airborne and spaceborne lidar measurements of water vapor profiles: a sensitivity analysis

Airborne and spaceborne lidar measurements of water vaporprofiles: a sensitivity analysis

Syed Ismail and Edward V. Browell

This paper presents an evaluation of the random and systematic error sources associated with differentialabsorption lidar (DIAL) measurements of tropospheric water vapor (H20) profiles from airborne andspaceborne platforms. The results of this analysis are used in the development and performance evaluationof the Lidar Atmospheric Sensing Experiment (LASE) H20 DIAL system presently under development at theNASA Langley Research Center for operation on a high altitude ER-2 (advanced U-2) aircraft. The analysisshows that a <10% H 20 profile measurement accuracy is possible for the LASE system with a vertical andhorizontal resolution of 200 m and 10 km, respectively, at night and 300 m and 20 km during the day. Globalmeasurements of H2 0 profiles from spaceborne DIAL systems can be made to a similar accuracy with avertical resolution of 500 m and a horizontal resolution of 100 km.

1. Introduction

Water vapor (H20) has a strong influence on manyatmospheric processes associated with meteorology,climate, and the global hydrologic cycle. Measure-ments of the spatial and temporal variability of watervapor are important in understanding the hydrologiccycle, global circulation and dynamics, storm phenom-ena, dynamics of the atmospheric boundary layer, andatmospheric radiative transfer. High vertical resolu-tion (<1 km) measurements over both regional andglobal scales are needed in these studies. Sparselydistributed radiosonde network stations provide theonly sources of global H20 measurements. Passiveremote sensing techniques from space provide globalcoverage of H20 distribution but do not provide goodvertical resolution.1 Lidar remote sensing techniquescan provide high resolution measurements of H20 dis-tributions. The Raman lidar technique2 can providehigh resolution (100 m) H20 measurements, but thistechnique is limited to relatively short (<7 km) verti-cal ranges. The differential absorption lidar3 4

(DIAL) method can provide long-range high resolutionmeasurements of H20. The DIAL method which hasthe potential to provide high resolution measurementsof H20 on a global scale if operated from spaceborneplatforms, is discussed in this paper.

Syed Ismail is with ST Systems Corporation (STX), Hampton,Virginia 23666, and Edward Browell is with NASA Langley Re-search Center, Hampton, Virginia 23665.

Received 9 December 1988.

Although H20 was first measured in 1966 with theDIAL technique,5 more extensive H20 measurementshave been made possible from ground-based6 710-12

and airborne8 9 platforms due to recent DIAL technol-ogy developments. As a precursor to the developmentof a spaceborne DIAL system for global measurementof H20 profiles, an advanced autonomous DIAL sys-tem called the Lidar Atmospheric Sensing Experiment(LASE)13 is being developed at the NASA LangleyResearch Center for flight on a high-altitude ER-2, anextended range U-2, aircraft. In recent years severalstudies1418 have also been conducted exploring thepotential for DIAL measurements from space.

In this paper we present a comprehensive sensitivityanalysis for a nadir-viewing DIAL system for range-resolved measurements of H20 profiles from an air-craft or space platform. This sensitivity analysis in-cludes potential error sources related to the detectionof the lidar signals and to the knowledge of the H20absorption cross section. The purpose of this study isto demonstrate the sensitivity of the DIAL H20 mea-surement accuracy to atmospheric and instrument pa-rameters and to show that the DIAL systems describedhave the potential to meet the measurement needs ofH20 in the atmosphere. Any development of a DIALsystem should be based on the knowledge of thesesensitivities to reduce or eliminate such errors. Be-cause signal errors are dependent on lidar system pa-rameters and range to the measurement region, exam-ples of signal error sensitivities for both airborne andspaceborne platforms are given. For a complete anal-ysis, we use the parameters for the LASE system.Parameters for the lidar atmospheric sounder and al-timeter (LASA) DIAL system,16"17 envisioned for oper-

1 September 1989 / Vol. 28, No. 17 / APPLIED OPTICS 3603

ation from a polar orbiting satellite are used in theanalysis for the measurements of tropospheric H20.Even though H20 absorption lines in the 720-nm bandare used in all calculations, the analysis of the DIALmeasurement accuracy discussed in this paper is appli-cable to other bands of H20 and to other molecules. Inthe following sections, a general description of theDIAL methodology is presented; the signal detectionand absorption cross section errors are evaluated inrelation to the LASE and LASA systems; and thecombined DIAL measurement errors are discussed.

II. DIAL Method Principles

The simplified DIAL equation3 for determining theaverage gas concentration, n, between ranges R, and R2is given by

1 1 S.n(R 1 )Soff(R 2)12Aex(R2 -R,) S0.(R2)S0ff(R1) (

where Son, and Soff are the lidar signal amplitudes at theon and off wavelengths; the lidar wavelengths are se-lected to be on the peak of the gas absorption line andin an unabsorbed region off the peak; and Aa is thedifferential absorption cross section between the onand off wavelengths. Eq. (1) assumes that the wave-length separation between the on and off laser lines issmall (AX < 0.1 nm). In this case, aerosol backscatterand atmospheric extinction'9 errors are considerednegligible, and the laser lines are selected so that thereare negligible interfering gas influences. 2 0

On the basis of Eq. (1) DIAL measurement errorscan be classified into two categories, signal detectionerrors and differential absorption cross section relatederrors. The signal errors treated in the present analy-sis are random errors due to uncertainties in the de-tected signals, background signal noise from solar radi-ation, and errors resulting from detector dark currentand amplifier noise. Analyses of signal errors for somerepresentative DIAL systems have been reported inthe literature.32122 Additional errors in the signaldetection and processing system for the LASE H20DIAL measurements are discussed in the last sectionof this paper.

Errors due to uncertainties in the knowledge of thedifferential cross section are caused by atmosphericeffects and laser system characteristics. The mainatmospheric effects include temperature sensitivities,pressure broadening and pressure shifts of H20 ab-sorption lines, and Doppler broadening (DB) of theRayleigh backscattered component of the lidar return.The laser system parameters that influence DIALmeasurements include the finite width of the laser linecompared to the width of H20 absorption line, thespectral purity of the laser line, and uncertainty in theknowledge of the laser line position. Because of thesesystematic influences it is necessary to evaluate thedifferential absorption cross section, Aa, as an effec-tive absorption cross section, Aaeff, which can then beused in the measurement of n using Eq. (1). Theconcept of an effective cross section and a method for

calculating it from the actual H20 absorption crosssection are discussed in a later section.

An H20 absorption line is broadened in the atmo-sphere by two processes, pressure broadening andDoppler broadening, which are represented by the wellknown Lorentz and Doppler profiles, respectively. Inthe lower atmosphere (<15 km altitude), pressurebroadening is the dominant spectral broadening influ-ence. The pressure broadened linewidth, y (halfwidthat halfmaximum HWHM), is given by23-25

p T0 \O.62

Po )(2)

where P is the atmospheric pressure and T the atmo-spheric temperature. The line strength S is given by

S=S( )(T 1.5 p[Ehc ( 1 1)], (3)

where E" is the energy of the lower state, h is Planck'sconstant, c is the speed of light, and k is Boltzmann'sconstant. The Doppler width, YD, is given by

YD = (2kT n2/m)1/2 (4)

where v0 (cm) is the line center position, and m is themass of the molecule. For DIAL measurements, theH20 absorption line shape can be represented by theVoigt profile,26 which in its limits goes to the Lorentzand Doppler profiles. The absorption cross section,a(v), using the Voigt profile representation, is given by

a(P) = 0o, Y y2+ep((-t)2 dt, (5)

whereS /1n2 1/2

YD 7 7 )

y = ' (ln2)"2;YD

x = - (1n2)/2.'YD

Ill. DIAL Signal Detection Errors

The lidar signal errors described in this section arethe random errors due to uncertainties in the knowl-edge of the detected signals, errors due to sunlight ormoonlight background, and errors due to detector darkcurrent and amplifier noise. For calculating the totalsignal detection error, the magnitude of the lidar sig-nals, background signal, and detector dark current arecomputed based upon a set of model atmosphere andDIAL parameters.

