Page 89 - 2022年第53卷第8期
P. 89
SpatialdownscalingofMSWEPdatasetbasedonATAK and
itsinfluenceonrainfallmerging
1,2
1
1,2
1,2
YUNZhaode,HUQingfang ,WANGYintang ,LILingjie ,
1
1
WANGLeizhi,CHENJiandong
(1.NanjingHydraulicResearchInstitute,StateKeyLaboratoryofHydrology - WaterResources&
HydraulicEngineeringScience,Nanjing 210029,China;
2.YangtzeInstituteforConservationandDevelopment,Nanjing 210098,China)
Abstract:Globalprecipitationdatasetsprovideanewwaytoobtainlarge - scaleprecipitationspatialdistribution,
butthelowspatialresolutionhasalwaysbeenoneofthemaingapsrestrictingtheirapplicationsatthebasinor
regionalscale.Therefore,spatialdownscalingmethodforglobalprecipitationdatasetsisofsignificantvaluebothin
theoryandpractice.UsingtwostatisticalmethodsincludingAreatoAreaKriging(ATAK)andInverseDistance
Weighted(IDW), Multi - SourceWeighted - EnsemblePrecipitation (MSWEP) gridded rainfallisspatially
downscaledfrom0.1° × 0.1°to0.02° × 0.02°intheHanjiangRiverBasin ,withoutconsideringthegroundrainfall
dataandanyauxiliaryinformation.TheresultsshowthatalthoughthemonthlyrainfallfieldsobtainedbyATAK
downscalingisnotratherdifferentfrom IDW intermsofstatisticalaccuracyindices,however,theyimprovethe
abilitytodescribethelocalvariationofmonthlyrainfallandovercomethesmoothingeffectofIDW toacertain
extent.Furthermore ,throughGeographicallyWeightedRegression(GWR),thedisaggregatedMSWEPdataby
ATAKandIDW andtherawMSWEPdatawithoutspatialdownscalingareusedasthebackgroundfieldstomerge
thegaugesobservedrainfallrespectively.Itisfoundthatthemonthlyfusedprecipitationobtainedusingthethree
differentkindsofbackgroundfieldsaregenerallysimilarinspatialpatterns ,andtheaccuracystatisticsarealso
quiteclose.However ,amongthethreekindsofmergedprecipitationdatasets,therearestillsomedifferencesin
thespatialcontinuityanddetails.Inthecaseofdensegroundrainfallgaugeswithinthestudyarea ,theinfluenceof
backgroundfield differenceson themergingresultsofMSWEP and gaugeobserved precipitation cannotbe
completelyeliminated , and even maybeamplified locally.Therefore, strengthen thecomparison ofspatial
downscalingmethodsforglobalprecipitationdataandchoosingtheappropriatemethodisvital.Theseconclusions
provideimportantreferencesforthespatialdownscalingofglobalprecipitationdatasetsandaccurateestimationof
rainfallfields.
Keywords:spatialdownscaling;rainfallmerging;MSWEP;ATAK;GWR;HanjiangRiverBasin
(责任编辑:耿庆斋)
— 9 7 —
6