An Optical Disdrometer for Measuring Size and Velocity of Hydrome

130

JOURNALOFATMOSPHERICANDOCEANICTECHNOLOGYV17

OLUME

AnOpticalDisdrometerforMeasuringSizeandVelocityofHydrometeors

¨

FFLERANGARTINO

-MML

Institutfu¨rMeteorologieundKlimaforschung,ForschungszentrumKarlsruhe,Universita¨tKarlsruhe,Karlsruhe,Germany

¨

RGOSSU

JJ

OsrvatorioTicine,SwissMeteorologicalInstitute,Locarno,Switzerland

(Manuscriptreceived15September1998,infinalform6May1999)

ABSTRACT

Thecharacteristicsofaprototypeopticaldisdrometerareprented.Particlesaredetectableinthediameter

rangefrom0.3to30mmhavingvelocitiesofupto20ms.Advantagesofthenewsystemare(i)itiseasy

Ϫ1

tohandle,robust,andlowcost,allowingaclusterofinstrumentstoinvestigatethespatialandtemporalfine-

scalestructureofprecipitation;(ii)itprovidesreliabledetectionoftherangeofsmalldrops;and(iii)itallows

thepossibilityofsnowmeasurements.ResultsofrainmeasurementsarecomparedwithdatafromaJoss–

WaldvogeldisdrometerandaHellmannraingauge.Furthermore,somesnowmeasurementsareprentedand

comparedwithresultsofarearchspectrometer.Theoverallagreementisgood.Therepeatabilityofparticle

sizeestimationwascheckedinthediameterrangebetween1.4and8.0mmandyieldedastandarddeviation

oflessthan5%.Fordropvelocitiesthestandarddeviationvariesbetween25%(0.3-mmdrops)and10%(5-mm

drops).Theopticaldisdrometercanalsorveasaprentweathernsor,detectinganddifferentiatingamong

rain,snow,drizzle,graupel,hail,andtheabnceofprecipitation.

1.Introductionniques.Ontheonehand,therearesingleparticlecoun-

Foryearsscientistsfrommanyfieldsofrearchhave

beeninterestedinmeasuringthesizeandvelocityof

particles.Alargenumberofdropsizinginstrumentsare

describedinliterature.Theycanbedividedintoveral

groups,dependingonthephysicalprincipleud.

Impacttechniquesarethebasisofthefirstgroupof

instruments.Earlyworkwithafiltermethodwasdone

byDiem(1956).Adisadvantageofthismethodwasthe

largeeffortneededtoevaluatethemeasurements.An-

otherinstrumentinthisgroupisthewell-knownJoss–

Waldvogeldisdrometer(JossandWaldvogel1967).It

iswidelyudasareferenceinstrumentforraininves-

tigations.

Acondgroupisbadonimagingtechniques.

Someexamplesofthearetheopticalarrayprobe

(Knollenberg1970),thethree-dimensionalholography

(BorrmannandJaenicke1993),thepluviospectrometer

(Franketal.1994)badonavideocamera,thevideo

distrometer(Scho¨nhuberetal.1994)withtwoline-scan

cameras,andtheparticlespectrometer(Barthazyetal.

1998),tomentionjustafew.

Athirdgroupusalargevarietyofscatteringtech-

ters,suchastheforwardscatteringprobe(Knollenberg

1981),pha-Dopplerinstruments(Bachalo1980;Dom-

nicketal.1993),orextinctionprobes(Hauretal.

1984;Grossklausetal.1998).Ontheotherhand,there

areinstrumentsthatprobesmallerorlargerparticlecol-

lectivesbymakinguofFraunhoferdiffraction(Gerber

1993;LawsonandCormack1995;Lo¨ffler-Mangetal.

1996;Lo¨ffler-Mang1998)orbymeasuringtheback-

scatterofradarwaves(Sheppard1990;Rogersetal.

1993;Lo¨ffler-Mangetal.1999).

Inthispaperthecharacteristicsofaprototypeoptical

disdrometer,badonsingleparticleextinction,arepre-

nted.Particlesaredetectableinthediameterrange

between0.3and30mm,havingvelocitiesofupto20

ms.Attributesofthesystemweprentareasfol-

Ϫ1

lows.

1)Itiseasytohandle,robust,andlowcost;therefore

itispossibletoinstallnetworksofdisdrometers,thus

making‘‘point’’measurementsmorereprentative

andinvestigatingsmall-scalevariability.

2)Smalldropsarereliablydetected.Thisisofinterest

forinvestigatingscavengingandchemicaleffects.