Summertime profiles of molecular density, H20 con-centration, atmospheric temperature, and pressure forthe mid-latitude region27 are used in these simulations.The assumed aerosol profile (Fig. 1) is a compositethree layer aerosol model containing a 0-2-km bound-ary layer with a ground visibility of 23 km,27 a tropo-spheric model representing clear conditions from 2-10km, and a background aerosol model above 10 km.ture profile, with a temperature uncertainty of d10 K,

3604 APPLIED OPTICS / Vol. 28, No. 17 / 1 September 1989

I 50J at SHz. ^ .2.' /R .. ., FOV - Lo..r DWvrgenc 0 ,S R

ON LINE p------- ~~ ~-- OFF LINE I

. . . - ACKGROUND SIGNAL- ARK CURRENT SIGNAL X

, 4 o ~~~~ 2 / I I o

2 C , /X / I4i 1'',0 , P,

I10 = 10 =PHOTOELECTRONS

1 0PER MICRO SEC.

10.

Fig. 2. Lidar signals, detector dark current, and background signallevels for the LASE system operating from an aircraft altitude of 16

km for ground surface reflectivity values of 0.8, 0.3, and 0.03.

LOG AEROSOL SCATTERING RATIO

Fig. 1. Assumed three layer aerosol model representing back-ground aerosol conditions with a ground visibility of 23 km. A

constant phase function' 9 value of 0.023 sr-' is assumed.

The DIAL parameters (Table I) used in these simula-tions represent parameters of the LASE system.'3Unless otherwise specified, nominal values of X = 728nm, Aau = 20.87 X 10-24 cm2 , and y = 0.1 cm-' areassumed in these calculations.

The lidar signal S (photoelectrons) from range R iscalculated using the expression

5()-(E/hv)Asn3QcAt R

S(R) 2R2 exp-2 adR (6)

where E [J] is the laser pulse energy, A is the effectivearea of receiver, 1i is the quantum efficiency of the

detector, A [sr-1 km-'] is the total atmospheric volumebackscattering coefficient, Q is the total optical effi-ciency of the receiver for the signal accumulation timeAt, and a is the atmospheric extinction coefficientincluding the absorption by H20. The backgroundsignal B (photoelectrons), assuming a Lambertian sur-face and zenith sun, is calculated from the expression.

= -. x- AwQsnAt exp(-2r),hp x 2 (7)

where L is the solar irradiance 2 8 value (1300 W m2

Aim 1) at 728 nm for a nominal two-way optical depth

2r of 0.5, w is the optical filter width (FWHM, Table I),p is the surface reflectance, and 0 is the receiver field ofview. The dark currentD (photoelectrons) is calculat-ed from the expression

D= ( NEP ) (8)

TransmitterEnergy per pulseRep RateWavelengthBeam divergencePulse widthAircraft altitudeAircraft velocityLaser spectral widthTuning errorLaser centroid measurement

uncertaintyLaser spectral measurement

resolutionSpectral purity

ReceiverArea (effective)Field of viewFilter bandwidth (FWHM)Optical transmittance (total)Detector efficiencyNoise eq. powerExcess noise factor

150 mJ (on & off)5 Hz726.5-732.0 nm0.73 mrad300 ns16-21 km200 m/s1.1 pm0.25 pm (la)+0.25 pm (la)

<0.5 pm

>0.99 (0.9995 expected)

0.11 m2

1.23 mrad0.4 nm (day); 1 nm (night)29% (day); 49% (night)80% APD (Si)2 X 10-14 W Hz-l/22.5

where NEP is the noise equivalent power.29 The NEPvalue used in this study represents the effective noiseoutput of the detector-amplifier system measured inthe absence of a lidar signal and day background. Themagnitudes of the atmospheric backscattered lidarsignals, the day background, and the dark currentvalues calculated using the LASE parameters (Table I)are shown in Fig. 2. The daytime background signalswith zenith sun conditions and three surface reflec-tance values30 of 0.03, 0.3, and 0.8 for water; vegetationor soil; and snow or low cloud conditions, respectively,are shown in this figure. The nighttime backgroundfor the worst case of a full moon condition is a factor of10-5 lower 3 l than the values given in Fig. 2.

The detectors used in the LASE system are siliconavalanche photodiodes (Si:APD). The Si:APD has ahigh quantum efficiency, -80% at 728 nm, and alsohigh detector noise compared to PhotomultiplierTubes (PMTs). An analysis32 of the performance ofSi:APDs for the LASE system showed that; in thiscase, because of the strong atmospheric signals, theadvantage due to the high quantum efficiency morethan offsets the disadvantages due to their higher


15

E

iiia)t-

Et

Table 1. LASE H20 DIAL Parameters

noise. In addition, Si:APDs29 have shown linear de-tector response over a wide input signal range (10-'4-l0-8 W) and show excellent impulse response charac-teristics.

The random error in the measurement of a signal S iscalculated from Poisson statistics as A. For an ava-lanche photodiode (APD) an additional factor calledthe excess noise factor29 F must be included to accountfor the nonuniform gain multiplication in the diode.With this factor, the random error for the same signalis given by F. Assuming that: (i) the random errorscan be combined from all sources [see Eq. (1)] accord-ing to the sum of the squares of individual errors and(ii) the s signals are not correlated, the relative error inthe DIAL measurement of n is given by

An 1 2 2 (S + B)F + D1 (/2

n 2Aan(R 2 -R 1) S

where i = 1,2 is for the ranges R, and R2, respectively,and j = 1,2 is for the on and off signals, respectively.Equation (9) is a modified form of Eq. (9) of Ref. 22. Itcan be used to optimize the lidar parameters, to selectthe optimum detector system, and in random errorcalculations.

The LASE detector and signal processing system isbeing developed to achieve two objectives: 1) to coverthe expected dynamic range >105 from the cloud orground off-line signals to the weak on-line signals froma clean atmosphere and 2) to prevent degradation ofthe signal-to-noise ratio (SNR) due to the addition ofthe digitization errors by no more than 10%. Toachieve these objectives the signal is divided into a 9:1ratio and channelled to two Si:APD detectors. Thelow signal detector is primarily intended to measurethe strong ground and cloud returns with a 12-bitdigitizer. The primary objective of this is for columnH20 measurements. The high signal (90% energy)detector is intended to measure atmospheric signals.The output of this detector is amplified and digitizedby two 12-bit digitizers. The signal from one digitizerA is amplified so that for the weakest atmosphericsignals the SNR is not degraded by >10% by the digi-tizer random error. The signal amplification for thesecond digitizer B is set so that digitizer B, at signallevels where A is saturated, does not degrade the SNRby >10%. In our computation of the errors, the ran-dom errors are multiplied by a factor of 1.1 to take intoaccount the digitizer error.

The nighttime random error profiles for four H20lines with two-way optical depths from 2.2 to 16.2 areshown in Fig. 3. This figure shows that stronger ab-sorption lines are more suitable for DIAL measure-ments at higher altitudes and that at least two lines areneeded to optimally make H20 measurements over theentire 0-10-km altitude range. Even though the ran-dom part of the DIAL measurement error (Eq. 9) de-pends only on the signals and the local optical depthacross the range cell, the minimum error for each of thefour profiles in Fig. 3 was at a one-way optical depth of-1. This result is consistent with the calculations ofRemsberg and Gordley (1978).22 The relationship of

10 -

S

0 5 o 15 20PERCENTAGE RANDOM ERROR

Fig. 3. Random error profiles for the LASE H 20 DIAL systemoperating with night background conditions. A midlatitude sum-

mer H 20 profile 27 is assumed.

