Mapping Shrub Coverage in Xilin Gol Grassland with Multi-Temporal Sentinel-2 Imagery

In recent decades, shrubs dominated by the genus Caragana have expanded in a large area Xilin Gol grassland, Inner Mongolia, China. This study comprehensively evaluated performances of multiple factors for mapping shrub coverage across grassland based on spectral and temporal signatures Sentinel-2 imagery, first time produced large-scale result this region. Considering regional differences gradients types sizes area, was divided into three subregions precipitation data, i.e., west, middle east regions. The estimation accuracy from dry- wet-year different vegetation indices (VIs) regression methods were compared each subregion, key phenological periods selected. We also four strategies, which pairwise combinations zoning (i.e., precipitation) non-zoning, full series VIs period. Results show that dry year (2017) is higher than wet (2018). optimal high spatial variability. terms non-zoning. root mean square error (RMSE), overall (OA) recall ‘zoning + (or period)’ strategy 0.052 (0.055), 76.4% (79.7%) 91.7% (94.6%), respectively, while ‘non-zoning 0.057 (0.060), 75.5% (74.6%) (88.6%), respectively. using better low-value prediction cover. Based period’, map whole region generated with RMSE 0.055, OA 80% 95%. not only provides data coverage, but reference dynamic monitoring area.