The Airborne Demonstrator for the Direct-Detection Doppler Wind Lidar ALADIN on ADM-Aeolus ...

15 Pages·2009·1.53 MB·English

2516 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME26
The Airborne Demonstrator for the Direct-Detection Doppler Wind Lidar
ALADIN on ADM-Aeolus. Part II: Simulations and Rayleigh Receiver
Radiometric Performance
ULRIKEPAFFRATH,CHRISTIANLEMMERZ,OLIVERREITEBUCH,ANDBENJAMINWITSCHAS
DeutschesZentrumfu¨rLuft-undRaumfahrt,Institutfu¨rPhysikderAtmospha¨re,Oberpfaffenhofen,Germany
INESNIKOLAUS
PhysicsSolutions,Munich,Germany
VOLKERFREUDENTHALER
Ludwig-Maximilians-UniversityofMunich,Munich,Germany
(Manuscriptreceived7April2009,infinalform21July2009)
ABSTRACT
IntheframeoftheAtmosphericDynamicsMissionAeolus(ADM-Aeolus)satellitemissionbytheEu-
ropeanSpaceAgency(ESA),aprototypeofadirect-detectionDopplerwindlidarwasdevelopedtomeasure
windfromgroundandaircraftat355nm.WindismeasuredfromaerosolbackscattersignalwithaFizeau
interferometerandfrommolecularbackscattersignalwithaFabry–Perotinterferometer.Theaimofthis
studyistovalidatethesatelliteinstrumentbeforelaunch,improvetheretrievalalgorithms,andconsolidate
theexpectedperformance.Thedetectedbackscattersignalintensitiesdeterminetheinstrumentwindmea-
surementperformanceamongotherfactors,suchasaccuracyofthecalibrationandstabilityoftheoptical
alignment.Resultsofmeasurementsandsimulationsforaground-basedinstrumentarecompared,analyzed,
anddiscussed.Thesimulatedatmosphericaerosolmodelswerevalidatedbyuseofanadditionalbackscatter
lidar.ThemeasuredRayleighbackscattersignalsofthewindlidarprototypeuptoanaltitudeof17kmare
comparedtosimulationsandshowagoodagreementbyafactorbetterthan2,includingtheanalysesof
differenterrorsources.FirstanalysesofthesignalattheMiereceiverfromhighcirruscloudsarepresented.In
addition,thesimulationsoftheRayleighsignalintensitiesoftheAtmosphericLaserDopplerInstrument
(ALADIN)AirborneDemonstrator(A2D)instrumentongroundandaircraftwerecomparedtosimulations
ofthesatellitesystem.Thesatellitesignalintensitiesabove11.5kmarelargerthanthosefromtheA2D
ground-basedinstrumentandalwayssmallerthanthosefromtheaircraftforallaltitudes.
1. Introduction mate(Bakeretal.1995;ESA1999).Satellite-basedlidar
systems offer the potential foradequateverticalresolu-
Atpresent,ourinformationonthethree-dimensional
tionandglobalcoverage.
wind field over the Northern Hemisphere oceans, the
WithinthecontextoftheEarthExplorercoreprogram
tropics,andtheSouthernHemisphereisincompletebe-
oftheEuropeanSpaceAgency(ESA),theAtmospheric
cause of insufficient measurement data. There are sig-
Dynamics Mission Aeolus (ADM-Aeolus) comprises
nificantareaswheremeasurementsdonotyieldreliable
adirect-detectionDopplerlidartomeasureglobalwind
data,andthereisastrongdemandforimprovementsin
fieldsfromsatellite,whichwillbethefirstEuropeanli-
global wind profiles, which are crucial for numerical
darinspaceandthefirstwindlidarinspaceworldwide.
weather prediction andstudiesrelatedtothe global cli-
The lidar system on ADM-Aeolus is the Atmospheric
LaserDopplerLidarInstrument(ALADIN),whichis
designed to provide global observations of wind pro-
Corresponding author address: Oliver Reitebuch, Deutsches
filesinthetroposphereandlowerstratospherefornu-
Zentrum fu¨r Luft- und Raumfahrt, Institut fu¨r Physik der At-
mospha¨re,Oberpfaffenhofen,82234Wessling,Germany. mericalweatherprediction(ESA2008;Stoffelenetal.
E-mail:[email protected] 2005).
DOI:10.1175/2009JTECHA1314.1
(cid:2)2009AmericanMeteorologicalSociety
DECEMBER2009 PAFFRATH ET AL. 2517
In the frame of the ADM-Aeolus program, a proto- the A2D on ground and aircraft are compared to sim-
typeinstrumentwasdeveloped—theALADINAirborne ulationsofthesatellite.
Demonstrator (A2D)—to validate the measurement The A2D instrument is introduced in section 2. The
principlewithrealisticatmosphericsignals from ground simulator is presented in section 3, and results of the
and aircraft before satellite launch. The instrument de- radiometricperformancearediscussedinsection4.
sign of the A2D is described by Reitebuch et al. (2009,
hereafterPartI).
