The acoustic wave ineachAOM is repetitively pulsed on for 1 μswith a per-iod of 667 μs. The pulse length, which determines thetransmitted pulse energy, is chosen as a trade-off be-tween signal-to-noise and range resolution in the li-dar return. This pulse width and period correspondto a range resolution of 150 m and a maximum rangeof 100 km. The effective maximum vertical range isvery much less than this because of the relativelylow transmitted pulse energy (∼500 nJ). Indeed thedata acquisition system only records for 50 μs afterthe pulse is transmitted, corresponding to a maxi-mum range of about 7 km. Vertical resolution couldbe improved by reducing the pulse width, but thiswould reduce pulse energy and signal-to-noise ratio,
10 June 2010 / Vol. 49, No. 17 / APPLIED OPTICS 3275
as well as compromising themaximum vertical rangebecause of background light. The pulse repetition ratecould be considerably higher without introducingrange ambiguities, but in our case is limited by thedata acquisition system.
To achieve reasonably reproducible tuning charac-teristics and longitudinal mode stability, light fromeach laser diode passes through two Faraday isola-tors, together providing 60 dB of optical isolation.Each laser operates in a single longitudinal modeand provides a mode-hop-free tuning range of morethan 200 GHz. They operate in several discrete re-gions over awavelength band from820 to834 nm.De-vice selection is necessary to have two lasers that tuneto the samewavelength range.Of 15devices,we foundthree that would tune continuously, and repeatably,over a range from 822.6 to 823:4 nm. The ability totune over this range has remained unchanged overthree years, and only requires that the diode tempera-ture and current are stabilizednearparticular values.The tuning of these diodes is subject to hysteresiswhenmode hops occur. Because of this, it is necessarythat the desired temperature and current be ap-proached in a certain direction. In establishing theseproperties, an optical spectrum analyzer proves use-ful. The optical amplifier is a Sachertechnik TA830tapered amplifier with injection current pulsed syn-chronously with the arrival of themaster laser pulsesof alternatingwavelength. Its ratedmaximumoutputpower is 500 mW.
A. On-Line Master Laser Control System
The zeroth-order on-line laser beam after the AOM iscoupled into a single-mode optical fiber, and then to a50∶50 fiber coupler, which serves as both a splitterand a combiner. An important function of the fiberis to isolate the alignment of the beams around theAOMs and isolators from alignment at the absorptioncell. One output port of the coupler directs the light toamultipass absorption cell with a 30 mpath length, aratiometric detector, and a feedback system to controllaser current and wavelength. Our multipass cell is ahomemade variant of a Herriot cell, where the light iscoupled into the space between themirrors via a smallperiscope. After 66 traversals of the space betweenthemirrors, the beam encounters the periscope againand is coupled out of the cell.
A dither modulation at 1:5 kHz is introduced viathe on-line laser current control so that the laser hasa peak-to-peak frequency modulation of 500 MHz.Phase-sensitive detection of the resulting amplitudemodulation at the output of the vapor cell, when thelaser is tuned through awater resonance, provides anerror signal in order to keep the laser locked to thewater resonance peak. However, there are unwantedsources of amplitude modulation with changing laserwavelength. First, the dither of laser injection currentalso modulates the laser power. Second, themultiple-beam interference fringes due to reflections withinthe fiber splitter have a greater contrast than manywater absorption lines. By comparing the optical
Fig. 1. Differential absorption lidar laser stabilization system showing optical paths (solid lines) and electronic signals (dashed lines).AOM, acousto-optic modulator; SOA, semiconductor optical amplifier; ISO, Faraday optical isolator; DAQ, data acquisition; PD, photo-diode; LPF, low-pass filter; BPF, bandpass filter; TP, test point. The blocks labeled TEST indicate the points at which step voltages are usedto test the response of the servo loops, as discussed in the text. The gray circles represent fiber coupling lenses.
3276 APPLIED OPTICS / Vol. 49, No. 17 / 10 June 2010
power entering the vapor cell with the power trans-mitted through it, the ratiometric detector eliminatesboth of these sources of error. In the output of theratiometric detector, these variations are reduced tobelow the level of the electronic noise.
Two other less significant sources of wavelengtherror have not been dealt with. First, there is somemultiple-beam interference due to the water-vaporcell itself, because there is a small amount of overlapof the laser spots on the mirrors. Some light at themirrors is scattered into paths that effectively short-cut one or more of the passes in the cell, and this ismanifested as fringes at the output of the ratiometricdetector, with a free spectral range corresponding tothe mirror separation. In our case, these fringes aremore than an order of magnitude weaker than theabsorption line at 822:9 nm that we use, and causenegligible offset error in the stabilization to the cen-ter of the water line. For weaker lines, however, thiscould be a major concern. Second, the pulses trans-mitted to the sky are shifted in frequency from theabsorption line center by the 80 MHz acoustic fre-quency of the AOM. In the lower troposphere, wherethe water absorption lines are at least 2 GHz wide(HWHM), this introduces a negligible systematicerror for our application (<0:3% difference betweenthe peak and actual absorption cross sections).
To minimize the cost and component count of thesystem, we use a single fiber coupler. Thismeans thatthe off-line laser also passes through the absorptioncell. However, because its wavelength is not modu-lated, it does not affect the on-line control system.Although a 16 GHz beat signal between the on-lineand off-line lasers is also present at the water-vaporcell, the low-speed photodiodes at the absorption celldo not respond to modulation at this frequency.
B. Off-Line Master Laser Control System
The off-line master laser is isolated, passed throughan AOM, and coupled into the fiber coupler in thesame way as the on-line laser. The second outputfrom the fiber coupler goes to a photodiode (NewFocus Model 1481-S) which has a frequency responsefrom DC to 25 GHz. This diode detects the beat fre-quency between the two lasers. We have chosen afixed frequency offset between the on- and off-line la-sers of 16 GHz. Since the on-line master laser al-ready includes a dither, the beat signal around16 GHz also has a frequency modulation amplitudeof 500 MHz at 1:5 kHz. The bandpass filter, with acenter frequency of 16 GHz and a −3 dB bandwidthof 500 MHz, converts this frequency modulation toan amplitude modulation whose phase depends onwhich side of the filter’s transfer function the beatis tuned to. The microwave power is detected by atunnel diode detector (Herotek DT2018), which hasa bandwidth much greater than 1:5 kHz, so thatphase sensitive amplification of the dither compo-nent of the microwave power can be used to generatean error signal. This is integrated and fed back to theoff-line laser injection current controller. Thus, the
off-line control system locks the frequency differenceof the lasers to the zero crossing of the derivative ofthe bandpass filter’s transfer function, in a similarway to the stabilization used for the on-line laser.
There are two lock points for the off-line laser, oneon each side of the absorption line. These are easilydistinguishable when setting up the system. Notethat there are only two such points, in contrast to[18], where there are four, because we dispense witha microwave oscillator.
Obtaining a different frequency offset (from16 GHz) between the lasers is simply a matter ofpicking a bandpass filter with passband centered atthe desired offset frequency. The bandwidth shouldbe reasonably narrow, and the filter transfer functionshould not have ripple or a wide flat region in thepassband. Of course, the photodiode that detectsthe beat signal needs to have a sufficiently largebandwidth—this will be the limiting factor in prac-tice. Photodiodes with bandwidth of up to 100 GHzare obtainable. The price of microwave componentstends to increase with increasing frequency, andwe have chosen the value 16 GHz as a compromisebetween cost and the need to make sure that theoff-line laser is tuned sufficiently far into the wingsof the water absorption line.
3. System Timing
For water vapor in the lower troposphere, the line-width of the molecular resonance at 822:9 nm isdominated by pressure broadening and ranges fromabout 5 GHz at sea level to about 2 GHz at the high-est altitude at which we expect to be able to measurewater-vapor concentrations with a transmitted pulseenergy of 500 nJ (about 4 km). The smaller of theselinewidth values imposes a constraint on the line-width of the on-line master laser, which should beless than 100 MHz [22]. However, the amplitude ofthe frequency dither applied to the on-line masterlaser is 500 MHz. To get around this difficulty, wesynchronize the extraction of the optical pulses bythe AOMs with the zero crossings of the dither signal(Fig. 2). This ensures that the laser frequency withinthe pulses is consistent from pulse to pulse. Providedthe pulse is short compared to the dither period, thefrequency chirp within the pulse can be neglected. Inour system, the pulse length is 1 μs and the ditherperiod is 1333 μs, resulting in an optical frequencychirp of less than 1 MHz.
The output pulse can be set to any phase of thedither signal, but there is an advantage to keepingit close to the zero crossing. When the optical pathis switched out of the cw beam by the AOM to formthe output pulse, a transient will appear on theoutput of the ratiometric detector, which, althoughshort in duration, will disturb the feedback controlloop at the integrator. If this transient is timed to co-incide with the zero crossing of the dither signal, thetransient will not be propagated through the controlsystem. In other words, the transient impulse due tooptical switching will be attenuated by the lock-in
10 June 2010 / Vol. 49, No. 17 / APPLIED OPTICS 3277
(which is an analogue multiplier), if it coincides withan instantaneous zero reference.