Furthermore,bychoosingdifferentversionsofthe

opticalsystemthemeasuringrangeofthensormay

bemodifiedtoalsoestimatethesizeofdrizzledrops

withdiametersdownto0.1mm.

3)Estimatesofthesizeandvelocityofsnowflakescan

Correspondingauthoraddress:MartinLo¨ffler-Mang,PMTechAG,

AmStorrenacker1a,D-76139,Karlsruhe,Germany.

E-mail:loeffler-mang@

᭧2000AmericanMeteorologicalSociety

F2000FFLER-MANGANDJOSSLO

EBRUARY

¨

131

T1.Specificationsoftheopticaldisdrometer.

ABLE

Sensorhead

Sizedϭ225mm,hϭ200mm

Weight5kg

Lardiodewavelength780nm

power3mW

Lightsheetsize30mmϫ1mmϫ160mm

Measuringarea48cm

2

Measuringrangediameter0.3–30mm(0.1–10mm)

velocity0.1–20ms

Ϫ1

Electronics

Size250mmϫ260mmϫ150mm

Weight5kg

Powersupply10–40VDCor100–230VAC

Powerconsumption10W

A/D–Converterresolution12bit

samplerate50ϫ10s

3

Ϫ1

Memorycapacity(internal)1monthofrain

Access,interfacePCviaRS232

beobtained.Thisinformationisufulfor‘‘prent

weathernsors’’andforinterpretingresultsfrom

weatherradarsystemsinwintertime,especiallyin

alpineregions,wherehydrometeorsintheradarvol-

umeusuallyconsistofsnow.

Section2describestheinstrument:generalattributes,

theopticalpartofthensor,thensorhousing,the

dataanalysis,andtheaccuracyachieved.Inction3

resultsfromfieldoperationsareprentedandcompared

withdatafromotherinstruments.Finally,ideasforthe

future,instrumentaldevelopment,andplannedappli-

cationsareoutlinedinction4.

2.Measuringsystemshowsthesignalsoftwoparticlesofdifferentsize.The

a.Attributesoftheinstrument

Thedisdrometerconsistsofanopticalnsorwithin

ahousingandsomeappropriateelectronicsincludingstartandtheendofasignalisimplementedinthesoft-

solidstatememory,whichallowsatleastonemonth’sware.

recordingofraindata.AttributesoftheinstrumentareGeometricalconsiderationsshowthattheeffective

summarizedinTable1.Thefollowingctionsexplainwidthofthelightsheetdependsontheparticlesize.To

themeasuringsysteminmoredetail.becompletelyinthelightsheet,largerparticleshavea

b.Opticalnsor

Thebasisoftheinstrumentisacommerciallyavail-

ablensor,producingahorizontalsheetoflight(30

mmwideand1mmhigh,160mmlong).Thelightsheet

isproducedbya780-nmlardiodewithapowerofTwodifferentprotectionshavebeentested.Atfirst,

3mW.Inthereceiverthelightsheetisfocudontoaahousing(Fig.2a)ofashapesimilartoaHellmann

singlephotodiode.Thetransmitterandreceiverareraingaugewasanalyzedfortherainmeasurements.

mountedinahousingforprotection(ection2c).InThen,forsnowmeasurementsatunnel-likehousingwas

theabnceofdropsthereceiverproducesa5-Vsignalud(Fig.2b).

attheoutputofthensor.ParticlespassingthroughtheTheHellmannhousinghasbeentested;forexample,

lightsheetcauadecreaofthissignalbyextinctiontheeffectsofwindwereinvestigatedingreatdetailby

andthereforeashortreductionofthevoltage.Thevolt-Nespor(1998).Furthermore,theHellmannhousinghas

agedecreadependslinearlyonthefractionofthelightarathersmalloutsidedimension,producingaminimum

sheetblocked.Figure1(upperpart)schematicallyofdisturbanceforrain,thoughonlyatverticalincidence.

F.1.Signalsofparticlesfallingthroughthelightsheet.(a)Small

IG

andlargeparticles,(b)rawsignalfromthensor,and(c)inverted

andamplifiedsignalafterthresholdingformeasuringpurpos.

amplitudeofthesignaldeviationisameasureofparticle

size,thedurationofthesignalallowsanestimateof

particlevelocity.Anappropriateconcepttodetectthe

smallerregioninhorizontaldirection.Therefore,toes-

timateconcentration,theeffectivewidthforeachpar-

ticleistakenintoaccount.

c.Sensorhousing

132

JOURNALOFATMOSPHERICANDOCEANICTECHNOLOGYV17

OLUME

F.2.(a)FrontandtopviewsofopticaldisdrometerinHellmann

IG

raingaugehousing.(b)Sideviewsofopticaldisdrometerintunnel

housing.

