简介：Basedonthedailymeantemperatureand24-haccumulatedtotalprecipitationovercentralandsouthernChina,thefeaturesandthepossiblecausesoftheextremeweathereventswithlowtemperatureandicingconditions,whichoccurredinthesouthernpartofChinaduringearly2008,areinvestigatedinthisstudy.Inaddition,multimodelconsensusforecastingexperimentsareconductedbyusingtheensembleforecastsofECMWF,JMA,NCEPandCMAtakenfromtheTIGGEarchives.ResultsshowthatmorethanathirdofthestationsinthesouthernpartofChinawerecoveredbytheextremelyabundantprecipitationwitha50-areturnperiod,andextremelylowtemperaturewitha50-areturnperiodoccurredintheGuizhouandwesternHunanprovinceaswell.Forthe24-to216-hsurfacetemperatureforecasts,thebias-removedmultimodelensemblemeanwithrunningtrainingperiod(R-BREM)hasthehighestforecastskillofallindividualmodelsandmultimodelconsensustechniques.TakingtheRMSEsoftheECMWF96-hforecastsasthecriterion,theforecasttimeofthesurfacetemperaturemaybeprolongedto192hoverthesoutheasterncoastofChinabyusingtheR-BREMtechnique.ForthesprinkleforecastsovercentralandsouthernChina,theR-BREMtechniquehasthebestperformanceintermsofthreatscores(TS)forthe24-to192-hforecastsexceptforthe72-hforecastsamongallindividualmodelsandmultimodelconsensustechniques.Forthemoderaterain,theforecastskilloftheR-BREMtechniqueissuperiortothoseofindividualmodelsandmultimodelensemblemean.