Abstract: Correlation analysis of LANDSAT TM data and its derived data, meteorological data and topographic data with the biomass of tropical forest vegetation of different ages (Young, middle-aged, near mature, mature and over mature forests) were explored in Xishuangbanna, Yunnan Province, China. The analysis included four steps. First, the biomass of each forest was determined from field inventory data and a GIS database based on the geo-coordinates of each forest sample site. Second, the LANDSAT TM images were geometrically corrected using topographic maps. The derived data were derived from the LANDSAT TM images using principal component analysis, tasseled cap transform and vegetation index analysis. Third, the data, including LANDSAT TM data and its derived data, topographic data, such as DEM and aspect, and climatic data, such as annual average temperature, annual average accumulative temperature above zero degree, annual average precipitation and humidity, were referenced to the same projection and coordinate system and interpolated across a grid at a resolution of 30 meter. The LANDSAT TM data and its derived data, the topographic data and the climatic data for the samples were achieved by using overlay analysis. Fourth, all of the data were overlaid onto the different-aged forests. Finally, correlations among the LANDSAT TM and its derived data, meteorological data, topographical data and forest biomass were analyzed for each forest type. Our results are as follows: 1) Correlations between the biomass of the young forest and LANDSAT TM1 and LANDSAT TM6 were significant at the 0.05 level and both correlations reached -0.33. 2) The correlation between the biomass of the middle-aged forest and precipitation was significant at the 0.05 level and reached 0.33. 3) The correlations between the biomass of the near mature forest and VI3, TM4 and Bright Index were significant at the level 0.05 and were 0.50, -0.45 and -0.45, respectively. 4) Correlations of the biomass of mature and over mature forest and the second principal component, which were significant at the 0.05 level, reached -0.46. 5) The highest correlation existed between the near mature forest biomass and VI3.

Key words: Tropical forest vegetation, Forest age, Biomass, Remote sensing, Correlation analysis