doppler lidar atmospheric wind sensor: reevaluation of a 355-nm incoherent doppler lidar

Doppler lidar atmospheric wind sensor: reevaluation of a355-nm incoherent Doppler lidar

David Rees and 1. Stuart McDermid

We reevaluate the performance of an incoherent Doppler lidar system operating at 354.7 nm, based on recentbut well-proven Nd:YAG laser technology and currently available optical sensors. For measurements in thelower troposphere, up to -5 km altitude, and also in the Junge-layer of the lower stratosphere, a wind

component accuracy of ± 2 m/s and a vertical resolution of 1 km should be obtained with a single pulse from a1-J laser, operating at Polar Platform altitudes (700-850 km) and high scan angles (55°). For windmeasurements in the upper troposphere (above -5 km altitude) and stratosphere (above and below the Jungelayer) the concentration of scatterers is much lower and higher energies would be required to maintain + 2 m/s

accuracy and 1 km vertical resolution, using single laser pulses. Except for the region in the vicinity of the tro-

popause (10 km altitude), a 5-J pulse would be appropriate to make measurements in these regions. Theworst case is encountered near 10 km altitude, where we calculate that a 15-J pulse would be required. To re-duce this energy requirement, we would propose to degrade the altitude resolution from 1 km to 2-3 km, andalso to consider averaging multiple pulses. Degrading the vertical and horizontal resolution could provide anacceptable method of obtaining the required wind accuracy without the penalty of using a laser of higher

output power. We believe that a Doppler lidar system, employing a near ultraviolet laser with a pulse energyof 5 J, could achieve the performance objectives required by the major potential users of a global space-bornewind observing system.

1. Introduction

The possibility of obtaining atmospheric wind pro-files, on a global basis, by direct measurement of Dopp-ler shifts has been considered for more than a decade.Passive interferometers (Hays et al.,' Killeen et al., 2

Rees et al.3

,4

) can observe natural airglow emissions toobtain wind measurements in the mesospheric andthermospheric regions (above 70-80 km). For the up-per troposphere and stratosphere regions (10- to 50-km altitude), observation of Doppler shifts of narrowabsorption lines in the backscattered solar spectrum isa powerful technique for measurement of winds overthe sunlit hemisphere. Suitable absorption lines aregenerated by molecular absorption within the earth'satmosphere, in addition to solar-generated Fraunhoferlines. A theoretical analysis of this technique has beenpresented by Hays,5 Rees et al. 6 and by Skinner et al. 7

Rees et al.6 and Rees8 have demonstrated, from highaltitude balloon flights, that a state-of-the-art triple-etalon Fabry-Perot interferometer can successfully

David Rees is with University College London, Department ofPhysics & Astronomy, London WC1E 6BT, U.K., and StuartMcDermid is with Jet Propulsion Laboratory, Table Mountain Fa-cility, Wrightwood, California 92397-0367.

Received 25 August 1989.0003-6935/90/284133-12$02.00/0.© 1990 Optical Society of America.

observe the absorption lines due to molecular oxygenand water vapour, considered by Hays,5 and can makewind measurements of comparable accuracy to balloonsondes from a platform above the earth's lower atmo-sphere.

An alternative passive technique for detecting wind-induced Doppler shifts in the atmospheric thermalemission spectrum has been suggested by McCleese etal.9 10 This approach uses gas correlation spectrosco-py, viewing the earth's limb, to determine the Dopplershift, probably using N20 and CO2 as the tracer gases.This would permit wind measurements over the alti-tude range from '20 to 120 km.

The highest priority requirements of operationalmeteorology for a truly global wind observing systemlare for global measurements of winds in the tropo-sphere and lower stratosphere (up to 25-30 km), withan altitude resolution of close to 1 km, particularlywithin the planetary boundary layer (PBL), and with awind accuracy of 4± 2 m/s. The passive absorption linetechnique cannot fulfill these requirements because oftwo limitations. First, although the best height reso-lution by a passive optical measurement is obtained byviewing at the limb, the altitude resolution of the pas-sive technique is limited to approximately 1/2 scaleheight, or 4 km, as an inherent property of limb-scan-ning geometry. Second, useful wind measurementscannot be made by the passive technique in the lowertroposphere, specifically within the PBL and up to 5-

1 October 1990 / Vol. 29, No. 28 / APPLIED OPTICS 4133

Table I. Summary of Menzles's1's Previous Assessment

Wavelength

Laser

530 nm

Nd:YAG, Doubled

350 nm

XeCl, H 2 Raman shifted

Linewidth <30 MHz

Telescope Primary area = m2

Receiver Tandem Fabry-PerotHRE gap = 30 cmHRE FSR = 500 MHzSpectral resolution = 50 MHz

Detector IPD with 12-ring anode

Optical Efficiency 0.020

Photocounts required 90 105

Photons required at >4.5 x 103 >5.25 x 103telescope primary

Required pulse energy -40 J.pulse& -100 J.pulse-for m.s- uncertainty

10-km altitude, because intense aerosol and Rayleighscattering makes the atmosphere, as viewed at thelimb, optically thick below 10-km altitude. Also,below -10-km altitude, the combination of pressure-broadening and optical thickness effects broaden thetarget absorption lines, reducing the sensitivity fordetermining accurate Doppler shifts and offsetting theincreased signal at lower altitudes in terms of the back-scattered sunlight.5

An active space-borne Doppler lidar system hasbeen seriously considered for making global wind ob-servations in the boundary layer, throughout the tro-posphere, and in the lower and middle stratosphere.Abreu12 and Hays et al.13 considered the performanceof an incoherent Doppler lidar system, based on detec-tion and frequency analysis of the backscattered signalfrom a frequency-doubled, single-mode Nd:YAG laserworking at 532 nm. They showed that the Rayleigh-scattered component could be separated from theaerosol-scattered component, and that the narrow lineobtained from the latter provided the best signal-to-noise ratio (SNR) for Doppler wind measurements.For several practical reasons, which are discussed be-low, we prefer the frequency-tripled Nd:YAG wave-length at 354.7 nm.

The technology of CO2 laser backscatter Dopplerlidars is well advanced, and a number of ground-basedand airborne systems have been developed, many ofwhich are described in the proceedings of the NASAGlobal Winds Workshop.'4 The comparative perfor-mance of several candidate laser/detector combina-tions in the role of a space-borne Doppler Wind LidarSystem which had been considered for use on the

Space Transportation System (STS-Shuttle), theSpace Station, or the polar-orbiting segments of theSpace Station (Polar Platform) was assessed by Men-zies.15 The incoherent systems he considered and hisconclusions are summarized in Table I. The specificconcern here is to reassess the incoherent DopplerWind Lidar System which Menzies'5 dismissed as hav-ing rather formidable energy and total power require-ments, compared with the CO2 isotope laser/hetero-dyne detection system. It is important to note thatMenzies' calculations were based on a target wind mea-surement uncertainty of 1 m/s. In this paper, thisrequirement has been relaxed to 2 m/s. This is inaccordance with more recent considerations of thewind measurement precision required for operationalmeteorology and this, in itself, leads to a significantreduction in the laser energies needed for a space-based system.