4. Wavelength Stability Measurements
Here we present the wavelength stability, which isone of the critical characteristics for any DIAL sys-tem, as well as show the transient behavior of thesystem. Some DIAL measurements are also pre-sented to demonstrate the reliability of the stabiliza-tion system.
To measure the residual optical frequency fluctua-tions, our procedure is as follows. First we apply astep perturbation to the on-line laser while measur-ing the change in the beat frequency, with the feed-back loops open. This enables us to relate the magni-tude of the error signals observed at TP1 and TP2(see Fig. 1) to the size of the frequency excursion.
Next, the feedback loops are closed and the noiseat TP1 and TP2, as shown in Fig. 3, is measured.In this figure, the conversion relating voltage to fre-quency has been applied so that the vertical scalerepresents the frequency excursion. From the mea-surement at TP2, we obtain a relative frequencyvariability (RMS) between the lasers of 1:2 MHz,and from that at TP1, an RMS value of 0:7 MHzfor the fluctuation in the frequency of the on-linelaser.
Here we measure noise that has frequency compo-nents that are mostly greater than the reciprocal ofthe feedback loop time constants (both ∼1 s). Fre-quency fluctuations on a time scale significantlyshorter than 1 s are uncorrected by the feedback.Thus, the noise at TP2 is the sum of uncorrelatedcontributions from both the on-line and off-linelasers. There are also contributions from the electro-nics. These time constants were gauged from the
Fig. 2. System timing coordinates the optical switching (AOM), electronic switching, dither, and data acquisition. The adjustable delays(Td) compensate for the response times of the electronic and optical components, such as the finite time for acoustic signals to propagatethrough the AOMs (∼400 ns) to effect the optical switching. In addition, the delays ensure that the data acquisition system is triggered∼5 μs before the laser pulses are transmitted, to enable background signal levels to be measured. The pulse lengths are controlled bymonostable multivibrators (HC4538), that are not shown explicitly.
Fig. 3. Laser wavelength variability measured from the error sig-nals at TP1 and TP2, with the feedback loops closed. TP1 showsthe fluctuations in the on-line laser frequency, and TP2 showsthose for the difference frequency. The signals shown are the vol-tages measured at the respective test points, scaled as described inthe text so that the vertical axis is in frequency units. The verticalseparation between the traces is an arbitrary DC shift introducedfor clarity.
Fig. 4. Response of the control loops to small-signal step pertur-bation at the on-line laser. The response of the on-line control sys-tem acts to cancel the effect of the perturbation as the area underthe error-signal curve is equal to the step amplitude. The instan-taneous shift in the beat frequency also produces an error signal inthe off-line control signal.
3278 APPLIED OPTICS / Vol. 49, No. 17 / 10 June 2010
transient behavior of the error signals when a stepperturbation was applied to the on-line laser, whileboth feedback loops were closed. A step voltage in-jected directly into the on-line laser diode injectioncurrent controller, at one of the points labeled TESTin Fig. 1, produced an ∼50 MHz instantaneous fre-quency shift in the on-line master laser and the beatnote simultaneously. The response of both controlsystems acts to cancel the frequency shift. The re-sponses at the test points TP1 and TP2 to a pertur-bation at the on-line laser are illustrated in Fig. 4.The relaxation times for the signals are both ∼1 s,and these are the loop time constants.
The off-line control system responds to the beatfrequency shift, even though there is no perturbationto the off-line wavelength; however, the output fromthe off-line integrator returns to its original value. Itis interesting to note in Fig. 4 that, even though theoptical and RF beat frequency perturbation are inthe same direction, the instantaneous responses ofthe two control systems are in opposite directions.This is because the integration constants of thetwo control loops are of the opposite sign becausethe water absorption behaves like a band-stop filter,while the 16 GHz filter is a bandpass filter.
The effect of multiple-beam interference in the op-tical path between the on-line laser and the photode-tector after the multipass absorption cell is tosuperpose small interference fringes on the water ab-sorption profile. Such interference is rather tempera-ture sensitive. This can give rise to an offset in thelocking point for the on-line laser, which, in turn, con-tributes to the fluctuations of the beat frequency be-tween the two lasers, on a time scale of severalseconds. We observe these fluctuations when theoff-line laser is unlocked and the on-line laser islocked; when both loops are locked, they are presentbut not directly visible. Such fluctuations are of a si-milar order of magnitude to the RMS fluctuationsshown in Fig. 3, i.e., about 1 MHz. Thus, this sourceof error is reduced to an acceptable level, also.
An important point is that these results were ob-tained at a sampling rate of 100 samples=s, capturingonly the low-frequency perturbations up to 50 Hz.Faster frequency fluctuations were also present butat a much lower level, evidenced by measurementsof the 16 GHz beat spectrum. A full measurementof the laser spectra is required to characterize the ra-pid fluctuations that contribute to the spectral wings.This knowledge is critical in DIAL because, if thespectral wings of the on-line laser extend well intoand beyond the wings of the absorption spectrum,the effective absorption cross section is reduced fromthat which would we obtain for a perfectly monochro-matic laser tuned to line center. Such a reductionleads to a systematic error in the retrieved water con-centration. This can be the case even if the width athalf-maximum of the laser spectrum is much lessthan that of the absorption spectrum, as the formermay have significantly more power in the far wingsthan Lorentzian-or Gaussian-shaped spectra. Thisis especially true of diode lasers, and others that havevery broad gain bandwidths.
To illustrate the problem, we show in Table 1 mea-surements of the maximum fractional absorption inthe multipass cell as the on-line master laser (withand without amplification) is tuned through the lineat 822:92 nm. The calculated absorption is thatwhich would be expected based on a measurementof the humidity by a capacitive sensor that was cali-brated against the saturated vapor pressure over asaturated salt solution (magnesium nitrate) [23,24].The relative humidity (RH) was 54% and the tem-perature was 296 K. The absorption cross sectionis deduced from the HITRAN database [25]. The ab-
Fig. 5. (Color online) (a) Measurements of water vapor mixing ratio (in grams per kilogram) using the DIAL, and (b) the raw backscattersignal for the off-line wavelength, showing the height to which aerosol scattering is seen. In (a), the arrows on the left-hand scale indicatethe approximate boundaries of height range for which there is a signal-to-noise ratio greater than 1. In (b), the arrow on the scale indicatesthe height of maximum signal, which is the point at which the transmitted beam fully intersects the field of view of the receiving telescope.
Table 1. Maximum Fractional Absorption for Water at 822:92 nmwith Relative Humidity 54%
Calculated from RH Measured
Amplified 0.1279 0.1224Master only 0.1282 0.1301
10 June 2010 / Vol. 49, No. 17 / APPLIED OPTICS 3279
sorption cross section of the master laser is slightlylarger than expected, but this is within the measure-ment uncertainty. The absorption cross section of theamplified light is about 5% less than expected, andthis reduction is attributable to the amplified spon-taneous emission from the optical amplifier, and theresulting lack of spectral purity. The two values forexpected absorption are different because the tem-perature and humidity changed slightly betweenthe measurements. Such a measurement can bemade straightforwardly each time the lidar is oper-ated, in order to establish the effective on-line ab-sorption cross section.
As a demonstration that this master laser systemworks reliably, we include DIAL measurements inFig. 5. These data were obtained with the lidar point-ing at the zenith from the Adelaide central businessdistrict, over two hours on the evening of 22 March2010. The time shown is local time. Figure 5(a) showsthe water-vapor mixing ratio in grams per kilogramand Fig. 5(b) shows the base-ten logarithm of raw re-turn signal for the off-line wavelength. The data wereoversampled in time at 50 ns intervals by the data ac-quisition system, and averaging over 500 ns has beenapplied, corresponding to 75 m intervals on the verti-cal scale. (The transmitted pulse width is 1 μs, so thatthe true range resolution is 150 m.) The data are alsoaveraged over 5 min intervals, in which there wereapproximately 1:5 × 105 transmitted pulses at eachwavelength. In Fig. 5(a), the signal to noise falls to1, above about 700 m. Figure 5(b) partly explains thisrather lowmaximum range; the return from the aero-sols has fallen to the background signal level by about1 km. The on-line signal falls somewhat faster, due toabsorption, and when this signal has fallen to thebackground, meaningful measurement of mixing ra-tio is no longer possible. This shallow depth of theaerosol scatterers is typical for this location. The ver-tical band of low signal in Fig. 5(b), at about 2210 h,was caused by the alignment between master andslave drifting. This was readjusted without affectingthe lock of the master laser system. There is a corre-sponding “hole” of noisy data in Fig. 5(a) at about thesame time. The receiving telescope had a diameter of40 cm, and the on-line wavelength was 822:92 nm.
