MODELING AND PREDICTION CONCERNING TIME SERIES OF FLOOD/DROUGHT RUNS USING THE SELF-EXCITING THRESHOLD AUTOREGRESSIVE MODEL

在线阅读 下载PDF 导出详情
摘要 WhenlinearregressivemodelssuchasARorARMAmodelareusedforfittingandpredictingclimatictimeseries,resultsareoftennotsufficientlygoodbecausenonlinearvariationsinthetimeseries.Inthispaper,anonlinearself-excitingthresholdautoregressive(SETAR)modelisappliedtomodelingandpredictingthetimeseriesofflood/droughtrunsinBeijing,whichwerederivedfromthegradedhistoricalflood/droughtrecordsinthelast511years(1470—1980).TheresultsshowthatthemodelingandpredictingwiththeSETARmodelaremuchbetterthanthatoftheARmodel.Thelattercanpredicttheflood/droughtrunswithalengthonlylessthantwoyears,whiletheformalcanpredictmorethanthree-yearlengthruns.ThismaybeduetothefactthattheSETARmodelcanrenewthemodelaccordingtotherun-turningpointsintheprocessofpredic-tion,thoughthetimeseriesisnonstationary.
机构地区 不详
出处 《气象学报:英文版》 1990年4期
出版日期 1990年04月14日(中国期刊网平台首次上网日期,不代表论文的发表时间)
  • 相关文献