ESTIMATION OF FOREST BIOMASS BASED ON MULITI-SOURCE REMOTE SENSING DATA SET – A CASE STUDY OF SHANGRI-LA COUNTY
Keywords: Texture, Regression model, Shangri-La county, Multi-source data, Forest biomass estimation
Abstract. Forest biomass is an important indicator for the structure and function of forest ecosystems, and an accurate assessment of forest biomass is crucial for understanding ecosystem changes. Remote sensing has been widely used for inversion of biomass. However, in mature or over-mature forest areas, spectral saturation is prone to occur. Based on existing research, this paper synthesizes domestic high resolution satellites, ZY3-01 satellites, and GLAS14-level data from space-borne Lidar system, and other data set. Extracting texture and elevation features respectively, for the inversion of forest biomass. This experiment takes Shangri-La as the research area. Firstly, the biomass in the laser spot was calculated based on GLAS data and other auxiliary data, DEM, the second type inventory of forest resources data and the Shangri-La vector boundary data. Then, the regression model was established, that is, the relationship between the texture factors of ZY3-01 and biomass in the laser spot. Finally, by using this model and the forest distribution map in Shangri-La, the biomass of the whole area is obtained, which is 1.3972 × 108t.