5. Conclusion
We describe a low-cost master laser system for DIALusing a separate master laser for each wavelength. Asynchronous dither and timing system locks thewavelength of the transmitted pulse to the desiredwavelength. The wavelength stability shows a RMSvariability of less than 1 MHz over 10 min. The sys-tem design lends itself to other types of laser diodesand optical amplifiers, and also to multiple off-linewavelengths.
This work was funded by the Australian ResearchCouncil and the Australian Bureau of Meteorology.Contributions by K. Bae, C. Baer, A. Heitmann,and P. Moran are gratefully acknowledged.
References
1. K. Minschwaner and A. E. Dessler, “Water vapor feedback inthe tropical upper troposphere: model results and observa-tions,” J. Climate 17, 1272–1282 (2004).
2. W. Ingram, “On the robustness of the water vapor feedback:GCM vertical resolution and formulation,” J. Climate 15,917–921 (2002).
3. S. Pal, A. Behrendt, H. Bauer, M. Radlach, A. Riede, M.Schiller, G. Wagner, and V. Wulfmeyer, “3-dimensional obser-vations of atmospheric variables during the field campaignCOPS,” IOP Conf. Ser.: Earth Environ. Sci. 1, 012031 (2008).
4. V. Wulfmeyer, H. Bauer, P. Di Girolamo, and C. Serio, “Com-parison of active and passive water vapor remote sensing fromspace: an analysis based on the simulated performance of IASIand space borne differential absorption lidar,” Remote Sens.Environ. 95, 211–230 (2005).
5. S. Heise, J. Wickert, G. Beyerle, T. Schmidt, and Ch. Reigber,“Global monitoring of tropospheric water vapor with GPSradio occultation aboard CHAMP,” Adv. Space Res. 37,2222–2227 (2006).
6. A. Ansmann and J. Bosenberg, “Correction scheme for spectralbroadening by Rayleigh scattering in differential absorptionlidar measurements of water vapor in the troposphere,” Appl.Opt. 26, 3026–3032 (1987).
7. V. Wulfmeyer and J. Bosenberg, “Ground-based differentialabsorption lidar for water-vapor profiling: assessment of accu-racy, resolution, and meteorological applications,” Appl. Opt.37, 3825–3844 (1998).
8. A. Behrendt, T. Nakamura, M. Onishi, R. Baumgart, and T.Tsuda, “Combined Raman lidar for the measurement of atmo-spheric temperature, water vapor, particle extinction coeffi-cient, and particle backscatter coefficient,” Appl. Opt. 41,7657–7666 (2002).
9. D. Bruneau, P. Quaglia, C. Flamant, M. Meissonnier, and J.Pelon, “Airborne lidar LEANDRE II for water-vapor profilingin the troposphere. I. System description,” Appl. Opt. 40,3450–3461 (2001).
10. A. R. Nehrir, K. S. Repasky, J. L. Carlsten, M. D. Obland, andJ. A. Shaw, “Water vapor profiling using a widely tunable, am-plified diode-laser-based differential absorption lidar (DIAL),”J. Atmos. Ocean. Technol. 26, 733–745 (2009).
11. J. L. Machol, T. Ayers, K. T. Schwenz, K. W. Koenig, R. M.Hardesty, C. J. Senff, M. A. Krainak, J. B. Abshire, H. E. Bravo,and S. P. Sandberg “Preliminary measurements with an auto-mated compact differential absorption lidar for the profiling ofwater vapor,” Appl. Opt. 43, 3110–3121 (2004).
12. H. Schwarzer, A. Börner, A. Fix, B. Günther, H-W. H., M.Raugust, F. Schrandt, and M. Wirth, “Development of a wave-length stabilized seed laser system for an airborne watervapour lidar experiment,” Proc. SPIE 6681, 66810H (2007).
13. E. V. Browell, S. Ismail, and W. B. Grant, “Differential absorp-tion lidar (DIAL) measurements from air and space,” Appl.Phys. B 67, 399–410 (1998).
14. M. Wirth, A. Fix, P. Mahnke, H. Schwarzer, F. Schrandt, andG. Ehret, “The airborne multi-wavelength water vapor differ-ential absorption lidar WALES: system design and perfor-mance,” Appl. Phys. B 96, 201–213 (2009).
15. G. J. Koch, J. Y. Beyon, F. Gibert, B. W. Barnes, S. Ismail,M. Petros, P. J. Petzar, J. Yu, E. A. Modlin, K. J. Davis, andU. N. Singh, “Side-line tunable laser transmitter for differen-tial absorption lidar measurements of CO2: design and appli-cation to atmospheric measurements,” Appl. Opt. 47, 944–956(2008).
16. K. Ertel, H. Linné, and J. Bösenberg, “Injection seeded pulsedTi:sapphire laser with novel stabilization scheme and capabil-ity of dual-wavelength operation,” Appl. Opt. 44, 5120–5126(2005).
3280 APPLIED OPTICS / Vol. 49, No. 17 / 10 June 2010
17. H. R. Khalesifard, A. Fix, G. Ehret, M. Schiller, and V.Wulfmeyer, “Fast-switching system for injection seeding of ahigh-power Ti:sapphire laser,” Rev. Sci. Instrum. 80, 073110(2009).
18. S. Schilt, R. Matthey, D. Kauffmann-Werner, C. Affolderbach,G. Mileti, and L. Thévenaz, “Laser offset-frequency locking upto 20 GHz using a low-frequency electrical filter technique,”Appl. Opt. 47, 4336–4344 (2008).
19. L. Hollberg, S. Diddams, A. Bartels, T. Fortier, and K. Kim,“The measurement of optical frequencies,” Metrologia 42,S105–S124 (2005).
20. J. Appel, A. MacRae, and A. I. Lvovsky, “A versatile digitalGHz phase lock for external cavity diode lasers,” Meas. Sci.Technol. 20, 055302 (2009).
21. U. Schünemann, H. Engler, R. Grimm, M. Weidemüller,and M. Zielonkowski, “Simple scheme for tunable frequencyoffset locking of two lasers,” Rev. Sci. Instrum. 70, 242–243(1999).
22. V. Wulfmeyer and C. Walther, “Future performance of ground-based and airborne water-vapor differential absorption lidar.II. Simulations of the precision of a near-infrared, high-powersystem,” Appl. Opt. 40, 5321–5336 (2001).
23. World Meteorological Organization, “Measurement of humid-ity,” in Guide to Meteorological Instruments and Methods ofObservation (World Meteorological Organization, 2008), pp.1.4–25.
24. A. Apelblat, “The vapour pressures of water over saturatedaqueous solutions of barium chloride, magnesium nitrate, cal-cium nitrate, potassium carbonate, and zinc sulfate, at tem-peratures from 283 K to 313 K,” J. Chem. Thermodyn. 24,619–626 (1992).
25. L. S. Rothman, I. E. Gordon, A. Barbe, D. C. Benner, P. F.Bernath, M. Birk, V. Boudon, L. R. Brown, A. Campargue,J.-P. Champion, K. Chance, L. H. Coudert, V. Danaj, V. M.Devi, S. Fally, J.-M. Flaud, R. R. Gamache, A. Goldmanm,D. Jacquemart, I. Kleiner, N. Lacome, W. J. Lafferty, J.-Y.Mandin, S. T. Massie, S. N. Mikhailenko, C. E. Miller, N.Moazzen-Ahmadi, O. V. Naumenko, A. V. Nikitin, J. Orphal,V. I. Perevalov, A. Perrin, A. Predoi-Cross, C. P. Rinsland,M. Rotger, M. Šimečková, M. A. H. Smith, K. Sung, S. A.Tashkun, J. Tennyson, R. A. Toth, A. C. Vandaele, and J.Vander Auwera, “The HITRAN 2008 molecular spectroscopicdatabase,” J. Quant. Spectrosc. Radiat. Transfer 110, 533–572(2009).
10 June 2010 / Vol. 49, No. 17 / APPLIED OPTICS 3281
176 APPENDIX H. PUBLICATIONS
H.2 Transmitter design for differential absorption watervapour LIDAR
• Dinovitser, A., Hamilton, M. W., and Vincent, R. A. (2009). Transmitter design for differ-ential absorption water vapour lidar. In Proceedings of the 8th International Symposium onTroposheric Profiling: Integration of Needs, Technologies and Applications, volume S13 -P09 of Profiling of water vapor and temperature. ftp://ftp.sma.ch/outgoing/dcr/ISTP/ABSTRACTS/data/1787618.pdf
A NOTE:
This publication is included on pages 178-181 in the print copy of the thesis held in the University of Adelaide Library.