The baseline system parameters which are consid-ered here are summarized in Table II. The sourcecould be a frequency-tripled Nd:YAG laser, based onan injection-locked, Q-switched, unstable-resonatordesign as described by Park et al.,16-1 8 or the diode-pumped, frequency-stabilized Nd:YAG laser de-scribed by Zhou et al.,19 which would appear to havean advantage in higher power conversion efficiency.

II. Lidar Components

A. Doppler Receiver

The detector system considered has several particu-lar features. A very high resolution, capacitance-sta-bilized, Fabry-Perot interferometer20'21 would be used

4134 APPLIED OPTICS / Vol. 29, No. 28 / 1 October 1990

I J.pulse- (1.79 x 1018 photons.pulseI)

20 MHz ,

I m2

0.5 mrad

I km (vertical)

Triple Fabry-Perot

-2.5 GHz, 500 m.s1

-25 MHz

0.20 (complete system)

32-ring IPD

17.5 MHz, 3 m.s1

25%

as a spectral analyzer. Additionally, two lower resolu-tion capacitance-stabilized etalons, used in conjunc-tion with an interference filter, would reduce the un-wanted signals from scattered sunlight and from therelatively broader Rayleigh scattered signal (discussedlater) to a low and acceptable level. A multichannelimaging photon detectorl' 22' 23 would provide effectiverange-gated low noise signal detection. The optics,window and photocathode of the detector would beoptimized for the spectral region near 355 nm.

In the NASA study of the operation of the LAWSspace-borne Doppler lidar, a conical scan, at an angleof between 450 and 600 to the nadir was selected as theprimary observing strategy. This is illustrated in Fig.19 of the NASA-LAWS report,2 4 where it is also de-scribed in detail. From a space platform at 700-850-km altitude, and with a laser pulse repetition frequen-cy of -10 Hz, a conical scan strategy makes it possibleto obtain a swath coverage providing twice-daily globalcoverage at the equator with a horizontal resolution of50-100 km. Although we could have considered otherstrategies, for example a push-broom; for consistency,we have used the same geometry as the basis for ourperformance calculations. For the incoherent windlidar system under consideration, the use of a tripleFabry-Perot interferometer within the detector makesboth daytime and nighttime measurements possible,again providing consistency with the strategy consid-ered in the NASA-LAWS report.24

A capacitance-stabilized etalon design was chosenfor the Fabry-Perot etalon elements of the detector, inpreference to the fixed-gap, zerodur spacer design con-sidered by Abreu12 and by Menzies.15 The design andperformance of the zerodur-spaced etalon is describedin detail by Rees et al.

2 5 and Killeen et al.2 6 The

capacitance-stabilized etalon design20' 21 is essential.A large, variable, but predictable Doppler shift of thelidar backscattered signal is introduced by the combi-nation of the orbital velocity of the space platform, therotation of the earth, and the variable viewing geome-

try of the lidar scan pattern over the earth's surface.This variable Doppler shift is as much as 10 km/s line-of-sight, 2 orders of magnitude greater than the peakwinds (100 m/s), which might be detectable in jetstreams in the upper troposphere and lower strato-sphere, or associated with stratospheric sudden warm-ing phenomena.

It was assumed by Abreu12 that it would be neces-sary to illuminate and detect at least one complete freespectral range (FSR) of the Fabry-Perot etalon used inthe incoherent detector system. A solid angle corre-sponding to the complete free spectral range of theetalon would also be detected in the focal plane. Thisrequirement was based on the assumption that itwould not be possible to predict accurately where thevariable line-of-sight Doppler shift would place thereturned signal (in absence of atmospheric winds).This assumption appears to have been accepted byMenzies,15 in his evaluation of comparative perfor-mance of incoherent and coherent Doppler lidar sys-tems.

There are two compelling reasons to use a tunablehigh resolution etalon (HRE) as the primary means ofspectral analysis of the backscattered signal. First,current state-of-the-art imaging photon detectors(IPD) can only operate in one of two modes. In a 2-Dimaging mode, the photon signals are processed se-quentially, limiting the maximum peak photon rate(total device) to -5 MHz. This, however, does notmeet the peak photon detection rate requirement formeasurements within the PBL. Using an anode struc-ture comprised of discrete elements within the IPD,much higher photon counting rates per anode elementcan be obtained, typically 20 MHz per channel for a 20-channel device. The anode elements of such devicescan also be shaped to match constant wavelength in-crements in the focal plane of a Fabry-Perot interfer-ometer (i.e., concentric, equal area annulae). We havetaken the present performance limits to be 20-MHzphoton counting rate for each of thirty-two equal areawavelength channels in the image plane of the Fabry-Perot interferometer. This is the specification of adevice currently under commercial development.2 7

As such, this new device is a modest technical improve-ment on the 20-channel IPD already demonstrated forDoppler Wind Lidar applications by Rees et al. 2 8

Such a detector can process photon signals at the peakrates required by the proposed lidar system.

Very narrow linewidths can be generated by single-mode lasers and the spectral width of the aerosol scat-tered atmospheric return signal is similarly very nar-row. To obtain the ± 2 m/s accuracy adopted asrequired for a wind measurement for operational me-teorology, and in view of the current technical perfor-mance limits of an IPD, the Doppler Wind Lidar mea-surement requirements can only be met if a small partof a free spectral range of the high resolution etalon isimaged onto the detector. Otherwise, the narrowspectrum of the aerosol-scattered signal is inadequate-ly sampled by the discrete channels of the detector.

