OPTICAL AND SAR IMAGERY INTEGRATION BASED ON CLOUD COMPUTING FOR LAND COVER MAPPING IN THE CERRADO

The land cover mapping is of great relevance for the environmental monitoring and land management. Time series from the synthetic aperture radar (SAR) of Sentinel-1 (S-1) and the MSI/Sentinel-2 (S-2) optical sensor provide promising conditions for the land cover mapping due to their spectral, spatial and temporal resolutions. Here, we explored the hypothesis that the combination of S-1 and S-2 time series allows higher accuracy in the land cover mapping in Cerrado biome. The images were classified using the Random Forest algorithm in the Google Earth Engine cloud processing platform. The classifications obtained using only the S-2 data (kappa = 89.99) showed higher accuracy than those with the S-1 data (kappa = 75.78). The classification efficiency increased by combining the S-1 and S-2 data (kappa = 93.07). The results found here suggest that the shortwave infrared band, the VH polarization from SAR data, and the Cellulose absorption index (CAI) and Hall Cover index were the most significant variables in the land cover mapping of the Cerrado biome.