Mapping Agricultural Intensification in the Brazilian Savanna: A Machine Learning Approach Using Harmonized Data from Landsat Sentinel-2

Agricultural intensification practices have been adopted in the Brazilian savanna (Cerrado), mainly transition between Cerrado and Amazon Forest, to increase productivity while reducing pressure for new land clearing. Due growing demand more sustainable practices, accurate information on geospatial monitoring is required. Remote sensing products artificial intelligence models pixel-by-pixel classification great potential. Therefore, we developed a methodological framework with spectral indices (Normalized Difference Vegetation Index (NDVI), Normalized Water (NDWI), Soil-Adjusted (SAVI)) derived from Harmonized Landsat Sentinel-2 (HLS) machine learning algorithms (Random Forest (RF), Artificial Neural Networks (ANNs), Extreme Gradient Boosting (XGBoost)) map agricultural considering three hierarchical levels, i.e., temporary crops (level 1), number of crop cycles 2), types second season double-crop systems 3) 2021–2022 municipality Sorriso, Mato Grosso State, Brazil. All were statistically similar, an overall accuracy 85 99%. The NDVI was most suitable index discriminating cultures at all levels. RF-NDVI combination mapped best level 1, levels 2 3, model XGBoost-NDVI. Our results indicate potential combining HLS data provide decision-makers intensification, aim toward development agriculture.