Inthefirsthousingthelightsheetisfoldedbytwoexitedthelightsheet.Thedistanceofinfluenceofa

mirrorstokeeptheinstrumentsizesmall.Thetunnelparticleisgivenbythelightsheetthicknessof1mm

housingmakesiteasytoasmblethensorandtoplustheparticlediameter(eFig.1).Theratioofthis

adjustthelightsheet.Finally,atfirstapproximationthedistanceandthesignaldurationyieldstheparticleve-

nsitivemeasuringareaisindependentofwindspeedlocity.

anddirection.Becauofthepossibilityofcoincidencesofparticles

d.Dataanalysis

Theanalysisofthesignal(Fig.1,middlepanel)con-forstrongrainandforrainwithalargenumberofdrops,

sistsofthefollowingsteps:removaloftheDCpart,coincidencesmayincreatoavaluethatisnotneg-

inversion,amplification,andfiltering(eFig.1,lowerligiblebutissosmallastobeeasilycorrected.Prob-

panel).ThenafastA/Dconversionisdone,followedabilitiesofcoincidencewerecalculatedforthreeex-

bythresholdingtodetectthestartofparticlesandtheir

maximumvalue.Theshadowsignaldurationandthe

timebetweentwoparticlesarealsorecorded.Thelast

threequantitiesarestoredforeachparticleforfurther

calculationofdistributions(size,velocity,energy,etc.)

aswellasintegralvalues(i.e.,rainrate,radarreflectiv-

ity,liquidwatercontent).Thetimeneededforthisanal-

ysisofoneparticleislessthan1ms.

Theopticalnsorwasoriginallydesignedforex-

tinctionmeasurementswithsignaldurationsofmore

than2ms.Forsnowflakesthisdurationisexceeded;

therefore,thedeviceneedsnocorrection.Thesizeofa

particleiscalculatedfromthemaximumreductionof

thesignal(reflectingtheblockedfractionofthelight

sheet).Theparticleisassumedtobesphericalwitha

diametercorrespondingtothewidthofthemaximum

blockedarea.Forsinglesnowflakes,whichmayhave

rathercomplicatedshapes,thisassumptionislesswell

fulfilledthanforraindrops.Butfortheenmblesmea-

suredduring60sormore,thestochasticalvariationis

reduced.

Calibrationisneededtodeterminethecharacteristics

oftheinternalelectronicsofthensor.Specialcareis

neededforsignalsoflessthan2-msduration.Notethat

typicalsizesandvelocitiesofraindropsleadtosignal

durationsbetween0.4and1.1ms.Becauoflimited

bandwidthoftheoff-the-shelfelectronicsud,there-

ductionofthemeasuredsignalamplitude(ascompared

tothetheoreticalamplitude)hastobecompensated.For

rainandhailmeasurements,thedevicewastherefore

calibratedwithparticlesofknownsize,fallingatter-

minalvelocity.Twenty-vendifferentsizesofglass

spheres,ethanol,andwaterdropsinsizerangesfrom

0.275to4.29mmwereud.Signalsofparticleswere

measuredinthelaboratoryafterafreefallinairofa

10-mheight.Fromeachsizesomehundredparticles

wereanalyzedindividually,thenthemedianvoltagefor

theparticlesizewascalculated.Thusanempiricalre-

lationbetweensizeandvoltagewasobtained.Thisre-

lationisudfortheevaluationofraindropsandhail.

Notethatthecalibrationalsotakesintoaccountthe

oblatenessoflargerraindrops.

Theparticlevelocityiscalculatedfromthesignal

duration.Thesignalstartswiththeparticleenteringthe

lightsheetandendswhentheparticlehascompletely

inthelightsheet,acorrectionisappliedtoestimatethe

realnumberconcentration(RaaschandUmhauer1984).