Second, and equally importantly, the available illu-


Table II. Ldar System Parameters

Laser Energy, E

Laser Linewidth

Telescope Area, A

Telescope FOV

Overlap Function, ((R)

Range Resolution, AR

Receiver

HRE FSR

HRE Spectral Resolution

Receiver Transmission

Detector

Detector Channel Width

Detector QEl

1 4 8 12 16 20 24 28 32IPD DETECTOR RING/CHANNEL NUMBER

Fig. 1. Illustration of the scheme used for illuminating the far fieldwith the backscattered laser light. Traditionally, a complete freespectral range of the Fabry-Perot etalon in the Doppler receiver isilluminated in the far field. This is illustrated in the top section ofthe figure. Only a small part of the laser light backscattered fromthis relatively large solid angle can be potentially transmitted by thereceiver-that close to the wavelength of the backscattered signal.A region of 0.2 FSR about this wavelength is illuminated, and thedetector optics are modified, so that this region in the far fieldilluminates the entire detector (thirty-two channels). The bottompanel shows a schematic of the recorded spectrum, thirty-two chan-nels each 3 m/s wide, with the aerosol scattered spectrum illuminat-

ing three adjacent channels.

mination from the laser is the most important cost-driver of a space-borne (or even ground-based) lidarfacility. If the laser beam only illuminates the regionof the target corresponding to that portion of the freespectral range where light can be transmitted and usedefficiently by the detector, an efficiency advantage ofthe order of the finesse of the high resolution etaloncan be obtained, compared with a system in which theentire free spectral range is illuminated. This is illus-trated in Fig. 1. The OPD of the tunable HRE canalways be optimized by compensating for variablespacecraft-ground Doppler shifts to ensure that thisoptimum illumination is maintained. To achieve thespectral resolution necessary to implement thisscheme, i.e., to ensure that the detected signal fromaerosol scattering always covers two or more discretespectral channels, the product of the etalon gap (inmm) and the finesse (reflective plus defect) must be atleast 6000. In terms of practical designs which couldbe readily space-qualified, with present state-of-the-art optical technologies, the etalon gap could be up to120 mm with a maximum achievable finesse on theorder of 100. The capacitance stabilization of theoptical path difference of an etalon is straightfor-ward20'2 ' and has been demonstrated in field instru-ments. We should note that Hays et al.13 consideredoptimizing the etalon plate separation of the HRE

from a theoretical point of view and deduced that a gapof 30 cm would be optimum. While their numericalanalysis was rigorous, it did not include considerations,such as the practical aspects of building and stabilizingan etalon of such a large plate separation, or the engi-neering requirements of mounting such a device in aspace-borne instrument or on a space platform. Froma practical point of view, the fabrication and spacequalification of an etalon of -6-10-cm plate separa-tion would not pose particular difficulties, and thecapacitance stabilization of such an etalon is straight-forward. Etalon plate separations significantly in ex-cess of 10 cm would generate a variety of engineeringproblems in designing and fabricating the etalon andits mount, to space standards. More importantly, theoperational flexibility introduced by capacitance sta-bilizing the HRE no longer makes it mandatory to uselarge gaps to obtain optimum performance.

At any one particular setting of the HRE gap, a line-of-sight wind velocity range of 100 m/s (560 MHz, or0.2 FSR of the HRE) can be expanded to fill the detec-tor and the Doppler-shifted backscattered signal canbe observed in an optimum way. Figure 1 shows thescenario for a high resolution etalon gap of 60 mm (i.e.,-500 m/s FSR). In the focal plane, 0.2 FSR of theHRE (which subtends 0.45 of the radius of one com-plete FSR) is expanded to fill the multiring detector(32 channels). The capacitance stabilization and PZTdrivers of each of the three etalons allow the appropri-ate 0.2 FSR (corrected for the line-of-sight componentof the spacecraft motion and earth rotation) to beselected near the optical axis. The implicit assump-tion is made that changes of the line-of-sight velocitydue to variations of orbit and spacecraft attitude, andthe relative orientation of the conical scan mechanism,are predictable to much better than +100 m/s and thatimmediate a posteriori attitude analysis will providethe line-of-sight baseline velocity to +1 m/s. Thus, forthe 32-channel detector each channel will have a spec-tral width on the order of 17.5 MHz or 3 m/s. For alaser linewidth of 20 MHz or less the spectral resolu-tion of the system is limited by the etalons within thedetector system. For the specified product of thefinesse and etalon gap of 6000 mm, the observed spec-tral width of the aerosol backscattered peak is -25MHz. Since the spectral width of each detector chan-nel is 17.5 MHz, there will always be some illuminationin at least two and possibly three channels of the detec-tor. This assures that the centroid of the backscat-tered spectrum can be determined to 2 m/s. Thisplaces requirements on the minimum number of de-tected photons per laser pulse and per range elementwhich are discussed later.

B. Detector Efficiency

The detection of incident photons occurs within aphoto-emissive detector, following transmissionthrough a number of optical components. The overalldetection efficiency (QD) is given by:

(QD) = RlTlT27'3T4RQEIPD,


where R, = reflection coefficient of the input tele-scope (0.9),

R2 = reflection coefficient of the camera tele-scope (0.9),

T = transmission coefficient of the interfer-ence filter (0.6),

T 2 = T 3 = T4 = transmission efficiency of the Fabry-Perot etalons (0.75 each), and

QEIPD = quantum efficiency of the MCP detec-tor (0.25 at 355 nm).

For these values, (QD) = 0.04 to 0.06, noting uncertain-ties in some of the values.

The system parameters of the laser and detector, assummarized in Table II, will therefore be used, incombination with an atmospheric model which is onlyslightly different to that used by Menzies.15 The val-ues are relatively conservative and represent realistictwo-pass attenuation losses plus conditions of relative-ly low dust loading of the atmosphere.

In agreement with Abreu,12 Hays et al.,13 Rosenbergand Sroga29 and Menzies,15 it is the aerosol-scatteredsignal which provides the best SNR for Doppler windsensitivity, since the aerosol backscattered signal is notDoppler broadened, except by wind shear and turbu-lence, parameters which are the ultimate objective ofthe measurements. The outgoing laser beam has aspectral width which corresponds to 2-4 m/s. Thedetector only detects a spectral region, close to theoptical axis of the interferometer, corresponding to 100m/s, with each of thirty-two channels having a spectralwidth of 3 m/s. The 100 m/s corresponds to the maxi-mum line-of-sight (l-o-s) winds which are expected.To follow the changing wavelength of the signal due tothe l-o-s component of the spacecraft motion and earthrotation, to an accuracy of + 1 m/s, the optical pathdifference of each of the three etalons has to be appro-priately stepped. The rate of stepping, synchronizedwith the reception of each backscattered laser pulse,may correspond to 1 FSR per pulse, for a 10-Hz pulserepetition rate, and a 10-s conical scan period.2 4 Suchinformation (based on spacecraft orbit, earth rotationand line-of-sight viewing direction relative to thespacecraft velocity vector) is expected to be readilyavailable on sophisticated space platforms, such as theproposed NASA/ESA Polar Platforms, and other ele-ments of the Space Station. This information is alsoavailable on-board spacecraft such as the NOAA-TIROS series. Corrections, due to earth oblateness,the slight ellipticity of the orbit required to avoidprecession of perigee in a sun-synchronous orbit, andthe effects of yaw steering of the platform can all becorrectly and accurately computed.

