Shrub biomass estimation in Mu Us Sandland using simple GO model and multi-angle observations

摘要

Forest biomass is an important indicator in carbon sequestration capacity and forest carbon budget evaluation, but there is less focus on shrub biomass. Multi-angle images provide the volume scattering information which could improve inversion accuracy. HJ-1 satellites are the sun synchronous recurrent frozen orbit small satellite constellation for environment and disaster monitoring and forecasting and were launched on September 6, 2008, China, with a repetition cycle of two days. Now 6 angle observations have been built up from several days of HJ-1A / B images. Images were first calibrated both geometrically and atmospherically. A simplified Geometric-Optical(SGM) model was used for the parameters estimation. The background estimation was conducted from the LiSparse-RossThin kernel weights and the Walthall empirical model. Then, the three kernel weights of the LiSparse-RossThin model were retrieved for each location. Finally, the shrub cover and radius were retrieved from SGM model. Biomass was estimated from the shrub cover and radius. 21 plots for background estimation and verification were measured during July 2009. The R2 between simulated cover and measured cover is 0.21, but the RMSE of the model fitting is 0.09. The distribution of the shrub cover estimated from multi-angle HJ images is coincidence with fine resolution Google Earth image. So is the biomass distribution. The weak relationship may be because the higher spatial heterogeneity, the view angles' limitation, and the time variance among images. The multi-angle HJ images are low cost which is useful in far and extended sandland monitoring and evaluation. The results reported here also showed that the development of HJ satellite data for wide range of mapping application.