S=

la

727, t 150mJ t 5HzSi APO DETECTOR

y=1 7- 0-'

ALBEDO__---- 0 0 WATER

-______ 0.30 LAND0.|080 SNOW/CLOUDS

A Z= 500 m E~e20 km _ _ ~ ~

PR A A10 1O

PERCENTAGE RANDOM ERROR 2

Fig. 4. Random error profiles for the LASE H20 DIAL systemoperating over various daytime background conditions.

the random error to the horizontal averaging (achievedby averaging a number of laser shots along horizontaldistance x) and vertical smoothing (along R) can alsobe obtained from Eq. (9). By averaging DIAL mea-surements from N independent shot pairs, where eachshot pair samples the same atmospheric volume andwhere the gas profile does not change across the N shotpairs, the standard error is expected to decrease as1/N. (This relation holds for near UV DIAL sys-tems3 3 and mid-IR DIAL systems3 4). If Ax is thehorizontal distance covered during N shot pair mea-surements, An/n 1/Ax. For night background con-ditions, where S >> B, increasing the range cell sizeincreases the amount of signal received, and thereforefrom Eq. (9), An/n 1/AR'5 . Therefore, the com-bined dependence of the measurement error on hori-zontal averaging and range cell size is given by

An -0.5(AR)-1_5n (10)

and for conditions where S << B (for example when theday background is unfiltered) An/n - Ax-0.5 AR-l .

The signal error profiles for the LASE system duringdaytime operation are shown in Fig. 4 for the three day


i

I

?*, 727-m 150mJ A S.S APO DETECTOR

e(X 10 um) 2-WAY OPT. THK_ _ _ - - 96.0 233__…______-_ 458 11 0

18.7 4.51O.8 2.6

IIZ-200B. X10k _.k . _ _

7' _ _-

.<_~ r t lof

as5

5

. -1

I.0I

5

-E

Fig. 5. Random error profiles for a spaceborne H 2 0 DIAL system aan altitude of 700 km. A summer H 2 0 model is assumed.

background conditions identified in Fig. 2. To maintain a measurement accuracy for the daytime conditions comparable to those at night, the horizontal averaging (Ax) and the range cell size (AR) are increased tcompensate for the lower optical throughput of thinterference filter and the increased noise due to daytime background. Over water, except for sun glintthere is very little influence of solar background (seFig. 2), and Eq. (10) is expected to hold over most of thrange. Figure 4 shows that with the LASE systemdaytime measurements of water vapor profiles are possible even under high background conditions, such athose over the clouds and snow.

Browell et al.16 have presented simulations whicishow the feasibility of making DIAL H20 profile measurements from space. To partially compensate fothe 1hR2 loss in the lidar signal, both the laser energ]and collection area need to be increased over thoseused in airborne systems. Using a nominal spacecrafaltitude of 700 km, a receiver area of 1.1 M2 , a laseenergy of 1 J and a detector quantum efficiency of 0.2it is found that the signals from a spaceborne lidaisystem are smaller by a factor of 115 compared to theLASE system. This requires a detector system oper.ating in the photon counting mode for measurement oJthe clear atmospheric returns and in the analog modEfor stronger signals from aerosol layers, the groundand clouds. Figure 5 shows signal error profiles, calcu.lated using the lidar parameters from Table II, for spaceborne H20 DIAL system. Larger spatial averag-ing (Az = 500 m; Ax = 100 km) and higher laser pulserepetition rates (>20 Hz) for spaceborne DIAL sys-tems are necessary to reduce measurement errors to alevel adequate for global scale investigations. Themain difference between the random error profiles ofthe LASE system (Fig. 3) and spaceborne system (Fig.5) occurs at high altitudes (>10 km). For example, at10 km, the random errors of the LASE and the space-borne systems are 5% and >20%, respectively. In thecase of the LASE system the decrease in the atmo-spheric backscattering with altitude is more than com-pensated by the 1IR2 increase in the signal. For thespaceborne system this range dependence is small and

consequently the combination of low atmospheric sig-nals and low differential optical depths results in largerandom errors. Therefore, for high resolution DIALH20 measurements at high altitudes, selection ofstronger H20 absorption lines from the 940-nm and1140-nm bands3 5 is necessary. Browell et al.16 showedthat spaceborne DIAL H20 profile measurements inthe daytime are possible using a narrow receiver field-of-view (0.1 mradian) and a Fabry-Perot interferome-ter to act as a filter in rejecting background.

In addition to the signal induced uncertainties dis-cussed above, absorption cross section errors also con-tribute to the total error budget. But unlike the signalerrors, which are range dependent, the cross section

t error is independent of the altitude of platform provid-ed that the platform is above 20 km because theoptical depth above this altitude is <0.01 for the 720-nm H20 band.

IV. Cross Section Errors

r- Errors in the DIAL measurement of gas profiles arealso caused by uncertainties in the knowledge of the

e gas absorption cross sections at the remote atmospher-ic scattering volume. This problem is most severewhen narrow absorption lines are used for DIAL mea-

e surements, e.g., H20. In addition to DIAL systeme characteristics, atmospheric effects can cause uncer-

tainties in the knowledge of the absorption cross sec-- tions. The following systematic effects can produces uncertainties in the absorption cross sections used in

the DIAL calculations:.h (1) Modification of the laser spectral profile by mo-- lecular absorption,r (2) Doppler broadening of the elastically backscat-tered signal and other atmospheric spectral broaden-

ing effects,t (3) Pressure shifts of absorption lines,

r (4) Temperature sensitivity of absorption lines,9 (5) Laser spectral purity,Lr (6) Laser wavelength uncertainty, and(7) Knowledge of laser spectral output.

These effects are caused by atmospheric influencesOf (1)-(4) or DIAL system influences (5)-(7). The spec-

tral widths of the lidar pulses are in general finitecompared to H20 absorption linewidths. With thiscondition, it is necessary to calculate the effective val-

Table II. Parameters of a Spaceborne Water Vapor DIAL System

TransmitterPulse energy 1 JDivergence 0.1 mradPulse rep. rate 20 Hz (on and off)Wavelength 727 nmAltitude 700 kmVelocity (ground) 7 km/s

ReceiverEffective area 1.1 m2

Field of view 0.1 mradFilter bandwidth (FWHM) 0.01 nm (day); 1.0 nm (night)Optical transmittance (total) 30% (day); 55% (night)Detector quantum efficiency 20%Noise eq. power 2 X 10-i W Hz-l/2


I I IX_ =727.- IJ t 20H.

A e-_______ 96.O.10O" z.11

----- ___------- 45 t.10:-^+

1 80

7*,

0-z c

._. . 1 z.- o msA Z500- m X=1 00 .km1

5 1~~~I0 1 52PERCENTAGE RANDOM ERROR

I.

I.0

5

L

ue of the absorption cross section, Ueff which is a convo-lution of the lidar spectrum with the absorption line.The value of aeff is then used in place of Aa in calculat-ing n from Eq. (1). Assuming that the off-line absorp-tion is negligible, the effective absorption cross sectionis given by the equation.

J G(v,z)ar(v,z) d'

aeff(Z) =

J G(v,z) dv

6-~

EF:1

ci

(11)

where G(v,z) is an altitude-dependent on-line lidarspectral intensity profile and a¢(v,z) is the Voigt ab-sorption cross section profile. Equation (11) is nonlin-ear, and it can be evaluated numerically for any laserspectral shape. In our evaluation of Eq. (11), we use200 spectral intervals across the H20 line profile, andmake computations at 1-km altitude intervals. Thesecalculations are for both the forward propagating andbackscattered beams because the spectral distributionG is expected to be different for the two cases. Toevaluate the error, we simulate a DIAL measurementretrieval. First, the lidar return signals are calculatedusing a set of lidar parameters and including the effectsof atmospheric influences. The effective absorptioncross section values are computed using the assumedlidar parameters and estimated atmospheric effects,which are in error compared to the real values used inthe signal calculations. Using the simulated signalsand the calculated absorption cross sections, we obtainthe retrieved DIAL H20 concentration profile. Theerror in the DIAL measurement is the difference be-tween the actual H20 profile used the in signal calcula-tions and the retrieved profile. For the sensitivityanalysis, each parameter is independently varied whileholding all other parameters constant. However acomplete evaluation must include the coupled effectsbetween the various parameters. Therefore followingthe parametric study an example is discussed whichcombines all parameters that are likely to influence theDIAL H2 0 profile measurements.

