a stabilized master laser system for differential absorption lidar · 2015-10-29 · a stabilized...

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,


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


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


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


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-


(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


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,


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


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.


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


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


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


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-


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%.


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


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


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.


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.


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.

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


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)


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



LaserAmplifier

Master Laser 2



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.


(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-


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.


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.,


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.


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


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


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.


(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.


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-


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


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


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


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.


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


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


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


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.


(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


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


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).


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


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


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.


(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)


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


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


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.


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


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.


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.


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.


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


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


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


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.


(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%.


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.


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


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


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


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


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


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.


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.


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.


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


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.


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.


Figure A.10: TA driver schematic


(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.


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

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


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.


Oscilloscope

Isolator


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


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


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.2.5 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


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)


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.


Oscilloscope

LaserT

I Current

Temp.PID

control


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


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


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


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.


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-


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


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.


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.


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.


(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.

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);


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);


% 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);


% 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);


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);


% 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);


trig.Trigger = 1;

trig.Slope = CsMl_Translate(’Positive’, ’Slope’);

trig.Level = 15;

trig.Source = 1;


trig.ExtCoupling = CsMl_Translate(’DC’, ’ExtCoupling’);

trig.ExtRange = 2000;

[ret] = CsMl_ConfigureTrigger(handle, trig);


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


157


else


end


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));


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));




else


end

if size(vps2)<2 %ditto


else

vps2=vertcat(vps2,datan(c:c+84,2));

end


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’);


%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.


%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))


==============================

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=0

Gl1=(296/T)ˆa(v,6)*P1*(a(v,4)*mr+a(v,3)*(1-mr)); %calculate the widths at the 3 pressures



p(1:401,1:5)=zeros;


for x= 1:401


p(x,3)= N*P1*a(v,2)*Voigt(p(x,1),Gl1,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


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)â(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)â(c,6)*(a(c,4)*mr+a(c,3)*(1-mr));

Yl2=(296/T)â(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.


%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=0




p(1:401,1:5)=zeros;


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);


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.


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


(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.


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


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).


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).


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

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:


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:


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

a stabilized master laser system for differential absorption lidar · 2015-10-29 · a stabilized...

Documents