摘要
叶面积指数(LAI)是植被最重要的结构参数,既是地表蒸散模型的重要输入参数,又是决定生物量和产量的关键因子。资源一号02D卫星作为我国自主建造并成功运行的民用高光谱卫星,其高光谱数据为叶面积指数这一参量的反演提供了更为精细的波段选择。文章以资源一号02D卫星高光谱数据为数据源,应用辐射传输模型生成不同植被和土壤背景条件下的反射率光谱,并通过特征选择减少数据冗余,筛选与叶面积指数相关的关键吸收反射信息,应用机器学习方法构建叶面积指数的反演模型,并对所构建模型的精度进行评价。结果表明:辐射传输模型与机器学习相结合的方法可以实现对叶面积指数的估算,并得到较高的模型精度(R2=0.77),可尝试应用于区域叶面积指数产品生产。
Leaf area index(LAI)is the most important structural parameter of vegetation.It is not only an important input parameter of surface evapotranspiration model,but also a key factor determining biomass and yield.The paper uses hyperspectral data of ZY-1-02D satellite to carry out the estimation of leaf area index by combining a radiative transfer model with a machine learning method.Firstly,we calculate reflectance spectra under different vegetation and soil background conditions by using the radiative transfer model.Then,we reduce redundant data by feature selection in order to extract key absorption and reflection information related to LAI.Finally,we construct the LAI inversion model by machine learning and we evaluate the accuracy of the model.The method of combining radiative transfer model with machine learning can realize the estimation of leaf area index and get a higher model accuracy(R2=0.77).It can be applied to regional leaf area index products.
作者
魏丹丹
赵世湖
肖晨超
崔航
刘书含
WEI Dandan;ZHAO Shihu;XIAO Chenchao;CUI Hang;LIU Shuhan(Land Satellite Remote Sensing Application Center,Ministry of Natural Resources,Beijing 100048,China)
出处
《航天器工程》
CSCD
北大核心
2020年第6期169-173,共5页
Spacecraft Engineering
基金
国家重大航天工程,特大型地质灾害防治专项(121133000000190007)。