(1) Modification of Laser Spectral Profile byMolecular Absorption

As a laser beam is propagated in an absorbing atmo-sphere, its spectral shape is altered3637 due to theabsorption. The effective H20 absorption cross sec-tion changes due to this effect and leads to an underes-timation in the DIAL determination of H20 concen-trations. The error from this effect is a function ofoptical depth and laser linewidth. The results of ouranalysis of this effect are shown in Fig. 6 for a H2 0absorption cross section of 70.1 X 10-24 cm2 . Thethree profiles show the effect for three laser linewidths(1, 2 and 4 pm), where the laser line consists of threeequal-amplitude, equally spaced monochromaticmodes across the specified linewidth. Figure 6 showsthe influence of optical depth, which increases towardslower altitudes, and larger laser linewidths on theDIAL H20 density error. Clearly larger linewidths arenot suitable for DIAL measurements at large optical

3

3-MODE LASERTOTAL LINEW1DTH

1 pm2 pm4 pm

II l

I II0 2 4 6 8

NUMBER DENSITY ERROR(-=S).0

Fig. 6. Influence of laser spectral width causing systematic errorsdue to the distortion of the laser profile. An H 2 0 cross section of 70.1

X 10-24 cm2 is assumed.

depths. Cahen and Megie36 also evaluated the influ-ence of laser linewidths on DIAL H20 measurementsand showed that if laser linewidths comparable to ab-sorption H20 linewidths are used, DIAL measurementerrors up to 25% can result. In the absence of atmo-spheric processes that broaden the laser linewidth, theconclusion would be to use the narrowest laserlinewidth to minimize the DIAL measurement errors.

(2) Doppler Broadening of the ElasticallyBackscattered Signal and Other AtmosphericSpectral Broadening Effects

The spectral distribution of laser energy is modifiedin the backscattering processes due to the motion ofatmospheric gases and aerosols. This Doppler broad-ening (DB) effect on lidar backscatter returns has beenextensively discussed in the literature.37 -44 Aerosolparticle velocities in the atmosphere are several ordersof magnitude lower than those of air molecules andhence the Doppler broadening effect due to aerosols isexpected to be negligible (<<0.1 pm). For simplicity,the velocity distribution of molecules can be assumedto be given by the Maxwellian function. Therefore,the velocity distribution function, f(v), in a given direc-tion is represented by

I1_m = 1/2 / mv 2 \f(V) exp-~ I'\2rkT) \_ /k (12)

where m is the average mass of the air molecules. TheDoppler shift for the case of the elastically backscat-tered radiation is given by the equation

X -Xo 2v________ = (13)

where X0 is the incident wavelength, and X is the Dopp-ler shifted wavelength. The wavelength distributionof the elastically backscattered photons is given for p <1 atm by

F(X) = A exp(- mT) (X _0 )2, (14)

where A is a normalizing constant. The DB halfwidth


I

I

'y is the distance from line center to the position wherethe intensity drops to A/2. Therefore, from Eq. (14)we get

2X 2kT ln2cN m (15)

The value of y for X = 727 nm at 300 K is -1.7 pm, andthe altitude variation of the DB full linewidth (2y) forthe given model temperature profile is shown in Fig. 7.This figure also shows the altitude variation of the H20absorption linewidth, assuming a Voigt line depen-dence. From Fig. 7 it is clear that the DB effect has asignificant influence on the effective absorption crosssection and that this influence is considerable at highaltitudes (>5 km) because of the combination of small-er H20 linewidths and the decrease of aerosol scatter-ing. The altitude variation of the systematic underes-timation of the water vapor number densities resultingfrom no corrections for the DB effect is shown in Fig. 8(solid line). As expected, the DB has a larger influenceat high altitudes, and the error also depends upon theaerosol-to-molecular backscatter ratio at all altitudes.The influence of the three layer aerosol model (Fig. 1)can be seen in Fig. 8. While the influence of DB can belarge (12% DIAL measurement error at 15-km alti-tude), it can be corrected to first order with someknowledge of the aerosol-to-molecular backscatter ra-tio and the temperature profile. Figure 8 (dotted line)shows the residual error due to Doppler broadeningeffects when the influence of DB is calculated using a50% error in the estimated aerosol-to-molecular back-scatter ratio45 (with a minimum 20% error in theknowledge of the total scattering ratio) and a 3% (10K) temperature error. These are combined togetherto produce the worst case undercorrected error. It isassumed that the aerosol-to-molecular backscatter ra-tio can be derived from the off-line return signal, and aclimatological model temperature profile can be usedfor the temperature profile. These calculations (Fig.8) show that the residual DB effect, after being partial-ly corrected, causes 1.5% error below 10 km altitudein the absence of large aerosol gradients.41 It has beenshown by Ismail and Browell43 that Doppler broaden-ing reduces the sensitivity of DIAL measurement er-rors to other lidar and atmospheric parameters. Thiseffect is discussed in detail later in this paper.

While Doppler broadening can cause a large system-atic error, other spectral broadening mechanisms likeBrillouin scattering (BS) and Raman scattering (RS)have smaller influences on DIAL measurements. TheBS effect produces two additional peaks in the elasti-cally backscattered signal. Their separation dependsupon the velocity of sound, and their intensity dependsupon the atmospheric pressure.46 The BS effect ismost pronounced at high pressures (>1 atm). Forsimplicity it can be modeled as additional broadeningof the Doppler profile. For H20 DIAL systems the BSeffect has negligible influence at high altitudes (>10km) due to low pressures and in the boundary layer dueto increased aerosol scattering and a larger H20linewidth. The BS effect has its maximum influence

E

25 _ 1 X l

I\ / DOPPLER BROADENING20 (AIR BACKSCATTER)

15

VOIG PROFILEi! 0-\ , (H20 LINEWIDTH)

10

0 1 2 3 4 5 6 7 8 9 10LINEWIDTH (FWHM), PM

Fig. 7. Comparison of the altitude dependence of the H20 absorp-tion linewidth and the Doppler-broadened Rayleigh backscattered

linewidth.

15B

5 0 D 1 5NsUM1BER DENSITY ERROR C-=C)

Fig. 8. Systematic error due to Doppler broadening of molecularbackscatter in the atmosphere. Also shown is the reduced error

profile when Doppler broadening effect is estimated.

in the free troposphere above the boundary layer. Anestimate of the magnitude of the BS on the Dopplerprofile was made from the data obtained by a highspectral resolution lidar system.40 This measurementindicates that the maximum spectral broadening of theDoppler profile due to BS is <20% at -1 km. Wefound that in our model calculations the BS increasedthe DIAL error by <0.5% which is much smaller thanthe random error or the DB error at this altitude.Thus, the BS error is negligible compared to othererror sources in DIAL H20 concentration measure-ments.

The Raman scattering (RS) effect generates shiftedwavelengths resulting from rotational and vibrationalinduced transitions. The vibrational Raman spec-trum is low intensity (10-3 of the unshifted backscat-tered energy47) and is shifted by >50 nm at 727 nm. Itcan be easily removed by broadband (20 nm) opticalfilters. The rotational Raman backscatter has an in-tensity -3.5% of the elastically backscattered energy47

from atmospheric molecules. The nearest Ramancomponent is within -0.65 nm of the incident wave-


length, and this effect can influence the DIAL mea-surements using optical filters with bandwidths largerthan 1 nm. The systematic error due to RS increaseswith optical depth and decreases with increased aero-sol backscattering. In a purely molecular atmosphereat large one-way optical depths ( > 2) with filterwidths >20 nm, systematic errors >10% are expected.For H20 DIAL systems operating at 727 nm it is pref-erable to filter the rotational RS using an optical filterwith <1 nm bandwidth. For such systems the error isnegligible (<0.5%) for optical depths up to 4. A gener-al discussion of the influence of the rotational Ramaneffect on DIAL measurements will be presented in aseparate paper.

