Page 130 - 2023年第54卷第3期
P. 130
abilitytodetectprecipitationextremesovermountainouscountryNepal[J].Atmosphere,2021,12(2):254.
[25] GADELHAAN,COELHOVHR,XAVIERAC,etal.Gridbox - levelevaluationofIMERGoverBrazilatvari
ousspaceandtimescales[J].AtmosphericResearch,2019,218:231 - 244.
[26] HUQ,YANGH,MENGX,etal.Satelliteandgaugerainfallmergingusinggeographicallyweightedregression
[J].ProceedingsoftheInternationalAssociationofHydrologicalSciences,2015,368:132 - 137.
[27] QURESHIS,KOOHPAYMAJ,FIROZJAEIM K,etal.EvaluationofSeasonal,Drought,andWetCondition
EffectsonPerformanceofSatellite - BasedPrecipitationDataoverDifferentClimaticConditionsinIran[J].Remote
Sensing ,2021,14(1):76.
[28] TIANB,CHENH,WANGJ,etal.AccuracyassessmentanderrorcauseanalysisofGPM (V06)inXiangjiang
rivercatchment [J].HydrologyResearch,2021,52(5):1048 - 1065.
[29] AGHAKOUCHAKA,MEHRANA.Extendedcontingencytable:Performancemetricsforsatelliteobservationsand
climatemodelsimulations[J].WaterResourcesResearch,2013,49(10):7144 - 7149.
TheinfluenceofbaselineprecipitationontheaccuracyofGPM dataevaluation:
Acasestudyof“20170610”extremtprecipitationeventinNanjing
1,2
1,2
1,2
1
1
3
ZHANGYe,HUQingfang ,HUANGYong,WANGYintang ,LILingjie ,YUNZhaode
(1.StateKeyLaboratoryofHydrology - WaterResourcesandHydraulicEngineering,
NanjingHydraulicResearchInstitute,Nanjing 210029,China;
2.YangtzeRiverConservationandGreenDevelopmentResearchInstitute,Nanjing 210029,China;
3.AnhuiInstituteofMeteorologicalSciences,Hefei 230031,China)
Abstract:Intheexistingliterature,therearetwomajormethodstoobtainreferenceprecipitationforevaluatingac
curacyofprecipitationretrievalsfromsatellitesuchasGPM :gaugesreferenceprecipitationandinterpolatesrefer
enceprecipitation.However ,existingstudiesarelimitedduetothelackofunderstandingabouttheeffectsofthe
twobenchmarkingmethodsinsatelliteprecipitationestimates (SPEs)evaluationandthestrategyandpriorityon
howtousethem.ThisworktakestheJune10 ,2017extremerainfalleventinNanjingasanexample.Weimple
mentaRandowextractionstudybyrandomlysamplingfromadensegaugenetworktoderivereferenceprecipitation
underdifferentgaugedensityscenarios.TheaccuracycharacteristicsofthreeGPM data(IMERGEarly,IMERG
LateandGSMaPNow)areanalyzedandcomparedusingthetworeferenceprecipitationestimationmethods.The
resultsshowthattheinfluenceofdifferentreferenceprecipitationestimationmethodsdependsonanalysisscale,
GPM datatype,evaluationmetric,andgaugesdensity.Takingtheinterpolatedreferencefromthehighestdensity
gaugesasground“truth”,theevaluationerrorsbasedonthetworeferenceprecipitationsarefurtherquantifiedun
derdifferentgaugesdensities.Acertaingauge - densitythresholdisidentifiedfortheevaluationof24 - hourcumula
tiverainfallandhourlyrainfall ,respectively.Ifraingaugedensityishigherthanthatthreshold,itismorereason
abletousetheinterpolation - basedreferenceestimationmethod ;otherwise,eitherinterpolation -orstation - based
referenceestimationmethodscanbeusedwithoutsignificantdifference.Thisstudydeepenstheunderstandingofer
rorcharacteristicsofGPMprecipitationproduct ,andprovidesaguidanceforreasonablydeterminingthegroundref
erenceforevaluationofSPEs.
Keywords:satellite - retrievedprecipitation;GPM;surfacereferenceprecipitation;accuracyverification;gaugedensity.
(责任编辑:于福亮)
— 3 8 —
0