Using TODWL and Optical Particle Countersto Investigate Aerosol Backscatter Signatures

from Organized Structures in the Marine Boundary Layer

D.A. BowdleUniversity of Alabama in Huntsville

G.D. Emmitt and S.A. WoodSimpson Weather Associates

Working Group on Space-Based Lidar WindsFrisco, Colorado, June 29 - July 1, 2004

CONTENTS

• EXPERIMENT

• ANALYSIS

• RESULTS

• SUMMARY

• Joint research project by ONR and NPOESS IPO

• Investigate data processing issues related to future space-based wind lidar operations

• Develop calibration/validation procedures for all wind profiling systems (ground-based, airborne, space-based)

• Conduct basic research on lower tropospheric winds and aerosols in the marine and continental boundary layers

MOTIVATION

TODWL Transceiver

2.0125 µm, coherent detection

4-6 mJ, 330 nsec (FWHM), 80 Hz

10 cm telescope

two axis scanner, 30 & 120 deg, side door mount

digitization rate 100 MHz

~7-10% total system efficiency

INSTRUMENTSAircraft Platform

NPS CIRPAS Twin Otter

Naval Postgraduate SchoolCenter for Interdisciplinary Remotely-piloted Aircraft Studies

Optical Particle Counters

TODWL Scanner

PCASP 0.1-3.0 m

FSSP2.5-51 mCAPS

0.45-118 m

OPERATIONS

LocationSchedule

Flight Plans MBL Database

• Straight and level 50-100 km runs

• Along-wind and cross-wind runs

• Multiple altitudes over same ground track

• Near surface, near and above inversion

• Series 1: February 9-15, 2002

• Series 2: March 12-15, 2002

•Series 3: February 8-21, 2003

Series 1:• Monterey area & San Joaquin River

Series 2:• Monterey area• Monterey to Boulder via Las Vegas

Series 3:• Monterey area, ocean & land

• 8 flights, approx 30 hours

• multiple scanning patterns

• concentrate on February 20, 2003

MEASUREMENTS*

beamdirection

beamdirection

(neglecting pitch)

scatteringvolume

Vi

Vac

Ri,j =

| V

ac |

(

j +

t i)

i

particleprobes

ground track,heading,

ground velocity(neglecting

yaw, sideslip)

Xj =

| V

ac |

t j

*Backscatter-related scan patterns• along-track RHI step-and-stare

• forward stare

• nadir stare

BACKSCATTER EQUATION* - 1

For a diffuse atmospheric target, with volume backscatter coefficient ,

( ) ( )2

,,

RcRtartar

XX

rr β

ρ⋅

=⋅Ψ

For calibrations against a hard target, with diffuse reflectance ,

( )ht

hthttartar R

τ

ρκρ

⋅=⋅Ψ ,X

r

*generalized from ACLAIM backscatter analysis [Steve Hannon, 1999]

Combining the above equations gives a non-dimensionalized formulation:

( )( )

( ) ( )( ) ( )

( ) ( ) ( )( ) ( )⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

⋅

⋅⋅

⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

⋅⋅⋅⋅

⋅⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

=⋅

⋅

hththththththt

ht

hthththththt RKfR

RKfRRc

BE

BE

fRSNRR

fRSNRR

,,,

,,,

2

,

,,

,,2

2

2

2

XX

XXX

XX

XX

X

Xrr

rrr

rr

rr

r

r

η

ηρκ

βτ

BACKSCATTER EQUATION - 2

When TODWL points straight forward, make the following assumptions:

( ) ( ) ( ) ( ) ( ) ( ) ( ) ( )( ) ( ) ( ) ( ) [ ]( ) [ ]

0ˆ ,1ˆ 1,ˆ where

,ˆ2exp2exp, ;,ˆ,

,,ˆ,,,, ; ;

0

02

0

≈≈≈

⎟⎟⎠

⎞⎜⎜⎝

⎛−⋅⋅⋅−=⋅=

⋅⋅===

∫

αβ

αβ

δδδ

δαδββ

δηηη

het

R

hethetopt

drrRzRKRzR

fRfRzzfRzBBzEE

XXXX

XXXX

rrrr

rrrr

Combine the terms that have no range dependenceExpress the backscatter equation using non-dimensional variables

( ) ( ) ( ) [ ]( )[ ]( )

( ) ( ) [ ]⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧

⎟⎟⎠

⎞⎜⎜⎝

⎛⋅⋅−⋅⋅

×⎭⎬⎫

⎩⎨⎧

⋅⋅−⋅⋅−

⋅⋅=

∫R

het

hththet

drrfRR

R

RzfRzzfRS

0

00

,ˆ2exp,,ˆ,ˆ

02exp

2exp,,ˆˆ,,ˆ

XXX

X

rrr

r

αβ δδδ

αα

ηβ

( ) ( ) ( )( ) ( )