C. Laser Source

The proposed laser source for this lidar system is afrequency tripled Nd:YAG laser. Throughout the1980s there has been significant progress, in particular,by the group at Stanford, lead by Professor Byer, in theachievement of stable, Fourier transform limited sin-gle axial mode operation of Nd:YAG oscillators. Forexample, Park et al.16-18 have measured experimental-

30

25

20E

o 15

-J< 10

5

0'1 .OE+05 1.OE+06 1.OE+07

AEROSOL NUMBER DENSITY (Particle.m'3 )

1 .OE+08

Fig.2. Aerosol distribution model, showing aerosol number densityas a function of altitude (0-30 km), as used in the atmospheric

backscatter calculations.

ly an optical bandwidth of 10 MHz from a Q-switchedNd:YAG oscillator. Such systems are rendered evenmore attractive with the advent of diode pumping andthe discovery of more efficient, higher damage thresh-old nonlinear crystals such as beta-barium borate.Even with readily available (in 1988) commercial sys-tems it is possible to obtain single axial mode ouptut atenergies up to 3 J/pulse at the fundamental wave-length (Quantel/Lightwave Electronics). It is antici-pated that such companies will continue to make ad-vances, especially in the area of diode pumping. Thus,the laser parameters used in this evaluation are be-lieved to be realistically achievable and the only signif-icant development would be the production of a diodearray large enough to pump the Nd:YAG laser. At thepresent time, flashlamp technology may be suitable forspace qualification of such a laser.

Ill. Performance Calculations

A. Atmospheric Backscatter

The Rayleigh component of the atmospheric back-scatter can be readily and accurately calculated.However, the backscatter from aerosol particles ismodel dependent, somewhat subjective, and there isno simple and adequate descriptive function. For ex-ample, the ratio of the backscatter coefficient at 694nm (ruby laser) to that at 10.6 ,m (CO2 laser) has beenmodeled as 20 to 4030 but was measured in one study as400 to 1000.31 In this study, the tables derived byDeirmendjian,32 from fundamental scattering theory,were used to establish effective backscatter cross sec-tions. These cross sections have then been combinedwith the aerosol number density profile developed byWright et al.33 which is shown in Fig. 2 for altitudes upto 30 km. The same model was used for some of thecalculations in the LASA Report 3 4 which considered alidar system on the Eos platform to measure aerosolsand some atmospheric chemical species. This modelgives a variation on the order of 2 orders of magnitude


| -

i

between the aerosol backscatter at near-UV wave-lengths and wavelengths -10 Am, in general agreementwith the conclusions of Megie and Menzies.35

For single scattering the volume scattering coeffi-cient, ,3sc, is defined by I.Sc = WIex, where ex is thevolume extinction coefficient and w is the albedo ofsingle scattering. In his monograph Deirmendjian32

has tabulated values of flex and w for various wave-lengths and aerosol size distributions. These values ofOex correspond to an aerosol number density of 100cm 3. For a polydispersion of spherical particles (Mietheory) the phase function, which indicates the angu-lar distribution of scattered radiation is given by

Pj(OnX) = 4r OA,, j = 1,2P.(O~~~nX)=4 ,,\

For 1800 backscatter this can be written as

PjwX =4 j(r,X)

Pj(~r) = f(X)

If the depolarization caused by the scattering is negli-gible, or if the detector used is polarization insensitive,then

P,,(X) = 41".

Values of the phase function, P/47r, have also beentabulated by Deirmendjian3 2 which therefore allowsthe calculation of the backscattering coefficient corre-sponding to 100 particles/cm3 , #X = P/rfls. The shor-test wavelength considered by Deirmendjian was 450nm. Using the above relationships and recalling thatthe backscattering coefficient is the product of thebackscatter cross section and the particle number den-sity, the backscatter cross section was calculated andplotted as a function of wavelength as shown in Fig. 3.The calculations used the Water Haze Model M which

E- 19

Z 0 3550

1 E-20

0

.. 1E-21

0

1 E-220.1 1 1 0

WAVELENGTH (Microns)Fig. 3. Aerosol backscatter cross section as a function of wave-

length for a marine type aerosol.

corresponds to a marine type aerosol and, thus, is mostrepresentative of the global atmosphere. From a shortextrapolation the aerosol backscatter cross section at354.7 nm is found to be 4.3 X 10-20 km2 .

The Rayleigh component of the atmospheric back-scatter was derived using the molecular number densi-ties from the U.S. Standard Atmosphere3 6 and theRayleigh scattering cross-section calculated from theformula 3 7 d(w)/dQ = 5.45(X/0.55)- 4 X 10-38. Thus,the Rayleigh backscatter cross section at 354.7 nm is3.49 X 10-37 km2 .

Table III shows some of the important values andintermediate results in the calculation of the atmo-spheric backscatter. The scattering ratio has beenincluded in this table for comparison with experimen-tal data. In lidar experiments at JPL-TMF, scatter-

Table ll. Atmospheric Backscatter Parameters

4138 APPLIED OPTICS / Vol. 29, No. 28 / 1 Octoler 1990

ing ratios between 1.05 and 1.10 have been observed inthe lower stratosphere for 355-nm radiation 3 8 whichagrees well with this model. Intense scattering fromthe high density of aerosols trapped in the planetaryboundary layer is also well modeled.

The final atmospheric parameter required to esti-mate the lidar signal returns is the two-way transmis-sion factor. For simplicity the tables from Elterman39

were used since they provide optical thicknesses forboth aerosol and Rayleigh extinctions (and ozone al-though this is negligible at 354.7 nm) from variousaltitudes to infinity, i.e., space. It is felt that this isfurther justified since the atmospheric attenuation at355 nm is dominated by the Rayleigh term and thusany differences in the aerosol model between thisstudy and Elterman's model will not cause significanterrors.

B. Lidar Returns

Using the data from Table III, the lidar equation canbe solved to give the number of photons per pulsearriving at the telescope primary. The efficiency ofthe detector is considered separately and in detailelsewhere in this paper. The form of the lidar equa-tion used is

S = E t(R) ,B T2 AR sec/ A/(R secO)2 ,

whereE = laser energy in photons per pulse,

t(R) = overlap function of the telescope and laserfields-of-view,

f = atmospheric backscatter coefficient,T = atmospheric transmission,A = receiver area,R = range,

AR = range resolution, and0 = conical scan angle (i.e., angle from nadir)

Several combinations of spacecraft orbit altitudeand conical scan angle have been suggested. In Men-zies's15 analysis he considered a scan angle of 550 froma platform altitude of 800 km since this would providetwice daily global coverage at the equator. This isrepresentative of the parameters expected for a free-flying satellite for wind measurements (i.e., Windsat).The LAWS report 2 4 also considered two other scenari-os. A Space Shuttle based system, which would orbitat 300 km, requires a slightly greater scan angle toobtain the same global coverage from the lower alti-tude, and a Space Station system, which would be at500-km altitude. The lidar equation has been evaluat-ed to estimate the lidar returns to the telescope prima-ry for all three of these scenarios (300 km, 600; 500 km,600; 800 km, 550) and the results are given in Table IV(a)-(c), respectively. The same set of lidar parame-ters, as listed in Table III, was used for all platformaltitudes even though it has been suggested that theSpace Station could accommodate a larger telescopeand a more powerful laser.

