基于水稻背景特性的植被指数参数修正研究

植被指数广泛应用在各种植被遥感监测中，但不同土壤背景会对基于植被指数的遥感监测精度产生影响，特别是以水为背景的水稻遥感监测。该研究旨在对各种背景调节植被指数的参数进行修正，以便更适合以水土混合物为背景的水稻参数估算。首先通过不同生育期水稻的冠层光谱构建在不同参数条件下的背景调节植被指数（WDVI、SAVI、SAVI2、TSAVI），然后以多种方程形式拟合以不同参数构建的各个植被指数与水稻叶面积指数LAI的关系，最后通过比较各拟合方程的决定系数（R2）得到各植被指数修正后的合适参数。结果表明：在使用植被指数估算水稻LAI时，其参数都需要修正。对WDVI，其修正后的参数α＝1.44；对SAVI，其修正后的参数L＝0.08；对SAVI2，其修正后的参数θ＝0.02；而对TSAVI，其修正后的参数a＝0.5，b＝0.02，X =0.02。另外，在各种拟合方程形式中，以指数和幂函数的拟合效果最佳。在以WDVI、SAVI、SAVI2和TSAVI为自变量，以LAI为因变量的各种估算模型中，TSAVI对LAI具有较高的估算精度，SAVI和SAVI2次之，WDVI最差。总之，在进行水稻遥感监测时，对植被指数的参数进行修正有利于提高监测精度。

Modification of vegetation indices based on rice background characteristics

Vegetation indices are widely used in many remote sensing monitoring of vegetation, but vegetation monitoring based on vegetation indices may be affected by variation of soil background, especially for rice which is cultivated in flooded soil. The objective of study is to modify the parameters of a number of background-adjusted vegetation indices with the purpose of more accurately estimating rice agricultural variables which have different backgrounds of various water coupled with different things in suspension. In the study, at the first step, vegetation indices with different parameters, which can reduce the effect of background, were derived from the spectra at different rice development stages. Second, the regression equations between vegetation indices and LAI using different forms were fitted. Finally, the optimal parameters were determined by selecting the highest coefficients of determination of fitted equations in different forms. The results indicated that the parameters of these vegetation indices all should be modified because of the special characteristics of rice background. The modified parameter were α =1.44 for WDVI, L=0.08 for SAVI, θ =0.02 for SAVI2, a =0.5, b=0.02, X =0.02 for TSAVI. In addition, among all the fitted equations taking WDVI, SAVI, SAVI2 and TSAVI as independent variables and LAI as dependent variable, TSAVI gave more accurate estimation than SAVI and SAVI2. WDVI behaved relatively poorly. In all, for remote sensing monitoring of rice, parameter modifications of vegetation indices can improve the monitoring precision.