( )( ) ( )z

c

z

z

zBzE

zBzEz

htht

ht

htopt

opt

htht00 2

ˆ βρκ

τ

η

ηβ ⋅

⎭⎬⎫

⎩⎨⎧

⋅⋅⋅

⋅⎪⎭

⎪⎬⎫

⎪⎩

⎪⎨⎧⋅

⎭⎬⎫

⎩⎨⎧⋅=

( ) ( )( )hthththt fRSNRR

fRSNRRfRRSN

,,

,,,,ˆ

2

2

X

XX r

rr

⋅

⋅= ( ) ( )

( )hthththet

hethet fRz

fRzfRz

,,

,,,,ˆ

η

ηη =

baselineterms

perturbationterms

ANALYTICAL APPROACH

For a pulsed coherent 2-m Doppler lidar, analysis of

ABSOLUTE BACKSCATTER VARIABILITY

• requires absolute backscatter calibration at range Rht;

• requires correction for nominal range response function;

• requires correction for atmospheric extinction;

• requires correction for atmospheric refractive turbulence;

• assumes system stability during a given data run;

RELATIVE BACKSCATTER VARIABILITY

• avoids all of the above requirements.

• exclude wild velocities

• exclude backscatter dropouts

• exclude major pulse tail artifacts

• account for aircraft pitch

ANALYTICAL METHODS

Dropouts & Anomalies

CorrelationFiltering

• TODWL time-range plots (Hovmuller)

• aerosol time-size plots

• scale analysis

• variance analysis

• for TODWL

- compute mean V & at each range

- compute residual V & at each pixel

• for OPC, compute mean, residual Nm

(Normalized) Turbulent Residuals

• 1-s data – good V’ and ’ most ranges

• 1-s data – poor OPC count statistics

• filtered V’ & ’ may not resolve waves

• filtered OPC improves count statistics

RESULTS*

SAMPLING CONDITIONS• sharp inversion ~450 m; winds below inversion NNW ~17 m/s; RH ~70% at ~30 m, ~90% --> 45% across inversion, ~30% above inversion

• horizontal legs at ~35 m (x1), ~400 m (x1), ~900 m (x3), 1400 m (x1)

HOVMULLER PLOTS IN RADIAL VELOCITY AND SNR• stratification by aircraft pitch eliminates unphysical “striping”, and markedly reduces the observed variation along individual coherent features

• stratified plots still exhibit residual non-coherent variation along features

• radial velocity variation across scene up to 8 m/s; along features <1 m/s

• SNR variation across scene (fixed range) up to 6 dB, along features TBD

• promising results from preliminary attempts to correct for atmospheric attenuation and lidar range response, even before pitch stratification

• velocity-backscatter correlations observed below, at, above inversion

OPTICAL PARTICLE COUNTERS• large particles, with poor count statistics, often dominate 2-m backscatter

*Planned graphics unavailable due to severe case of Microsoft fever

CONCLUSIONS

ATMOSPHERIC FEATURES

• turbulent waves in aerosol and velocity, multiple scales

• aerosol-velocity correlations will bias DWL LEO winds, even in clear air

• nature & magnitude of bias will depend on shot integration strategy

ANALYTICAL CHALLENGES

• beam elevation offset, pitch fluctuations, altitude fluctuations

• measured vs. modeled absolute backscatter

• OPC operational status

• OPC count statistics

SCIENCE POTENTIAL

• substantial information content remains untapped in TODWL database

RECOMMENDATIONS - 1

INSTRUMENTATION AND OPERATIONS

TODWL – modify programmed scans to account for pitch offset in mounting

TODWL – add option for automatic dither in beam elevation

TODWL – improve frequency, quality of ground-based radiometric calibrations

OPC – verify PCASP, FSSP, CAPS operational status on every flight

OPC - add flight-level sensor that has higher volume sampling rate

OPC ANALYSIS METHODS

replace contiguous-point temporal smoothing by feature-composited averaging

replace measured size distributions from individual OPC’s by aerosol-model- constrained composites from FSSP, PCASP, CAPS forward, CAPS backward

augment composited size distributions using Monte Carlo & Poisson statistics

RECOMMENDATIONS - 2

ANALYSIS POTENTIAL – MEAN CONDITIONS

• Backscatter: model using cabin data (OPC); derive from TODWL

• Attenuation: model using cabin data (OPC, T, RH); derive from TODWL

• Coherence Length: model using cabin data (V, T, RH); derive from TODWL

ANALYSIS POTENTIAL – TURBULENT CONDITIONS

• scale analysis: power spectrum, structure function, autocorrelation

• analysis of variance: composite wave, inter-wave, intra-wave, sensor, sampling

• aerosol microphysics: identify and quantify sources of aerosol variability

ACKNOWLEDGMENTS

• This work was funded by the Office of Naval Research through the Center for Interdisciplinary Remotely-piloted Aircraft Studies and by the Integrated Program Office of NPOESS

• SPAWAR and ONR 35/SBIR Program provided the lidar and supported its integration into the CIRPAS Twin Otter

• IPO co-funded the lidar adaptation to the Twin Otter.

• IPO solely funded the mission planning, flight hours, data collection, and the post-flight installation of the lidar in a trailer for inter-flight research.