Mann-kendall和Sen趋势分析算法介绍

AnalysisofchangesinmeteorologicalvariablesusingMann-KendallandSen'sslopeestimatorstatisticaltestsinSerbiaMilanGocic⁎,SlavisaTrajkovicFacultyofCivilEngineeringandArchitecture,UniversityofNis,A.Medvedeva14,18000Nis,SerbiaabstractarticleinfoArticlehistory:Received16February2012Accepted21October2012Availableonline26October2012Keywords:meteorologicalvariablestrendanalysisMann-KendalltestSen'sslopeestimatorTheannualandseasonaltrendsofsevenmeteorologicalvariableswereanalyzedfortwelveweatherstationsinSerbiaduring1980–2010.Thenon-parametricMann-KendallandSen'smethodswereusedtodeterminewhethertherewasapositiveornegativetrendinweatherdatawiththeirstatisticalsignificance.Theoccur-renceofabruptchangeswasdetectedusingcumulativesumchartsandbootstrapping.Inthepresentstudy,theincreasingtrendswereindicatedinbothannualandseasonalminimumandmaximumairtemperatures'series.Therelativehumiditydecreasedsignificantlyinsummerandautumn,whilethevaporpressurehadasignificantincreasingtrendinspring,summerandautumn.Besides,nosignificanttrendsweredetectedinsummerandwinterprecipitationseries.Ingeneral,theresultsofusingtheMann-KendallandSen'stestsdemonstratedthegoodagreementofperformanceindetectionofthetrendformeteorologicalvariables.©2012ElsevierB.V.Allrightsreserved.1.IntroductionClimatechangeisreferredtoaslargevariationsinclimateaver-ageswhichexistfordecadesorevenlongerperiods.Althoughclimatechangeoccurredonaglobalscale,itsimpactsoftenvaryfromregiontoregion(TrajkovicandKolakovic,2009).Therefore,theanalysisofchangesinmeteorologicalvariablesrepresentsanimportanttaskinclimatechangedetection.Inrecentyears,variousstudieshavebeendonefordetectingpos-sibleclimatetrendsandchangesacrosstheworld.However,mostofthesestudieshavefocusedonchangesintemperature(maximum,minimumormean)andprecipitationonly.YunlingandYiping(2005)analyzedclimatechangetrendsandcharacteristicsduring1960–2000at19stationsalongtheLancangRiver(China)accordingtothearchivaldataofmonthlyairtemperatureandprecipitationseries.Theyfoundincreasesintemperatureandde-creasesinprecipitation.Karaburunetal.(2011)analyzedtheevolutionofannual,seasonalandmonthlymean,minimumandmaximumtem-peraturesinIstanbulfrom1975to2006byusingMann-KendalltestandSen'smethod.TemperaturechangesintheKingdomofSaudiArabiafor29yearsusingdataobtainedfrom29meteorologicalstationswereinvestigat-edbyElNesretal.(2010).Theyfoundthatawarmingtrendwasinthemaximum,minimumandaveragetemperaturesthroughouttheyearexceptinthewintermonthsofNovembertoJanuarywherenon-significantcoolingtrendswereobserved.InSwitzerland,Ceppietal.(2012)performedthetrendanalysisoftemperaturesfrom1959to2008,usinga2×2kmgriddeddataset.Theyfoundthattheseasonaltrendsareallpositiveandmostlysignif-icantwithanannualaveragewarmingrateof0.35°C/decade.TrendsofstreamflowintheTarimRiverBasinduringthepast50yearsareanalyzedbyTaoetal.(2011).Theresultsindicatedthattemperature,precipitation,relativehumidityandactualvaporpressureshowpositivetrends,whilewindspeed,sunshinedura-tionandpotentialevapotranspirationshownegativetrends.ThesimilarresultswereobtainedbyXuetal.(2010).TheyconcludedthatbothmeanannualairtemperatureandprecipitationhadanincreasingtrendfortheTarimRiverBasinduringthepastfivedecades.Singhetal.(2008)examinedthetrendandvariabilityofseasonalandannualrainfallandrelativehumidityonthebasinscaleforthenorthwestandcentralpartsofIndiausingMann-Kendallstatisticaltest.Inthisresearch,theyconcludedthatthemajorityofriverbasinsinIndiahadanincreasingtrendintherelativehumiditybothonsea-sonalandannualscales.ThesimilarresultswereachievedbyVincentetal.(2008)whoanalyzedsurfacetemperatureandrelativehumiditytrendsinCanadafortheperiod1953–2005.Jiangetal.(2010)analyzedwindspeedchangesbasedontwoob-servationaldatasetsinChinafrom1956to2004andconcludedthattheyallshowdecreasingtrendsoverbroadareasofChina.Someresearchersanalyzedthewatervaporandconcludedthattherewasanincreaseinatmosphericwatervapor(Philiponaetal.,2005;Trenberthetal.,2005;Willettetal.,2007).InIrannumerousstudieshaveexaminedthechangesinmeteoro-logicalvariables(ModarresandSilva,2007;TabariandHosseinzadehTalaee,2011a,2011b;Tabarietal.,2011a).Ingeneral,theirresultsGlobalandPlanetaryChange100(2013)172–182⁎Correspondingauthor.Tel.:[phone];fax:[phone].E-mailaddress:[e-mail](M.Gocic).0921-8181/$–seefrontmatter©2012ElsevierB.V.Allrightsreserved.:ficantchangeinmeteorologicalvariablesatmanysta-tionsinIran.Thepurposeofthispaperistoanalyzethevariabilityofsevenme-teorologicalvariablesat12stationsinSerbiaduring1980–2010.Be-sides,theobjectivesofthisstudyare:(1)toanalyzeanddiscussthetrendcharacteristicsofmeteorologicalvariablesindetail;(2)toquantifythesignificanceofchangesbyusingtheMann-KendalltestandtheSen'sslopeestimatorafterremovingtheinfluenceofsignifi-cantlag-1serialcorrelationfromthetimeseries;and(3)todetecttheoccurrenceofabruptchangesbyusingcumulativesumchartsandbo