Shrub biomass estimation in semi-arid sandland ecosystem based on remote sensing technology

Abstract

Global land degradation and sustainable development has become a serious challenge for the terrestrial ecosystems. Shrub plays a crucial role in global ecosystem protection, ecological reconstruction, which is especially important in arid and semi-arid sandland ecosystem. Shrub above ground biomass (AGB) is a proxy of carbon sequestration capacity. Shrub AGB in Mu Us Sandland was estimated using different methods based on Landsat Thematic Mapper (TM) data, topography data, combined with in situ survey data. Linear regression model, multiple stepwise regression model, machine learning model and geometric optical model were used to estimate shrub biomass in combination with in situ data, respectively and their effects were validated and compared. Results showed that shrub AGB predicted from one multiple stepwise regression model with Ratio Vegetation Index (RVI) and Brightness from K-T transformation as input variables reached highest accuracy. For both high and low shrub coverage regions, shrub AGB distribution maps derived from this multiple stepwise regression model achieved higher precision. All these findings will provide a scientific support for ecological sustainable development in eco-vulnerable ecosystems.