2. Instrumentdescription
To evaluate the measurement capability of the in-
strument and to predict its performance, the detected TheA2Dincludesafrequency-tripledNd:YAGlaser
signalintensitywasanalyzed.Therandomerrorofwind operatingatawavelengthof354.89 nm(Schro¨deretal.
measurements of the ALADIN instrument is mainly 2007),aCassegraintelescopewithadiameterof0.2 m,
determinedbythesignalintensityresultingfromphoton acoaxiallaserbeampathwithrespecttothetelescope,
noise. A simulator was developed to represent the andtwospectrometers(Fig.1)todetecttheaerosoland
ALADIN instrument for performance analyses and to molecularbackscattersignal(Durandetal.2005;PartI).
improvetheprocessingalgorithms.Theobjectiveofthis Details oftheA2Ddesignandcomparisonstothesat-
paperistocomparetheexpectedsignalintensitiesfrom elliteinstrumentcanbefoundinPartI.
simulations with measurements to validate the radio- The emitted photons of the laser pass the transmit
metricperformance. optics with a transmission t , which includes three
T
The radiometric performance of direct-detection mirrors, and for the airborne systems,the aircraft win-
Dopplerlidarswasdeterminedbysimulations,beginning dow. The backscatter signal from the atmosphere is
in 1979, for different spaceborne instruments (Abreu collected by the telescope, reflected by mirrors, and
1979; Menzies 1986; Rees and McDermid 1990; McGill passesthefrontopticswithareceivetransmissiont .
R
etal.1999).Theradiometricperformancewasvalidated Thebackscattersignalispartlytransmittedthroughthe
forattenuatedbackscatter,asshownbyTaoetal.(2008), Fizeau interferometer and imaged as a fringe onto the
formeasurementsofaground-basedinstrumentandthe detector.Thesignalstrengthdependsonthewavelength-
current satellite lidar on the Cloud-Aerosol Lidar and dependent transmission of the Fizeau interferometer
Infrared Pathfinder Satellite Observation (CALIPSO). t (l) and the peak transmission of 0.406, which is de-
Fiz
Acomparisonofmeasuredandmodeledsignalintensities scribedinsection3e.Thistransmissiont (l),whenin-
Fiz
was shown by Fischer et al. (1995) for a ground-based tegrated over the imaged spectral range, yields the
wind lidar, concluding that the measured radiometric spectralefficiencyof12.7%.Furthermore,apeaktrans-
signal intensities are within the range of the modeled missionoftheFizeauinterferometerof40.6%hastobe
values. A performance validation was presented by taken into account. The aperture of the Fizeau in-
Gentry and Chen (2003) for a mobile wind lidar at terferometer is circular; because of the truncation at
a wavelength of 355 nm, where simulations and mea- asquaredetectorplane,thesignalisreducedbyafactor
surementscorrespondtoeachother. of2/p,whichiscalledthepupiltruncationratio(Paffrath
A lidar simulator was introduced by Veldman et al. 2006).
(1999)toanalyzetheperformanceoftheADM-Aeolus AfterreflectionattheFizeauinterferometer,theback-
instrumentduringitsinitialdesignphase(ESA1999).It scatter signal is directed toward the Fabry–Perot inter-
wasfurtherappliedbyMarseilleandStoffelen(2003)for ferometer.TheDopplerfrequencyshiftofthemolecular
the performance prediction regarding different atmo- backscatter signal is measured with the double-edge
spheric conditions. A simulator of the direct-detection method using a Fabry–Perot interferometer (Garnier
lidarfortheADM-Aeolusinstrumentwasdevelopedby and Chanin 1992; Gentry et al. 2000), which is charac-
Leikeetal.(2001)andupdatedtoincorporatetheactual terizedbyanewmethodtoseparatelightdependingon
ALADIN satellite design. This simulator provided the polarizationcalledthesequentialtechnique(Fig.2).
basisoftheA2Dsimulator,whichwasusedtovalidate In the sequential Fabry–Perot interferometer of the
theA2Dinstrument(Paffrath2006). A2D with a mean spectral reflection of 82%, the in-
MeasurementswiththeA2Dwereperformedin2007 coming photons are directed first to filter A with a
and 2008 from ground. It is planned to extend the resulting mean spectral transmission of about 18%.
analysisoftheradiometricperformance withmeasure- Thus,82%ofthephotonsarereflectedoffthefirstfilter
mentsfromaircraftinadownward-lookingperspective, tothesecondfilterB,whereagain18%oftheincoming
as for the ALADIN satellite. In this study, the radio- photons are transmitted, resulting in a total trans-
metric performance of simulations and ground-based mission of 15% to filter B. Such a scheme is more ef-
measurementsisanalyzed.Additionally,simulationsof ficient than conventional nonsequential Fabry–Perot
2518 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME26
FIG.1.BlockdiagramoftheA2Dreceiverwiththecorrespondingopticaltransmissionco-
efficientsusedforsimulations.ThesquaredetectorplaneoftheRayleighreceiverisdividedinto
twosectionsforfiltersAandB.TheefficiencyoftheMie(Rayleigh)receiveris0.6%(1.5%).
