Thermal Dynamics and Vulnerability in Brazil's Cerrado: A Comparative Analysis of Agricultural and Urban Expansion Impacts on Land Surface Temperature in Humid and Semi-Arid Sub-Regions

This study investigates the impacts of agricultural expansion and urbanization on land surface temperature (LST) dynamics across humid and semi-arid sub-regions of Brazil's Cerrado biome, with a focus on the Anápolis region as a humid-zone case study. Using multi-temporal satellite data (Landsat 5/7/8, Sentinel-1, Sentinel-2) combined with climate models (NASA POWER, IMERG, ERA5) from 1999 to 2024, the research quantifies LST variations and their relationship with land-use changes in contrasting climatic regimes—humid areas like Anápolis and southern Goiás (>1200 mm/year precipitation, denser vegetation) versus semi-arid areas like northern Goiás or western Bahia (<800 mm/year, sparser vegetation). The analysis is anticipated to identify significant LST increases, particularly in areas converted to croplands (soybeans, corn) and urban zones, with potential temperature rises up to 4°C in the most affected semi-arid regions. A novel Environmental Thermal Vulnerability Index (IVTA) is developed using machine learning (Random Forest, XGBoost) to assess vulnerability patterns, integrating four key variables: ΔLST, precipitation (P), soil roughness (SR), and vegetation cover (NDVI). The IVTA, formulated as IVTA = (LST * a) - (P * b) + (SR * c) + ((1 - NDVI) * d), leverages IMERG for P and Sentinel-1 SAR for SR (reflecting land alteration intensity), with coefficients adjusted per sub-region to reflect climatic and anthropogenic differences. This index is expected to provide improved predictive accuracy over traditional linear models, offering a three-tier classification of vulnerability (low, moderate, high). The study will likely demonstrate how climate extremes, particularly El Niño events, exacerbate thermal stress and agricultural impacts, with greater intensity in semi-arid zones. Future projections under SSP scenarios estimate potential LST increases up to 2.5°C by 2050, with approximately 30% of the study area potentially reaching high vulnerability status. The findings are expected to highlight urbanization, precipitation deficits, and soil alteration as critical drivers of thermal vulnerability, varying by sub-region, while providing science-based recommendations for sustainable land management—such as ecological corridors in humid zones and climate-smart urban planning in semi-arid areas. This research contributes a scalable methodological framework for assessing thermal vulnerability in tropical agricultural frontiers, supporting climate adaptation strategies in the Cerrado and similar biomes facing rapid land-use changes.