基于Triple-Collocation的地面观测与卫星遥感数据融合的雪深反演

许剑辉; 舒红

doi:10.13203/j.whugis20130727

基于Triple-Collocation的地面观测与卫星遥感数据融合的雪深反演

许剑辉¹,
舒红^1,

¹ 武汉大学测绘遥感信息工程国家重点实验室,湖北武汉,430079

基金项目: 国家自然科学基金资助项目(41171313,41331175)；地理空间信息工程国家测绘地理信息局重点实验室开放研究基金资助项目(201329)；湖北省自然科学基金资助项目(2014CFB725)

详细信息

作者简介:
许剑辉,博士生,主要从事时空统计与数据同化研究。

通讯作者:
舒红

中图分类号: P237.9
计量
- 文章访问数: 01459
- HTML全文浏览量: 069
- PDF下载量: 0787
出版历程
- 收稿日期: 2013-12-01
- 修回日期: 2015-04-04
- 发布日期: 2015-04-04

摘要

摘要: 目的针对单一被动微波遥感反演雪深的精度和空间分辨率不足的问题,提出一种星地多源数据融合的雪深反演方法。以北疆每日站点观测雪深、AMSR E遥感反演雪深和SSM/I遥感反演雪深数据为研究对象,首先利用地统计方法结合地面站点观测数据估计北疆区域的雪深,然后采用Triple Collocation方法分别估计三个雪深数据的误差方差,最后结合最小二乘原理实现星地雪深观测数据的融合。对融合雪深与地面雪深观测数据进行验证分析,结果显示,与AMSR E和SSM/I遥感反演雪深相比,融合的雪深与地面观测雪深的相关性更高;融合的雪深的精度有一定程度的提高。实验结果证明了多源数据融合方法的有效性。
- 雪深 /
- AMSR E /
- SSM/I /
- TripleCollocation /
- 最小二乘原理

HTML全文

参考文献(1)

[1]