A Dinovitser, A., Hamilton, M. W., & Vincent, R. A. (2009) `Transmitter design for differential absorption water vapour lidar', in Proceedings of the 8th International Symposium on Tropospheric Profiling: Integration of Needs, Technologies and Applications, volume S13 - P09 of Profiling of water vapor and temperature, Delft, The Netherlands
182 APPENDIX H. PUBLICATIONS
H.3 Towards low-cost water-vapour differential absorp-tion lidar
• Hamilton, M. W., Dinovitser, A., and Vincent, R. A. (2009). Towards low-cost water-vapour differential absorption lidar. In Proceedings of the 8th International Symposium onTroposheric Profiling: Integration of Needs, Technologies and Applications, volume S13 -O02 of Profiling of water vapor and temperature. ftp://ftp.sma.ch/outgoing/dcr/ISTP/ABSTRACTS/data/1662333.pdf
A Hamilton, M. W., Dinovitser, A., & Vincent, R. A. (2009) `Towards low-cost water vapour differential absorption lidar', in Proceedings of the 8th International Symposium on Tropospheric Profiling: Integration of Needs, Technologies and Applications, volume S13 - 002 of Profiling of water vapor and temperature, Delft, The Netherlands
A NOTE:
This publication is included on pages 184-187 in the print copy of the thesis held in the University of Adelaide Library.
188 APPENDIX H. PUBLICATIONS
H.4 Towards low-cost water-vapour differential absorp-tion lidar
• Hamilton, M., Atkinson, R., Dinovitser, A., Peters, E., and Vincent, R. A. (2008). Towardslow-cost water-vapour differential absorption lidar. In Society of Photo-Optical Instru-mentation Engineers (SPIE) Conference Series, volume 7153 of Society of Photo-OpticalInstrumentation Engineers (SPIE) Conference Series. http://adsabs.harvard.edu/abs/2008SPIE.7153E...6H
A Hamilton, M., Atkinson, R., Dinovitser, A., Peters, E., & Vincent, R. A. (2008). `Towards low-cost water-vapour differential absorption lidar', in U.N. Singh, K. Asai & A. Jayarman (eds) Lidar Remote Sensing for Environmental Monitoring IX, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, volume 7153, Noumea, New Caledonia
NOTE:
This publication is included on pages 190-198 in the print copy of the thesis held in the University of Adelaide Library.
It is also available online to authorised users at:
http://dx.doi.org/10.1117/12.804740
Bibliography
Abrarov, S., Quine, B., and Jagpal, R. (2010). Rapidly convergent series for high-accuracycalculation of the voigt function. Journal of Quantitative Spectroscopy and Radia-tive Transfer, 111(3):372 – 375. http://www.sciencedirect.com/science/article/B6TVR-4X7YNPK-2/2/9f7c12bf4acc9d155867caf3ff21f069.
Amediek, A., Fix, A., Wirth, M., and Ehret, G. (2008). Development of an OPO system at 1.57µm for integrated path DIAL measurement of atmospheric carbon dioxide. Applied Physics B:Lasers and Optics, 92:295–302. http://dx.doi.org/10.1007/s00340-008-3075-6.
Ansmann, A. (1985). Errors in ground-based water-vapor dial measurements due to doppler-broadened rayleigh backscattering. Applied Optics, 24(21):3476–3480. http://adsabs.harvard.edu/abs/1985ApOpt..24.3476.
Aumann, H., Chahine, M., Gautier, C., Goldberg, M., Kalnay, E., McMillin, L., Revercomb,H., Rosenkranz, P., Smith, W., Staelin, D., Strow, L., and Susskind, J. (2003). AIRS/AM-SU/HSB on the Aqua mission: design, science objectives, data products, and processing sys-tems. Geoscience and Remote Sensing, IEEE Transactions on, 41(2):253 – 264. http://dx.doi.org/10.1109/TGRS.2002.808356.
Baron, P., Mendrok, J., Dupuy, E., and Kasai, Y. (2008). Tropospheric water vapor retrieval froma nadir THz/FIR sounder. In Larar, A. M., Lynch, M. J., and Suzuki, M., editors, Proc. SPIE,,volume 7149, page 714909. SPIE. http://link.aip.org/link/?PSI/7149/714909/1.
Behrendt, A., Wagner, G., Petrova, A., Shiler, M., Pal, S., Schaberl, T., and Wulfmeyer,V. (2005). Modular lidar systems for high-resolution 4-dimensional measurements of wa-ter vapor, temperature, and aerosols. In U. N. Singh & K. Mizutani, editor, Society ofPhoto-Optical Instrumentation Engineers (SPIE) Conference Series, volume 5653 of Soci-ety of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, pages 220–227.http://adsabs.harvard.edu/abs/2005SPIE.5653..220B.
Berland, B. (2003). Photovoltaic Technologies Beyond the Horizon: Optical Rectenna Solar Cell,Final Report, 1 August 2001-30 September 2002. National Renewable Energy Laboratory.http://www.nrel.gov/docs/fy03osti/33263.pdf.
Black, E. D. (2001). An introduction to Pound–Drever–Hall laser frequency stabilization. Ameri-can Journal of Physics, 69(1):79–87. http://link.aip.org/link/?AJP/69/79/1.
Bohren, C. F. and Huffman, D. R. (1998). Absorption and Scattering of Light by Small Particles.John Wiley & Sons inc. http://adsabs.harvard.edu/abs/1998asls.book.....B.
Bureau Of Meteorology (BOM) (2011). South Australia’s improving weather services. http://www.bom.gov.au/NexGenFWS/.
Bosenberg, J. (1998). Ground-based differential absorption lidar for water-vapor and temperatureprofiling: Methodology. Appl. Opt., 37(18):3845–3860. http://ao.osa.org/abstract.cfm?URI=ao-37-18-3845.
199
200 BIBLIOGRAPHY
Braun, W. C. (1985). Simplified calculations for accuracy of a lidar dial system to measure at-mospheric H2O vapor and temperature. Appl. Opt., 24(1):109–117. http://ao.osa.org/abstract.cfm?URI=ao-24-1-109.
Browell, E. V., Ismail, S., and Ferrare, R. A. (2000). Lase water vapor, aerosol, and cloud mea-surements during recent field experiments. In Tenth ARM Science Team Meeting Proceedings.
Browell, E. V., Ismail, S., and Grant, W. B. (1998). Differential absorption lidar (DIAL) mea-surements from air and space. Applied Physics B: Lasers and Optics, 67:399–410. http://dx.doi.org/10.1007/s003400050523.
Browell, E. V., Ismail, S., and Grossmann, B. E. (1991). Temperature sensitivity of differentialabsorption lidar measurements of water vapor in the 720-nm region. Appl. Opt., 30(12):1517–1524. http://ao.osa.org/abstract.cfm?URI=ao-30-12-1517.
Browell, E. V., Wilkerson, T. D., and Mcilrath, T. J. (1979). Water vapor differential absorp-tion lidar development and evaluation. Appl. Opt., 18(20):3474–3483. http://ao.osa.org/abstract.cfm?URI=ao-18-20-3474.
Bruneau, D., Quaglia, P., Flamant, C., Meissonnier, M., and Pelon, J. (2001a). Airborne LidarLEANDRE II for Water-Vapor Profiling in the Troposphere. I. System description. AppliedOptics, 40:3450–3461. http://adsabs.harvard.edu/abs/2001ApOpt..40.3450B.
Bruneau, D., Quaglia, P., Flamant, C., and Pelon, J. (2001b). Airborne Lidar LEANDRE II forWater-Vapor Profiling in the Troposphere. II. First results. Applied Optics, 40:3462–3475.http://adsabs.harvard.edu/abs/2001ApOpt..40.3462B.
Buck, A. L. (1981). New equations for computing vapor pressure and enhancement factor. Journalof Applied Meteorology, 20:1527–1532. http://adsabs.harvard.edu/abs/1981JApMe..20.1527B.
Bykov, A., Lavrentieva, N., Mishina, T., Sinitsa, L., Barber, R., Tolchenov, R., and Tennyson,J. (2008). Water vapor line width and shift calculations with accurate vibration-rotationwave functions. Journal of Quantitative Spectroscopy and Radiative Transfer, 109(10):1834– 1844. http://www.sciencedirect.com/science/article/B6TVR-4RJYVDN-1/2/
90ec60c20b5577cee81532e1309896ad.
Cahen, C., Megie, G., and Flamant, P. (1982). Lidar monitoring of the water vapor cycle in thetroposphere. Journal of Applied Meteorology, 21(10):1506–1515. http://adsabs.harvard.edu/abs/1982JApMe..21.1506C.
Calpini, B., Haefele, A., Renaud, L., Ruffieux, D., Dinoev, T., and Simeonov, V. (2011). TheRaman Lidar at Payerne, a new perspective. http://www.multimedia.ethz.ch/episode_play/?doi=10.3930/ETHZ/AV-d597853d-362a-40d0-ba9e-0584724000d6.
Clemesha, B. R., KENT, G. S., and WRIGHT, R. W. H. (1966). Laser probing the lower atmo-sphere. Nature, 209(5019):184–185. http://dx.doi.org/10.1038/209184a0.