The atmospheric model used assumes a low, back-ground level aerosol loading. An exception to this

Table IV. Ldar Returns at Telescope Primary

Altitude Rayleigh Aerosol(km) (photon/pulse) (photon/pulse)

(a) Spacecraft Platform 300 kmScan Angle 600

0 2573 2486

5 11244 634

10 12816 70

15 9373 108

20 5400 342

25 2763 192

(b) Spacecraft Platform 500 kmScan Angle 600

0 926 895

5 4048 228

10 4614 25

15 3374 39

20 1944 123

25 995 69

(c) Spacecraft Platform 800 kmScan Angle 550

0 666 643

5 2237 126

10 2322 13

15 1601 19

20 895 57

25 447 31

would occur following a volcanic eruption, or otherprocess, which would enhance the atmospheric aerosolloading. In this case, the aerosol returns would begreatly enhanced, even though there would be a partialoffset at lower altitudes due to the reduction in atmo-spheric transmission.


C. Spectral Distribution of the Lidar Signal

The signal return numbers shown in Table IV illus-trate that it is necessary to distinguish between theaerosol and Rayleigh scattered signals above the PBL.Although there is an adequate aerosol signal return perpulse; it could be drowned by the statistical noise of theRayleigh scattered signal.

The spectral intensity distributions of both the Ray-leigh and the aerosol returns can be represented by aGaussian profile,

1(v) = IO exp[-(v - v0)210.36Av2 ] [1/(4 ln2) = 0.36].

However, the halfwidth Av is different for each case.At thermal equilibrium, the molecules in the air have aMaxwellian distribution of velocities, each of whichwill cause a Doppler shift of the incident radiation.The result is an overall Doppler broadening of theRayleigh lidar return. The Doppler width is given by,

ACOD = (2w0/c)V(2RTln2/M)

where R is the gas constant, T is the temperature, andM is the molecular weight. In frequency units andsubstituting for R and c,

AvD = 7.16 X 10-7voV/(T/M) (Hz)

Thus, at 354.7 nm (8.46 X 1014 Hz) with a mean molec-ular weight for air of 28.96 gm/mole and using thetemperature at 5-km altitude of 255 K, the width(FWHM) of the Rayleigh return is 1.80 GHz. Sincethe Doppler width varies with a it is not overly sensi-tive to the atmospheric temperature profile. For therange of temperatures from sea level to 25 km, thevariation in the width of the Rayleigh return is lessthan 15%.

Due to the low thermal velocity of the heavier aero-sol particles, the spectral distribution of the aerosolbackscatter essentially follows that of the laser source.As discussed above this is characterized by a linewidthon the order of 20 MHz.

Taking these halfwidths and using the magnitude ofthe signal returns from Table IV for the areas underthe respective profiles, the backscatter spectrum canbe synthesized. As an example, the atmospheric back-scatter spectrum corresponding to the signal returnedfrom 5-km altitude and for a scan angle of 550 from an800 km oribt in Fig. 4. Figure 4(a) shows the completespectrum, demonstrating the spectral broadening ofthe Rayleigh signal compared to the aerosol return.The region, corresponding to approximately 100 m/s,that would be expanded to fill the Doppler detector isindicated in Fig. 4(b).

The small free spectral range of the HRE, coupledwith its very high spectral resolution, act to reduce thenoise effects of the Rayleigh signal, except for that partwhich falls within the bandwidth of the aerosol scat-tered signal. The free spectral range (60 mm OPD,-2.5 GHz) encompasses most of the Rayleigh scat-tered signal and this effectively results in a constantbackground signal underlying the aerosol return as isshown in Fig. 4(b). For this example the Rayleighsignal component is -20% of the aerosol signal. The

8

7

6-J

>

-J-Jz0j

5

4

3

2

-2000 -150 -1000 -500 0 500 1000FREQUENCY (MHz)

8

7

-:I_J 5w> 4-J

1 3

2ui

0 L_-300 -200 -100 0 100

FREQUENCY (MHz)

1500 2000

200 300

Fig. 4. Spectral distribution of the combined aerosol and Rayleighscattering return, calculated for 5-km altitude in the assumed modelatmosphere. The upper panel shows the complete spectrum over arange of + 2000 MHz. The lower panel shows, on an expanded scale,the spectrum within one free spectral range of the detector of the

proposed Doppler Wind Lidar System.

remainder of the signal is spectrally distinguished bythe HRE and falls in detector channels outside thosesampling the aerosol return. Rayleigh signals frommore than one FSR away from the central aerosolreturn are additionally attenuated by the spectralshape (assumed Gaussian) of the Rayleigh signal (seeFig. 4 and the equations discussed earlier) and by theattenuation due to the second and third etalons of thetriple etalon FPI in the detector. The worst caseinterference is likely to occur around 10-15-km alti-tude where the Rayleigh signal is approximately equalto the aerosol signal in the detector channel corre-sponding to the peak return. Within the PBL, wherethe aerosol loading is highest, the Rayleigh componentfalls to <1% of the signal in the aerosol return channel.

