The study was carried out in the Triângulo Mineiro mesoregion, Brazil, which is predominantly characterized as a Cerrado biome. The region comprises 66 municipalities and covers an area of 90,545 km², with an average altitude of 769 m (Figure 1). This region stands out in agriculture, with agribusiness being one of the predominant activities in the region due to the suitability of the soils and favorable climatic conditions (Araújo; Araújo Sobrinho, 2020).

Figure 1 Source: The authors (2023).

The region’s climate is classified as Aw according to the Köppen classification, characterized as tropical with alternating wet and dry seasons, and Semi-dry Tropical of South-Central Brazil (Tr*scb) according to the Novais and Machado (2023), whit an average annual rainfall of 1374 mm and an average annual temperature of 24.7ºC (Fuzzo et al., 2024).

The NASA Power system (Stackhouse et al., 2015) was developed to provide meteorological information for direct use in architecture, power generation and agrometeorology. It compiles information from multiple direct data sources and those derived from gridded data systems. For example, real-time daily data for temperature and relative humidity are obtained from the Global Model and Assimilation Office (GEOS-4) system, and rainfall data are obtained from the Global Precipitation Climate Project (Maldonado Júnior et al., 2019).

Data between 1981 and 2021 was retrieved, via http requests. The data used were total daily rainfall (mm) and average, minimum and maximum daily temperatures (ºC). Virtual stations (Figure 1) were used for all 66 municipalities in the Triângulo Mineiro region, Brazil.

The climatological water balance was calculated according to Thornthwaite and Mather (1955). Soil water storage (STR), water deficit (DEF) and water surplus (SUR) of the soil-plant-atmosphere system were estimated.

To determine potential evapotranspiration, the equation proposed by Thornthwaite (1948) was used (Equations, 1, 2, 3, 4 and 5), which presents good accuracy in estimating PET for the region (Rosa et al., 2023).

PET = ETp × Cor (1)

Cor = ND 30 × N 12 (2)

I =   0,2 × Tn 1,514 (3)

ETp = -415,85 + 23,24 × T - 0,43 × T 2 ,   f o r   T   ≥ 26,5 ° C (4)

ETp = 16 × 10 × T I ,   f o r   0 ° C   ≤ T   ≤ 26,5 ° C (5)

In which:

PET - potential evapotranspiration (mm day^-1);

The calculated water balance uses the available water capacity (AWC) of 100 mm, which is the value that represents the characteristics of the soils in the region (Latossolo) and the used value in studies for soil moisture (Novais et al., 2018; Lorençone et al., 2022).

The moisture indices proposed by Thornthwaite (1948) became popular among the scientific community and are a widely accepted climate classifier and were therefore adopted to characterize the places studied as humid and dry. The calculations of aridity, water and humidity indices were determined according to Equations 6, 7 and 8.

I w = SUR PET × 100 (6)

I a = DEF PET × 100 (7)

I m = I w   - 0,6 × I a (8)

In which:

I_w - water index;

To classify the region's climate, Thornthwaite's climate classification (1948) was used based on the moisture index (Table 1), for each city studied.

Table 1

Climatic types		Moisture index (Im)
Class	Type
A	Perhumid	100 ≤ Im
B4	Humid	80 ≤ Im < 100
B3	Humid	60 ≤ Im < 80
B2	Humid	40 ≤ Im < 60
B1	Humid	20 ≤ Im < 40
C2	Subhumid	0 ≤ Im < 20
C1	Dry subhumid	-33,3 ≤ Im < 0
D	Semi arid	-66,7 ≤ Im < 33,3
E	Arid	-100 ≤ Im < -66,7

Source: Thornthwaite (1948). Elaborated by the authors (2023).

The representative concentration pathways (RCPs) receive their names according to the levels of radioactive forces, in W/m² (Hartin et al., 2015). In this work, the RCP 2.6, RCP 4.0, RCP 6.0 and RCP 8.5 scenarios were used to analyze projections for the end of the 21st century, between the years 2081 and 2100, using the BCC-CSM2MR global climate model developed in the Beijing Climate Center (BCC), obtained by the WorldClim (2023).