Cowell, E. W., Alimardani, N., Knutson, C. C., Conley, J. F., Keszler, D. A., Gibbons, B. J., andWager, J. F. (2011). Advancing MIM Electronics: Amorphous Metal Electrodes. AdvancedMaterials, 23(1):74–78. http://dx.doi.org/10.1002/adma.201002678.
Dechiaro, L. F. and Chemelli, R. G. (2003). Anomalous linewidth narrowing in microwave beatfrequency spectra between adjacent longitudinal modes in semiconductor lasers. IEEE Journalof Quantum Electronics, 39:1372–1380. http://adsabs.harvard.edu/abs/2003IJQE...39.1372D.
BIBLIOGRAPHY 201
Demtrder, W. (2003). Laser Spectroscopy; 3rd ed. Springer, Berlin.
Dinovitser, A., Hamilton, M. W., and Vincent, R. A. (2009). Transmitter design for differential ab-sorption water vapour lidar. In Proceedings of the 8th International Symposium on TroposhericProfiling: Integration of Needs, Technologies and Applications, volume S13 - P09 of Profilingof water vapor and temperature. ftp://ftp.sma.ch/outgoing/dcr/ISTP/ABSTRACTS/data/1787618.pdf.
Dinovitser, A., Hamilton, M. W., and Vincent, R. A. (2010). Stabilized master laser system fordifferential absorption lidar. Applied Optics, 49:3274–+. http://adsabs.harvard.edu/abs/2010ApOpt..49.3274D.
Divakarla, M. G., Barnet, C. D., Goldberg, M. D., McMillin, L. M., Maddy, E., Wolf, W., Zhou,L., and Liu, X. (2006). Validation of Atmospheric Infrared Sounder temperature and watervapor retrievals with matched radiosonde measurements and forecasts. Journal of GeophysicalResearch (Atmospheres), 111(10):D09S15. http://adsabs.harvard.edu/abs/2006JGRD..11109S15D.
DOE (2011). Department Of Energy ARM Climate Research Facility. http://www.arm.gov/.
Drayson, S. (1976). Rapid computation of the voigt profile. Journal of Quantitative Spectroscopyand Radiative Transfer, 16(7):611 – 614. http://www.sciencedirect.com/science/article/B6TVR-46BSNRS-3G/2/3672bc92d096c414ab5bdd341b57fd72.
Edwards, P. N. (2001). A Brief History of Atmospheric General Circulation Modeling, volume 70of International Geophysics. Academic Press. http://www.sciencedirect.com/science/article/pii/S0074614200800509.
Eichinger, W. E., Cooper, D. I., Forman, P. R., Griegos, J., Osborn, M. A., Richter, D., Tellier,L. L., and Thornton, R. (1999). The Development of a Scanning Raman Water Vapor Lidar forBoundary Layer and Tropospheric Observations. Journal of Atmospheric and Oceanic Tech-nology, 16:1753. http://adsabs.harvard.edu/abs/1999JAtOT..16.1753E.
Elterman, L. (1966). Aerosol measurements in the troposphere and stratosphere. Appl. Opt.,5(11):1769–1775. http://ao.osa.org/abstract.cfm?URI=ao-5-11-1769.
Evenson, K. M., Jennings, D. A., and Petersen, F. R. (1984). Tunable far-infrared spectroscopy.Applied Physics Letters, 44:576–578. http://adsabs.harvard.edu/abs/1984ApPhL.
.44..576E.
Feltz, W. F., Smith, W. L., Knuteson, R. O., Revercomb, H. E., Woolf, H. M., and Ben How-ell, H. (1998). Meteorological Applications of Temperature and Water Vapor Retrievals fromthe Ground-Based Atmospheric Emitted Radiance Interferometer (AERI). Journal of AppliedMeteorology, 37:857–875. http://adsabs.harvard.edu/abs/1998JApMe..37..857F.
Fiocco, G. and Smullin, L. D. (1963). Detection of scattering layers in the upper atmosphere (60-140 km) by optical radar. Nature, 199(4900):1275–1276. http://dx.doi.org/10.1038/1991275a0.
Fox, R. W., Robinson, H. G., Zibrov, A. S., Mackie, N., Marquardt, J., Magyar, J., and Hollberg,L. W. (1993). High-sensitivity spectroscopy with diode lasers. In Y. C. Chung, editor, Societyof Photo-Optical Instrumentation Engineers (SPIE) Conference Series, volume 1837 of Societyof Photo-Optical Instrumentation Engineers (SPIE) Conference Series, pages 360–365. http://adsabs.harvard.edu/abs/1993SPIE.1837..360F.
202 BIBLIOGRAPHY
Girolamo, P. D., Behrendt, A., Kiemle, C., Wulfmeyer, V., Bauer, H., Summa, D., Drnbrack, A.,and Ehret, G. (2008). Simulation of satellite water vapour lidar measurements: Performanceassessment under real atmospheric conditions. Remote Sensing Data Assimilation Special Is-sue. Remote Sensing of Environment, 112(4):1552 – 1568. http://www.sciencedirect.com/science/article/B6V6V-4R118DD-1/2/3b8bb195205ab039eedee1067637006f.
Goldberg, M. D., Qu, Y., McMillin, L. M., Wolf, W., Zhou, L., and Divakarla, M. (2003).AIRS near-real-time products and algorithms in support of operational numerical weatherprediction. IEEE Transactions on Geoscience and Remote Sensing, 41:379–389. http://adsabs.harvard.edu/abs/2003ITGRS..41..379G.
Greenspan, L. (1977). Humidity fixed points of binary saturated aqueous solutions. Journal ofresearch of the National Bureau of Standards. A, Physics and chemistry, A(81a):89–96. http://nvlpubs.nist.gov/nistpubs/jres/081/1/V81.N01.A06.pdf.
Gregory, I. S., Evans, M. J., Page, H., Malik, S., Farrer, I., and Beere, H. E. (2007). Analy-sis of photomixer receivers for continuous-wave terahertz radiation. Applied Physics Letters,91(15):154103. http://link.aip.org/link/?APL/91/154103/1.
Grzeschik, M., Bauer, H.-S., Wulfmeyer, V., Engelbart, D., Wandinger, U., Mattis, I., Althausen,D., Engelmann, R., Tesche, M., and Riede, A. (2008). Four-dimensional variational data anal-ysis of water vapor raman lidar data and their impact on mesoscale forecasts. Journal of Atmo-spheric and Oceanic Technology, 25(8):1437–1453. http://journals.ametsoc.org/doi/abs/10.1175/2007JTECHA974.1.
Hamamatsu (2001). Metal Package Photomultiplier Tube R7400U series, R7400U-20. http://www.hamamatsu.com. photonics K.K., Electron Tube Center.
Hamilton, M., Atkinson, R., Dinovitser, A., Peters, E., and Vincent, R. A. (2008). Towards low-cost water-vapour differential absorption lidar. In Society of Photo-Optical InstrumentationEngineers (SPIE) Conference Series, volume 7153 of Society of Photo-Optical Instrumenta-tion Engineers (SPIE) Conference Series. http://adsabs.harvard.edu/abs/2008SPIE.7153E...6H.
Hamilton, M. W., Dinovitser, A., and Vincent, R. A. (2009). Towards low-cost water-vapourdifferential absorption lidar. In Proceedings of the 8th International Symposium on TroposhericProfiling: Integration of Needs, Technologies and Applications, volume S13 - O02 of Profilingof water vapor and temperature. ftp://ftp.sma.ch/outgoing/dcr/ISTP/ABSTRACTS/data/1662333.pdf.
Hansen, K. (2008). Water vapor confirmed as major player in climate change. http://www.nasa.gov/topics/earth/features/vapor_warming.html.
Heise, S., Wickert, J., Beyerle, G., Schmidt, T., and Reigber, C. (2006). Global monitoring oftropospheric water vapor with GPS radio occultation aboard CHAMP. Advances in SpaceResearch, 37:2222–2227. http://adsabs.harvard.edu/abs/2006AdSpR..37.2222H.
Herotek, I. (2008). Tunnel diode detectors. http://www.herotek.com/datasheets/pdf/Tunnel_Diode_Detectors_0.1-40GHz.pdf.
Herriott, D., Kogelnik, H., and Kompfner, R. (1964). Off-axis paths in spherical mirror interferom-eters. Appl. Opt., 3(4):523–526. http://ao.osa.org/abstract.cfm?URI=ao-3-4-523.
Hindle, F., Cuisset, A., Bocquet, R., and Mouret, G. (2008). Continuous-wave terahertz by pho-tomixing: applications to gas phase pollutant detection and quantification. Comptes RendusPhysique, 9:262–275. http://adsabs.harvard.edu/abs/2008CRPhy...9..262H.
