摘要
叶面积指数(Leaf Area Index,LAI)作为表征不同作物生长状况的基本参数,是农业精细化管理及农田生态系统建模的关键。我国农田作物种植比较离散,受地表空间结构非均一性和反演模型非线性等因素影响,不同尺度遥感数据估算的作物LAI存在一定的差异,即农田作物LAI的遥感反演普遍存在尺度效应问题。以包头遥感综合验证场农业示范区为研究区,利用无人机高光谱数据结合PROSPECT+SAIL模型构建典型农作物区多类型作物的查找表(Look-Up-Table,LUT)反演农田LAI,研究查找表用于玉米、马铃薯、向日葵、瓜地等不同作物LAI反演的适用性和精度;通过无人机高光谱数据聚合获得多尺度遥感数据源,结合Taylor展开理论和计算几何模型,提出了一种既考虑类间差异又考虑类内异质性的尺度转换模型,定量描述多种作物混合的非均一地表LAI反演过程中的尺度效应特征。结果表明:基于分类和参数敏感性分析的LUT方法能很好地应用于包头典型农作物区多类型混合作物LAI反演,总估算精度为相关系数R^2=0.82、均方根误差RMSE=0.43m^2/m^2。随着反演尺度的增加,作物类间差异造成的反演偏差明显高于类内异质性,利用本文所提出的尺度转换模型均能较好纠正低分辨率LAI反演的尺度效应问题。
Leaf Area Index（LAI）is a key structural characteristic of crops and plays a significant role in precision agricultural management and farmland ecosystem modeling.However,LAI retrieved from different resolution data contain a scaling bias due to the spatial heterogeneity and model non-linearity,that is,there is scale effect during multi-scale LAI estimate.In this article,the typical farmland of Baotou test site in Inner Mongolia is taken as the study area,based on the combination of PROSPECT model and SAIL model,a multiple dimensional Look-Up-Table（LUT）is generated for multiple crops LAI estimation from unmanned aerial vehicle hyperspectral data.Based on Taylor expansion method and computational geometry Model,a scale transfer model considering both difference between inter-and intra-class is constructed for scale effect analysis of LAI inversion over inhomogeneous surface.The results indicate that,（1）the LUT method based on classification and parameter sensitive analysis is useful for LAI retrieval of corn,potato,sunflower and melon on Baotou test site,with correlation coefficient R^2 of 0.82 and root mean square error RMSE of 0.43m^2/m^2.（2）The scale effect of LAI is becoming obvious with the decrease of resolution,and scale difference between inter-classes is higher than that of intra-class.which can be corrected efficiently by the scale transfer model established based Taylor expansion and computational geometry.
出处
《遥感技术与应用》
CSCD
北大核心
2017年第3期427-434,共8页
Remote Sensing Technology and Application
基金
国家863计划项目“遥感载荷性能与数据质量检测技术”(2013AA122102)
