Lewang_population 定量遥感班课件

《Lewang_population 定量遥感班课件》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、利用高空间分辨率遥感数据进行人口估算王乐副教授(终身教授),研究生主任纽约州立大学布法罗分校地理系美国国家地理信息分析中心lewang@buffalo.eduResearchInterests•LandCoverandLandUseClassification•ObjectExtraction（IKONOS,LiDAR)•HyperspectralRemoteSensing•CoastalMangroveForestsMonitoring•InvasiveSpeciesSpreadModeling•UrbanSprawlMo

2、deling•PopulationEstimationfromRemoteSensingU.S.CensusBureauwwwcensusgovwww.census.govBackground•Censusdataarebyagggggregatedunits•ThesmallestU.S.censusunitisblocksProblems.Visualizepppopulation2.Estimatepppopulationsdistributionwithinblockswithinarbitraryboundarie

3、seviousPopulationEstimationMethodsA.DemographicApproaches•ComponentMethodII•AdministrativeRecords•RtiRatioCClorrelatition•HousingUnit(HU)B.RemoteSensingApproaches1.Arealinterpolation2.Statisticalmodeling:•fromurbanareas•fromlanduseareas•fromdwellingunits•fromimagep

4、ixelcharacteristicsMotivationHousingunitsareinsidebuildingsObjectivesDevelopeffectivemethodsforautomaticallyextractinghousingunitsandassociatedstructuralinformationDevelopeffectivemethodsforautomatedclassificationofdetailedurbanlandusetypes;Developanautomatedmetho

5、dforblock‐levelpopulationestimatesfromremotesensing.StudyAreaAustin,Texas:•3rdfastestgggrowinglargecityintheUS.withapopulationof750,000Data•LiDAR–OptechALTM1225system–41+millionpointsover16mi2ofthecity–Firstandlastpulsereturns(elevationw/intensity)•DigitalOrthophot

6、o•GISLayers–LandUseMap–BuildingfootprintAirborneLiDARAustin,TXData(Cont(Contd)’d)LaadndUseCCasslassNuubemberofPeecetrcentPlPopulatitionDitibtiDistributionBuildingsLevel1Level2Single-familyy((SF))1452979%esidential8%(R)Mobilehomes(MH)240%30%15%Multi-family(MF)8605%3

7、%Commercial(CO)8405%Business13%Office(OF)4422%9%(B)3972%8%Industrial(IN)3972%14%Civic(CI)6954%OpenSpace(OS)OpenSpace(OS)1241%titutionalResidential(R)Business(B)Institutional(I)MixedR/B(I)Transportation(TR)620%MixedR/IMixedB/IData(Cont(Contd)’d)10000y=0.879x+1809.RR

8、=Residential9000R²=0.844B=Business8000tsI=Institutional7000E=EmptyUnit6000g5000R/BR/B/IHU4000R/IIdealRegressionB3000HousinERegressionLineH2000I10