BIBLIOGRAPHY 203
Hollberg, L, O., Velichansky, V, L., and Weimer, C, S. (1996). High-Accuracy Spectroscopy withSemiconductor Lasers: Application to laser-frequency Stabilization. John Wiley & Sons inc.http://tf.nist.gov/general/pdf/1186.pdf.
Honeywell (2010). HIH-4000 Series Humidity Sensors. www.honeywell.com/sensing.
Johnson, E., Meyer, R. C., Hopkins, R. E., and Mock, W. H. (1939). The measurement of lightscattered by the upper atmosphere from a search-light beam. J. Opt. Soc. Am., 29(12):512–517.http://www.opticsinfobase.org/abstract.cfm?URI=josa-29-12-512.
JPL (2002). Airs mission description. http://airs.jpl.nasa.gov/mission/description/.NASA Jet Propulsion Laboratory.
Kameyama, S., Imaki, M., Hirano, Y., Ueno, S., Kawakami, S., and Nakajima, M. (2008). De-velopment of 1.6 micron CW modulation ground-based DIAL system for CO2 monitoring. InSingh, U. N., Asai, K., and Jayaraman, A., editors, Lidar Remote Sensing for Environmen-tal Monitoring IX, volume 7153, page 71530L. SPIE. http://link.aip.org/link/?PSI/7153/71530L/1.
Kiehl, J. T. and Trenberth, K. E. (1997). Earths annual global mean energy budget. Bull. Amer.Meteor. Soc, pages 197–208.
Koch, G. J., Barnes, B. W., Petros, M., Beyon, J. Y., Amzajerdian, F., Yu, J., Davis, R. E., Ismail,S., Vay, S., Kavaya, M. J., and Singh, U. N. (2004). Coherent differential absorption lidarmeasurements of CO2. Applied Optics, 43(26):5092–5099. http://ao.osa.org/abstract.cfm?URI=ao-43-26-5092.
Koch, G. J., Beyon, J. Y., Gibert, F., Barnes, B. W., Ismail, S., Petros, M., Petzar, P. J., Yu, J., Mod-lin, E. A., Davis, K. J., and Singh, U. N. (2008). Side-line tunable laser transmitter for differen-tial absorption lidar measurements of co2: design and application to atmospheric measurements.Applied Optics, 47:944–956. http://adsabs.harvard.edu/abs/2008ApOpt..47..944K.
Kursinski, E. R., Hajj, G. A., Schofield, J. T., Linfield, R. P., and Hardy, K. R. (1997). Observingearth’s atmosphere with radio occultation measurements using the global positioning system. J.Geophys. Res., 102(D19):23429–23465. http://dx.doi.org/10.1029/97JD01569.
Kursinski, R., Hajj, G., Leroy, S., and Herman, B. (2001). The GPS Radio Occultation Technique.In L.-C. Lee, C. Rocken, & R. Kursinski, editor, Applications of Constellation Observing Sys-tem for Meteorology, Ionosphere & Climate, pages 53–+. http://adsabs.harvard.edu/abs/2001acos.conf...53K.
Lazzarotto, B., Quaglia, P., Simeonov, V., Larcheveque, G., van den Bergh, H., and Calpini, B.(1999). A raman differential absorption lidar for ozone and water vapor measurement in thelower troposphere. International Journal of Environmental Analytical Chemistry, 74(1):255–261. http://www.informaworld.com/10.1080/03067319908031430.
Leon-Hidalgo, M. C., Gozalvez-Zafrilla, J. M., Lora-Garcia, J., and M Arnal-Arnal, J. (2009).Study of the vapour pressure of saturated salt solutions and their influence on evaporation rateat room temperature.
LICEL GmbH (2002). Optical transient recorder. http://www.licel.com.
Luntama, J.-P., Kirchengast, G., Borsche, M., Foelsche, U., Steiner, A., Healy, S., von Engeln, A.,O’Clerigh, E., and Marquardt, C. (2008). Prospects of the EPS GRAS Mission For OperationalAtmospheric Applications. Bulletin of the American Meteorological Society, 89(12):1863–1875. http://journals.ametsoc.org/doi/abs/10.1175/2008BAMS2399.1.
204 BIBLIOGRAPHY
Machol, J. L., Ayers, T., Schwenz, K. T., Koenig, K. W., Hardesty, R. M., Senff, C. J., Krainak,M. A., Abshire, J. B., Bravo, H. E., and Sandberg, S. P. (2004). Preliminary measurements withan automated compact differential absorption lidar for the profiling of water vapor. AppliedOptics, 43:3110–3121. http://adsabs.harvard.edu/abs/2004ApOpt..43.3110M.
McManus, J. B., Kebabian, P. L., and Zahniser, M. S. (1995). Astigmatic mirror multipassabsorption cells for long-path-length spectroscopy. Appl. Opt., 34(18):3336–3348. http://ao.osa.org/abstract.cfm?URI=ao-34-18-3336.
Mehuys, D., Welch, D., and Goldberg, L. (1992). 2.0 w cw, diffraction-limited tapered amplifierwith diode injection. Electronics Letters, 28(21):1944 –1946.
Melfi, S. H., J. D. Lawrence, J., and McCormick, M. P. (1969). Observation of Raman scatteringby water vapor in the atmosphere. Applied Physics Letters, 15(9):295–297. http://link.aip.org/link/?APL/15/295/1.
Milman, A. S. (1978). Calculation of optical depths from an integral of the Voigtfunction. Journal of Quantitative Spectroscopy and Radiative Transfer, 20(6):593– 597. http://www.sciencedirect.com/science/article/B6TVR-46CBW6N-18/2/
288c4ab0c9c0267aed545c3cb17f1095.
Miloshevich, L. M., Vomel, H., Whiteman, D. N., and Leblanc, T. (2009). Accuracy assessmentand correction of Vaisala RS92 radiosonde water vapor measurements. Journal of GeophysicalResearch (Atmospheres), 114.
Moore, A. S. J., Brown, K. E., Hall, W. M., Barnes, J. C., Edwards, W. C., Petway, L. B., Little,A. D., Luck, J. W. S., Jones, I. W., Antill, J. C. W., Browell, E. V., and Syed, I. (1996). Devel-opment of the Lidar Atmospheric Sensing Experiment (LASE) -An Advanced Airborne DIALInstrument. In Ansmann, A., Neuber, R., Rairoux, P., and Wandinger, U., editors, Advances inAtmospheric Remote Sensing with Lidar, pages 281–288. Springer-Verlag.
Moore, W. J. (1984). Basic Physical Chemistry. Prentice Hall International.
Mueller, E. R. (2003). Terahertz radiation: applications and sources. The industrial physicist,9(4):27.
Muth, C., Lee, P., Shiue, J., and Allan Webb, W. (2004). Advanced Technology MicrowaveSounder on NPOESS and NPP. In Geoscience and Remote Sensing Symposium, 2004. IGARSS’04. Proceedings. 2004 IEEE International, volume 4, pages 2454 – 2458 vol.4. http://dx.doi.org/10.1109/IGARSS.2004.1369789.
Nagali, V., Chou, S. I., Baer, D. S., and Hanson, R. K. (1997). Diode-laser measurements oftemperature-dependent half-widths of H2O transitions in the 1.4 µm region. ournal of Quanti-tative Spectroscopy and Radiative Transfer, 57:795–809.
Nehrir, A. R., Repasky, K. S., and Carlsten, J. L. (2009a). Design and Testing of a Compact WidelyTunable Cascaded Master Oscillator Power Amplifier (MOPA) Diode Laser Based Micro PulseDifferential Absorption Lidar (DIAL) for Water Vapor Profiling in the Lower Troposphere. In8th International Symposium on Tropospheric Profiling, pages S13–P08. ftp://ftp.sma.ch/outgoing/dcr/ISTP/ABSTRACTS/data/1787618.pdf.
Nehrir, A. R., Repasky, K. S., Carlsten, J. L., Obland, M. D., and Shaw, J. A. (2009b). Watervapor profiling using a widely tunable, amplified diode-laser-based differential absorption lidar(dial). Journal Of Atmospheric And Oceanic Technology, 26:733–745.
NOAA (2000). Global positioning system for water vapor measurement. online. http://www.etl.noaa.gov/technology/archive/gps.
BIBLIOGRAPHY 205
Obland, M. D. (2007). Water vapor profiling using a widely tunable amplified diode laser differ-ential absorption lidar (DIAL) [electronic resource] /. PhD thesis, Montana State University.http://etd.lib.montana.edu/etd/2007/obland/OblandM0507.pdf.
Odashima, H., Zink, L. R., and Evenson, K. M. (1999). Tunable far-infraredspectroscopy extendedto 9.1 THz. Optics Letters, 24:406–407. http://adsabs.harvard.edu/abs/1999OptL...24..406O.
Olivero, J. and Longbothum, R. (1977). Empirical fits to the Voigt line width: Abrief review. Journal of Quantitative Spectroscopy and Radiative Transfer, 17(2):233– 236. http://www.sciencedirect.com/science/article/B6TVR-46V0D4P-K/2/
6ea61622d51016571492f70ccb7df928.
