基于垂直植被指数的东北黑土区玉米LAI反演模型研究  

Retrieval model for estimating corn LAI in black soil region of Northeast China based on Perpendicolar Vegetation Index

作　　者：李凤秀[1] 张柏[1] 宋开山[1] 王宗明[1] 刘焕军[1] 杨飞[1] 

机构地区：[1]中国科学院东北地理与农业生态研究所

出　　处：《干旱地区农业研究》2008年第3期33-38,54,共7页Agricultural Research in the Arid Areas

基　　金：中国科学院知识创新重要方向性项目(KZCX3-SW-356);中国长春净月潭遥感站网络台站基金资助项目

摘　　要：本文旨在探讨以不同波段组合垂直植被指数所建立的高光谱模型对玉米叶面积指数（Leaf Area Index，LAI）的反演精度。在不同水肥耦合作用条件下，实测玉米冠层的高光谱反射率与叶面积指数数据以及裸土的高光谱反射率数据，在高光谱红光波段（631～760nm）与近红外波段（760～1050nm）逐波段构建土壤线，并在此基础上构建垂直植被指数（Perpendieolar Vegetation Index，PVI），找出与LAI具有最佳相关性波段组合PVI，建立玉米LAI估算模型。结果显示，采样波段间隔越窄，反演精度越高，在采样波段间隔1．4nm的PVI（R677，R918）反演2004年的玉米LAI模型中，最佳回归方程是指数函数，精度达91．1％，标准差为0．1997，RMSE=0．0399，通过了0．01极显著验证。采用高光谱数据构建的PVI植被指数对玉米LAI的估算可以取得较高的精度。An experiment was carried out to evaluate the precision of hyperspectral reflectance models based on Perpendicolar Vegetation Index （PVI） which is formed by various bands for monitoring corn leaf area index （LAI）. Corn was cultivated under different water-fertilizer coupled control conditions and corn LAI was collect- ed simultaneously with LI -COR LAI- 2000, while corn canopy reflectance data and bare soil reflectance data were collected with ASD spectroradiometer （350-1 050 nm）. At first, each band of NIR and red was applied to establish soil-line, based on which PVI could be established, then to find out the best band for PVI; and then, PVI with the best reflectance band was applied to regress against corn LAI. The result showed that the sampling wave band gap was narrower, and the accuracy of retrieved was better. The best corn LAI model retrieved by sampling 1.4 nm interval PVI in 2004 was exponential model, and the accuracy of estimation exponential model of LAI established with the wave went up to 91.1%, and the standard error was 0. 1997, RMSE = 0. 0399, which is qualified for 0.01 level. The PVI vegetation index established by hyperspectral data can highly estimate the corn LAI.

关 键 词：高光谱 玉米LAI PVI 土壤线 

分 类 号：S127[农业科学—农业基础科学]