interferometers, where a beam splitter halves the in- ground were performed with zenith-looking geometry,
comingfluxtoeachfilter. whereonlyverticalwindsaffectthesignal.Theincoming
The sequential routing technique results in different photonsintheRayleighreceiverpatharereducedbythe
peaktransmissionsforfiltersAandB.Thetransmissions filterpeaktransmission(0.293ofAand0.196ofB)and
ofbothfiltersforzerowindspeedandaDoppler-shifted the spectral efficiency. The spectral efficiency of the
spectrumfor 250 m s21(thelargevalueistakenforil-
lustration) are shown in Fig. 3, where the Rayleigh
spectrumwithzerowindspeediscenteredclosetothe
crosspointofbothfiltercurves.Inthepresenceofwind
speed,however,theRayleighspectrumisshiftedtoward
one maximum of the filter curves, resulting in a differ-
ence of intensity ratio of the two filters. The Doppler
shiftcanbedeterminedfromeithertheratioofthein-
tensitiesA/B(FlesiaandKorb1999;Gentryetal.2000)
orfromthecontrastfunction(A2B)/(A1B),assug-
gested by Chanin et al. (1989) and used for ALADIN
(Dabasetal.2008;Tanetal.2008).
To analyze the radiometric performance, the signal
intensities that are transmitted through filters A and B
aresummedup.Thistotalsignalisonlyslightlyaffected
bythewindspeed.Windsof10 m s21alongthelineof
sightleadtoashiftoftheRayleighspectrumof0.02 pm FIG. 2. Illustration of the spectral efficiency of the sequential
Fabry–Perot interferometer for a mean reflection of 82%: the
and therefore only lead to small variations in the total
transmittedsignaloffilterA(B)with18%(15%)efficiencyresults
signal of A and B of less than 2%. To validate the ra-
inthetotalspectralefficiencyof33%forabroadbandRayleigh
diometric performance, the A2D measurements from spectrum.
DECEMBER2009 PAFFRATH ET AL. 2519
FIG.3.Principleofthedouble-edgemethodwithfiltersAandBandanatmosphericsignal
spectrumwiththenarrowbandMieandthebroadbandRayleighsignalfor0and250ms21.
GrayareasindicatethetransmittedsignaltotheRayleighdetectorfor0ms21.
Fabry–Perot interferometer describes the ratio of ulator is characterized by a high vertical atmospheric
transmitted to incoming photons and is determined by layerresolutionof15m.Simulationsareperformedwith
the spectral width of the Rayleigh signal, the filter singleoraccumulatedlaser-pulsespectrawithaPoisson-
spectral widths, and the filter spectral spacing. Alto- distributed random number of detected photons from
gether,thetransmissionfactorsoftheRayleighreceiver atmospheric scattering processes. The input parameters
arethespectralefficiency(0.33),thepeaktransmission of the simulator are adapted to the actual instrumental
(mean0.244),thereceiveopticstransmission(0.22),and parameters.
thequantumefficiencyofthedetector(0.85),andthey
a. Backscattersignal
result in a Rayleigh receiver efficiency of 1.5%. The
correspondingefficiencyis0.6%fortheMiereceiver. The lidar equation is used to determine the back-
The instrument parameters of the A2D are listed in scattersignaldetectedbyalidarsystem.Thenumberof
Table1.TheA2Ddetectorisanaccumulationcharged signal electrons on the detector per laser pulse from
coupleddevice(ACCD)thatiscapableofaccumulating adistancerfromthelidarsystemisgivenby
electronicchargesfromseverallaser-pulsereturns.The
incomingphotons at theACCD are convertedto elec- N (l, r)5N DR(A)t(l)T t t m b(r)T2(r), (1)
trons with a quantum efficiency of 0.85. The signal is e L r2 p R T eff
imagedontoalightsensitiveareaof16316pixels.An
whereAistheareaofthereceivertelescope;DRisthe
electro-optic modulator (EOM) in the front optics is
depthofthesensingvolume;bistheatmosphericback-
used to reduce the backscatter light close to the in-
scatter coefficient; and T2 is the atmospheric two-way
strumentupto1 kmtoavoidasaturationoftheACCD.
transmission,includingmolecularandaerosolextinction.
Duringmeasurementsin2007,thetransmissionofthe
The instrumental parameters are the wavelength-
EOMwasreducedto0.75forallaltitudes.However,the
dependenttransmissiont(l),thefilterpeaktransmission
nominaltransmissionof100%wasachievedin2008.
T (T ,T ,andT ),thereceiveopticstransmissiont ,
p A B Fiz R
thetransmitopticstransmissiont ,andthequantumef-
T
3. Simulations:Atmosphereandinstrument ficiencym ofthedetector.ThetermN istheequivalent
eff L
number of emitted photons by the laser, which are de-
The ALADIN prototype simulator was developed
rivedfrom
to represent the A2D operated on ground and air-
craft.Thesimulatorincludesthelasertransmitter,the l
receiver,thedetectionunit,andtheinteractionofthe NL5 hLc EL, (2)
transmitted light with the atmosphere. This enables
the study of the radiometric and wind measurement wherel isthewavelengthofthelaser,cisthespeedof
L
performance for different atmospheric states and the light,hisPlanck’sconstant,andE istheenergyofthe
L
improvementofthewindretrievalalgorithms.Thesim- laserpulse(Table1).