Organisation Internationale de Metrologie Legale (OIML) (1996). The scale of relative humidityof air certified against saturated salt solutions. International Recomendation OIML R 121Edition 1996 (E).
Ottaway, D. J., Veitch, P. J., Hollitt, C., Mudge, D., Hamilton, M. W., and Munch, J. (2000). Fre-quency and intensity noise of an injection-locked Nd:YAG ring laser. Applied Physics B: Lasersand Optics, 71:163–168. http://adsabs.harvard.edu/abs/2000ApPhB..71..163O.
Pal, S., Behrendt, A., Bauer, H., and Radlach, M. (2008). 3-dimensional observations of atmo-spheric variables during the field campaign cops. In 14th International Symposium for the Ad-vancement of Boundary Layer Remote Sensing. IOP Publishing. http://iopscience.iop.org/1755-1315/1/1/012031/.
Peach, G. (1981). Theory of the pressure broadening and shift of spectral lines.Advances in Physics, 30(3):367–474. http://www.informaworld.com/10.1080/
00018738100101467.
Philbrick, C. R., Wright, T., Snyder, M., Hallen, H., NC, R., Brown, A. M., Verghese, S., andEdwards, P. S. (2010). Characteristics of atmospheric aerosols based on optical remote sensing.In 12th Conference on Atmospheric Chemistry, The 90th American Meteorological SocietyAnnual Meeting. http://ams.confex.com/ams/pdfpapers/163441.pdf.
Pierce, D. W., Barnett, T. P., Fetzer, E. J., and Gleckler, P. J. (2006). Three-dimensionaltropospheric water vapor in coupled climate models compared with observations from theairs satellite system. Geophys. Res. Lett., 33(21):L21701. http://dx.doi.org/10.1029/2006GL027060.
Ponsardin, P., Higdon, N. S., Grossmann, B. E., and Browell, E. V. (1994). Spectral control of anAlexandrite laser for an airborne water-vapor differential absorption lidar system. Appl. Opt.,33(27):6439–6450.
Posthumus, J., Deninger, A., and Lison, F. (2005). Distributed feedback diode lasers:Spectral properties and current applications. Toptica Photonics. http://www.
topticaphotonicsinc.com/products/itemlayer/51/DFB-diode-lasers.pdf.
Refaat, T. F. I. (2000). Advanced atmospheric water vapor DIAL detection system. PhD thesis,Old Dominion University.
Rosen, H., Robrish, P., and Chamberlain, O. (1975). Remote detection of pollutants using reso-nance Raman scattering. Applied Optics, 14:2703–2706.
206 BIBLIOGRAPHY
Rothman, L. S., Gordon, I. E., Barbe, A., Benner, D. C., Bernath, P. F., Birk, M., Boudon, V.,Brown, L. R., Campargue, A., Champion, J.-P., Chance, K. V., Coudert, L. H., Sung, K., and etal. Toth, R. A. (2009). The hitran 2008 molecular spectroscopic database. Journal of Quantita-tive Spectroscopy and Radiative Transfer, 110(9-10):533–572.
Rothman, L. S., Rinsland, C. P., Goldman, A., Massie, S. T., Edwards, D. P., Flaud, J.-M.,Perrin, A., Camy-Peyret, C., Dana, V., Mandin, J.-Y., Schroeder, J., McCann, A., Gamache,R. R., Wattson, R. B., Yoshino, K., Chance, K., Jucks, K., Brown, L. R., Nemtchinov, V., andVaranasi, P. (1998). The HITRAN Molecular Spectroscopic Database and HAWKS (HITRANAtmospheric Workstation): 1996 Edition. Journal of Quantitative Spectroscopy and RadiativeTransfer, 60:665–710.
Schilt, S., Matthey, R., Kauffmann-Werner, D., Affolderbach, C., Mileti, G., and Thevenaz, L.(2008). Laser offset-frequency locking up to 20 GHz using a low-frequency electrical filtertechnique. Applied Optics, 47:4336–+. http://adsabs.harvard.edu/abs/2008ApOpt..47.4336S.
Schneider, M., Romero, P. M., Hase, F., Blumenstock, T., Cuevas, E., and Ramos, R. (2010).Continuous quality assessment of atmospheric water vapour measurement techniques: FTIR,Cimel, MFRSR, GPS, and Vaisala RS92. Atmospheric Measurement Techniques, 3(2):323–338. http://www.atmos-meas-tech.net/3/323/2010/.
Schotland, R. M. (1964). The determination of the vertical profile of atmospheric gases by meansof a ground based optical radar. In Third Symposium on Remote Sensing of the Atmosphere,page 215224. University of Michigan, Ann Arbor, Michigan.
Schotland, R. M. (1966). The determination of the vertical profile of atmospheric gases by meansof a ground based optical radar. In Proceedings of the 4th Symposium on Remote Sensing of theEnvironment, pages 273–283. University of Michigan, Ann Arbor, Michigan.
Schotland, R. M. (1974). Errors in the lidar measurement of atmospheric gases by differentialabsorption. Journal of Applied Meteorology, 13(1):71–77. http://adsabs.harvard.edu/abs/1974JApMe..13...71.
Schreier, F. (2009). Comments on A Common Misunderstanding about the Voigt Line Profile.Journal of the Atmospheric Sciences, 66(6):1860–1864. http://journals.ametsoc.org/doi/abs/10.1175/2009JAS2906.1.
Schwarzer, H., Borner, A., Fix, A., Gunther, B., Hubers, H.-W., Raugust, M., Schrandt, F., andWirth, M. (2007). Development of a wavelength stabilized seed laser system for an airbornewater vapour lidar experiment. In Society of Photo-Optical Instrumentation Engineers (SPIE)Conference Series, volume 6681 of Society of Photo-Optical Instrumentation Engineers (SPIE)Conference Series. http://adsabs.harvard.edu/abs/2007SPIE.6681E..13S.
Sherwood, S. C., Roca, R., Weckwerth, T. M., and Andronova, N. G. (2010). Tropospheric wa-ter vapor, convection, and climate. Reviews of Geophysics, 48:G2001+. http://adsabs.harvard.edu/abs/2010RvGeo..48.2001S.
Silver, J. A. (1992). Frequency-modulation spectroscopy for trace species detection: theory andcomparison among experimental methods. Appl. Opt., 31(6):707–717. http://ao.osa.org/abstract.cfm?URI=ao-31-6-707.
Simeonov, V., Dinoev, T., Serikov, I., Froidevaux, M., Bartlome, M., Calpini, B., Bobrovnikov, S.,Ristori, P., van den Bergh, H., Parlange, M., and Archinov, Y. (2010). Lidars as an operationaltool for meteorology and advanced atmospheric research. In EGU General Assembly Con-ference Abstracts, volume 12 of EGU General Assembly Conference Abstracts, page 14001.http://adsabs.harvard.edu/abs/2010EGUGA..1214001S.
BIBLIOGRAPHY 207
Smullin, L. D. and Fiocco, G. (1962). Optical echoes from the moon. Nature, 194(4835):1267–1267. http://dx.doi.org/10.1038/1941267a0.
Synge, E. H. (1930). A method for investigating the higher atmosphere. Philosophical Magazine,7(9:60):1014–1020.
Taubman, M. S. and Hall, J. L. (2000). Cancellation of laser dither modulation from opti-cal frequency standards. Optics Letters, 25:311–313. http://adsabs.harvard.edu/abs/2000OptL...25..311T.
Thorlabs (2011). www.thorlabs.hk/Thorcat/4800/4813-S01.pdf.
Toptica Photonics Inc. (2012). cw Terahertz Generation Spectroscopy Kit. http://www.toptica.com.
Trenberth, K., Jones, P., Ambenje, P., Bojariu, R., Easterling, D., Tank, A. K., Parker, D.,Rahimzadeh, F., Renwick, J., Rusticucci, M., Soden, B., and Zhai, P. (2007). Observations:Surface and Atmospheric Climate Change. Cambridge University Press, Cambridge, UnitedKingdom and New York, NY, USA.
Turner, D. D., Tobin, D. C., Clough, S. A., Brown, P. D., Ellingson, R. G., Mlawer, E. J., Knuteson,R. O., Revercomb, H. E., Shippert, T. R., Smith, W. L., and Shephard, M. W. (2004). The qmeaeri lblrtm: A closure experiment for downwelling high spectral resolution infrared radiance.Journal of the Atmospheric Sciences, 61(22):2657–2675. http://dx.doi.org/10.1175/JAS3300.1.
UCAR (2002). International H2O Project (IHOP) 2002. http://www.eol.ucar.edu/dir_off/projects/2002/IHOP.html.
