草地遥感估产中不同尺度信息源关联方法对比及评价  

Evaluation and Comparing Methods of Correlating Information Source of Various Scales in Estimation of Grass Production

作　　者：陶伟国[1] 徐斌[1] 缪建明[2] 杨秀春[1] 覃志豪[1] 

机构地区：[1]中国农业科学院农业资源与农业区划研究所,北京100081 [2]农业部草原监理中心,北京100026

出　　处：《中国草地学报》2006年第4期68-74,共7页Chinese Journal of Grassland

基　　金：国家自然科学基金项目(30571078)

摘　　要：选用NDVI、EVI、MSAVI、OSAVI、AFRI五种植被指数分析比较了样点法、纯光谱法、绿度分层法三种关联方法,研究表明:(1)绿度分层法是最好的关联方法,得到的决定系数超过了0.9,但还需要进一步的生态分区才能更多的表达产量的局部空间特征;(2)对分辨率为1000m的MODIS影像来说,样点法中的“点-点”关联与“点-面”关联经统计检验没有显著差别,植被指数之间拟合程度差别不大;(3)在纯光谱法中,计算出了基于MODIS数据的K-T变换系数阵,但没得到预期的结果;(4)植被指数中EVI、MSAVI、OSAVI是比较好的估产指标。The key to using vegetation indices derived from remotely sensed data to predict grass biomass lies in association of remotely sensed data and ground data.Thus three correlating methods including sample point,greenness strata and pure spectral approach with NDVI,EVI,MSAVI,OSAVI and AFRI were analyzed and compared.The major conclusions from this study are: 1) Greenness strata is the best method(R^2>0.9),but suitable landscape ecological regionalization is necessary for expressing more local saptial information of grass production;2) For 1000m MODIS resolution,point to point and point to face is not different evidently after statistical test,and the variance of coefficient of determination is smaller among the five kinds of vegetation index;3) In pure spectral approach,the coefficient matrix of tasseledcap transform is computed.But the result is not satisfactory;4) The EVI,MSAVI and OSAVI are the best indices for estimating grass production.

关 键 词：植被指数 草原估产 穗帽变换 MODIS 尺度转换 

分 类 号：S127[农业科学—农业基础科学]