2520 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME26
TABLE1.InstrumentparametersoftheA2Dandthesatelliteinstrumentusedforsimulations.
Valueinsimulation
Instrument Symbol Parameter A2D Satellite
Instrument z Altitude 0km 408km
i
Laser l Laserwavelength 354.9nm* 355nm
L
E Laserpulseenergy 57mJ* 120mJ
L
FWHM Laserlinewidth 0.021pm* 0.021pm
L
u Laserdivergence 200mrad(2007)* 12mrad
L
100mrad(2008)*
Receiver u FOV 100mrad 19mrad
R
d Telescopediameter 0.2m* 1.5m
Tel
t Transmitopticstransmission 0.98* 0.66
T
t Receiveopticstransmission 0.22* 0.42
R
t EOMtransmission 0.75(2007)* 1.0
EOM
1.00(2008)*
Fizeauinterferometer T Peaktransmission 0.406* 0.60
Fiz
Spectralefficiency 0.127** 0.135
FilterFWHM 0.059pm* 0.059pm
FilterUSR 0.69pm* 0.69pm
FilterFSR 0.92pm** 0.92pm
Fabry–Perotinterferometer T PeaktransmissionA 0.293* 0.68
A
T PeaktransmissionB 0.196* 0.61
B
Spectralefficiency 0.33** 0.37
FilterFWHMA 0.749pm* 0.70pm
FilterFWHMB 0.752pm* 0.70pm
FilterFSR 4.6pm* 4.6pm
Spectralspacing 2.603pm* 2.300pm
ACCD m Quantumefficiency 0.85* 0.85
eff
*Measured.
**Derivedfromcalculations.
b. Theatmosphere from the atmospheric temperature T(z) and pressure
p(z)profile,
The atmospheric backscatter and extinction coeff-
icients are calculated from the reference model atmo- (cid:2)296K(cid:3)(cid:2) p(z) (cid:3)
b 5 N s , (3)
sphere (RMA) climatology data, which were derived Mol T(z) 1:013 3 105Pa L Mol
fromfieldcampaignsbefore1991(Vaughanetal.1995).
These data were used for different satellite lidar simu- withN 52.47931025 m23moleculespervolume,the
L
lations(MarseilleandStoffelen2003;DiGirolamoetal. Loschmidt’s number referenced to a temperature of
2008; Ehret et al. 2008). The aerosol backscatter co- 296 K,andapressureof1.0133105 Pa.
efficientsofthemedianmodelagreewithinanorderof Thetwo-waytransmissionisderivedfrom
magnitude and better with aerosol measurements in
" #
ðz
Europeduringthelastfewyears(Wandinger2003).The T2(z)5exp (cid:2)2 ta(z)dz , (4)
temperature and pressure profiles of this study were
z
i
takenfromtheU.S.StandardAtmosphere,1976,which
representsanidealizedstateoftheearth’satmosphere, where the total extinction a is the sum of the aerosol
referring to a period with moderate solar activity for extinction a and molecular extinction a . The alti-
A Mol
various climatic conditions (Champion 1985). Option- tude of the instrument is z, and the altitude of the at-
i
ally, temperature, pressure, cloud cover, backscatter mospherictargetisz.
t
coefficients,andwindprofilesfromobservationscanbe The molecular extinction is calculated from a 5
Mol
usedasinputtothesimulator. b 8p/3. The aerosol extinction is derived from a 5
Mol A
Themolecularbackscattercoefficientisderivedfrom 50b ,wheretheextinction-to-backscatterratioorlidar
A
theRayleighbackscattercrosssectionperairmolecule ratioisassumedtobeaconstantvalueof50sr,whichcan
s (58.444 3 10232 m2 sr21 at 355 nm; Collis and beassumedasameanforcontinentalaerosol(Vaughan
Mol
Russell 1976) and the number of molecules N per et al. 1995; Winker et al. 1996; Marseille and Stoffelen
Mol
unitvolumedependingonaltitudez,whichiscalculated 2003).
DECEMBER2009 PAFFRATH ET AL. 2521
c. Spectraldistributionofthebackscattersignal
Thespectraforscatteringonaerosolsandmolecules
arecalculatedindependenceonwavelengthandatmo-
spheric temperature. The spectrum of laser light is
broadened for molecular scattering resulting from the
molecular thermal motion, which may be described by
a Gaussian line profile function with a standard de-
viation(SD)s of
R
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2l kTN
s 5 L A, (5)
R c m
air
wherem isthemeanmolecular airmass(2.93 1022
air
kg mol21), k is the Boltzmann constant, and N is the
A
Avogadro constant. The spectrum of laser light scat-
teredbyaerosolsisassumedtobeequaltothefullwidth
FIG. 4. The signal-loss factor f(r) resulting from a laser di-
athalfmaximum(FWHM)ofthelaser-pulsespectrum vergenceof100(dashedline)and200mrad(boldline)forare-
FWHML(seeTable1). ceiverFOVof100mrad.