The moisture indices calculated in the various tested scenarios made it possible to generate maps using the interpolation method, with the IDW model (Valjarević et al., 2022), nearest neighbor and resolution of 0.25° (25 km), using the Qgis software.

The moisture index proposed by Thornthwaite (1948) is based on the relationship between water availability and evapotranspiration, considering factors such as temperature and rainfall (Souza et al., 2013). Based on this, a current characterization of these factors, with historical data, becomes necessary.

The air temperature in the Triângulo Mineiro region showed a pattern in monthly variation (Figure 2). Monthly air temperature values varied between 16.2 and 28.0°C, with an average annual air temperature of 22.4°C (±1.02ºC). High air temperatures in the region occurred between January and March and from September to December, with an average of 23.9°C, with October being the hottest month in the region with 25.0°C. However, the air temperature decreases between April and August, with an average of 20.3°C, July standing out with an average temperature of 19.0 ºC.

Figure 2 Source: The authors (2023).

The average annual rainfall accumulated in the Triângulo Mineiro was 1,318.8 mm (± 52.0 mm) (Figure 2 and Figure 3A). The annual distribution of precipitation in the region varies between months, with rainfall concentrated in the hottest periods of the year. The lowest rainfall values were observed from May to September. However, rainfall increased from January to March and from November to December, representing 91.6% of the total volume of rainfall for the year in the region.

Figure 3 Source: The authors (2023).

According to Sanches et al. (2017) the reduction in rainfall and relative humidity in the months of October, together with the increase in temperatures (maximum, average and minimum) are directly related to the permanence of tropical atmospheric systems over the Brazilian continent, that is, atmospheric blocking, which prevents the penetration and advance of polar systems. Corroborating the results, Fuzzo et al. (2023) carried out rainfall mapping for the southern region of the Triângulo Mineiro, Brazil, with 20 years of rainfall data, and identified periods of rainy seasons, with emphasis on the month of January, and dry seasons, especially the month of July.

The annual potential evapotranspiration (PET) for the Triângulo Mineiro region is 1,123.74 mm (±149.2 mm) (Figure 3A), with greater dispersion of data in relation to rainfall. The PET determines the demand for water in each region, being widely used in different fields (Xiang et al., 2020), such as hydrology and climatology. Regarding climate characterization, the high PET values present in the region directly interfere with the water balance in the soil, altering the water deficit and surplus in the area (Figure 3B).

The interannual distribution patterns of water surplus (SUR) and water deficit (DEF) in the studied municipalities were similar (Figure 3B). The DEF for the region presented an average of 195.04 mm (± 77.54 mm) and the SUR average of 391.04 mm (± 113.71 mm). Figure 4 shows the spatial distribution of the water deficit and surplus for the Triângulo Mineiro region.

Figure 4 Source: The authors (2023).

The increase in DEF and reduction in EXC in the Triângulo Mineiro region occurred in the western direction (Figure 4). The highest water deficit values were observed as latitude decreased and longitude increased. This region is known as Pontal do Triângulo Mineiro, Brazil, and is characterized by high temperatures and low thermal amplitude (close to 2ºC) (Mello et al., 2007).

The topographic decline towards the Paraná river (at the junction of the Parnaíba and Grande rivers) directly influences the temperature distribution and consequent evapotranspiration in the study area. At the top of the relief located to the east, temperatures are below 19ºC on average throughout the year, while in Pontal, to the west, the altitude is below 400 m and the continentality causes the average annual temperature to rise to 26ºC (Novais et al., 2018). Therefore, such conditions directly affect the climate classification of the region.

According to the humidity index (Iu), the Triângulo Mineiro region presented five types of climates: that is, B1 (humid), B2 (humid), B3 (humid), C2 (subhumid) and C1 (dry subhumid) (Figure 5). The greater number of climatic subtypes is a characteristic of the Thornthwaite classification system due to its greater sensitivity (Phumkokrux; Trivej, 2024).

Figure 5 Source: The authors (2023).

