利用光学遥感数据_GIS及人工神_省略_网络模型分析区域.pdf

《利用光学遥感数据_GIS及人工神_省略_网络模型分析区域.pdf》由会员上传分享，免费在线阅读，更多相关内容在工程资料-天天文库。

1、第14卷第6期地学前缘(中国地质大学(北京);北京大学)Vol.14No.62007年11月EarthScienceFrontiers(ChinaUniversityofGeosciences,Beijing;PekingUniversity)Nov.2007UTILIZATIONOFOPTICALREMOTESENSINGDATAANDGISTOOLSFORREGIONALLANDSLIDEHAZARDANALYSISUSINGANARTIFICIALNEURALNETWORKMODEL利用光学遥感数据、G

2、IS及人工神经网络模型分析区域滑坡灾害12,＊BiswajeetPradhan,SaroLee1.CilixCorporation,LotL4-I-6,Level4,Enterprise4,TechnologyParkMalaysia,BukitJalilHighway,BukitJalil,57000,KualaLumpur,Malaysia2.GeoscienceInformationCenter,KoreaInstituteofGeoscienceandMineralResources(KIGAM)30

3、,Kajung-Dong,Yusung-Gu,Tae-jon,KoreaBiswajeetPradhan,SaroLee.UtilizationofopticalremotesensingdataandGIStoolsforregionallandslidehazardanalysisusinganartificialneuralnetworkmodel.EarthScienceFrontiers,2007,14(6):143-152Abstract:Theaimofthisstudyistoevaluate

4、landslidehazardanalysisatSelangorarea,MalaysiausingopticalremotesensingdataandaGeographicInformationSystem(GIS).Landslidelocationswereidentifiedinthestudyareafrominterpretationofaerialphotographsandfieldsurveys.Topographical,geologicaldataandsat-elliteimage

5、swerecollected,processedandconstructedintoaspatialdatabaseusingGISandimageprocessing.Thereareabout10landslideoccurrencefactorsthatwereselectedas:topographicslope,topographicaspect,topographiccurvatureanddistancefromdrainage;lithologyanddistancefromlineament

6、;landcoverfromTMsatelliteimages;thevegetationindexvaluefromLandsatsatelliteimages;precipitationdata.Thesefactorswereanalyzedusinganadvancedartificialneuralnetworkmodeltogeneratethelandslidehazardmap.Eachfactorsweightwasdeterminedbytheback-propagationtraini

7、ngmethod.Thenthelandslidehazardindiceswerecalculatedusingthetrainedback-propagationweights,andfinallythelandslidehazardmapwasgeneratedusingGIStools.Landslidelocationswereusedtoverifyresultsofthelandslidehazardmapandtheverificationresultsshowed82.92%accuracy

8、.Theverificationresultsshowedsufficientagreementbetweenthepresump-tivehazardmapandtheexistingdataonlandslideareas.Keywords:landslide;hazard;artificialneuralnetwork;GIS;MalaysiaCLCnumber:P642.22Document