d. Effectofthelaserbeamexceedingthereceiver
fieldofview
backscatter signal is reduced by a factor f(r) depend-
Ifthelaserbeamdivergence(u 5200mradin2007 ing on the range r, which is calculated from the ratio
L
and100mradin2008)givenasa63svalueislargerthan of the received power P , restricted by the FOV of
rec
thefieldofview(FOV;u 5100mrad)ofthereceiver, the receiver with radius r , and the laser beam
R FOV
then the transmitted laser beam power is partially lost powerP withradiusr (Witschas2007).Thepower
total L
and the backscatter signal is reduced. The laser beam of the laser beam is calculated by integration over the
intensity distribution was determined to be Gaussian two-dimensional beamprofileintensities in thetwodi-
with low M2 values of 1.9 in 2007 and 1.2 in 2008. The rectionsx(r)andy(r):
ð1rFOVð1rFOV (cid:5) (cid:2)9[x2(r)1y2(r)](cid:3)(cid:6)
I (r)e (cid:2) dxdy
P 0 2r2(r)
f(r)5 rec 5 (cid:2)rFOV (cid:2)rFOV L , (6)
P ð1‘ð1‘ (cid:5) (cid:2)9[x2(r)1y2(r)](cid:3)(cid:6)
total I (r)e (cid:2) dxdy
0 2r2(r)
(cid:2)‘ (cid:2)‘ L
whereI isthemaximumlaserintensityatdistancerand functions. The Airy function is written as (Vaughan
0
r isthe3s radiusofthelaserbeam,whichisapproxi- 2002)
L
matedbyr 50.5u r.ThediameteroftheFOVofthe
L L
receiver can be calculated from rFOV 5 0.5uRr 1 t(l)5[11Fsin2(u/2)](cid:2)1, (7)
r . The function f(r) is shown in Fig. 4 for a re-
Tel(r50)
ceiver FOV of 100 mrad and laser divergences of 100
whereFisthecoefficientoftheFinesse,whichisderived
(dashed line) and 200 mrad (bold line). Even with the fromF5(2F /p)2andthereflectivefinesseofF 5FSR/
r r
laser beam within the FOV of the receiver, there is
FWHM.Thefreespectralrange(FSR)isdefinedasthe
about5%lossofbackscatter lightatthereceiverfrom
spectraldistanceofthetransmissionmaximawithFSR5
14-kmaltitudebecauseoftheGaussianbeamprofile.A l2/(2nd) for perpendicular incidence of light (Vaughan
laserdivergenceof200mradresultsinalossofpowerof 2002). The phase u 5 4pnd(cosd)l21 is linked to the
about45%at14-kmaltitude.
wavelength, where d is the angle of incidence, d is the
distance of the two etalon plates of the interferometer,
e. TheMieandRayleighspectrometer
andnistherefractiveindexofthemediumbetweenthe
FortheMiereceiverwiththeFizeauinterferometer, plates. The transmission of the Mie spectrometer de-
the filter transmission curves are assumed to be Airy pendingonwavelengthlis
2522 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME26
(cid:5) (cid:2)p(cosd)l(cid:3)(cid:6)(cid:2)1 atmosphere, the background light during daytime op-
t (l)5 11Fsin2 , (8) eration, and an electronic detection chain offset. The
Fiz FSR
Fiz detectionchainoffsetisaconstantelectricvoltageatthe
whereFSR andFWHM arethefreespectralrange analog digital converter. The background light is re-
Fiz Fiz
and full width at half maximum of the Fizeau in- duced by several filters in the A2D front optic. The
terferometer,respectively,accordingtoTable1.Forthe remaining background light and the detection chain
case where the laser beam is out of the FOV and the offsetaredeterminedfromanadditionalmeasurement
EOM transmissionsmallerthan1,thenumberof elec- andremovedduringsignalprocessing.
tronsattheMiedetectorN iscalculatedby TovalidatetheRayleighradiometricperformance,the
e,Fiz
signal electrons on the Rayleigh receiver were analyzed
N (l, r)5N (l, r)f(r)t , (9)
e,Fiz e EOM for cloud-free sky on different days with similar atmo-
spheric temperature profiles. The measured Rayleigh
withN (l,r)fromEq.(1),thefactorf(r)[Eq.(6)],and
e signal level only depends slightly on atmospheric tem-
the EOM transmission t (Table 1). The signal
EOM perature.Furthermore,forhigheraltitudes,pureRayleigh
photons that are not transmitted through the Fizeau
signal isexpected at the receiver without impact of Mie
interferometer are reflected to the Rayleigh receiver
signal.FirstanalysesoftheMiesignalattheMiereceiver
(N ). The transmission of each filter of the Fabry–
e,FP arepresentedfromhighercirrusclouds.Multipurposeli-
Perot interferometer can be described by the Airy
dar system (MULIS) measurements are available from
function,
2007, and there was one event with cirrus clouds during
(cid:5) (cid:2)p(cosd)l(cid:3)(cid:6)(cid:2)1 thecampaign.Furtheranalysesareplannedinthefuture.