Fornormalrain,coincidencesarenotaproblem.But

F2000FFLER-MANGANDJOSSLO

EBRUARY

¨

133

tremeexamples,consideringdropswithdiametersbe-

tween0.3and5.5mm.Foranextremelystrongstrat-

iformrain(withsizedistributionparameterN

0

mmmandrainrateRϭ100mmh)itis10%,

ϭ8000

Ϫ3Ϫ1Ϫ1

foramostintensive‘‘drizzle’’(N

0

ϭ30000m

Ϫ3Ϫ1

mm,

Rϭ30mmh)9%,andforanextremeconvective

Ϫ1

shower(Nmm,Rϭ300mmh)itis

0

ϭ1400m

Ϫ3Ϫ1Ϫ1

5%.Similarmaximumvaluesofcoincidenceprobability

arefoundinsnow.

e.Accuracy

Anumberofeffectsinfluencetheaccuracyofdeter-

miningsizeandvelocity.Thehomogeneityofthelight

sheetwascheckedwitha1.0-mm-diameterwire.Fora

meansignalof130mV,avariationoflessthanϮ5mV

wasfoundwhenmovingtheverticalwirethroughthe

wholeareaofthelightsheet.Thenoisuperimpod

onthesignalisontheorderofϮ3mV.TheA/Dcon-

versionhasaresolutionof12bits(4096steps)for10

Vandasamplerateof50ϫ10s.Thesamplerate

3

Ϫ1

limitstheaccuracyoftheestimateofsignalduration;

thatis,itmainlyinfluencesvelocityerrors.

Therepeatabilitywascheckedwithtwodifferentsizes

ofsteelballs(5.52and8.03mmindiameter)andwith

asmallglasssphere(1.45mm).Thesameparticlewas

thrown100timesthroughthelightsheetwithavelocity

ofapproximately1ms

Ϫ1

.Inafirstexperimentthepo-

sitionwasalwaysinthemiddleofthesheetneartothe

transmittingnsor.Then,inacondexperiment,the

particleswererandomlypasdthroughthewholemea-

suringarea.Thefirstexperimentyieldedforallthree

particlesizesastandarddeviationofsizedetermination

below3%,thecondwassmallerthan5%.

Whenusingthe20sizeintervalsoftheJoss–Wald-

vogeldisdrometertoclassifydrops,theoverallerrorin

estimatingthediameterinthewholerangeofthein-

strumentdoesnotexceedϮ100

␮

mplusϮ5%.Forthe

velocitymeasurementsofraindrops,theerrorsarewith-

in25%forthesmallestdrops(0.3mm)and10%for

largestdrops(5mm).

Whencalculatingintegralvaluessuchasrainrateand

radarreflectivity,theinstrumentalerrorsarereducedby

averaging,asverifiedbysimulations.Thestochastic

variationcaudbythequantizationofraindropsmay

exceedtheinstrumentalerrors.Itsmagnitudedepends

onthesamplesize,anattributevalidforalldropsizing

instruments(Smithetal.1993).

Fortheconsiderationsmisadjustmentoftheoptics,

waterordustonnsorwindows,dropsorting,ora

poorlydefinedmeasuringareacaudbyhighhorizontal

windspeedwerenottakenintoaccount.Theerror

sourcescanbereducedtoanegligiblesizebyproper

installationandmaintenanceoftheinstrument.

3.Resultsandcomparisonwithstandard

instruments

a.Rainmeasurements

Firstmeasurementsofraindropspectrawerecon-

ductedfromMaytoJuly1997.Theopticaldisdrometer

wasplaced50cmfromaJoss–Waldvogeldisdrometer

ontheroofofasmallbuildingintheForschungszen-

trumKarlsruhe.Tofacilitatecomparisonoftwoin-

struments,theresultsoftheopticaldisdrometerwere

calculatedintimeintervalsof1minforthe20class

ofdropsizeoftheJoss–Waldvogeldisdrometer.For

illustrationaconvectiveeventon21May1997inthe

earlymorninghourswaschon.Foreachinstrument

the10-minmeannumberdensitywascalculatedasa

functionofdropdiameter(Fig.3).Thereisgoodagree-

mentbetweenthemeasurementsoftheopticalandthe

Joss–Waldvogeldisdrometerinthediameterrange

from0.7to2mm.

Theca,however,showsdifferencesintherangeof

smalldrops.Strongwindsand/oracousticnoimay

inhibitdetectionofthelowerendofthedropsizewith

theJoss–Waldvogeldisdrometer(thisdeviceoriginally

wasdesignedforthedeterminationofZ–Rrelations).

Thispartofthespectrumhaslessinfluenceonrainrate

andradarreflectivity,butmaybeimportantforcloud

physicsaspectsandwashoutofpollutantsbyrain.

Forlargerdropstheconcentrationisratherlow.

Therefore,thestandarddeviationistoolargeforasafe

interpretation(eerrorbars).Notealsothattheinstru-

mentsdidnotmeasuretheidenticaldrops.