(3) Pressure ShiftIn addition to spectral broadening effects (Eq. 2),

changes in atmospheric pressure cause small shifts (-1pm/atm at 727 nm) in the center position of H20

absorption lines. Bosenberg48 has measured shifts fornumerous H20 absorption lines in the 727-nm region,which exhibit an average shift of 0.7 pm/atm. Recent-ly Mandin et al.49 have published the complete lowpressure (nearly vacuum conditions) absorption spec-tra of H20 lines in the 725-nm band. Approximatepressure shift values of these lines can be obtained bycomparing the line positions determined by Mandin etal.4 9 with the ground atmospheric pressure H20 linepositions from the AFGL35 compilation. A compari-son of these line shifts with Bosenberg's measurementsshow general agreement to within -20%. If not prop-erly compensated for, these pressure shifts can causesignificant errors in DIAL H20 measurements. Thepressure shifts for given altitude, cause the same effectas detuning the laser wavelength off the absorptionline center which is discussed later. But unlike thedetuning error, the pressure shift error is altitude de-pendent. If the pressure shift is accurately measured,the DIAL measurement error from it can be fully cor-rected. Figure 9 shows altitude profiles of the errorscaused by a pressure shift of 1.0 pm/atm when the laserline is adjusted to coincide with the peak of H20 line atambient pressures equivalent to altitudes of 0, 4, and20 km. The worst uncorrected error occurs when thelaser line is tuned to the peak of the H20 line at groundpressure and the lidar measurements are made at highaltitudes (an error of >16% will result above 14 kmaltitude). Zuev et al.5 0 show that for the 694.38-nmH20 absorption line (pressure shift 0.91 pm/atm), aDIAL concentration measurement error of 32% willresult at 20 km if the laser is tuned to the groundpressure line position. Their results agree with theresults shown in Fig. 9. However these calculationsare overestimates of the error because the DB influ-ence was ignored. Ismail and Browell4 3 show thaterrors due to pressure shift effects are overestimatedby up to 45% when the DB influence is omitted. Whenthe laser line coincides with the peak of the H20 ab-sorption line at an intermediate altitude of 4 km, theresults (Fig. 9) show that very low pressure shift errors(<1%) result over the 0-10 km altitude range of the

E-

5 10 15NUMBER DENSITY ERRO R (-=~)

Fig. 9. Systematic errors caused by tuning the laser to H20 linecenter at an altitude of 0, 4, and 20 km. A pressure shift of 1.0 pm/

atm is assumed.

DIAL measurements. Even these small errors can beremoved if the pressure shift is known and accountedfor in the DIAL data reduction. Figure 9 also showsthat moderate DIAL errors (3%) are caused at groundlevel when the laser is tuned to the H20 line peak at anatmospheric pressure representative of 20-km alti-tude. To minimize the pressure shift error over abroad altitude range, the laser should be tuned to-wards a lower pressure absorption line center position.This analysis suggests that knowledge of the pressureshift for the H20 absorption line used in the DIALmeasurement is necessary to reduce this systematicerror effect to a negligible level (<0.5% from 0-10 kmaltitude). For the LASE H20 DIAL system a spectro-scopic measurement experiment54 will measure thepressure shifts to an accuracy of <0.05 pm/atm.

(4) Temperature Sensitivity of H20 LinesFor DIAL measurements the H20 lines are selected

so that the center line cross section, o, has a smalltemperature dependence (Aao < 1% for a ±10-Kchange in temperature). Usually the temperaturesensitivity of the line center absorption cross section iscalculated because most DIAL systems operate withnarrow laser lines near the peak of H20 absorptionlines. Previously, temperature sensitivity calcula-tions have been made for H20 concentration25 andmixing ratio7 measurements. Using the Lorentz pro-file, it can be shown that the condition for temperatureinsensitive lines for H20 concentration measurementsis given by

E"hcN 0.88k

and for mixing ratio calculations byE"hc

N 1.88k

(16)

(17)

where TN is the temperature at which there is notemperature sensitivity for lines having a ground stateenergy level of E" (cm-'). These simplified relation-ships are approximately valid only near ground atmo-spheric conditions. For more accurate results a Voigt


model for the H20 absorption line is necessary. Nu-merical calculations using the Voigt profile model forDIAL H20 concentration and mixing ratio measure-ments and using the most recent H20 line parametersare reported by Browell et al. (1989).51 These calcula-tions show that for a midlatitude summer atmospherictemperature model and for DIAL measurements nearground the temperature insensitive absorption lineshave E" values in the range 100-250 cm-' for H20concentration measurements and E" in the range 250-450 cm-' for H20 mixing ratio measurements. Theselines cause an error of <1% over a 10 K temperatureuncertainty. More details on optimum H20 line selec-tions in the 720-nm region and their altitude depen-dence are given in Ref. 51.

(5) Spectral Purity of LasersIn the DIAL method, the on-line laser energy is

assumed to be absorbed by H20 molecules, and theH20 concentration is retrieved on the basis of a calcu-lated effective absorption cross section. Many, if notmost, laser systems produce small amounts of energyoutside desirable spectral limits. The spectral purityis defined, here, as the ratio of the energy within theacceptable spectral limits to the total energy transmit-ted. In our analysis we assume that the spectral impu-rity is a component of the laser spectral output that isunabsorbed by H20. In some lasers this output resultsfrom a broadband emission arising from a process likeamplified spontaneous emission (ASE).

Since the spectral impurity is an unabsorbed compo-nent of laser energy, the effective absorption crosssection is lowered by its presence. The magnitude ofthe unabsorbed component of laser energy comparedto the total energy left in the beam depends on thespectral purity, a, of the transmitted beam and theoptical depth for the absorbed portion of the laseroutput to the range of interest. The unabsorbed ener-gy is treated like off-line energy, and therefore, the on-line signal return from an optical depth (due to H20absorption) is given by

Son = Sffl(l - a) + ae-2T, (18)

where it is assumed that the on- and off-line laserenergies are equal initially. For the evaluation of thesystematic errors due to spectral purity, the effectivecross sections are calculated with and without thespectral purity, and the calculated uncorrected errorsthat result for spectral purity values of 0.95, 0.99, and0.9995 are shown in Fig. 10. Small amounts of spectralimpurity can produce large systematic errors. Formost DIAL applications, lasers with spectral purity>0.99 are necessary if no knowledge of the actualamount of spectral impurity can be obtained. Recent-ly developed Alexandrite lasers, which are being incor-porated into various DIAL systems35253 have beendemonstrated to have high spectral purity (0.9999).53If the spectral purity of a laser system is known or if itcan be calibrated from in situ measurements, DIALmeasurements are still possible if the spectral purity islower than 0.99.9

15

10

BF

5

00

5 10 1 5NUMBER DENSITY ERROR(-5%)2o

Fig. 10. Influence of spectral purity (caused by ASE) on H2 0 DIALmeasurements.

(6) Laser Wavelength Uncertainty and TuningError

For the operation of most DIAL systems, the on-linelaser must be tuned near the center of a gas absorptionline. The DIAL measurement error depends upon theuncertainty in the knowledge of the laser position. Inaddition, the magnitude of this error also dependsupon the detuning of the laser from the H20 line cen-ter. Figures 11(a) to 11(c) show the DIAL measure-ment errors associated with laser line position uncer-tainty and laser tuning errors with respect to the centerof the absorption line. In Figure 11(a) the tuning erroris assumed to be zero. Because of the symmetry of theH20 line, the concentration measurement error de-pends on the magnitude of the laser line position un-certainty and not on its sign. Figure 11(b) shows thatfor a tuning error of +0.25 pm the concentration mea-surement error depends upon both the magnitude aswell as the direction of the position measurement er-ror. For a laser detuning of +0.25 pm, because of thesymmetry of the H20 line, a zero concentration mea-surement error is reached for either a 0 or -0.5 pmposition measurement error. On the other hand, forthe position measurement errors of +0.25 pm and-0.25 pm, the magnitude and direction of the DIALerrors are different. Figure 11(c) shows that as thetuning error increases to 0.5 pm, the DIAL systembecomes more sensitive to position measurement un-certainty. Since the laser line position uncertaintycannot be made zero, a combination of a low tuningerror and a low position measurement error is neededto minimize the DIAL measurement error. For theLASE system the tuning error and the position mea-surement error are both specified to be <0.25 pm (1).

(7) Wavemeter ResolutionMeasurement of the laser spectral profile is needed

both in the tuning of the laser system to maintain itwithin specified spectral position limits and also forcalculating the effective cross section in Eq. (11). Inthe LASE system the laser spectral profile is measuredby a wavemeter which samples a small fraction of thelaser output. The centroid of the laser spectral profile


POSITION UNCERTAINTY0 00 pm

-- -0.25 pm0.50 pm

O~~~~~~~~~~~~~~~~~1 _ /.25/

I ..