t (l)5 11Fsin2 . (10)
FP FSR Becauseoftheimpactofthetelescopeoverlapupto
FP
2 km and the attenuation of the signal close to the in-
ThenumberofelectronsontheRayleighdetectorfor strument resulting from the EOM, measurements and
filterAiscalculatedusingEqs.(1)and(10)withf(r)and simulations were evaluated above 2-km altitude. The
theEOMtransmissiontEOM: range bins close to the instrument up to 2 km are
315.6 m, and range bins for higher altitudes are from
N (l,r)5N (l, r)f(r)t . (11)
A e,FP EOM 631.2to1262.4 m.Therangebinwidthdependsonthe
ACCDintegrationtimest ,whichcanbeamultipleof
The reflected signal of Rayleigh filter A is transmitted int
2.104ms(315.6 m).Duringsignalprocessing,thesignal
through filter B, and the number of electrons on the
electronsattheRayleighreceiverACCDfromfiltersA
Rayleighdetectorforthisfilteriscalculatedwith
andBaresummedupandscaledtoarangebinwidthof
NB(l, r)5TB(l)[1(cid:2)TA(l)]NA(l, r). (12) 315.6 m by a factor of tint/2.104 ms to have a uniform
verticalresolutionfrommeasurementsandsimulations.
The vertical signal profiles presented in this paper are
theresultof630accumulatedlaserpulsesduring14 sat
4. Radiometricperformance
thedetector.Atotalof10%of700pulsesfromalaser
Ground-basedmeasurementswereperformedinJuly with50-HzrepetitionratearelostbecauseoftheACCD
2007attheRichardAßmannObservatoryoftheGerman readout.Tovalidatetheradiometricperformanceofthe
Weather Service (DWD; Deutscher Wetterdienst) in instrument,differentdayswithoutcloudsarecompared.
Lindenberg,65 kmsoutheastofBerlin(528139N,148089E, Simulationsofverticalprofilesofsignalelectronsfor
97m MSL) and in October 2008 at the German Aero- differentatmosphericconditionsareintroducedinsec-
space Center (DLR; Deutsches Zentrum fu¨r Luft- und tion4a.Section4bshowsresultsofmeasuredbackscatter
Raumfahrt) in Oberpfaffenhofen (488049N, 118169E, coefficients by an aerosol lidar compared with atmo-
620 mMSL). sphericmodelsforaerosolcontent.A2Dmeasurements
ThemaindifferencebetweentheA2Dmeasurements fromgroundarevalidatedwithsimulationsinsections4c,
in2007and2008wasthelaserdivergence,whichwas200 4d, and 4e. Simulations of the A2D from ground and
mradin2007andthuslargerthanthe100-mradfieldof airborneplatformsandthesatelliteinstrumentareshown
viewofthereceiver,andthetransmissionoftheEOM, insection4f.
whichwas75%in2007(Table1).In2008,thelaserbeam
a. Simulations
divergencewas100mrad,withinthefieldofviewofthe
receiver,andtheEOMtransmissionwas100%. The signal electrons on the Rayleigh detector have
The measured signal electrons at the A2D detector been calculated for different aerosol content (median
arise from the laser light that is backscattered by the and lower-quartile models from the RMA; Fig. 6) and
DECEMBER2009 PAFFRATH ET AL. 2523
FIG. 5. Simulations of the number of signal electrons on the
Rayleighdetectorfordifferentatmospherictemperatureprofiles FIG. 6. Simulations of the number of signal electrons on the
(U.S.StandardAtmosphere,1976;midlatitudesummer;andarctic Rayleigh detector for the U.S. Standard Atmosphere, 1976 tem-
winter;thinblacklines).Theratioofthenumberofelectronsfrom perature profile and different aerosol models. The ratio of the
the U.S. Standard Atmosphere, 1976 to the midlatitude summer numberofelectronsregardingthemedian(dottedline)andlower-
(bolddashedline)andtoarcticwinter(dottedline)isuptoafactor quartile(boldline)aerosolmodelsisuptoafactorof1.4(dashed
of1.1. line).
quartile aerosol models with the temperature profile of
different atmospheric temperatures (U.S. Standard At- theU.S.StandardAtmosphere,1976showdifferencesin
mosphere,1976;midlatitudesummer;andarcticwinter; signal up to a factor of 1.4. The aerosol backscatter co-
Fig. 5). Temperature and pressure determine the mo- efficients were determined from measurements by an
leculardensityandthusthemolecularbackscattersignal aerosolbackscatterlidarin2007.Therewerenoaccom-
[Eq.(3)]. panyingmeasurementsofthebackscattercoefficientsin
Thetemperaturescanvaryoveralargerange,andthe 2008.Itisassumedthatthedifferencesofsignalin2008
U.S. Standard Atmosphere,1976 model isclose tomea- presented in this paper, up to a factor 1.4 (40%), can
sured temperature profiles during the A2D observation arisefromdifferencesinatmosphericaerosolcontent.