For1hthetimeriesoftherainrate,calculated

fromthedropsizedistributions,isprentedforboth

disdrometersinFig.4.Itshowsastrongconvective

eventintheafternoonof5July1997between1600and

1700CETwithadurationofthree-quartersofanhour,

reachingrainratesof20to30mmh

Ϫ1

.Thequantitative

agreementbetweenthetwodisdrometersisquitegood,

consideringtheaccuracyoftheinstruments(around

10%fortherainrate).

Theintercomparisonofdailyrainsums,measuredby

bothdisdrometersandaHellmannraingauge,arecom-

paredinthehistogramofFig.5.For10daysofthe

timespanbetweenMayandJuly1997dataofallthree

measuringdevicesareavailable.Formostofthedays,

thedailysumsagreewithin1mmofrainamount.In8

outof10castheJoss–Waldvogeldisdrometershowed

slightlyhigher(10%)valuesthantheopticaldisdro-

meterandtheHellmannraingauge.Theagreementbe-

tweenthelasttwoinstrumentsisnotsurprisingbecau

theopticaldisdrometerandHellmanngaugealsohave

nearlyidenticalhousingswithsimilarwindeffects.

Whenlookingatdailyrainsums,theobrvedsample

ofdropsislargeandstochasticvariationscaudbythe

quantizationofrainbydropsbecomenegligible.Then

theobrvedvariationsreflectsystematicinstrumental

differences.Thevariationsofrainamountmeasured

withtheopticaldisdrometeraresimilartothoofthe

134

JOURNALOFATMOSPHERICANDOCEANICTECHNOLOGYV17

OLUME

F.3.Meanspectralnumberdensityvsdropdiameter,0611–0620CET21May1997;

IG

comparisonofopticaldisdrometer(solidline)andJoss–Waldvogeldisdrometer(dashedline).

Joss–WaldvogeldisdrometerandtheHellmannrainimately29000measureddropsof5July1997isplotted.

gauge,knowntobeintheorderof10%whenmeasuringEachdropresultsinapointofthediagram.Thedashed

dailyrainsums.Nespor(1998)discusdtheinfluencelineinFig.6alsoshowstheempiricalrelationfrom

ofthewind.Atlasetal.(1973)afterthemeasurementsfromGunn

TheopticaldisdrometermeasuressizeandvelocityandKinzer(1949).Themeasurementsscattersignifi-

ofsingleparticles,allowingavelocity–sizecorrelation.cantlyaroundtheempiricalcurve.Thisscatteriscaud

InFig.6,velocityversusdropdiameterfortheapprox-byturbulenceofairclotothegroundinfluencingthe

F.4.Timeriesofrainrateduring1600–1700CET5Jul1997;comparisonofoptical

IG

disdrometer(solidline)andJoss–Waldvogeldisdrometer(dashedline).

F2000FFLER-MANGANDJOSSLO

EBRUARY

¨

135

F.5.Dailyrainsumsof10daysduringMay–Jul1997;comparisonofopticaldisdrometer,

IG

Joss–Waldvogeldisdrometer,andconventionalHellmannraingauge.

F.6.Dropvelocityvsdropdiameterfor5Jul1997:eachpointreprentsthevelocityand

IG

thesizeofasingledrop(totalnumberofdrops:ഠ29000);comparisonwithempiricalcurveof

Atlas(1973)afterGunnandKinzer(1949):ϭ9.65Ϫ10.3e.

␷

Ϫ0.6D

136

JOURNALOFATMOSPHERICANDOCEANICTECHNOLOGYV17

OLUME

F.7.(a)Spectralnumberdensityvssnowflakediameterfor5-hinterval,0900–1400CET

IG

1Mar1998;comparisonofopticaldisdrometer(solid)andETHparticlespectrometer(dashed).

(b)Meanvelocityofparticlesinsizeclassvssnowflakediameterfortheperiodin(a),comparison

ofopticaldisdrometer(squares),ETHparticlespectrometer(cross),andempiricalcurve(dashed

line,LocatelliandHobbs1974).

fallvelocityand,probablymoreimportant,byinstru-Thecutoffatlowvelocitiesresultsfromthemaximum

mentallimitations,suchastheuncertaintiescaudbyvalueofdetectedsignaldurations,whichwas2.54ms

quantizationandthresholding.Thecutoffatapproxi-(127A/Dconversiontimesteps)forrainmeasurements.

mately0.3-mmdropdiameterresultsfromthetriggerSomehundreddatapointscanbeenintherange

leveludtodetectthedrops(atriggerlevelisnecessarydirectlyabovethecutoffline.Thesignalsweremainly

toparatedropsignalsfromnoi);theline-by-lineproducedbydropssplashingonthehousingand,there-

structureofthedatapointscomesfromthe20-stimefore,havingunrealisticallylowvelocities.Theycould

␮

resolutionoftheA/Dconversion.beremovedbyplausibilityconsiderations.

