以浙江省内的一景Landsat 8影像和96块野外调查数据为数据源,提取出植被指数、纹理特征、地形因子并进行优选,对优选后的变量因子进行分组,生成植被指数、植被指数+纹理特征相结合的2种自变量集,采用偏最小二乘回归法构建乔木林地上生物量估算模型,并对估算结果进行对比分析。结果显示:在构建的模型中,植被指数+纹理特征集构建的模型精度均要优于植被指数集构建的模型,说明多光谱波段的纹理特征具有改善模型估算效果的作用。采用的变量筛选方法较好地考虑了变量间的相关性及共线性问题,可以提高所构建模型的稳定性。

With the Landsat 8 image and 96 field survey data in Zhejiang Province as the data source,the vegetation indices,texture features,terrain factors were extracted and optimized,and the preferred variables were selected.The factors were grouped into two sets of independent variables:vegetation indices,vegetation indices+ texture features.The least squares regression method was used to construct the aboveground biomass estimation model of arbor forest,and the estimation results were compared and analyzed.The results show that in the constructed models,the model accuracy of vegetation indices + texture features set is better than the model constructed by the vegetation indices set.The result indicates that the texture features of the multispectral bands have the effect of improving the model estimation accuracy.In addition,the variable screening method adopted in this paper takes into account the correlation between variables and the collinearity problem,which can improve the stability of the constructed model.