According to data obtained from meteorological towers, Brazil has significantly increased temperature in the past 20 years, particularly in the North and Midwest regions. Vapor pressure deficit and evapotranspiration were also analyzed, showing an increase across the entire country, confirming that the air is becoming drier. This warming trend is part of the global climate change phenomenon caused by the rise of greenhouse gases in the atmosphere, fires, poor soil management practices, deforestation, and logging. The increase in temperature and dryness has profoundly impacted Brazil’s climate and ecosystems, leading to intensified extreme weather events and changes in the distribution of both animal and plant species. This study highlights the importance of utilizing meteorological tower data to monitor and understand the effects of climate change in Brazil. It emphasizes the need for immediate action to address its causes and mitigate its negative impacts.

In this study, data collected from 2000 to 2019 on vapor pressure deficit ($VPD$) and its relationship with micrometeorological variables (fire occurrences, aerosol concentration, temperature, and carbon flux) were analyzed in six locations situated in different Brazilian ecosystems: Rio Branco, AC; Manaus, AM; Alta Floresta, MT (within the Amazon Rainforest); Baia das Pedras, MT (Pantanal); Fazenda Miranda, MT (Cerrado); and Petrolina, PE (northeastern semiarid region). Temporal series analysis of $VPD$ was conducted by determining the principal component of singular spectrum analysis (SSA) for this variable in all locations. It was observed that the main component of SSA for $VPD$ is sensitive to local land-use changes, while no evidence of large-scale influences related to global climate change was observed. A strong coupling between $VPD$ values and local maximum temperature with monthly fire occurrence and logarithmic aerosol concentration profiles was also observed. The results of the study are discussed in the context of the ecosystems’ carbon sequestration capacity. The combined results of the study indicate a scenario in which local land-use changes can compromise the capacity of Brazilian ecosystems to absorb carbon.

Evapotranspiration is crucial for the water balance, especially in irrigated agriculture. It can be measured directly using lysimeters or estimated through models such as the Penman-Monteith (PM) model, recommended by the FAO. This study aimed to analyze evapotranspiration measurements from a constant water table lysimeter and verify their validity by comparing the measurements, taken between May and November 2023, with the estimates from the PM model. The experiment was conducted in Botucatu, SP, which has a tropical savanna climate (Aw), characterized by hot, rainy summers and cold, dry winters. For comparison, the following validation indicators were used: root mean square error (RMSE), mean absolute error (MAE), Willmott's index of agreement (d), mean bias error (MBE), Pearson’s correlation coefficient (r), relative mean bias error (rMBE), relative root mean square error (rRMSE), and the confidence coefficient (c). The results showed that in August, the PM model underestimated evapotranspiration measurements by 2.28%, the smallest variation in the analyzed period. In October, which had the greatest variation, the model overestimated the measurements by 46.39%. This highlights the need for seasonal adjustments when using the PM model in the region.

This study develops an integrative modeling approach combining gray-box modeling and machine learning (ML) to predict micrometeorological dynamics in the Amazon rainforest, specifically at the Jaru Biological Reserve in Rondônia, Brazil. A physics-based “dry model,” describing temperature dynamics primarily through radiative forcing and cooling processes, provides a baseline subsequently corrected by ML-derived additive terms. A fully connected neural network with 641 parameters successfully learned nonlinear corrections, significantly reducing prediction errors by over 60% in the dry season and by approximately 55.5% in the wet season. To ensure model reliability, we employed bootstrap ensemble modeling with 100 trained networks to quantify prediction uncertainty, revealing well-calibrated confidence intervals for the dry season (94.2% empirical coverage for nominal 90% intervals) but systematic biases during the wet season (mean bias: 2.08 K), highlighting fundamental limitations when key moisture-related processes are omitted from the physics-based framework. Systematic comparison with Gaussian process regression and polynomial approximations validated our uncertainty quantification approach while demonstrating that the neural network bootstrap ensemble provides superior robustness across diverse seasonal conditions. To enhance interpretability, neural corrections were approximated using symbolic (polynomial) regression, revealing dominant nonlinear temperature–radiation interactions in both dry and wet conditions. Our findings demonstrate marked seasonal contrasts in radiative forcing, atmospheric clarity, and cooling efficiency, underscoring the need for seasonally tailored modeling strategies. This combined physics-informed and data-driven methodology offers reliable and interpretable models essential for managing environmental impacts amid increasing anthropogenic and climatic pressures on the Amazon.