基于机器学习的冬小麦种植区土壤盐分反演制图
打开文本图片集
中图分类号：TP181;TP79;S153.6 文献标识码：A DOI编码：10.3969/j.issn.1006—6500.2025.10.005
Remote Sensing Machine Learning-Based Inversion and Mapping of Soil Salinity in Winter Wheat PlantingAreas
WANG Yanmei¹， XIAO Hui²， CHEN Kun²， CAO Hongtao1，DONG Yuchen³， ZHENG Mandi² （1.InstituteofeinaebeiEoloicalCvlationeveloptanjalUivesityanjO87a; stituteofAgriclturalesoucsndEnviont，ianjadyofgriculuralienesanj9，ina;3.i GreenManure Biotechnology Development CompanyLimited，Tianjin 3O16O6，China）
Abstract：InrdertoexplorequantitativemodelsforsoilsalinityintewinterheatplantingareasofJinghaiDistrict，Tianjinandto graspthedistributionofsoilsalinization，soilelectricalconductivity（EC）datawereobtainedthroughfieldsampling.Basednulti sourceremotesensingdataincudingSntinel-2，sisptralidices（NDVAV，EVIetc.）andtspectralbands（ueen red，ed-edge，arfrard，tc.）red.cialpoetalisassedtoreutipleeareblih highcontributionratestosoilsalinity.Thesoilsalinitypredictionperformanceoffvemachinelearningmodels，includingradomfor est，decisiontre，ndetremegadientbtigtree（GBost），asompaedndalydtodentifyeptialprdictiodel formapingsoilsalinitydistributioninthestudyarea.Theresultsshowedthatprincipalcomponentanalysis methodscreenedout7 characteristicarablessuchasB8A，CIDVIandNDVITeXGBostodeldemonstratedoutstandingpeformanceindiction accuracy and fitting effectiveness （ R2 =0.93，RMSE=O.14），making it the preferred choice for creating a spatial distribution map of soil salinityiwinterwheat-growingareasofTianjin’sJinghaiDistrict.Themapclearlyshowedseveresalinizationcharacteristicsprevalentthroughouttheregion.Fieldinvestigationsconfimedthatthesefidingsalignwithlocalsoilconditions.Thisstudydeostrates thefeasibilityofaplyingmachineleaingmethods toconstructsoilsalinityinversionmodelsinTanjin'sJinghaiDistrct，providing valuable references for developing targeted soil improvement strategies and crop cultivation approaches.
Key words： rmulti-sourceremotesensing; soilsalinization; spectralindex; machinelearning
在我国高度重视粮食安全的大环境下，土壤盐渍化对作物生长的影响问题也引起了社会各界的广泛关注。（剩余9644字）
