The comparison of the estimation of the biomass of tropical forest vegetation with the different forest vegetation type by using multi-variant liner regression

The estimation of the biomass of tropical forest vegetation is very important for global change research. The estimation is explored by using multi-variant linear regression in Xishuangbanna, Yunnan province, P.R. of China. It includes four steps. Firstly, the biomass for each sample is calculated by using the field inventory data of each sample. Secondly, the remote sensing data including Landsat TM data and its derived data, the topographic data such as DEM, slope and aspect, and the climatic data such as temperature and precipitation are referenced to the same projection and coordination. Thirdly, all of the data are stratified as the lithocarpus forest, the other hard broad-leaf forest and the other broad-leaf forest. Finally, The multi-variant regression analysis is used to estimate the biomass for each forest type. It is shown that the higher accuracy can be obtained for the hard broad-leaf forest compared to the lithocarpus and the other broad-leaf forest.