D. Solar Background

Since this system must be capable of making mea-surements during both the daytime and nighttime, thepossible impact of solar radiation should be consid-ered. The magnitude of the solar spectral irradianceat the top of the atmosphere is given by Thekaekara40


1

and for 354.7 nm this is -1220 mW/m 2 /nm. For thepurpose of this consideration, it is assumed that theattenuation of the solar radiation by the atmosphere isequally balanced by the scattered sky radiance. Forreflections from the Earth's land surface Lambertianscattering can be assumed such that,

I(A) = Io(X)p cosO/hr,

where p is the albedo and 0 is the solar zenith angle.Scattering from the ocean surface is a function of seastate and wind speed and the solid angle of the scatter-ing can be much less than for the Lambertian model.This has been studied for laser backscattering from theocean by Bufton et al.41 and the enhanced reflectanceof the ocean surface, for a narrow range of angles, wasdefined by an effective Lambertian reflectance. Anenhancement of a factor of ten over the Fresnel reflec-tance of the ocean surface was found in this study,4'giving a maximum effective Lambertian reflectance of0.2. Typical values of the land surface albedo are alsoin the region of 0.2. Therefore, for this example thesun is assumed to be at zenith and reflected from thesurface with an albedo of 0.2. Thus, the zenith-up-ward spectral radiance is 76 mW/m2/sr/nm. The mag-nitude of the received background solar radiation isgiven by the telescope receiver function,

SBG = I(A)M22AR/c J t(X)dX I(X)AQ2AR/ct(X)AX

where Q is the solid angle subtended by the telescope.For small angles, as in the lidar receiver, Q a 7r02/4,where k is the full plane angle of the telescope field-of-view. Using the system parameters in Table II, theirradiance at the telescope primary can be calculatedyielding a value of,

7.4 X 10-2 photons/pulse/range element/MHz.

Thus, for instrument spectral channel widths on theorder of 20 MHz, the solar background is on the orderof 1 photon/pulse. This should be compared with thelidar returns in Table IV, which then indicate thatsolar background will not limit the performance of thissystem.

E. Determination of Doppler Shift: Spectral Position

The system described above assumes that, given thevery narrow bandwidth of the laser source and thesmall broadening due to the Mie scattering process,the l-o-s wind component can be uniquely determinedwith a precision on the order of d2 m/s by detecting avery small number of photons per range element andper laser pulse. This assumption is justified by thefollowing brief discussion.

The precision to which the position (frequency) ofthe aerosol scattered signal can be measured is a func-tion of the spectral width of the signal, the noise associ-ated with that signal, and the resolving power of thespectrometer.4243 The system described here uses aphoton counting detection scheme which is governedby Poisson statistics. For the case where a spectralline is sampled at equidistant frequency increments,

Gagne et al.4 2 have shown that the centroid of the line

can be determined from the infinite summation,i=. i=- -1

v ={ viSi Si}

where vi = the frequency at the ith channel and Si = thenumber of photons counted in that channel.

It has further been shown42 that the variance (2) ofthis determination of the centroid (for a Dopplerbroadened line) is given by,

2 = Av2 18N ln(2)-'

where Av = FWHM of the spectral line and N = is thetotal number of photons counted.

For this study, the required wind measurement un-certainty, which corresponds to the variance in thedetermination of the centroid of the backscatteredsignal, is 2 m/s and the anticipated FWHM of thissignal is 3 m/s (-20 MHz). Application of the aboveequation indicates that, for these selected parameters,and in the absence of background noise, a single pho-ton is sufficient to determine the centroid position tothe precision required. In practice, an allowance mayhave to be made for the finite resolution of the interfer-ometer but, since this is carefully matched to thelinewidth of the aerosol return, this will not be a signif-icant factor. This shows that the determination of thecentroid frequency of the backscattered signal is not alimiting step in the wind measurement, even for verysmall numbers of detected photons per laser pulse.

F. Determination of Doppler Shift: SNR

Using the Poisson statistics of the photon countingdetection system, the SNR of the aerosol return for asingle channel of the receiver is given by,

SNRA = NA * (NA + NR + NS +_ND)0.5

where NA = number of counts due to aerosol scatter-ing,

NR = number of counts due to Rayleigh scatter-ing,

NS = number of counts due to solar background,and

ND = number of counts due to detector noise.To obtain the numbers of counts due to aerosol and

Rayleigh scattering in a single, 17.5 MHz wide channel,the calculated signals listed in Table IV must be con-sidered along with the spectral distribution of the lidarreturn shown in Fig. 4. Thus, for the Rayleigh returnwe find that the signal per channel, near the centralwavelength, is approximated by the total Rayleigh sig-nal multiplied by the ratio of the receiver channelwidth (17.5 MHz) and the FWHM of the Rayleighpeak (1.8 GHz). The worst case for the aerosol returnwould occur when it straddles two adjacent spectralchannels and, therefore, the aerosol signals from TableIV were halved for the SNR claculation. The solarand detector background levels are negligible. In hisanalysis, Menzies'5 considered that the threshold sig-nal level for the detection of the Doppler shift corre-


25

20

S 15

gb'i

- 10-J

5

0.0

Fig. 5. Signal-to-nsis of the detected sources. Values ar

sponded to a SI!used here.

For a platforangle, which r(produce the loyaerosol return ipulse with 1-kn1.5 at 5 km, anc<1. To determ15, and 25 km)dlude: increasmultiple pulsestal resolution, dgrading the winple, by increasicalculated SNRkm, 1.0 at 15 kncalculated SNRFig. 5 for singlepulse energy ofthe threshold fpulse, keeping avalues stated eakm. We note h(does not includeble and subvisiWhile a lidar wican view throulsuch clouds wotfrom this altituexpense of signmay allow us tokm without incrHowever, the ol(vertical, horizoaround 10 kmsubvisual cirrus

G. Eye Safety

For a space-biearth's surface,

cern. For the wavelength being considered, 354.7 nm,the American National Standards Institute (ANSI)has recommended a maximum permitted exposure(MPE) of 6 X 10-3 J/cm 2 which is similar to the valuefor far infrared wavelengths. This should be com-

-s-i Jpared with the MPE for visible and near infrared-*5 ~J wavelengths of -5 X 10-7 J/cm2 . For the laser ener-

gies that would be required for a space-borne windsensor, only lasers at wavelengths >1400 nm or <400nm can meet the eye safety requirements.

IV. Summary

The performance of a space-borne, near-ultraviolet2 3 4 5 6 7 8 Doppler wind lidar system has been reevaluated using

SNR available optical detector technologies and an instru-Loise ratio calculated for the Doppler wind analy- mentation/observing strategy which optimizes the de-single-pulse) lidar signal scattered from aerosol tection and use of the atmospheric aerosol scatteredshown for a lidar output of 1 J and 5 J per pulse. lidar signal.