I ,' / (a)-5 0 e

NUMBER DENSITY ERROR(-5:)

E

p2

I I I I'

?I= 728 nm, =20.9X10' cm RESOLUTION

______+ 0.5 pm1.0 pm

I I I IE.RO10 8 I 4 2 C

NUMBER DENSITY ERROR (7.)

1~

I'

POSITION UNCERTAINTY I0.00 pm

- - - - --- 0.25 pm I0.50 pm

R.

/ •,

(b)

I I.5 0 5

NUMBER DENSITY ERROR(-Z:-)

25 4 D

\'\,\\s~~~ /_/+.5

POSITION UNCERTAINTY \\ /0.00 pm

------- 025 pm (C)*--1- 0.50 p0

) Sac , , _ I ,~~~~~~~~~~~~~~~~~~~

NUMBER DENSITY ERRORf(-=)

Fig. 11. H2 0 DIAL measurement error caused by an uncertainty inthe knowledge of laser spectral position (a) no detuning error (b) 0.25

pm detuning error and (c) 0.5 pm detuning error.

measured by the wavemeter is assumed to be the posi-tion of the laser. The DIAL error resulting from un-certainties in the laser position measurement havealready been discussed in the earlier section. In addi-tion to the position measurement error, the laser spec-tral profile is broadened due the finite spectral resolu-tion of the wavemeter prior to recording. Thisspectral broadening makes the calculated effectivecross section lower than its actual value, causing theDIAL measurements to be overestimated. Figure 12shows the results of systematic overestimates in theDIAL concentration measurements for wavemeter res-olutions of 0.5 and 1.0 pm. The influence of wave-

Fig. 12. DIAL systematic error caused by spreading of the laserspectrum by the wavemeter optical resolution.

meter resolution increases with altitude because of thedecrease in the H20 absorption linewidth. In contrastto other systematic effects like the DB effect, spectralpurity, and laser linewidth (which cause an underesti-mate of DIAL measurements), a wavemeter resolutionof 0.5 pm (for the LASE system) causes only a slightoverestimate of the density measurement. When oth-er systematic influences are not completely corrected,the residual influence of the wavemeter resolution par-tially compensates for the other systematic effects.However, if a correction for the other systematic ef-fects is made to achieve a high accuracy (<5% error)H20 measurement, then a wavemeter resolution •0.5pm is required, or the wavemeter data must be decon-volved to reconstruct the laser spectral information foreach laser pulse.

V. Combined DIAL Measurement Error

The combined DIAL measurement error is the sumof the absolute values of the total random error and thenet systematic error. In the signal detection error weinclude the influence of the detector system, amplifierand digitizer effects, and background signals. In thetreatment of systematic effects, we chose to evaluatethe influence of each parameter separately; however,in general the selection of any parameter influencesthe DIAL system's sensitivity to other parameters. Inthe example of the combined DIAL errors in Fig. 13,the random errors (LASE nighttime case with Aa val-ues of 18.7 X 10-24 cm 2 and 90.6 X 10-24 cm 2 ) are addedto the combined systematic error arising from laserline distortion due to H20 absorption, the residualerror due to DB, laser detuning of +0.25 pm, lasercentroid measurement error of +0.25 pm, and a spec-tral purity of 0.99. The errors have been combined insuch a way that the worst case error is estimated. Thewavemeter resolution error is not included because itreduces other systematic errors, and it is assumed thatthe pressure shift error is fully corrected. The tem-perature sensitivity error is not included because it isexpected to be small. Using the parameters of theH20 absorption lines at 727.8078 nm and at 726.5594nm in air at STP,25 and using a climatological tempera-


1

C

E

L's

Ff

1 5

1 0

Q

5

0

1 0

E

p2

5

0

I 6

1 2

a

4

Io

_1 0

the DIAL H20 errors due to temperature effects for thetwo lines are <0.4% in the 0-5 km altitude region and<0.5% in the 5-10 km region, respectively. To analyzeDIAL H2 0 data it is necessary to correct for the DBeffect. This is because the DB effect can cause largeerrors (up to 10% below 10 kin) even in clean atmo-spheric regions, and this effect can easily be correctedto first order. For the LASE system spectral purity isexpected to be high (>0.9995), and the improvement inthe combined error for this spectral purity is repre-sented by the broken line profiles in Fig. 13. Thisfigure shows that by using two H20 absorption lineswith complementary measurement regions in the lowand mid-troposphere a total error of <10% over therange 0-10 km can be achieved by the LASE H20DIAL system with horizontal and vertical resolutionsof 10 km and 200 m, respectively, during nighttime.This result also applies to daytime measurements withhorizontal and vertical resolutions of 20 km and 300 m,respectively.

For spaceborne H20 DIAL measurements in the720-nm region, the additional optical depth between20 km and space is expected to be negligible (<0.01).Hence all systematic errors discussed in this paper arealso applicable to spaceborne measurements.- Forspace use further refinements in DIAL technology canreduce systematic errors in the DIAL H20 measure-ments. In the low altitude region (0-2 km), the ran-dom errors of the spaceborne system (Fig. 5) are com-parable with the random errors of the LASE system(Fig. 3). Consequently, for low altitude regions, spa-ceborne measurements of H2 0 profiles with a verticalresolution of 500 m and a horizontal resolution of 100km can be achieved with <10% error during nighttime.Similar measurement accuracies can be made duringdaytime with resolutions of 500 m and 250 km, respec-tively. At higher altitudes (2-10 km) further averag-ing is necessary to achieve a <10% combined error.

VI. Discussion and Conclusions

In the absence of large systematic errors in the DIALmeasurement H2 0, the random errors from the lidarsignal determine the optimum altitude region forDIAL measurements, and these errors prescribe theamount of horizontal and vertical averaging needed toobtain accurate DIAL measurements. The altitudedependence of the random errors depends upon sever-al factors including detector type, atmospheric back-ground level, atmospheric aerosol distribution, ab-sorption line strength, and the distribution ofatmospheric H20.

Detector systems with a combination of high quan-tum efficiency, low excess noise factor, and low noiseequivalent power values are desired for H20 DIALmeasurements. Generally for airborne applications,signals are strong 103-104 photoelectrons/gs) and de-tectors with high quantum efficiency like Si:APD ap-pear to be optimum3 2 in these conditions because theirhigher quantum efficiency values (0.8-0.9) comparedwith that of the PMT's (0.04-0.18) more than offsetthe higher noise of the ADPs. For spaceborne applica-

J:B7

p2

TOTAL ERROR ()

Fig. 13. Combined errors from random and systematic effects inthe LASE H20 DIAL system for two absorption lines. Total DIAL

measurement errors are shown for two levels of spectral purity.

tions the signals are expected to be much weaker andlow noise detectors with simultaneous analog and pho-ton counting techniques will be necessary. In ouranalysis of the LASE system we include the total noisefrom the APD-detector-amplifier system.

We have not considered the influence of signal over-load effects in the detector-amplifier system due toclouds. For the LASE system the presence of cirrusclouds can cause significant signal levels 5-60 times45

the off-line atmospheric signal levels to be expectedfrom low altitudes (<1 km). In addition, atmosphericreturns from the near-field can also cause signal over-load effects. (Even though the full beam overlap be-tween the transmitter and receiver is expected at arange of >6 km, the signal profile calculation showsthat the strongest near-field signal received will befrom -0.5 km range). The LASE system is beingdesigned to be insensitive to the influence of these highsignal conditions, and it is expected that the timedifferences between the near-field region where satu-ration can occur (10-16 km) and the DIAL measure-ment region (<10 km) permits sufficient recoverytime. However, in the case of the strongest cirrusclouds some degradation of the DIAL measurement isexpected. In the case of spaceborne DIAL measure-ments, no near-field signal effects are expected and theinfluence of the cirrus clouds has less impact due to thelower signal levels due to the greater ranges involved.Narrow band (FWHM • 0.4 nm) interference filtersand receiver field-of-view values of <1 mrad are need-ed to suppress day background for LASE applicationand Fabry-Perot interferometer filters along with nar-row field-of-view (-0.1 mrad) receivers will be re-quired for daytime DIAL measurements from space.

