Assessment of multi-sensor approach to savanna landscape mapping in Ghana

Land Use Land Cover (LULC) maps play an important role in land cover change assessment. In this study existing LULC maps of 2006 and 2015 were used to develop a standardized LULC classification procedure for continuous mapping and land management. The same procedure was used to produce 2023 LULC map for the study area. Landsat 8, Sentinel-1, and Sentinel-2 were used in combination with a Random Forest algorithm to assess the potential of multi-sensor Earth observations in mapping savanna ecological zones in the Google Earth Engine (GEE). The classification results yielded an overall accuracy of 73.32% and kappa coefficient of 0.6342 when integrating Landsat 8 and Sentinel-2 data. In addition, an overall accuracy of 80.21% and kappa coefficient of 0.7225 were obtained for the combined Landsat 8, Sentinel-2, and Sentinel-1 data. The results demonstrated that using Sentinel-1 data in addition to multispectral data improved the classification accuracy by almost 7%.