Subtype B1 has the greatest significance in the region, covering 40.70% of the total area, while B3 had the lowest presence, covering only 2.12% of the Triângulo Mineiro region. The B1 subtype is characterized by being present in subtropical climate regions with a humid climate (Karim et al., 2021), characteristic of the study region. In agreement with the results obtained for soil water deficit, subtypes C1 and C2, characterized by being subhumid and dry subhumid, covered a large part of the western region of the Triângulo Mineiro, as previously described.

In a study developed by Lorençone et al. (2022), which adopted the moisture index, it was observed that the state of Mato Grosso do Sul presented four types of climates, namely B1 (humid), B2 (humid), B3 (humid) and C2 (subhumid). The authors further report that it was already expected due to the sensitivity of Thornthwaite's classification criteria.

In relation to possible climate change scenarios, between the period 2081-2100, the end of the 21st century, there was variation in the distribution of the moisture index in the Triângulo Mineiro region (Figure 6).

Figure 6 Source: The authors (2023).

The main changes in the current scenario (Figure 2), with the different climate scenarios, were the significant reduction of the wetter climate classes and the advancement of the drier subclasses, starting from the first scenario studied (RCP 2.6). Future changes in climate will be characterized by a significant contraction in the humid region and an expansion of the arid/humid transition zones (Ma et al., 2019; Baronetti et al., 2022), like what was observed in the Triângulo Mineiro region.

In work carried out by Sanches et al. (2017) on the effect of climate change on the cities of Uberaba (1961-2015) and Capinópolis (1971-2015), located in the study region of this work, the authors suggest that rising temperatures are affecting the decrease in rainfall and causing longer dry seasons. Furthermore, the delay in the rainy season combined with the expansion of cultivation areas, population growth and rising temperatures has severely affected the availability of water resources and worsened water stress.

When analyzing climate change scenarios, a reduction in moisture index classes is observed (Table 2). In the RCP 2.6 scenario, four classes are present, with a substantial reduction in humid subtypes for the region.

Table 2

	B3	B2	B1	C2	C1
	Humid	Humid	Humid	Subhumid	Dry subhumid
Current (%)	2,12	20,57	40,70	28,21	8,4
RCP 2.6 (%)	-	2,14	44,14	24,14	29,58
RCP 4.5 (%)	-	-	23,04	36,55	40,42
RCP 6.0 (%)	-	-	9,35	45,29	45,37
RCP 8.5 (%)	-	-	4,46	25,87	69,68

Source: The authors (2023).

The RCP 4.5 and 6.5 scenarios were similar, presenting three classes, with an increase in subhumid classes and a predominance of subtype C1, with values greater than 40.0% of the area (Table 2). Rahimi et al. (2019), in a study for Southwest Asia, observed that the changes in moisture indices proposed by Thornthwaite's classification (Thornthwaite, 1948), because of climate changes associated with the RCP 4.5 and RCP 8.5 scenarios, suggest a general trend of drying and warming in most of the region, compared to historical conditions.

For the RCP 8.5 scenario, characterized as one with high levels of greenhouse gases in the atmosphere, there is a great predominance of class C1 - Dry subhumid, with almost 70.0% of the region occupied by this subtype. This highlights a major concern, as the reduction in humidity in the region could directly impact the main economic activity in the region, agriculture, which could lead to a decrease in local agricultural productivity (Srivastava et al., 2018).

Feng et al. (2014), in a more in-depth analysis, suggest that changes in precipitation played a slightly more important role in causing changes in climate type during the 20th century. However, projected changes in temperature play an increasingly important role and predominate over changes in climate type as warming becomes more pronounced in the 21st century. In line, Sylla et al. (2015) predict that global warming will increase evapotranspiration and the frequency and intensity of extreme climate events in the 21st century, altering soil water storage, which was in fact observed in the projections modeled in the present work.

Given this, the Brazilian Cerrado may be directly affected by climate change, as the biome could face direct consequences in terms of biodiversity, water resources, agriculture and the regional economy. Environmental conservation will be essential to mitigate adverse effects, requiring effective environmental management policies, sustainable agricultural practices and collaborative efforts between governments, local communities and environmental organizations. Furthermore, it is essential to promote public awareness about the importance of implementing measures to reduce greenhouse gas emissions to further minimize the impacts of climate change on this ecosystem.