Generally, H20 lines with cross section of 20 X10-24 cm2 are suitable for DIAL measurement in thelow troposphere (<5 km). For an optimum choice ofan absorption line, the distribution of H20 in the re-gion of measurement must be considered as discussedby Browell et al. (1985).16 Stronger H20 lines areneeded for DIAL measurements at higher altitudes.While H20 lines in the 720-nm band are adequate for


A-n 727.m 150mJ Rt 5Hz,a (X1 om .) SPECTRAL PURITY

+ + + 96. 0 0.9998.0 0.991 .7 0.9995

AZ=200.m axial0.kmC_

I .k0_'.

I I I

l

- --

5

.:

I .

1 0

5o

I . 20

Remote Sensing: Instrumentation and Techniques, OSA Tech.Digest Series, 18, 22-25 (1987).

13. E. V. Browell, et al., "Development of a High-Altitude H2 0

Airborne DIAL System-The Lidar Atmospheric Sensing Ex-

periment (LASE)," Abstracts, Thirteenth International LaserRadar Conference, Toronto, Canada, 11-15 Aug. (1986).

14. E. V. Browell and S. Ismail, "Spaceborne Lidar Investigations of

the Atmosphere," in Proceedings, ESA Workshop on SpaceLaser Applications and Technology, Les Diablerets, Switzer-land, ESA SP-202, 181-188, May (1984).

15. M. Endemann, et al., "Orbiting Lidars for Atmospheric Soun-

ding," Final Report, Vol. I, Battelle-Institut E. V., Frankfurt,

FRG (Dec. 1984).16. E. V. Browell, S. Ismail, M. P. McCormick, and T. J. Swissler,

"Spaceborne Lidar Systems for Measurement of AtmosphericWater Vapor and Aerosols," in Proceedings, AIAA/NASA EOSConference, Virginia Beach,VA, Paper No. 85-2091, Oct. 8-10

(1985).17. "NASA, Lidar Atmospheric Sounder and Altimeter," EOS Re-

port, Vol. 2 (1987).

18. M. J. Kavaya, S. W. Henderson, E. C. Russell, R. M. Huffaker,

and R. G. Frehlich, "Monte Carlo Computer Simulations ofGround-based and Space-based Coherent DIAL Water VaporProfiling," Appl. Opt. 28, 840-851 (1989).

19. E. V. Browell, S. Ismail, and S. T. Shipley, "Ultraviolet DIAL

Measurements of 03 Profiles in Regions of Spatially Inhomo-geneous Aerosols," Appl. Opt. 24, 2827-2836 (1985).

20. E. V. Browell and P. Woods, First Int. DIAL Workshop Report,

Virginia Beach, VA, Nov. 1985, NASA Ref. Pub. (1989).

21. R. T. Thompson, Jr., "Differential Absorption and ScatteringSensitivity Predictions," NASA Contract. Rep. 2627 (1976).

22. E. E. Remsberg and L. L. Gordley, "Analysis of Differential

Absorption Lidar from the Space Shuttle," Appl. Opt. 17, 624-630 (1978).

23. R. A. McClatchey, et al., "AFCRL Atmospheric Absorption Line

Parameters Compilation," Air Force Cambridge Research Lab-oratory, Rep. No. TR-0096 Jan. (1973).

24. W. S. Benedict and L. D. Kaplan, "Calculation of Line Widths in

H 2 0-N 2 Collisions," J. Chem. Phys. 30, 388-398 (1959).

25. T. D. Wilkerson, G. Schwemmer, B. Gentry, and L. P. Giver,

"Intensities and N2 Collision-Broadening Coefficients Mea-

sured for Selected H2 0 Absorption Lines Between 714 and 732

nm," J. Quant. Spectrosc. Radiat. Transfer 22, 315-331 (1979).26. S. R. Drayson, "Rapid Computation of the Voigt Profile," J.

Quant. Spectrosc. Radiat. Transfer 16, 611-614 (1976).

27. McClatchey et al., "Optical Properties of the Atmosphere," Air

Force Geophys. Lab., Hanscom AFB, AFCRL-72-0497 (1972).

28. M. P. Thekaekara, "Extraterrestrial Solar Spectrum, 3000-

6100A at 1-A Intervals," Appl. Opt. 13, 518-522 (1974).

29. P. P. Webb, R. J. McIntyer, and J. Conradi, "Properties of

Avalanche Photodiodes," RCA Review 35, 234-278 (1974).

30. D. E. Bowker, R. E. Davis, D. L. Myrick, K. Stacy, and W. T.

Jones, "Spectral Reflectances of Natural Targets for use inRemote Sensing Studies," NASA Ref. Pub., 1139, June (1985).

31. M. L. Wright, E. K. Proctor, L. S. Gasiorek, and E. M. Liston, "A

Preliminary Study of Air-Pollution Measurement by ActiveRemote-Sensing Techniques," NASA-CR-132724, June (1975).

32. R. L. Kenimer, "Predictions of Silicon Avalanche Photodiode

Performance in Water Vapor Differential Absorption Lidar,"Proc. Soc. Photo. Opt. Instrum. Eng. 889, 126-135 (1988).

33. M. J. T. Milton and P. T. Woods, "Pulse Averaging Methods for

a Laser Remote Monitoring System Using Atmospheric Backs-

catter," Appl. Opt. 26, 2598-2603 (1987).

34. W. B. Grant, A. M. Brothers, and J. R. Bogan, "Differential

Absorption Lidar Signal Averaging," Appl. Opt. 27, 1934-1938

(1988).35. L. S. Rothman, et al., "The HITRAN Database: 1986 Edition,"

Appl. Opt. 26,4058-4097 (1987).

36. C. Cahen and G. Megie, "A Spectral Limitation of the Range

Resolved Differential Absorption Lidar Technique," J. Quant.Spectrosc. Radiat. Transfer 25, 151-157 (1981).

37. C. L. Korb and C. Y. Weng, "The Theory and Correction of

Finite Laser Bandwidth Effects in DIAL Experiments," Ab-stracts, Eleventh International Laser Radar Conference, Mad-ison, WI, (1982).

38. G. Fiocco and J. B. Dewolf, "Frequency Spectrum of Laser

Echoes from Atmospheric Constituents and Determination ofthe Aerosol Content of Air," J. Atmos. Sci. 25, 488-496 (1968).

39. G. Fiocco, G. Benedetti-Michelangeli, K. Maischberger, and E.

Madonna, "Measurement of Temperature and Aerosol to Mole-cule Ratio in the Troposphere by Optical Radar," Nature (Lon-don), Phys. Sci., Lett. 229, 78-79 (1971).

40. S. T. Shipley et al., "High Spectral Resolution Lidar to MeasureOptical Scattering Properties of Atmospheric Aerosols. 1:Theory and Instrumentation," Appl. Opt. 22,3716-3724 (1983).

41. S. Ismail, E. V. Browell, G. Megie, P. Flamant, and G. Grew,

"Sensitivities in DIAL Measurements from Airborne and Space-

borne Platforms," Abstracts, Twelfth International Laser Ra-dar Conference, Aix-en-Provence, France (1984).

42. A. Ansmann, "Errors in Ground Based Water-Vapor DIAL

Measurements Due to Doppler-Broadened Rayleigh Backscat-tering," Appl. Opt. 24, 3476-3480 (1985).

43. S. Ismail and E. V. Browell, "Influence of Rayleigh-Doppler

Broadening on the Selection of H20 DIAL System Parameters,"Abstracts, Thirteenth International Laser Radar Conference,Toronto, Canada, Aug. 11-15 (1986).

44. 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).

45. P. B. Russell, B. M. Morley, J. M. Livingston, G. W. Grams, and

M. Patterson, "Improved Simulation of Aerosol, Cloud, andDensity Measurements by Shuttle Lidar," NASA ContractorReport 3473, Nov. (1981).

46. A. H. Lao, P. E. Schoen, and B. Chu, "Rayleigh-BrilliouinScattering of Gases with Internal Relaxation," J. Chem. Phys.64, 3547-3555 (1976).

47. H. Inaba, "Detection of Atoms and Molecules by Raman Scat-

tering and Resonance Fluorescence," in Laser Monitoring of theAtmosphere, E. D. Hinkley, Ed. (Springer, New York, 1976), pp.153-236.

