Global warming is intensifying and increasing the frequency of heat stress, particularly in tropical and subtropical regions. The warmest September on record in 2020 saw record-high temperatures in various locations, including South America, with Brazil's Midwest and North experiencing potentially fatal hyperthermia risks. These extreme heat events coincided with significant deforestation and forest fires, especially in the Amazon. Deforestation rates in the Brazilian Amazon reached their highest year-to-year increase since 2010 between August 2019 and July 2020, exceeding the targets set by the Paris Agreement and Brazil's National Policy. The Amazon rainforest's biodiversity and its role in regional climate regulation are critical. Progressive deforestation could cause abrupt climate shifts and intensified extreme weather events, exacerbating heat exposure's impacts on human health, work, and daily life, particularly for vulnerable groups. This study utilizes numerical climate modeling with the Brazilian Earth System Model (BESM-2.5) to evaluate the combined effects of Amazon savannization (deforestation) and climate change on heat stress risks and human health.

Existing literature highlights the growing concern over heat stress due to climate change and its impact on human health and productivity (Kjellstrom et al., 2018; Andrews et al., 2018). Studies show projected air temperature changes driven by global warming scenarios (Chou et al., 2014). Research on Amazon deforestation indicates its potential to cause significant regional climate change, including rainfall reduction and temperature increases (Nobre et al., 2009, 2016; Lovejoy & Nobre, 2019). The impacts of heat stress on human health are well-documented, ranging from heat-related morbidity and mortality to reduced physical and psychological performance (Kenny et al., 2010; Tawatsupa et al., 2012; Parsons, 2019). The wet-bulb globe temperature (WBGT) index is widely used to assess heat stress, considering factors like air temperature, humidity, solar radiation, and wind speed (ISO 7243, 2017; Havenith & Fiala, 2015). Previous studies have established thresholds for various levels of heat stress risk and survivability (Kjellstrom et al., 2015; Mora et al., 2017).

This study employed the Brazilian Earth System Model (BESM-OA2.5), a coupled ocean-atmosphere model, to conduct numerical climate modeling experiments. The model uses the Brazilian Global Atmospheric Model (BAM) and the GFDL Modular Ocean Model version 4p1 (MOM4p1). The surface model SSib48 was employed for heat flux and soil condition calculations. Two surface cover boundary conditions were used: (1) a forested Amazon Basin, and (2) a savannized Amazon Basin (representing complete deforestation). Three sets of experiments were conducted following the CMIP5 protocol: one historical run (1981-2010) with historical greenhouse gas concentrations and two future climate change scenarios (RCP4.5 and RCP8.5) for 2071-2100. The WBGT index was calculated for both outdoor and in-shade conditions using methods described by Liljegren et al. (2008) and Bernard (1999), respectively, incorporating temperature, humidity, solar radiation, and wind speed where relevant. The WBGT data were bias-corrected using a variable normalization approach. The results were then analyzed to assess the impact of savannization and climate change scenarios on heat stress levels. The study also incorporated a Social Vulnerability Index (SVI) to assess the vulnerability of populations to heat stress, based on factors such as human capital, urban infrastructure, and income and work. The ERA5 reanalysis dataset from ECMWF was used for meteorological data.

Simulations revealed that savannization of the Amazon Basin led to increased air temperatures, decreased relative humidity, and reduced precipitation, particularly within the basin. These effects were amplified under RCP4.5 and RCP8.5 climate change scenarios. The WBGT index showed the most dramatic impacts of savannization. In the hottest month, average daily maximum in-shade WBGT values exceeded the extreme risk threshold (≥34 °C), reaching 37 °C (RCP4.5) and 41 °C (RCP8.5). These high WBGT values were not observed in forested simulations, even under RCP8.5. Outdoor WBGT values reached 46 °C in the RCP8.5 savannization simulation. The combined effects of deforestation and climate change contributed to extreme in-shade WBGT values, increasing by up to 11.5 °C under RCP8.5 compared to historical conditions. Histograms of WBGT distributions illustrated the magnified risk of extreme heat under savannization. Under in-shade conditions, savannization in the historical period resulted in a WBGT distribution similar to the RCP8.5 scenario at the end of the century. Outdoor WBGT values in RCP8.5 savannization simulations exceeded 40 °C (risk to survivability) on 7% of days. This suggests that by the end of the century, more than six million people will be exposed to extreme heat stress risk (WBGT ≥ 34 °C) under RCP4.5 and more than eleven million under RCP8.5 (in-shade conditions) compared to historical conditions where such heat stress is not expected. These numbers include approximately 5.7 million highly vulnerable individuals in the RCP8.5 scenario. Considering outdoor conditions (WBGT > 40 °C), approximately 5 million people could be impacted by the end of the century, with two million deemed highly vulnerable.

The findings demonstrate that large-scale deforestation of the Amazon rainforest significantly magnifies the risk of extreme heat exposure associated with climate change. The study's results directly address the research question by quantifying the combined effects of deforestation and climate change on heat stress using a robust modeling approach. The magnitude of the projected increase in WBGT values (up to 11.5 °C under RCP8.5) underscores the severity of the threat. The results highlight the disproportionate impact on vulnerable populations in Northern Brazil. The equivalence between the effects of savannization in the historical period and RCP8.5 climate change at the end of the century emphasizes the urgent need to mitigate deforestation and greenhouse gas emissions. The study's implications extend beyond the immediate health impacts to include economic consequences, particularly in sectors like agriculture and construction, which rely heavily on outdoor labor. These findings are consistent with observed data on extreme warming in deforested areas of Brazil. The synergistic effects of deforestation-related vulnerabilities, such as unplanned urbanization and lack of infrastructure, further amplify the negative effects of climate change.

This study demonstrates the significant synergistic effects of Amazon savannization and climate change on heat stress levels in Brazil, particularly impacting vulnerable populations. The magnitude of the projected impacts underscores the urgency of addressing both deforestation and climate change to mitigate the risks to human health, well-being, and economic productivity. Future research could focus on more detailed vulnerability assessments, exploring adaptation strategies, and investigating the health system's capacity to respond to heat-related emergencies in vulnerable regions.

The study's projections rely on climate model outputs, which inherently have uncertainties. The use of a single climate model (BESM-OA2.5) limits the generalizability of the findings. The social vulnerability index used might not capture all relevant factors influencing individual-level vulnerability to heat stress. The population projections do not incorporate future population growth or changes in demographics.