Erosão e reflorestamento na Caatinga: aplicação de modelos RUSLE3D e USPED combinado com inteligência artificial

Abstract

Land use and land cover changes (LULC), expansion of farming, and reduction of native vegetation have exposed soil to erosion, compromising local ecosystem productivity and sustainability. This study aimed to analyze the impact of LULC alteration on erosive processes in the Caatinga and assess the applicability of AI-driven reforestation strategies in erosion control. In the first article, the Revised Universal Soil Loss Equation ThreeDimensional model (RUSLE3D) was applied to the Seridó River Basin (SRB) to analyze soil erosion between 1988 and 2018. The study revealed a reduction in native vegetation, with a 15.12% decrease in Caatinga forest cover, and over 35% of the area experienced LULC changes, primarily due to farming expansion. There was a direct relationship between reduced vegetation cover and increased erosion. The second article proposed forest restoration in the Caatinga using optimization algorithms to identify priority areas for reforestation. Applying the Unit Stream Power Erosion Deposition model (USPED) and AI, which aimed for a 20% increase in forest cover, reduced soil loss from 105,000 to 75,000 tons per year. The proposed reforestation improved soil stability, expanding stable areas from 39.48% to 43.07%, but led to increased fragmentation, potentially impacting carbon storage and edge effects. Thus, AI demonstrated promise in optimizing forest restoration, emphasizing the need to balance erosion mitigation with maintaining connectivity among vegetation fragments. The research underscores the critical role of forests in soil erosion mitigation and maintaining essential ecosystem services. Soil erosion models and AI techniques have proven powerful in guiding environmental management decisions.