F2000FFLER-MANGANDJOSSLO

EBRUARY

¨

137

F.8.Ten-minmeansofradarreflectivityvstime,0900–1400CET1Mar1998;comparison

IG

ofopticaldisdrometer(solid),ETHparticlespectrometer(dashed),andETHX-bandDopplerradar

100mabovegroundlevel(dotted).

b.Snowmeasurementsinginstrumentsestimatethevelocitiesclearlyabovethe

DuringFebruaryandMarch1998,anumberofrain

andsnoweventswereinvestigatedinLinthalinthe

SwissAlps.On1March1998apredominantlystrati-

formsnowfallwasrecordedwiththeopticaldisdrometer

(withtunnelhousing),theETHparticlespectrometer

(SwissFederalInstitute,Zu¨rich;Barthazyetal.1998),

andverticallypointingX-bandDopplerradar(Mosi-

mannetal.1993).For5htheechotopwasdetected

bytheradarataheightof2.7kmabovegroundlevel.

Atippingbucketmeasured5mmofprecipitation.

Forbothdisdrometersspectralnumberdensitieswere

calculatedasfunctionsofparticlesize.Distributions

wereintegratedover10min.Inaddition,themeanpar-

ticlevelocityforeachsizeclasswascalculatedassuming

asphericalshape.Theparticlesizewasassumedtobe

themaximumhorizontaldimensionofthesnowflake

en.Duringthe5hofobrvationbothinstruments

showedsimilarbehavior.

Fortheintercomparisonthemeansizedistribution

forthe5-hperiodisshowninFig.7a,themeanparticle

velocitiesinFig.7b.Thenumberdensities(size)ofthe

twoinstrumentsagreewellinmostsizeclass.Dif-

ferencesoccuratthesmalldiameterend,wherethe

opticaldisdrometerwasnotnsitiveinthefirsttwo

classcaudbythetriggerlevelud.Thiserrorhas

beenrecognizedandcorrected.Atthelargediameter

endonlytheopticaldisdrometermeasuredasingle

snowflakeinthesizeclassof12–14mm.Comparedto

anempiricalvelocity–sizerelationϭ0.8D(Fig.

␷

0.16

7b),takenfromLocatelliandHobbs(1974),theesti-

matedvelocitiesshowsignificantdifferences.Bothsiz-withreflectivitiesfromtheverticallypointingX-band

empiricalcurve,especiallyforthelargersnowflakes.

Thiscouldbesimplybecauthesnowflakesinvesti-

gatedinSwitzerlandweredifferentfromthofound

byLocatelliandHobbs.Ontheotherhand,iftheob-

rvedsnowflakesweremorebroadthanhigh(thever-

ticalextentofthesnowflakesisshorterthanthemea-

suredwidth),thiswouldresultinanoverestimateofthe

fallspeedofparticleslargerthanthethicknessofthe

lightsheet.Itsthicknessintheopticaldisdrometeris1

mm,intheETHparticlespectrometeronly0.15mm.

Thisdifferencemaycontributetothedifferencebetween

thetwoinstrumentsandthedashedrelation(forden-

dritesaccordingtoLocatelliandHobbs1974).Itwould

alsoexplainwhythespeedofsmallersnowflakeses-

timatedwiththeopticaldisdrometercomesclortothe

empiricalcurvethanwiththeETHparticlespectrom-

eter.

Finally,radarreflectivityfactorswerederivedfrom

theparticlesizedistributionsassuggestedbySmith

(1984),withtheassumptionthattheradarcrossction

ofanirregularparticleisthesameasthatofasphere

ofthesamemass(thenheudanartificefromMarshall

andGunn1952).Whenusingtheequivalentdiameter

ofmelteddrops,thedielectricfactorhastobemultiplied

by1.18tocompensateforthedensityofwaterwith

respecttoice,resultinginthevalue0.208insteadof

0.176.Forthemassdeterminationofsnowflakesagain

arelationaccordingtoLocatelliandHobbs(1974)was

ud:Mϭ0.059D

2.1

.Averagesover10minofradar

reflectivityversustimeareshowninFig.8,together

138

JOURNALOFATMOSPHERICANDOCEANICTECHNOLOGYV17

OLUME

F.9.Schematicconceptforusingvelocityandsizeinformationtodetectthedifferenttypes

IG

ofhydrometeors.

