Agricultural intensification practices have been adopted in the Brazilian savanna (Cerrado), mainly in the transition between Cerrado and the Amazon Forest, to increase productivity while reducing pressure for new land clearing. Due to the growing demand for more sustainable practices, more accurate information on geospatial monitoring is required. Remote sensing products and artificial intelligence models for pixel-by-pixel classification have great potential. Therefore, we developed a methodological framework with spectral indices (Normalized Difference Vegetation Index (NDVI), Normalized Difference Water Index (NDWI), and Soil-Adjusted Vegetation Index (SAVI)) derived from the Harmonized Landsat Sentinel-2 (HLS) and machine learning algorithms (Random Forest (RF), Artificial Neural Networks (ANNs), and Extreme Gradient Boosting (XGBoost)) to map agricultural intensification considering three hierarchical levels, i.e., temporary crops (level 1), the number of crop cycles (level 2), and the crop types from the second season in double-crop systems (level 3) in the 2021–2022 crop growing season in the municipality of Sorriso, Mato Grosso State, Brazil. All models were statistically similar, with an overall accuracy between 85 and 99%. The NDVI was the most suitable index for discriminating cultures at all hierarchical levels. The RF-NDVI combination mapped best at level 1, while at levels 2 and 3, the best model was XGBoost-NDVI. Our results indicate the great potential of combining HLS data and machine learning to provide accurate geospatial information for decision-makers in monitoring agricultural intensification, with an aim toward the sustainable development of agriculture.
The Brazilian Savanna presents a complex agricultural dynamic and cloud cover issues; therefore, there is a need for new strategies for more detailed agricultural monitoring. Using a hierarchical classification system, we explored the Harmonized Landsat Sentinel-2 (HLS) dataset to detect soybean in western Bahia, Brazil. Multispectral bands (MS) and vegetation indices (VIs) from October 2021 to March 2022 were used as variables to feed Random Forest models, and the performances of the complete HLS time-series, HLSS30 (harmonized Sentinel), HLSL30 (harmonized Landsat), and Landsat 8 OLI (L8) were compared. At Level 1 (agricultural areas × native vegetation), HLS, HLSS30, and L8 produced identical models using MS + VIs, with 0.959 overall accuracies (OA) and Kappa of 0.917. At Level 2 (annual crops × perennial crops × pasturelands), HLS and L8 achieved an OA of 0.935 and Kappa > 0.89 using only VIs. At Level 3 (soybean × other annual crops), the HLS MS + VIs model achieved the best performance, with OA of 0.913 and Kappa of 0.808. Our results demonstrated the potential of the new HLS dataset for medium-resolution mapping initiatives at the crop level, which can impact decision-making processes involving large-scale soybean production and agricultural sustainability.
Water erosion is one of the main problems faced in coffee cultivation, as it promotes environmental degradation and crop yield decrease. Erosion estimates support the planning of conservation management practices and allowing determining the rates of soil losses. Thus, the objective of this paper was to estimate the soil loss by water erosion using the Erosion Potential Method in a sub-basin predominantly covered by coffee cultivation and then to compare the results with the Soil Loss Tolerance limits. The study area is the Coroado Stream Sub-basin, located at Alfenas Municipality, south of Minas Gerais, Brazil. The sub-basin presented an Erosion Coefficient of 0.272, indicating a predominance of low-intensity erosion. The total soil loss estimate was 1,772.01 Mg year-1 with an estimated average of 1.74 Mg ha-1 year-1. Soil Loss Tolerance limits range from 4.75 to 7.26 Mg ha-1 year-1 and, according to the Erosion Potential Method, only 1.0% of the sub-basin presented losses above the limits. The areas with the highest slopes and bare soil concentrated the highest losses rates and should be prioritized in the adoption of mitigation measures. The Erosion Potential Method estimated soil losses in tropical edaphoclimatic conditions in a fast, efficiently and at low cost, supporting the adoption of conservation management practices.
Climate change is expected to alter the environmental suitability of land use and land cover (LULC) classes globally. In this study, we investigated the potential impacts of climate change on the environmental suitability of the most representative LULC classes in the southern Brazilian semiarid region. We employed the Random Forest algorithm trained with climatic, soil, and topographic data to project future LULC suitability under the Representative Concentration Pathway RCP 2.6 (optimistic) and 8.5 (pessimistic) scenarios. The climate data included the mean annual air temperature and precipitation from the WorldClim2 platform for historical (1970–2000) and future (2061–2080) scenarios. Soil data were obtained from the SoilGrids 2.1 digital soil mapping platform, while topographic data were produced by NASA’s Shuttle Radar Topography Mission (SRTM). Our model achieved an overall accuracy of 60%. Under the worst-case scenario (RCP 8.5), croplands may lose approximately 8% of their suitable area, while pastures are expected to expand by up to 30%. Areas suitable for savannas are expected to increase under both RCP scenarios, potentially expanding into lands historically occupied by forests, grasslands, and eucalyptus plantations. These projected changes may lead to biodiversity loss and socioeconomic disruptions in the study area.
Farmers in the Brazilian Cerrado are increasing grain production by cultivating second crops during the same crop growing season. The release of PlanetScope (PS) satellite images represents an innovative opportunity to monitor double cropping production. In this study, we analyzed the potential of six PS monthly mosaics from the 2021/2022 crop growing season to discriminate double cropping areas in the municipality of Goiatuba, Goiás State, Brazil. The four multispectral bands of the PS images were converted into normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), green–red normalized difference index (GRNDI), and textural features derived from the gray-level co-occurrence matrix (GLCM). The ten most important combinations of these attributes were used to map double cropping systems and other land use and land cover classes (cultivated pasture, sugarcane, and native vegetation) of the municipality through the Random Forest classifier. Training and validation samples were obtained from field campaigns conducted in October 2021 and April 2022. PS mosaic from February 2022 was the most relevant data. The overall accuracy and Kappa index of the final map were 92.2% and 0.892, respectively, with an accuracy confidence of 81%. This approach can be expanded for mapping and monitoring other agricultural frontiers in the Cerrado biome.
27 October 2020 in Revista Brasileira de Ciências Agrárias - Brazilian Journal of Agricultural Sciences