48. J. Bosenberg, "Measurements of the Pressure Shift of WaterVapor Absorption Lines by Simultaneous Photoacoustic Spec-troscopy," Appl. Opt. 24, 3531-3534 (1985).

49. J.-Y. Mandin, J.-P. Chevillard, C. Camy-Peyret and J.-M.Flaud, "The High Resolution Spectrum of Water Vapor Be-tween 13200 and 16500 cm- 1," J. Mol. Spectrosc. 116, 167-190

(1986).50. V. V. Zuev, Yu. N. Ponomarev, A. M. Solodov, B. A. Tikhomirov,

and 0. A. Romanovsky, "Influence of the Shift H20 AbsorptionLines with Air Pressure on the Accuracy of the AtmosphericHumidity Profiles Measured by the Differential-AbsorptionMethod," Opt. Lett. 10, 318-320 (1985).

51. E. V. Browell, S. Ismail, and B. Grossmann, "Temperature Sen-sitivity of Absorption Lines for H20 DIAL Measurements,"Manuscript in Preparation (1989).

52. C. Cahen, J. L. Lesne, J. Benard, and P. Ponsardin, "A Meteoro-logical (Humidity, Temperature, Aerosols) Mobile DIAL Sys-

tem: Concepts and Design," Abstracts, Thirteenth Interna-tional Radar Conference, Toronto, Canada, 11-15 Aug. (1986).

53. G. K. Schwemmer et al., "A Lidar System for Measuring Atmo-

spheric Pressure and Temperature Profiles," Rev. Sci. Instrum.58, 2226-2237, (1987).

54. B. Grossmann and E. V. Browell, Abstracts, Fourteenth Inter-

national Laser Radar Conference, San Candido, Italy, 20-25June (1988).


DIAL measurements in the 0-10 km altitude region forairborne systems, stronger lines like those in the 940-nm and 1140-nm bands35 are needed for high resolu-tion spaceborne measurements at >10 km altitude.

Systematic errors in DIAL measurements arise dueto uncertainties in the knowledge of the differentialabsorption cross section between the DIAL wave-lengths. These uncertainties are caused by either in-strumental or atmospheric effects. The instrumentaleffects related to the laser spectral width, spectralpurity, tuning error, and uncertainties in the knowl-edge of spectral profiles are discussed in this paper.These parameters can be controlled during the devel-opment of a DIAL system. DIAL measurement errorsresulting from the modification of the laser spectralprofile due to absorption from H20 in the atmospherecan be evaluated and corrected if necessary duringdata analysis. Pressure shift errors can be minimizedby tuning the laser to coincide with the H20 line posi-tion at the desired altitude or a slightly higher altitude(lower pressure). Temperature sensitivity of H20lines can be minimized by selecting the suitable tem-perature insensitive H20 lines. The secondary errorsintroduced by the residual pressure shift and tempera-ture effects can be corrected during data analysis pro-vided that the pressure shift characteristics of the lineare known and an estimate of temperature profile canbe made.

Doppler broadening plays a significant role in influ-encing DIAL H20 measurements, and this effect mustbe accounted for in the data analysis. In the cleanatmosphere it can be removed to first order and inextreme cases of a turbid aerosol layer, it can be ig-nored. In regions of large vertical inhomogeneity inthe aerosol structure the ability to correct the DBeffect depends upon the knowledge of the amount ofaerosol scattering that can be derived from either theoff-line signal4 4 or by an independent measurement ofaerosol scattering (which may not be practical in allcases). Another aspect of the DB effect which be-comes significant43 at high altitudes (>10 km), is thatit reduces the DIAL measurement sensitivity to pres-sure shifts, changes in the spectral profile in the atmo-sphere, and laser detuning effects. Another spectralbroadening mechanism that can affect DIAL measure-ments is the rotational Raman scattering by air mole-cules. The influence of this effect on DIAL measure-ments of H20 in the 720-nm region can be madenegligible by selecting narrow band (FWHM < 3 nm)optical filters.

This analysis of random and systematic influencesin DIAL H20 measurements has been used in thedevelopment of the LASE system operating in the 728-nm region and in the analysis of spaceborne DIAL H20measurements. The results of this study can be usedin testing of DIAL systems, field selection of DIALparameters, planning data analysis procedures, and inthe interpretation and validation of DIAL H20 mea-surements. The method discussed in this paper canbe easily extended to DIAL H20 measurements in the940-nm region, suitable for H20 measurement at high-

er altitudes (>10 km). The approach is also applica-ble to the DIAL measurement of other atmosphericgases. We show that a <10% H20 profile measure-ment accuracy is possible for LASE with a vertical andhorizontal spatial resolution of 200 m and 10 km, re-spectively, during nighttime and 300 m and 20 km,respectively, during daytime. Global measurementsof H20 profiles from spaceborne DIAL systems can beachieved with similar accuracy with spatial resolutionsof 500 m in the vertical and 100 km in the horizontal.We show that the H20 DIAL systems considered inthis analysis can meet the measurement needs for H20in the troposphere on regional and global scales.

We wish to thank W. B. Grant, A. F. Carter, W. M.Hall, R. L. Kenimer and B. Grossmann for their help-ful comments during many discussions. We alsothank G. Grew and S. Yeh for some of the initialcomputer program developments and S. Kooi for thecomputer graphics support. This research was par-tially supported by NASA Contracts NAS1-16115 andNAS1-18460 with STX, Hampton, VA.

References

1. C. Prabhakara and G. Dalu, "Passive Remote Sensing of theWater Vapor in the Troposphere and its Meteorological Signifi-cance," in Atmospheric Water Vapor, A. Deepak, T. D. Wilker-son, and L. H. Ruhrke, Eds. (Academic, New York, 1980).

2. S. H. Melfi, D. Whiteman, R. Ferrare, and V. Falcone, "Observa-tion of Frontal Passages Using a Raman Lidar," Abstracts,Fourteenth International Laser Radar Conference, San Can-dido, Italy, 24-26 June (1988).

3. R. M. Schotland, "Errors in the Lidar Measurement of Atmo-spheric Gases by Differential Absorption," J. Appl. Meteorol.13, 71-77 (1974).

4. R. T. H. Collis and P. B. Russell, in Laser Monitoring of theAtmosphere, E. D. Hinkley, Ed., (Springer, New York, 1976), p.117-151.

5. R. M. Schotland, "Some Observations of the Vertical Profile ofWater Vapor by Means of a Ground Based Optical Radar," inProceedings, Fourth Symposium on Remote Sensing of Envi-ronment, 12-24 Apr. 1966 (U. Michigan, Ann Arbor, 1966).

6. E. V. Browell, T. D. Wilkerson, and T. J. McIlrath, "WaterVapor Differential Absorption Lidar Development and Evalua-tion," Appl. Opt. 18, 3474-3483 (1979).

7. C. Cahen, G. Megie, and P. Flamant, "Lidar Monitoring ofWater Vapor Cycle in the Troposphere," J. Appl. Meteorol. 21,1506-1515 (1982).

8. E. V. Browell, "Remote Sensing of Tropospheric Gases andAerosols with an Airborne DIAL System," in Optical and LaserRemote Sensing, D. K. Killinger and A. Mooradian, Eds.,(Springer-Verlag, New York, 1983), pp. 138-147.

9. E. V. Browell, A. K. Goroch, T. D. Wilkerson, S. Ismail, and R.Markson, "Airborne DIAL Water Vapor and Aerosol Measure-ments Over the Gulf Stream," Abstract, Twelfth InternationalLaser Radar Conference, Aix-en-Provence, France, 13-17 Aug.(1984).

10. V. E. Zuev, Yu S. Makushkin, V. N. Marichev, A. A. Mitsel, andV. V. Zuev, "Lidar Differential Absorption and Scattering Tech-nique," Appl. Opt. 22, 3733-3741 (1983).

11. W. B. Grant, J. S. Margolis, A. M. Brothers, and D. M. Tratt,"CO 2 DIAL Measurements of Water Vapor," Appl. Opt. 26,3033-3042 (1987).

12. J. Bosenberg, "A DIAL System for High Resolution Water Va-por Measurements in the Troposphere," in Laser and Optical


airborne and spaceborne lidar measurements of water vapor profiles: a sensitivity analysis

Documents