While 1 J/pulse laser would be adequate, under1R of unity. The same definition was worst conditions, for wind speed measurements up to 5

km altitude, assuming observations from a Polar orbit-im altitude of 800 km and a 550 scan ing spacecraft at 800 km altitude, laser energies of 5 J/:present the conditions which would pulse would be required for measurements between 5vest signals, the calculated SNR of the and 25 km. Near the tropopause (10 km) a lasern a single channel, and for a 1-J laser energy of 15 J/pulse would be needed unless there werevertical resolution, is 4.0 for the PBL, subvisual cirrus clouds present or the spatial resolu-

I 1.1 at 20 km. All other altitudes are tion was degraded. The wind velocity accuracyine the wind at the other altitudes (10, achievable is h2 m/s and the available power budgetthere are several options. These in- for such a laser allows implementation of a conical scaning the laser pulse energy, averaging strategy for full vector wind determination with betterand, thereby, decreasing the horizon- than 100-km horizontal resolution over a swath -1500egrading the range resolution, and de- km wide, centered on the suborbit track.id measurement accuracy. For exam- At an energy rating of 5 J/pulse, a frequency-tripledng the laser pulse energy to 5 J the Nd:YAG laser, operating at 354.7 nm, is the primeis 8.9 for the PBL, 3.4 at 5 km, 0.6 at 10 contender with no serious problems due to slightly

a, 2.3 at 20 km, and 1.7 at 25 km. The reduced efficiency compared with the fundamentalI, as a function of altitude, is shown in (1060-nm) or frequency-doubled (532-nm) emission;laser pulses of 1 J and 5 J. For a laser 354.7 nm is inherently eye-safe at anticipated power5 J, the only altitude region for which levels and surface intensities, due to absorption withinr detection is not reached for a single the cornea.ll parameters and requirements at the Such a Doppler wind lidar system is competitiverlier, is that near the tropopause at 10 with the CO2 Doppler wind lidar, which has been con-

3re that the aerosol model we have used sidered previously (Menzies,15 NASA-LAWS24). Wee any clouds. Cirrus clouds, both visi- also note that both of these reports ended with a dis-al, tend to form at the tropopause. missal of the incoherent Doppler lidar due to a high

11 not penetrate visible cirrus clouds it pulse power requirement. The system discussed in;h subvisual cirrus. The presence of this report makes a more efficient use of the outputild greatly enhance the aerosol returns laser power by matching the usable and detectablede region, although somewhat at the field-of-view of the receiver (Fabry-Perot interferome-al from lower altitudes. This factor ter) with that of the transmitter. In eye-safe regionsachieve the threshold level SNR at 10 this is permissible while at other laser wavelengths iteasing the laser pulse energy above 5 J. would not be. The wind accuracy requirements con-)tion to degrade the spatial resolution sidered, 2 m/s, is slightly degraded in comparisonntal, or both) is available for a region with the LAWS target (1 m/s), however, the achiev-(8-13 km approx.) in the absence of able accuracy is consistent with requirements for oper-

ational meteorology.Advantages would accrue from the lower output

power requirement which is a major system cost driverrne laser system directed towards the for the Spaceborne Doppler Wind Lidar. Minor gainseye safety is a factor of particular con- would result from the lower power since development


of the laser would require a smaller extrapolation ofthe product of power and required lifetime comparedwith systems already demonstrated. Although it wasnot considered in detail, it should be noted that it isalso feasible to use a higher repetition rate laser, atreduced pulse energy. However, the total laser powerand dissipation cannot be reduced if wind accuracyand spatial resolution are to be maintained.

A number of detailed cost/benefit studies referencedin this paper have already highlighted the wide-spreadscientific advantages of global wind measurements,which would provide accurate and uniform data on themotion field to complement available satellite tem-perature data. The lack of uniform and accurateknowledge of the atmospheric motion field is now be-lieved to be the major cause of uncertainty in short-term and medium-term numerical meteorological pre-diction models.

The work described in this paper was carried out atUniversity College London supported by grants fromthe U.K. Science and Engineering Research Counciland at the Jet Propulsion Laboratory, California Insti-tute of Technology, through an agreement with theNational Aeronautics and Space Administration. Weare grateful to R. T. Menzies for critically reading thismanuscript and providing many useful suggestions.

References1. P. B. Hays, T. L. Killeen, and B. C. Kennedy, "The Fabry-Perot

Interferometer on Dynamics Explorer," Space Sci. Instrum. 5,

395-416 (1981).2. T. L. Killeen, B. C. Kennedy, P. B. Hays, D. a. Symanow, and D.

H. Ceckowski, "Image Plane Detector for the Dynamics Explor-

er Fabry-Perot Interferometer," Appl. Opt. 22, 3503-3513(1983).

3. D. Rees, T. J. Fuller-Rowell, R. Gordon, T. L. Killeen, P. B.

Hays, L. E. Wharton, and N. W. Spencer, "A Comparison of The

Wind Observations from the Dynamics Explorer Satellite with

the Predictions of a Global Time-Dependent Model," Planet.Space Sci. 31, 1299-1314 (1983).

4. D. Rees et al., "The Westward Thermospheric Jet-Stream of the

Evening Auroral Oval," Planet. Space Sci. 33, 425-456 (1985).

5. P. B. Hays, "High-Resolution Optical Measurements of Atmo-

spheric Wind from Space. 1. Lower Atmosphere Molecular

Absorption," Appl. Opt. 21, 1136-1141 (1982).

6. D. Rees, P. A. Rounce, P. Charleton, T. J. Fuller-Rowell, I.

McWhirter, and K. Smith, "Thermospheric Winds During theEnergy Budget Campaign: Ground-Based Fabry-Perot Obser-

vations Supported by Dynamical Simulations with a Three-Dimensional, Time-Dependent Model," J. Geophys. Res. 50,

202-211 (1982).

7. W. R. Skinner, P. B. Hays, and V. J. Abreu, "Optimization of a

Triple Etalon Interferometer," Appl. Opt. 26,2817-2827 (1987).

8. D. Rees, "Balloon-Based Interferometric Techniques," in Pro-

ceedings of the NASA Symposium on Global Wind Measure-ments, W. E. Baker and R. J. Curran, Eds. (Deepak Publishing,Hampton, VA, 1985), pp. 109-114.

9. D. J. McCleese and J. S. Margolis, "Remote Sensing of Strato-

spheric and Mesospheric Winds by Gas Correlation ElectroopticPhase-Modulation Spectroscopy," Appl. Opt. 22, 2528-2534

(1983).

10. D. M. Rider, J. T. Schofield, and D. J. McCleese, "Electrooptic

Phase Modulation Gas Correlation Spectroradiometry," inTechnical Digest, Toptical Meeting on Laser and Optical Re-mote Sensing: Instrumentation and Techniques (Optical So-ciety of America, Washington, DC, 1987), pp. 226-229.

11. C. P. Arnold, et al. "Results of an Observing System SimulationExperiment Based on the Proposed WINDSAT Instrument," inProceedings of the NASA Symposium on Global Wind Mea-surements, W. E. Baker and R. J. Curran, Eds. (A. DeepakPublishing, Hampton, VA, 1985), pp. 81-88.

12. V. J. Abreu, "Wind Measurements from an Orbital PlatformUsing a Lidar System with Incoherent Detection: An Analy-sis," Appl. Opt. 18, 2992-2997 (1979).