periodinJuly2007andOctober2008.Twofurthertem- In addition, the aerosol content does not only de-
peratureprofileshavebeenconsidered:thearcticwinter crease the Rayleigh signal by extinction above aerosol
profileandthemidlatitudesummerprofilewithamaxi- layers,butitalsoincreasetheRayleighreceiversignals,
mum temperature difference of 30 K at an altitude of whichiscalledcrosstalk,inthecaseofMiebackscatter
10km,whichresultsinadifferenceofafactor1.1ofthe signal from aerosols. From the Fizeau interferometer,
Rayleigh signal (Fig. 5). The differences in signal from about5%oftheincomingphotonsaretransmitted(with
expected temperature differences during several mea- apeaktransmissionof40.6%andaspectralefficiencyof
surementperiodsisabout5 K,whichresultsinavariation 0.127; see Fig. 1) and 95% are reflected toward the
ofsignalofabout61%(Table3). Rayleighreceiver(section3e).Thus,theRayleighsignal
Theincreasedaerosolcontentintheboundarylayeris levels in the boundary layer with aerosols are signifi-
acauseofhigherextinctionoflaserlightforlidarsystems cantlyincreasedbecauseofcrosstalkoftheMiesignal.
operatingfromground,whichleadstoadecreaseinsignal Inaddition,thefirst2 kmarestronglyinfluencedbythe
from higheraltitudes.Themedianaerosolmodelrepre- telescopeoverlap.Inthefollowing,analysesofthesig-
sentstheatmosphereduringmostdaysinJuly2007quite nalattheRayleighreceiverwereperformedforcloud-
well(section4b).TheRayleighsignalinFig.6,withthe freeconditionsabove2 km.ThesignalattheRayleigh
median aerosol model (dotted line), shows lower signal receiverincreasesignificantlybyMiesignalfromclouds;
fromhigheraltitudes(e.g.,10 km)resultingfromextinc- hence,onlyeventswithcloud-freeskyareconsideredto
tionfromaerosolsintheboundarylayercomparedtothe determinetheRayleighradiometricperformance.
lower-quartilemodel(boldline).Dayswithloweraerosol
b. MULISmeasurementsofaerosolbackscatter
contentarerepresentedbythelower-quartilemodeland
coefficients
show increased Rayleigh signal from higher altitudes
(Fig. 6,boldline) resulting from the lower extinction in TheMULISfromtheUniversityofMunichisamo-
theboundarylayer.Simulationsofthemedianandlower- bilebackscatterlidarwithanNd:YAGlaseroperatingat
2524 JOURNAL OF ATMOSPHERIC AND OCEANIC TECHNOLOGY VOLUME26
threewavelengths(1064,532,and355 nm).Thevertical
profiles of the backscatter and extinction coefficients
at 355 nm were derived from MULIS measurements
(Freudenthaleretal.2009)bymeansofthemethodde-
scribed by Klett (1985), assuming a height-independent
extinction-to-backscatterratioof55sr.Thevolumeand
aerosoldepolarizationwasmeasuredat532 nm.Forthis
study,theverticalprofileswitharesolutionof7.5mare
averagedover10min.
Thecomparisonsoftheaerosolbackscattermodelsof
theRMAand MULISmeasurements duringJuly2007
(0628and1226UTC8July;0333,0613,and0849UTC
14July;0750UTC15July;and1010and1830UTC17 FIG.7.AerosolbackscattercoefficientfromMULISmeasure-
July)isshowninFig.7,wheretheMULISobservations ments at 355nm during different days in July 2007 (thin black
areindicatedasthinblacklines.Allthesemeasurements lines).ThemeanaerosolbackscattercoefficientofMULISisrep-
resentedbythedottedline.TheRMAlower-quartile(dashedline
represent different aerosol loadings during periods
ontheleft),RMAmedian(boldline),andRMAhigher-quartile
without clouds. The mean value of all MULIS mea-
model atmosphere (short-dashed line on the right) are in good
surements is close to the RMA median aerosol model.
agreementwiththeMULISmeasurements.
Most of the MULIS measurements fall between the
lower-quartile and the higher-quartile aerosol models,
electronsattheMiereceiverwereobservedfromacir-
which vary by a factor of 10 in aerosol backscatter co-
ruscloudat9-kmaltitude,whichisafactorofabout10
efficientupto2.5 km.Althoughtheaerosolbackscatter
lower than from simulations. The high signal from the
coefficient of the median and lower-quartile models
zenith-pointing measurement may arise from specular
differs by a factor up to 5, the effect on the Rayleigh
reflectance, which occurs for specific ice crystal orien-
signalisonlyafactorof1.4(Fig.6).
tations (Noel and Sassen 2005). To avoid specular re-
c. FirstresultsontheMieradiometricperformance flectance, the MULIS lidar was pointing 28 off zenith,
buttheA2Dwaspointingtowardzenithtoavoidimpact
TheradiometricperformanceoftheMiereceiverwas
ofhorizontalwind.A2Dmeasurementswitha28–48off-
not analyzedin detailuptonowbecause oflargevari-
zenith-pointingsystemhavetobeperformedtoexclude
ations in the backscatter signal arising from aerosol
specularreflectance.
variations,depolarization,andspecularreflectancefrom
Hard target measurements were performed with a
clouds or, in the case of a hard target, of unknown al-
surfaceofunknownalbedoinadistanceof1–2 km,which
bedo.Inafirststep,aroughestimateoftheMiesignal
isstillinthetelescopeoverlaprange.Hence,thesemea-
fromcirruscloudsat8–10-kmaltitudewasinvestigated
surementswerenotconsideredwithrespecttotheMie
and compared to MULIS measurements at 2008–2018
radiometric performance. The reflectance of the sea
UTC8July.FromMULISmeasurements,abackscatter
coefficientof231025 m21 sr21andalidarratioof13sr surface from airborne observations with the A2D and
theanalysisoftheMiereceiversignalaredescribedby
at355 nmwereretrieved.Thiscorrespondsquitewellto
Lietal.(2010).