ZhaoLiang,Zhu Yuxiang,Yang Hong,etal.ADynamicApproachtoRetrievingSnowDepthBasedontheTechnologyofIntegratingSatelliteRemoteSensingandinsitu Data[J].犃犮狋犪犕犲狋犲狅狉狅犾狅犵犻犮犪犛犻狀犻犮犪,2013,4(71):769 782(赵亮,朱玉祥,杨弘,等.一种基于卫星遥感与地面测站数据融合技术的雪深动态反演方法[J].气象学报,2013,4(71):769 782)[2] LiuHai,ChenXiaoling,SongZhen,etal.StudyofCharacteristicParametricSelectionand ModelCon structionforSnow Depth Retrievalfrom MODISImage[J].犌犲狅犿犪狋犻犮狊犪狀犱犐狀犳狅狉犿犪狋犻狅狀犛犮犻犲狀犮犲狅犳犠狌犺犪狀犝狀犻狏犲狉狊犻狋狔,2011,36(1):113 116(刘海,陈晓玲,宋珍,等.MODIS影像雪深遥感反演特征参数选择与模型研究[J].武汉大学学报·信息科学版,2011,36(1):113 116)[3] LiuYan,RuanHuihua,ZhangPu,etal.KrigingInterpolationofSnowDepthattheNorthofTians hanMountainsAssistedby MODISData[J].犌犲狅犿犪狋犻犮狊犪狀犱犐狀犳狅狉犿犪狋犻狅狀犛犮犻犲狀犮犲狅犳犠狌犺犪狀犝狀犻狏犲狉狊犻狋狔,2012,37(4):403 405(刘艳,阮惠华,张璞,等.利用MODIS数据研究天山北麓Kriging雪深插值[J].武汉大学学报·信息科学版,2012,37(4):403 405)[4] FosterJL,HallD K,EylanderJ,etal.BlendedVisible,PassiveMicrowaveandScatterometerGlob alSnowProducts[C].The64thEasternSnowCon ference,Newfoundland,Canada,2007[5] HallDK,FosterJL,KumarS,etal.ImprovingtheAccuracyoftheAFWA NASA(ANSA)Blen ded Snow cover Product over the Lower GreatLakesRegion[J].犎狔犱狉狅犾狅犵狔犪狀犱犈犪狉狋犺犛狔狊狋犲犿犛犮犻犲狀犮犲狊犇犻狊犮狌狊狊犻狅狀狊,2012,9(1):1141 1161[6] FosterJL,HallDK,EylanderJB,etal.ABlen ded GlobalSnow Product Using Visible,PassiveMicrowaveandScatterometerSatelliteData[J].犐狀狋犲狉狀犪狋犻狅狀犪犾犑狅狌狉狀犪犾狅犳犚犲犿狅狋犲犛犲狀狊犻狀犵,2011,32(5):1371 1395[7] LiuY,Peters LidardCD,KumarS,etal.Assimi latingSatellite basedSnow DepthandSnow CoverProductsforImprovingSnowPredictionsinAlaska[J].犃犱狏犪狀犮犲狊犻狀犠犪狋犲狉犚犲狊狅狌狉犮犲狊,2013,54:208 227[8] StoffelenA.TowardtheTrueNear surface WindSpeed:ErrorModelingandCalibrationUsingTri ple Collocation[J].犑狅狌狉狀犪犾狅犳犌犲狅狆犺狔狊犻犮犪犾犚犲狊犲犪狉犮犺:犗犮犲犪狀狊(1978 2012),1998,103(C4):7755 7766[9] ScipalK,HolmesT,DeJeuR,etal.APossibleSolutionfortheProblem ofEstimatingtheErrorStructureofGlobalSoilMoistureDataSets[J].犌犲狅狆犺狔狊犻犮犪犾犚犲狊犲犪狉犮犺犔犲狋狋犲狉狊,2008,35(24):L24403[10]LerouxDJ,KerrYH,RichaumeP,etal.Spatial DistributionandPossibleSourcesofSMOSErrorsattheGlobalScale[J].犚犲犿狅狋犲犛犲狀狊犻狀犵狅犳犈狀狏犻狉狅狀犿犲狀狋,2013,133:240 250[11]FangH,WeiS,JiangC,etal.TheoreticalUncer tainty Analysisof Global MODIS,CYCLOPES,andGLOBCARBONLAIProductsUsingaTriple Collocation Method[J].犚犲犿狅狋犲犛犲狀狊犻狀犵狅犳犈狀狏犻狉狅狀犿犲狀狋,2012,124:610 621[12]YilmazMT,Crow W T,AndersonMC,etal.AnObjective Methodology for Merging Satellite andModel basedSoilMoistureProducts[J].犠犪狋犲狉犚犲狊狅狌狉犮犲狊犚犲狊犲犪狉犮犺,2012,48(11):W11502,doi:10.1029/2011WR011682[13]Che Tao.Studyon Passive Microwave RemoteSensingofSnowandSnowDataAssimilationMeth od[D].Lanzhou:ColdandAridRegionsEnviron mentaland Engineering ResearchInstitute,CAS,2006(车涛.积雪被动微波遥感反演与积雪数据同化方法研究[D].兰州:中国科学院寒区旱区环境与工程研究所,2006)犜犺犲犜狉犻狆犾犲犆狅犾犾狅犮犪狋犻狅狀犫犪狊犲犱犉狌狊犻狅狀狅犳犐狀狊犻狋狌犪狀犱犛犪狋犲犾犾犻狋犲犚犲犿狅狋犲犛犲狀狊犻狀犵犇犪狋犪犳狅狉犛狀狅狑犇犲狆狋犺犚犲狋狉犻犲狏犪犾犡犝犑犻犪狀犺狌犻1 犛犎犝犎狅狀犵11 StateKeyLaboratoryofInformationEngineeringinSurveyingMappingandRemoteSensing,WuhanUniversity,Wuhan430079,China犃犫狊狋狉犪犮狋:Becauseoftheinsufficientaccuracyandspatialresolutionofsnowdepthproductsretrievedbypassivemicrowaveremotesensing,anew multi sourcesdatafusionapproachisdevelopedforre trievingsnowdepth.Thedatafromdifferentsourcescontainsvisible,passivemicrowavesatelliteandin situdata.Thedailyin situ,AMSR EandSSM/Iretrievedsnowdepthproductsareusedinthisstudy.First,combiningin situsnow depth,thesnow depthof Northern Xinjiangisestimatedthroughgeostatisticalanalysis.Thentheerrorvariancesofeachproductarecalculatedusingatriplecollocation(TC)method.Finally,thenewsnowdepthproductsareobtainedby mergingin situ,AMSR EandSSM/IsnowdepthdatainaleastsquarescriterionwheretheoptimalweightsofeachproductaredeterminedwiththeTC basederrorvariances.Themergedsnowdepthisvalidateda gainstin situsnowdepthandexhibitsahighercorrelationwithin situobservationsthanthatwitho riginalAMSR EandSSM/Isnowdepth.Theresultswithhigheraccuracydemonstratetheeffective nessofourapproach.犓犲狔狑狅狉犱狊:SnowDepth;AMSR E;SSM/I;TripleCollocation;LeastSquareMethod犉犻狉狊狋犪狌狋犺狅狉:XUJianhui,PhDcandidate,specializesinspatio temporaldataanalysisanddataassimilation.E mail:xujianhui306@163.com犆狅狉狉犲狊狆狅狀犱犻狀犵犪狌狋犺狅狉:SHU Hong,PhD,professor.E mail:shu_hong@whu.edu.cn犉狅狌狀犱犪狋犻狅狀狊狌狆狆狅狉狋:TheNationalNaturalScienceFoundationofChina,No.41171313;theKeyProjectofNationalNaturalScienceFoundationofChina,No.41331175;theOpenResearchFundoftheKeyLaboratoryofGeo informaticsofNationalAdministrationofSurveying,MappingandGeoinformation,No.201329;theHubeiProvincialNaturalScienceFoundationofChina,No.2014CFB725/ZRY2014000982.