13. P. B. Hays, V. J. Abreu, J. Sroga, and A. Rosenberg, "Analysis of

a 0.5 Micron Space-Borne Wind Sensor," in Preprint Volume,

Conference on Satellite/Remote Sensing Applications, 25-29June, Clearwater Beach (American Meteorological Society, Bos-

ton, MA, 1984), pp. 266-271.14. W. E. Baker and R. J. Curran, Eds., Proceedings of the NASA

Symposium on Global Wind Measurements (A. Deepak Pub-lishing, Hampton, VA, 1985).

15. R. T. Menzies, "Doppler Lidar Atmospheric Wind Sensors: A

Comparative Performance Evaluation for Global MeasurementApplications from Earth Orbit," Appl. Opt. 25, 2546-2553(1986).

16. Y. K. Park, G. Giuliani, and R. L. Byer, "Stable Single-Axial-

Mode Operation of an Unstable-Resonator Nd:YAG Oscillator

by Injection Locking," Opt. Lett. 5, 96-98 (1980).

17. Y. K. Park and R. L. Byer, "Electronic Linewidth Narrowing

Method for Single Axial Mode Operation of Q-SwitchedNd:YAG Lasers," Opt. Commun. 37, 411-416 (1981).

18. Y. K. Park, G. Giuliani, and R. L. Byer, "Single Axial Mode

Operation of a Q-Switched Nd:YAG Oscillator by Injection See-

ding," IEEE J. Quantum Electron. QE-20, 117-125 (1984).19. B. Zhou, T. J. Kane, G. J. Dixon, and R. L. Byer, "Efficient

Frequency-Stable Laser-Diode-Pumped Nd:YAG Laser," Opt.

Lett. 10, 62-64 (1985).20. D. Rees, P. A. Rounce, I. McWhirter, A. F. D. Scott, A. H.

Greenaway, and W. Towlson, "Observations of Atmospheric

Absorption Lines from a Stabilised Balloon Platform and Mea-

- surements of Stratospheric Winds," J. Phys. E 15, 191-206(1982).

21. D. Rees, I. McWhirter, P. B. Hays, and T. Dines, "A Stable,

Rugged Capacitance-Stabilised Piezo-Electric Scanned Fabry-Perot Etalon," J. Phys. E 14, 1320-1325 (1981).

22. D. Rees, I. McWhirter, P. A. Rounce, and F. E. Barlow, "Minia-

ture Imaging Photon Detectors: II Devices with TransparentWindows," J. Phys. E 14, 229-233 (1981).

23. I. McWhirter, D. Rees, and A. H. Greenaway, "Miniature Imag-

ing Photon Detectors III. An Assessment of the Performance ofthe Resistive Anode IPD," J. Phys. E 15, 145-150 (1982).

24. "LAWS: Laser Atmospheric Wind Sounder," Earth ObservingSystem Volume IIg, NASA, Washington, DC (1987).

25. D. Rees, T. J. Fuller-Rowell, A. Lyons, T. L. Killeen, and P. B.

Hays, "Stable and Rugged Etalon for the Dynamics ExplorerFabry-Perot Interferometer. I: Design and Construction,"Appl. Opt. 21, 3896-3902 (1982).

26. T. L. Killeen, P. B. Hays, N. W. Spencer, and L. E. Wharton,

"Neutral Winds in the Polar Thermosphere as Measured fromDynamics Explorer," Geophys. Res. Lett. 9, 957-960 (1982).

27. R. Hertel, ITT; private communication (1987).

28. D. Rees, I. McWhirter, and D. Wade, "Development of a Dopp-

ler Wind Lidar System for Atmospheric Wind Measurements,"in Proceedings of the Eighth ESA Symposium on EuropeanRocket and Balloon Programmes and Related Research, ESA

SP 276 (1987), pp. 99-106.


29. A. Rosenberg and J. Sroga, "Development of a 0.5 n IncoherentDoppler Lidar for Space Application," in Proceedings of theNASA Symposium on Global Wind Measurements, W. E. Bak-er and R. J. Curran, Eds. (Deepak Publishing, Hampton, VA,1985), pp. 157-162.

30. J. W. Fitzgerald, "Effect of Relative Humidity on the AerosolBackscattering Coefficient at 0.694 and 10.6-,um Wavelengths,"Appl. Opt. 23, 411-418 (1984).

31. R. L. Schwiesow, R. E. Cupp, V. E. Derr, E. W. Barrett, R. F.Pueschel, and P. C. Sinclair, "Aerosol Backscatter CoefficientProfiles Measured at 10.6 m," J. Appl. Meteor. 20, 184-194(1981).

32. D. Deirmendjian, Electromagnetic Scattering on Spherical Po-lydispersions (Elsevier, New York, 1969).

33. M. L. Wright, E. K. Proctor, L. S. Gasiorek, and E. M. Liston, "APreliminary Study of Air Pollution Measurement by ActiveRemote Sensing Techniques," NASA Contractor Report CR-132724 (1975).

34. "LASA: Lidar Atmospheric Sounder and Altimeter," EarthObserving System Volume Id, NASA, Washington, DC (1987).

35. G. Megie and R. T. Menzies, "Complementarity of UV and IRDifferential Absorption Lidar for Global Measurements of At-mospheric Species," Appl. Opt. 19, 1173-1183 (1980).

36. "U.S. Standard Atmosphere," NOAA-S/T 76-1562, U.S. Gov-ernment Printing Office, Washington, DC (1976).

37. R. T. H. Collis and P. B. Russel, "Lidar Measurement of Parti-cles and Gases," in Laser Monitoring of the Atmosphere, E. D.Hinkley, Ed. (Springer Verlag, Berlin, 1976).

38. I. S. McDermid, JPL-TMF unpublished results (1987).39. L. Elterman, "UV, Visible and IR Attenuation for Altitudes to

50 km, 1968," AFCRL-68-0153 (1968).40. M. P. Thekaekara, "Extraterrestrial Solar Spectrum, 3000-6100

A at 1 A Intervals," Appl. Opt. 13, 518-522 (1974).41. J. L. Bufton, F. E. Hoge, and R. N. Swift, "Airborne Measure-

ments of Laser Backscatter from the Ocean Surface," Appl. Opt.22, 2603-2618 (1983).

42. J. M. Gagne, J. P. Saint-Dizier, and M. Picard, "Mbthode d'e-chantillonage des fonctions dterministes en spectrscopie: ap-plication a un spectrombtre multicanal par comptage photoni-que," Appl. Opt. 13, 581-588 (1974).

43. G. Hernandez, Fabry-Perot Interferometers (Cambridge U.P.,Cambridge, 1986).


doppler lidar atmospheric wind sensor: reevaluation of a 355-nm incoherent doppler lidar

Documents