the RMA model values for cirrus clouds with 1.4 3
1025 m21 sr21and14sr.Thevolumedepolarizationratio
d. RayleighradiometricperformanceofA2D
variesbetween0.1and0.3atthecloud altitudederived
in2007
fromMULISmeasurements.Ameandepolarizationra-
tio of 0.25 can be assumed for simulations for altitudes Measured A2D signal electrons from different days
of8–9km. and simulations are compared in Fig. 8, where the
The summated number of 122 000 Mie signal elec- Rayleighbackscattersignalisdetectedupto17-kmal-
tronswithacirruscloudandabackscattercoefficientof titude. The A2D measurements were selected for one
the RMA is comparable to simulations with MULIS day with lower aerosol content (14 July 2007) and an-
backscatter with 131 000 signal electrons. The corre- otherdaywithhigheraerosolcontent(17July2007),as
sponding A2D zenith-pointing measurements result in measured by MULIS. On 14 (17) July 2007, the A2D
217 000 Mie signal electrons from the cirrus cloud, measurements were averaged over 5 (15) min and
whichisafactorof1.6higherthanfromsimulationswith compared to simulations with the instrumental param-
MULIS backscatter. During a 158 off-zenith-pointing eters from Table 1. Simulations were performed with
measurement (1955–1957 UTC), only 12 250 signal measured MULIS aerosol backscatter coefficients
DECEMBER2009 PAFFRATH ET AL. 2525
FIG.8.(left)SignalelectronsontheRayleighdetectorfromA2Dmeasurementsononeday
with lower aerosol content (14 Jul 2007; black bold line) and one day with higher aerosol
content (17 Jul 2007; dashed bold line) are compared to simulations. Input parameters of
simulationsare measured temperatures by radiosondewith aerosolbackscattercoefficients
measuredbyMULIS(dottedlines).(right)Theratioofthenumberofsimulatedtomeasured
electronsontheRayleighdetectorfrom14(blackdottedline)and17Jul2007(dasheddotted
line)areshown.
averaged over 10 min and temperature profiles from showing that theday-to-dayvariationof thealignment
radiosonde,withthecorrespondingtimesinTable2. wasalsosignificantlyimproved.
TheA2Dsignalelectronsmeasuredon14and17July Simulations using the median aerosol backscatter
2007areafactorof2.5–6lowerthanthecorresponding coefficient and the temperature of the U.S. Standard
simulations(Fig.8,right).Factorsupto6atloweralti- Atmosphere,1976,ascomparedtoA2Dmeasurements,
tudes arise from broadening effects of the laser diver- differ by a factor of about 2 for altitudes above 4 km
genceresultingfromatmosphericturbulenceandsmall (Fig. 9, right). A factor of up to 4 difference for alti-
differencesinalignmentinthetransmitandreceivepath tudesbelow4 kmarisesfrombroadeningeffectsofthe
from day to day. Both strongly affect the telescope laserdivergenceresultingfromatmosphericturbulence
overlap function (Wandinger and Ansmann 2002). Be- and small misalignments in the transmit and receive
cause of insufficient agreement of the simulated and path. The fluctuations in signal are not larger than
observed overlap function, it has not been included in 3.5%, which is the SD of the current measurement at
the simulations up to now. Small differences between 1040–1103 UTC 20 October 2008. Concluding, it was
bothdaysarecausedbydifferencesinalignmentofthe shown that measurements withthe A2D differ to sim-
transmitandreceivepathopticsin2007. ulationsin2007byafactorof2.5andin2008byafactor
of2.0foraltitudesabove4 km.
e. RayleighradiometricperformanceofA2Din2008
The systematic differences of measurements and
In2008,thelaserdivergencewasreducedtobebelow simulationsarisefromuncertaintiesoftheatmospheric
100 mrad (63s, 99.7%) and is therefore within the re- conditions and the instrument parameters. Error con-
ceiver FOV. The EOM refractive index fluid was re- tributions to the measurements are the variation in
filled, which leads to a nominal transmission of 100%. alignmentofthetransmitandreceivepathandthetur-
Verticalprofilesofthenumberofsignalelectronson12 bulentbroadeningofthelaserbeam.Atmospherictur-
and 20 October 2008 are all upon each other (Fig. 9), bulencecancausethelaserbeamtobebroadenedand
TABLE2.TimetableofmeasurementsfromA2D,MULIS,andradiosonde.
Date A2Dmeasurements Inputsimulations
14Jul2007 0413–0418UTC Radiosonde0600UTC;MULIS0328–0338UTC
17Jul2007 0708–0723UTC Radiosonde1200UTC;MULIS1010–1020UTC
12Oct2008 1)0903–1000and2)1138–1146UTC U.S.StandardAtmosphere,1976RMAmedianaerosolmodel
20Oct2008 1040–1103UTC U.S.StandardAtmosphere,1976RMAmedianaerosolmodel