水稻叶面积指数的多光谱遥感估算模型研究

LAI是生态系统研究中最重要的结构参数之一,它是估计多种植冠功能过程的重要参数。通过两年的水稻田间试验,使用美国ASD背挂式野外光谱辐射仪(ASDFieldSpec),获取1999～2000年两年晚稻整个生育期的光谱数据,采用计算机测算图斑面积法测定LAI;根据已有的卫星传感器通道波段(MSS、RBV、SPOT、TM、CH)和它们的组合(比值植被指数、归一化差植被指数),以及具有物理意义的光谱区域(蓝区、绿区、黄边、红光吸收谷、红边、紫区、可见光区、近红外区、全部波段)等共有27个变量构建多光谱变量组,采用5个单变量线性与非线性拟合模型,用1999年试验数据为训练样本,建立水稻LAI的多光谱遥感估算模型。结果表明:适用于水稻LAI估算的多光谱变量是植被指数变量好于波段变量;RVI与NDVI比较,RVI好于NDVI。用2000年试验数据作为测试样本数据,对其精度进行评价和验证,非线性模型的精度高于线性模型的精度,其中以SPOT3/SPOT2为变量的对数模型,拟合R2与预测R2达到了最大,其RMSE和相对误差(%)为最低,因此,认为它是估算LAI的最佳模型。

The Study on Multi-spectral Remote Sensing Estimation Models about LAI of Rice

Through two years, rice farm experiment under the different nitrogenous level, by using ASD- FieldSpec spectral data of late rice in whole developing time have been collected. According to satellite sensors, spectral channels and spectral regions that have physical content, variable groups of multiple spectra can be constructed. Using linear and non-linear regression methods, the estimated models about LAI of rice have been built on the basis of the experiment data in 1999 acted as train sample and their precision has been evaluated and tested on the basis of the experiment data in 2000. In the multiple spectra variables,the vegetation indices are better than wide wave band variables and RVI is better than NDVI and the precision of non-linear regression methods is higher than one of linear regression methods in estimated LAI models. In all models of estimating LAI, the logarithm model of SPOT3/SPOT2 that is independent is the best one, because its root mean square error and relative error were least.