资源描述:
《改善水文气象预报的统计后处理》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、JournalofWaterResourcesResearch水资源研究,2012,1,161-168doi:10.4236/jwrr.2012.14023PublishedOnlineAugust2012(http://www.hanspub.org/journal/jwrr.html)StatisticalPost-ProcessingtoImprove*HydrometeorologicalForecasts#QingyunDuan,AizhongYeCollegeofGlobalChangeandEarthSystemSc
2、ience,BeijingNormalUniversity,Beijing#Email:qyduan@bnu.edu.cnReceived:Apr.11th,2012;revised:Apr.23rd,2012;accepted:May2nd,2012Abstract:Hydrologicforecastsbasedondirectoutputsfromahydrologicmodelcontainsignificantuncer-taintyfromvarioussources,includingmodelinputs,init
3、ial/boundaryconditionsandmodelstructure/modelparameters.Theuncertaintyleadstovariousbiasesinthehydrologicforecasts.Beforeissuingfinalhydro-logicforecaststotheforecastusers,itisnecessarytoremovethesebiases.Astatisticalpost-processorisaneffectivetoolthancanbeusedtoremov
4、evariousbiasesfromthehydrometeorologicalforecasts.Inthispaper,webrieflydescribetwopracticalpost-processingmethods:1)Theensemblepre-processorforpost-process-ingquantitativeprecipitationandtemperatureforecasts;and2)Thegeneralizedlinearmodelsforpost-proc-essingstreamflow
5、forecasts.WedemonstratedtheeffectivenessofthesetwomethodsinChina’sHuaiRiverbasinandtheFrenchBroadRiverbasinintheUS.Resultsclearlyshowthatpost-processingcansignificantlyimprovetherawhydrometeorologicalforecasts.Aninterestingobservationisthatpost-processingcanachi-eveth
6、esamedegreeofimprovementinstreamflowsimulationasmodelcalibration.Thissuggeststhat,forbasinswherecalibrationcannotbedoneproperlyduetodataissues(i.e.,streamflowregulations),wecanusepost-processingtocompensateforthelackofmodelcalibration.Keywords:EnsembleHydrologicalFore
7、cast;StatisticalPost-ProcessingMethods;EnsemblePre-Processor;GeneralizedLinearRegressionModel(GLM)*改善水文气象预报的统计后处理#段青云,叶爱中北京师范大学全球变化与地球系统科学研究院,北京#Email:qyduan@bnu.edu.cn收稿日期:2012年4月11日;修回日期:2012年4月23日;录用日期:2012年5月2日摘要:水文预报存在由模型输入、初始与边界条件、模型参数与结构等因素带来的不确定性,这些不确定性的存在给基于模
8、型输出结果而得出的预报带来预报均值和区间的偏差,在发布最终水文预报前我们必须要消掉这些偏差,统计水文气象后处理器是对水文气象模型直接输出结果进行后处理、从而达到消掉预报偏差的一个有效方法。本文简单介绍两个实用的统计后处理方法:1)对降水或气温集合预