基于高光谱与多光谱植被指数的洪河沼泽植被叶面积指数估算模型对比研究

A Comparative Study on Estimation Model for Leaf Area Index of Vegetation in Marshes in Honghe National Nature Reserve based on Hyperspectral and Multispectral Vegetation Indices

以洪河国家级自然保护区为研究区,2009年8月中旬,在研究区野外实测沼泽植物冠层的光谱反射率和叶面积指数(LAI),将地面实测的植物高光谱反射率以Landsat-5 TM波段范围为基准进行波谱重采样,以重采样后的光谱反射率计算多光谱植被指数,用几种常见的高光谱和多光谱植被指数建立估算沼泽植被叶面积指数的统计回归模型,对比这些模型的精度,选出最优模型。研究结果表明,用各植被指数建立的估算沼泽植被叶面积指数的回归模型分别为二次函数、对数函数或指数函数;各模型对沼泽植被叶面积指数的反演精度差别较大;在全波段高光谱植被指数中,用全波段归一化植被指数H-FNDVI(R930,R515)构建的估算沼泽植被叶面积指数的模型最佳;在常规高光谱植被指数中,用修正简单比率H-MSR构建的估算沼泽植被叶面积指数的模型最佳;在多光谱植被指数中,用多光谱归一化植被指数M-NDVI构建的估算沼泽植被叶面积指数的模型最佳。对比发现,由多光谱数据提取的植被指数构建的模型对研究区LAI的估算效果不太理想,而从实测高光谱数据提取的窄波段特有植被指数构建的估算沼泽植被叶面积指数模型表现出较明显的优势,表明窄波段植被指数更适合用来监测沼泽植被叶面积指数。

The Honghe National Nature Reserve was taken as the study area. The canopy reflectance and the leaf area index （LAI） of vegetation in marshes were measured in the Honghe National Nature Reserve in the middle of August, 2009. The hyperspectral vegetation indices were calculated based on the canopy reflec- tance. The hyperspectral data measured in field was resampled based on Landsat TM. The multispectral vege- tation indices were calculated by the multispectral data that was came from the resampling of the hyperspec- tral data. The statistic regression models were established between leaf area index measured in the field and the different forms of hyperspectral and multispectral vegetation index （VI）. The optimal models were select- ed by comparing the precision of these models. The result showed that the regression models for estimating leaf area index established by vegetation indices included quadratic function, logarithmic function and expo- nential function. The inversion precision of these regression models was quite different. In the full-band hyper- spectral vegetation indices, the optimal regression model for estimating leaf area index of vegetation in marsh- es was established by the full band normalized difference vegetation index H-FNDVI（R930, R5~5）. In the conven- tional hyperspectral vegetation indices, the optimal regression model was established by the modified simple ratio index H-MSR. In the multispectral vegetation indices, the optimal regression mode[ was established by multispectral normalized difference vegetation index M-NDVI. After comparing, the performance of the multi- spectral vegetation indices was not ideal. But the performance of narrow band vegetation indices extracted from the ground hyperspectral data was obviously predominant for estimating leaf area index of vegetation in marshes and all of them performed better in estimate accuracy than their broad band counterparts. So it was more suitable for narrow-band vegetation indices to estimate leaf area index of vegetation in marshes.