Observed monthly precipitation trends in China 1951-2002

Theor.Appl.Climatol.77,39–45(2004)DOI

10.1007/s00704-003-0018-3

40M.Gemmeretal.

1990sandthewatershortageintheYellowrivercatchmentbothofwhichhavebeenaggravatedbyhumanactivities(Kingetal.,2001).Therefore,thedetailedanalysisofprecipitationvariationsareimportantfortheassessmentofclimateinducedrisksandcountermeasures.2.Dataandmethodology

Thedatasetof160NationalMeteorologicalObservatory(NMO)stationswithlong-termmonthlyprecipitationdatainChinahavebeenanalysedinthisstudy.TheyhavebeenprovidedbytheNationalClimaticCentre(NCCC)oftheChinaMeteorologicalAdministration(CMA)andcontainvaluesfromJanuary1951toDecem-ber2002.ThelocationofthestationscanbeseeninFig.1.Thedensityofstationsislowerinthesparselypopulatedhighmountainousanddesertareaofwestandnorth-westChina.

ThehomogeneityoftheprecipitationrecordswasanalysedbycalculatingthevonNeumannratio(N),thecumulativedeviations(Q=nÀ0.5

andR=nÀ0.5),andtheBayesianprocedures(UandA)(Buishand,1982;Maniak,1997).Thedatasetsofallstationswhichhavebeenusedinthepresentstudyarehomogeneouswithasig-ni canceabovethe95%con dencelevel.

MonthlyprecipitationhasbeenanalysedbyapplyingtheMann-Kendalltrendtestforallthe160stations.Con dencelevelsof90%,95%,and99%weretakenasthresholdstoclassifythesig-ni canceofpositiveandnegativeprecipitationtrends.Trendsatasigni cancebelowthe90%con- dencelevelwerenotconsidered.

Theobservedtrendswerespatiallyinterpo-latedbyapplyingtheInverseDistanceWeighted(IDW)interpolationmethod.IDWcreatesaras-tersurface.Therastercellvaluesarecalculatedbyaveragingthevaluesofstationdatainthevicinityofeachcell.Stationdatainthispaperrefertothecon dencelevelsoftrendsatthe160stations.IDWimpliesthateachstationhasalocalin uencethatdecreaseswithdistance(DeBy,2001).Theinterpolatedrastersurfaceisbasedonaweightedaverageofthe

station

Fig.1.LocationofthestationsinChina

ObservedmonthlyprecipitationtrendsinChina1951–200241

values.Thevalueofeachcellisin uencedmostlybynearbypointsandlessbymoredistantpoints.TheIDWmethodrequiresthechoiceofapowerparameterandasearchradius.Thepowerparametercontrolsthesigni canceofcalculatedstationvaluesontheinterpolatedvalues.Byde ningahighpower,moreemphasisisplacedonthenearestpoints,andtheresultingsurfacewillhavemoredetail.ThepowerparameterintheIDWinterpolationwassettosix.Thechoiceofarelativelyhighpowerensuresahighdegreeoflocalin uenceandgivestheoutputsurfaceincreaseddetail.Thesearchradiushasbeenspe-ci edbychoosingfourneighbouringstations,i.e.thespatialinterpolationhasbeencarriedoutwithina xedradiusoffourneighbouringsta-

tions(seeBill,1999).Itisnotmeaningfultoanal-ysetheregionalstructureofprecipitationtrendsforthewholeofChinabyapplyingandcomparingdifferentinterpolationmethods.Applyingotherinterpolationmethods(e.g.kriging)resultsinsimilaroutputmaps.Thisisduetothelimitednumberofavailablestations,thescaleofthepro-jectionandthemagnitudeoftheresearcharea.Interpretationoftheresultsofdifferentinter-polationmethodswouldonlyreallybeusefulforsmallerareasorwithahigherstationdensity.3.Observedprecipitationtrends

Positiveandnegativeprecipitationtrends,atallcon dencelevels,existforeachmonth(Fig.

2).

Fig.2.Observedprecipitationtrendsatthe90,95,and99per-centcon dencelevel

42M.Gemmeretal.

January,July,September,andNovemberarethemonthswhichshowthehighestnumberoftrendsinbothpositiveandnegativedirections.ThesmallestnumberoftrendsoccursinMarch,October,andDecember.Sch

本文来源：https://www.bwwdw.com/article/xdkj.html