Applications, challenges and perspectives for monitoring agricultural dynamics in the Brazilian savanna with multispectral remote sensing

Author(s)

Parreiras, Taya Cristo ; Bolfe, Edson Luis ; Pereira, Paulo Roberto Mendes ; de Souza, Abner Matheus ; Alves, Vinicius Fernandes

• Document type: Journal article
• Source: REMOTE SENSING APPLICATIONS-SOCIETY AND ENVIRONMENT; v. 37, p. 18-pg., 2025-01-21.

Abstract

Land use and cover changes significantly impact landscape configuration, climate change, and society. The processes of expansion, conversion, intensification, diversification, and reduction materialize these changes in the agricultural environment. The Cerrado, or Brazilian Savanna, is a biodiversity hotspot, extremely important for water production, and one of the most important biomes for global food production. In this sense, monitoring agricultural dynamics in this environment plays a crucial role in sustainable planning, assessment of environmental impacts, and food security. In this study, we propose to analyze the evolution of the role of multispectral orbital remote sensing in mapping and monitoring agricultural dynamics processes in the Cerrado. Therefore, a narrative review of the literature based on studies developed in the biome was carried out to identify advances in tools, processes, and resources, as well as evaluate the challenges and perspectives for the future. Among other relevant results, monitoring these processes has become faster, more frequent, and more accurate, mainly through the combined use of high temporal resolution time series of spectral data and machine learning algorithms. Promising results have been obtained with Harmonized Landsat Sentinel-2 (HLS) data. The consolidation of deep neural networks has contributed substantially to detecting and delimitating complex intensification and diversification systems, such as central irrigation pivots and intercropping. However, there are challenges and obstacles to be faced, such as expanding the use of Sentinel-2 data, establishing means for sharing field data, and expanding studies to more fragmented landscapes, especially agricultural production on small properties. (AU)

• FAPESP's process: 24/13150-5 - Spatiotemporal analysis of agricultural dynamics in a region of high production diversity using multisensor imagery and machine learning
• Grantee: Taya Cristo Parreiras
• Support Opportunities: Scholarships in Brazil - Doctorate
• FAPESP's process: 22/09319-9 - Center of Science for Development in Digital Agriculture - CCD-AD/SemeAr
• Grantee: Silvia Maria Fonseca Silveira Massruhá
• Support Opportunities: Research Grants - Science Centers for Development