[Effects of topographic correction on remote sensing estimation of forest biomass]

Based on four commonly used models (Cosine model, C model, C + SCS model, and Minnaert model), the topographic effects in Landsat-5 image of Maoershan region in Heilongjiang Province acquired on July 21, 2007 were calibrated on the platform of IDL language. The 4 models were validated from the aspects of visual differences and quantitative statistical features of the images. After the correlation analysis on the corrected remote sensing data and the forest biomass data, the biomass retrieving models were constructed. Furthermore, the effects of different topographic factors on the estimation of forest biomass were studied. The results showed that due to its liner presumption, the topographic correction combined with K-T transformation was not suitable for forest biomass estimation, and the correlations between the remote sensing data and the forest biomass fluctuated significantly. The parameters of the transformation needed to be adjusted in accordance with the information of land surface. The information content of vegetation index was significantly increased after topographic correction, and the correlation between vegetation index and forest biomass was enhanced greatly. Among the four models, Cosine model over-corrected the shaded areas in image, C model and C + SCS model had good correction performance by using semi-empirical parameters, while Minnaert model decreased the error of biomass estimation and improved the precision of remote sensing retrieving models effectively.