radarmeasured100mabovegroundlevel.Theresults

ofthedisdrometersagreewellwitheachother,but

sometimesthereisadiscrepancyofupto5dBofthetoobtaininformationontheprecipitationprocess.

valuesmeasuredwithradar.Thedifferencemaybe

caudbysomewetnessofthesnoworevenmoreby

changesoftheparticletype.Firstestimationshave

shownthattheuofdifferentmass–sizerelations(re-

latedtodifferentparticletypes)hasastronginfluence

ontheradarreflectivityderivedfromparticlesizedis-

tributions.Forasinglesizedistribution,variationsof

upto20dBarepossible.Obviouslythereisaneedfor

additionalrearchonhowtoderiveradarreflectivity

factorsforsnowflakes.

4.Summaryandfuturedevelopment

Anopticaldisdrometerwasprentedandresults

fromrainandsnowwerecomparedwithdatafroma

Joss–Waldvogeldisdrometer,aHellmannraingauge,

andaparticlespectrometer.Theoverallagreementis

good.

Theerrorindeterminingsizeinthewholerangedoes

notexceedϮ100mplusϮ5%.Thiswasverifiedwhen

␮

comparingdropconcentrationinthe20sizeclassof

theJoss–Waldvogeldisdrometer.Theerrorlimitsare

expectedtoalsobevalidforsizeclassabovetho

recognizedbytheJoss–Waldvogeldisdrometer.Forthe

velocitymeasurements,theerrorforthesmallestdrops

(0.3mm)yieldsvaluesof25%andgoesdownto10%

forthelargestdrops(5mm).Fordailyrainsumsthe

standarddeviationbetweeninstrumentsisaround10%.anonymousreviewers.

Futureworkmayincludeaspectsofinstrumentalim-

provementaswellastheapplicationoftheinstrument

1)Instrumentalwork:

Rinvestigatethevalueofthevelocityinformation

toidentifyandeliminateedgeeffects,

Restimateeffectsofdropssplashingonthehousing,

and

Rtestamodifiednsortoextendthemeasuring

rangetowardsmalldrizzledrops.

2)Applications:

Rcollectmoreexperienceinmeasuringrainand

snow,

Rmakeuofthecombinedvelocity–sizeinfor-

mationfordetectingthetypeofhydrometeor(e

Fig.9),

Ranalyzestrongprecipitationeventsanddeducea

measureforsoilerosionbylargeraindrops,and

Rinvestigatethereprentativityofpointmeasure-

mentsbycombiningtheresultsofanumberof

identicalinstrumentsoperatedsimultaneously.

Acknowledgments.WewouldliketothankF.Fiedler

formakingthedevelopmentoftheinstrumentpossible.

Furthermore,thankstoE.Barthazyforpreparingthe

snowmeasurementswiththeETHparticlespectrometer

andDopplerradar,andtoJ.Handwerkerforsimulating

instrumentalerrors.Additionally,wewouldliketoex-

pressourappreciationforthehelpfulcommentsofthree

F2000FFLER-MANGANDJOSSLO

EBRUARY

¨

139

REFERENCES

Atlas,D.,R.Srivastava,andR.Sekhon,1973:Dopplerradarchar-

acteristicsofprecipitationatverticalincidence.Rev.Geophys.

SpacePhys.,11,1–35.

Bachalo,W.D.,1980:Methodformeasuringthesizeandvelocity

ofspheresbydual-beamlightscatterinterferometry.Appl.Opt.,

19,363–389.

Barthazy,E.,W.Henrich,andA.Waldvogel,1998:Sizedistribution

ofhydrometeorsthroughthemeltinglayer.Atmos.Res.47–48,

193–208.

Borrmann,S.,andR.Jaenicke,1993:Applicationofmicroholography

forground-badinsitumeasurementsinstratuscloudlayers:

Acastudy.J.Atmos.OceanicTechnol.,10,277–293.

Diem,M.,1956:MessungenderGro¨ßederRegentropfeninnatu¨r-

lichenRegenundbeiku¨nstlicherBeregnung(Measurementsof

raindropsizeinnaturalandartificialrain).Beitr.Naturk.Forsch.

Su¨ddeutschland,15,75–90.

Domnick,J.,C.Tropea,andR.Wagner,1993:Aminiaturizedmi-

conductorfiberopticpha-Doppleranemometer(DFPDA)with

applicationstoliquidsprays.Meas.Sci.Technol.,4,411–415.