W. Smith, S., Revercomb, H., Woolf, H., Huang, H., Larar, A., Zhou, D., Kireev, S., Tian, J.,and Liu, X. (2008). Ultra-spectral remote sounding: background and future. In Larar, A. M.,Lynch, M. J., and Suzuki, M., editors, Proc. SPIE,, volume 7149, page 714902. SPIE. http://link.aip.org/link/?PSI/7149/714902/1.
Wagner, G., Shiler, M., and Wulfmeyer, V. (2004). Design and Development of a High-PowerTI:Sapphire Laser Transmitter. In G. Pappalardo & A. Amodeo, editor, 22nd Internation LaserRadar Conference (ILRC 2004), volume 561 of ESA Special Publication, pages 203–+. http://adsabs.harvard.edu/abs/2004ESASP.561..203W.
Wandinger, U. (2005). Introduction to Lidar, pages 1–18. Springer Berlin / Heidelberg. http://dx.doi.org/10.1007/0-387-25101-4_1.
Wang, L.-G., Tate, D. A., Riris, H., and Gallagher, T. F. (1989). High-sensitivity frequency-modulation spectroscopy with a GaAlAs diode laser. J. Opt. Soc. Am. B, 6(5):871–876. http://josab.osa.org/abstract.cfm?URI=josab-6-5-871.
Weibring, P., Smith, J. N., Edner, H., and Svanberg, S. (2003). Development and testing of afrequency-agile optical parametric oscillator system for differential absorption lidar. Reviewof Scientific Instruments, 74:4478–4484. http://adsabs.harvard.edu/abs/2003RScI...74.4478W.
Wexler, A. (1977). Vapor pressure formulation for ice. J. Res. Natl. Bur. Stand., 80A:5–19.
White, J. U. (1976). Very long optical paths in air. J. Opt. Soc. Am., 66(5):411–416. http://dx.doi.org/10.1364/JOSA.66.000411.
208 BIBLIOGRAPHY
Whiteman, D. N., Demoz, B., di Girolamo, P., Comer, J., Veselovskii, I., Evans, K., Wang, Z.,Cadirola, M., Rush, K., Schwemmer, G., Gentry, B., Melfi, S. H., Mielke, B., Venable, D., andvan Hove, T. (2006). Raman Lidar Measurements during the International H2O Project. PartI: Instrumentation and Analysis Techniques. Journal of Atmospheric and Oceanic Technology,23:157. http://adsabs.harvard.edu/abs/2006JAtOT..23..157W.
Whiteway, J., Daly, M., Carswell, A., Duck, T., Dickinson, C., Komguem, L., and Cook, C.(2008). Lidar on the Phoenix mission to Mars. J. Geophys. Res., 113(E3):E00A08. http://dx.doi.org/10.1029/2007JE003002.
Whiting, E. E. (1968). An empirical approximation to the Voigt profile. Journal of QuantitativeSpectroscopy and Radiative Transfer, 8(6):1379 – 1384. http://www.sciencedirect.com/science/article/B6TVR-46D1XPV-1K/2/014371c1ed02db489c6d14ccb4fda23d.
Wickert, J., Beyerle, G., Konig, R., Heise, S., Grunwaldt, L., Michalak, G., Reigber, C., andSchmidt, T. (2005). GPS radio occultation with CHAMP and GRACE: A first look at a newand promising satellite configuration for global atmospheric sounding. Annales Geophysicae,23:653–658. http://adsabs.harvard.edu/abs/2005AnGeo..23..653W.
Wiederhold, P. R. (1997). Water vapor measurement. Marcel Dekker. http://books.google.com.au/books?id=mm5YY1nOJF4C.
Wieman, C. E. and Hollberg, L. (1991). Using diode lasers for atomic physics. Review of ScientificInstruments, 62:1–20. http://adsabs.harvard.edu/abs/1991RScI...62....1W.
Winker, D., Couch, R., and McCormick, M. (1996). An overview of LITE: NASA’s Lidar In-space Technology Experiment. Proceedings of the IEEE, 84(2):164 –180. http://dx.doi.org/10.1109/5.482227.
Wirth, M., Fix, A., Mahnke, P., Schwarzer, H., Schrandt, F., and Ehret, G. (2009a). The airbornemulti-wavelength water vapor differential absorption lidar WALES: system design and perfor-mance. Applied Physics B: Lasers and Optics, 96:201–213. http://adsabs.harvard.edu/abs/2009ApPhB..96..201W.
Wirth, M., Fix, A., Reichardt, J., Begbie, R., Engelbart, D., Vmel, H., Calpini, B., Romanens,G., A., A., Wilson, K. M., Vogelmann, H., and Trickl, T. (2009b). Intercomparison of Air-borne Water Vapour DIAL Measurements with Ground Based Remote Sensing and Radioson-des within the Framework of LUAMI 2008. In Proceedings of the 8th International Symposiumon Troposheric Profiling: Integration of Needs, Technologies and Applications, volume S07 -P01 of Profiling of water vapor and temperature. http://cerberus.rivm.nl/ISTP/data/1659777.pdf.
World Meteorological Organization (WMO) (2008). Guide to Meteorological Instruments andMethods of Observation,, volume 8, chapter Chapter 4, Measurement of Humidity, pages 1.4–25. World Meteorological Organization, P.O. Box No. 2300,, CH-1211 Geneva 2, Switzerland,7 edition. http://www.wmo.int/pages/prog/www/IMOP/publication.
World Meteorological Organization (WMO) (2010). Manual on the Global Data-processing andForecasting System, volume 1 of WMO-No. 485, chapter Appendix II-2: Observational datarequirements for GDPFS centres for global and regional exchange. World Meteorological Or-ganization, P.O. Box No. 2300,, CH-1211 Geneva 2, Switzerland. http://www.wmo.int/pages/prog/www/DPFS/Manual_GDPFS.html.
Wulfmeyer, V. (1998). Ground-based differential absorption lidar for water-vapor and temperatureprofiling: Development and specifications of a high-performance laser transmitter. Appl. Opt.,37(18):3804–3824. http://ao.osa.org/abstract.cfm?URI=ao-37-18-3804.
BIBLIOGRAPHY 209
Wulfmeyer, V., Bauer, H., Girolamo, P. D., and Serio, C. (2005). Comparison of active and passivewater vapor remote sensing from space: An analysis based on the simulated performance of iasiand space borne differential absorption lidar. Remote Sensing of Environment, 95(2):211 – 230.http://www.sciencedirect.com/science/article/pii/S0034425705000179.
Wulfmeyer, V., Bauer, H.-S., Grzeschik, M., Behrendt, A., Vandenberghe, F., Browell, E. V.,Ismail, S., and Ferrare, R. A. (2006). Four-Dimensional Variational Assimilation of WaterVapor Differential Absorption Lidar Data: The First Case Study within IHOP 2002. MonthlyWeather Review, 134:209–+. http://adsabs.harvard.edu/abs/2006MWRv..134..209W.
Wulfmeyer, V. and Bosenberg, J. (1998). Ground-based differential absorption lidar for water-vapor profiling: Assessment of accuracy, resolution, and meteorological applications. Appl.Opt., 37(18):3825–3844. http://ao.osa.org/abstract.cfm?URI=ao-37-18-3825.
Wulfmeyer, V. and Walther, C. (2001). Future Performance of Ground-Based and Airborne Water-Vapor Differential Absorption Lidar. I. Overview and Theory. Applied Optics, 40:5304–5320.http://adsabs.harvard.edu/abs/2001ApOpt..40.5304W.
Yeh, P., Wu, I.-F., Jiang, S., and Dagenais, M. (1993). High-power high-gain monolithicallyintegrated preamplifier/power amplifier. Electronics Letters, 29(22):1981 –1983.
Young, A. T. (1982). Rayleigh scattering. Physics Today, 35:42–48. http://adsabs.harvard.edu/abs/1982PhT....35a..42Y.
Young, S. A. and Elford, W. G. (1979). Stratospheric aerosol optical thickness measurements at35° S. Nature, 278(5704):540–541. http://dx.doi.org/10.1038/278540a0.
Yu, J., Rambaldi, P., and Wolf, J.-P. (1997). Dual-wavelength diode-seeded Ti:Sapphire laser fordifferential absorption lidar applications. Appl. Opt., 36(27):6864–6868. http://ao.osa.org/abstract.cfm?URI=ao-36-27-6864.
Zhou, X., Liu, X., Jeffries, J. B., and Hanson, R. K. (2005). Selection of nir H2O absorptiontransitions for in-cylinder measurement of temperature in ic engines. Measurement Scienceand Technology, 16(12):2437. http://stacks.iop.org/0957-0233/16/i=12/a=006.
Zuev, V. V., Ponomarev, Y. N., Solodov, A. M., Tikhomirov, B. A., and Romanovsky, O. A. (1985).Influence of the shift H2O absorption lines with air pressure on the accuracy of the atmospherichumidity profiles measured by the differential-absorption method. Opt. Lett., 10(7):318–320.http://ol.osa.org/abstract.cfm?URI=ol-10-7-318.
ThesisDIAL