Frank,G.,T.Ha¨rtl,andJ.Tschiersch,1994:Thepluviospectrometer:

Classificationoffallinghydrometeorsviadigitalimageproc-

cessing.Atmos.Res.,34,367–378.

Gerber,H.,1993:In-cloudmeasurementsofeffectivedropletradius.

J.AerosolSci.,24,583–584.

Grossklaus,M.,K.Uhlig,andL.Has,1998:Anopticaldisdrometer

foruinhighwindspeeds.J.Atmos.OceanicTechnol.,15,

1051–1059.

Gunn,R.,andG.Kinzer,1949:Theterminalvelocityoffallforwater

dropletsinstagnantair.J.Meteor.,6,243–248.

Haur,D.,P.Amayenc,B.Nutten,andP.Waldteufel,1984:Anew

opticalinstrumentforsimultaneousmeasurementofraindrop

diameterandfallspeeddistributions.J.Atmos.OceanicTech-

nol.,1,256–269.

Joss,J.,andA.Waldvogel,1967:EinSpektrographfu¨rNiederschlags-

tropfenmitautomatischerAuswertung(Aspectrographforrain

dropswithautomaticalanalysis).PureAppl.Geophys.,68,240–

246.

Knollenberg,R.G.,1970:Theopticalarray:Analternativetoscat-

teringorextinctionforairborneparticlesizedetermination.J.

Appl.Meteor.,9,86–103.

,1981:Techniquesforprobingcloudmicrostructure.Clouds:inraindropsizeobrvations.J.Appl.Meteor.,32,1259–1269

TheirFormation,OpticalPropertiesandEffects,P.V.Hobbs

andA.Deepak,Eds.,AcademicPress,15–91.

Lawson,R.P.,andR.H.Cormack,1995:Theoreticaldesignand

preliminarytestsoftwonewparticlespectrometersforcloud

microphysicsrearch.Atmos.Res.,35,315–348.

Locatelli,J.D.,andP.V.Hobbs,1974:Fallspeedsandmassof

solidprecipitationparticles.J.Geophys.Res.,79,2185–2197.

Lo¨ffler-Mang,M.,1998:Alar-opticaldeviceformeasuringcloud

anddrizzledropsizedistributions.Meteor.Z.,7,53–62.

,K.D.Beheng,andH.Gysi,1996:Dropsizedistributionmea-

surementsinrain—Acomparisonoftwosizingmethods.Meteor.

Z.,5,139–144.

,M.Kunz,andW.Schmid,1999:Ontheperformanceofalow-

costK-bandDopplerradarforquantitativerainmeasurements.

J.Atmos.OceanicTechnol.,16,379–387

Marshall,J.S.,andK.L.S.Gunn,1952:Measurementofsnow

parametersbyradar.J.Meteor.,9,322–327.

Mosimann,L.,M.Steiner,andW.Henrich,1993:Predictionofsnow

crystalshapeandrimingbyverticallypointingDopplerradar.

Atmos.Res.,29,85–98.

Nespor,V.,1998:Windinducederrorofprecipitationgauges.Op-

erationaluofradarforprecipitationmeasurementsinSwit-

zerland.FinalRep.NRP31,91–101.[AvailablefromtheSwiss

FederalInstituteofTechnologyZu¨rich,ETHZentrum,CH-8092

Zu¨rich,Switzerland.]

Raasch,J.,andH.Umhauer,1984:Errorsinthedeterminationof

particlesizedistributionscaudbycoincidencesinopticalpar-

ticlecounters.Part.Charact.,1,53–58.

Rogers,R.R.,D.Baumgardner,S.A.Ethier,D.A.Carter,andW.

L.Ecklund,1993:Comparisonofraindropsizedistributions

measuredbyradarwindprofilerandbyairplane.J.Appl.Me-

teor.,32,694–699.

Scho¨nhuber,M.,U.Urban,J.P.VPoiaresBabtista,W.L.Randeu,

andW.Riedler,1994:Measurementsofprecipitationcharacter-

isticsbyanewdistrometer.Proc.Conf.onAtmosphericPhysics

andDynamicsintheAnalysisandPrognosisofPrecipitation

Fields,Rome,Italy.

Sheppard,B.,1990:Measurementofraindropsizedistributionsusing

asmallDopplerradar.J.Atmos.OceanicTechnol.,7,255–268

Smith,P.L.,1984:Equivalentradarreflectivityfactorsforsnowand

iceparticles.J.ClimateAppl.Meteor.,23,1258–1260.

,Z.Liu,andJ.Joss,1993:Astudyofsampling-variabilityeffects