CLIMATE CHANGE AND RURAL WORKERS THERMAL COMFORT: HISTORICAL AND FUTURE IMPACTS

The aim of the present research was to propose a bioclimatic mapping to classify the thermal comfort and discomfort of rural workers within the state of Minas Gerais, considering historical and future scenarios. Monthly historical series (1976-2014) of minimum, mean and maximum temperature-humidity index (THI), determined through the values of air temperature (minimum, mean and maximum) and relative humidity from 48 weather stations located in the state of Minas Gerais were used to analyze the trends through Mann-Kendall and linear regression assays. The bioclimatic mapping of human comfort, obtained via geostatistical analysis, was developed as a function of the minimum, medium and maximum THI for the historical period (1976-2014) and future scenario (2024). Results indicate an overall trend of increasing in thermal discomfort conditions throughout the mesoregions of the state of Minas Gerais during the weather seasons, being more incisive in summer and spring.


INTRODUCTION
Climate change can generate negative impacts on human beings, especially workers in some agribusiness activities that require higher metabolic activ ity.Impacts of the thermal environ ment on humans have been widely studied because of the harmfu l effects on performance and health (Ou et al., 2014) and, under extreme conditions, leading to death (Loughnan et al., 2010).
Under heat stress conditions, humans present changes in physiological parameters, such as increases in heart rate, body temperature, blood pressure, and sweat production (Shen & Zhu, 2015).Heat-related disorders such as sunstroke, cramps and exhaustion have also been reported (Zheng et al., 2012).Extreme conditions of cold stress also cause health problems (McMichael et al., 2006).
Indexes of the thermal environ ment have been used to generate bioclimat ic maps applied to the analysis of the effect of the external environ ment on humans (Park et al., 2014).A mong the various indexes, the temperature and humid ity index (THI) (Tho m, 1959) has been widely used to classify the level of thermal comfort of humans, in view of the easy obtaining of meteorological data used as input into the equation.
In this context, the goal of the present study was to propose a bioclimat ic zoning to classify the thermal comfo rt and discomfort of rural workers in the State of Minas Gerais, using the temperature and humidity index (THI), considering a historical period and a future scenario.

MATERIAL AND MET HODS
For the development of bioclimat ic zoning of human co mfort for the State of Minas Gerais, it was used data from 48 weather stations located in the State of Minas Gerais belonging to the National Institute of Meteorology (INM ET) (Figure 1).The database is composed of the monthly average values of minimu m, mediu m and maximu m dry-bulb temperatures (t db,min , t db,mean and t db,max , respectively) and air relat ive humidity (RH).The State of Minas Gerais has an area of approximately 7% of the Brazilian territory, with 582,586 km² and wide climatic variab ility (Tonietto et al., 2006).The average values of annual precipitation vary from 700 to 1,100 mm in the driest regions (north, northeast and east) and from 1,200 to 1,500 mm in the rain ier regions (South, Triângulo Mineiro, northwest and Serra do Espinhaço and Mantiqueira) (Mello & Viola, 2013).In the lower regions of north and east of the State, the monthly average temperatures reach 27°C, while in the higher regions are observed temperatures around 13°C.
The weather stations used in this study are listed in Table 1, and the historical series of t db,min , t db,mean , t db,max and RH were of 39 years, covering the period fro m 1976 to 2014.THI was calculated through the (Equation 1) (Thom, 1958), based on t db,min , t db,mean , t db,max (°C) and dew point temperature (t dp ), that was determined by (Equation 2) (Vianello & Alves, 2012) as a function of the actual water vapor pressure (e, hPa).

Historical series trend analysis
The trends of the historical series of THI min , THI mean , THI max over a period of 39 years, for each studied station, were verified using the non-parametric Mann-Kendall test (Mann, 1945;Kendall, 1975) proposed by Sneyers (1975), and the linear regression analysis, methodologies commonly used (Minuzzi, 2010, Ávila et al., 2014, Tian et al., 2016).
The non-parametric Mann-Kendall test considers that, since there is stability of the time series (Hypothesis H 0 ), the succession of values occurs independently and the probability distribution remains unchanged (simp le random series).
A time series will p resent a tendency of increase or decrease of a certain variable if the value of the Mann-Kendall coefficient is positive (MK> 0) or negative (MK <0), respectively.For this purpose, the Z test should be applied at the 5% level of significance, wh ich provides Z 0.975 = 1.96.It is rejected H 0 if the Mann-Kendall test, │MK│>Z1 -α/2 , is greater than 1.96, indicating a significant trend in the data time series (AVILA et al., 2014).The S statistic was determined by (Equation 3), where n is the number of observations, Xj and Xi are the sequential values of the data and sgn (ϕ) is the signal function.The signal function assumes a value of 1 if ϕ > 0; 0 if ϕ = 0; and -1 if ϕ < 0.
(3) According to Kendall (1975), assuming the hypothesis that the data are identically distributed and independent, the average and the variance of the Z statistic are given by (Equations 4 and 5), respectively.The m is the number of associated classification groups, each one associated with a ti, wh ich corresponds to the THI values of the historical series of THI min , THI mean and THI ma x . (4) (5) The value of Z can be calculated as follo ws: The linear regression analysis was applied to obtain trends through the parametric significance t test over the angular coefficient () (Longobardi & Villani, 2010).This test considered the linear regression between the random variable Y (THI series) and time (X).The trend for a period of 10 years (2024) was calcu lated using the adjusted linear equations, multip lying β by 10.

Bioclimatic zoning
The bioclimatic zoning was carried out through the interval maps of THI min , THI mean and THI max for the State of Minas Gerais, considering the historical period fro m 1976 to 2014, and indicating trends for 2024, a decade after the end of the data historical series.
Statistical analysis and linear regression were performed by the R program (R Develop ment Core Team, 2014).The creation of the maps was made by the ArcGIS for Desktop 10.4 program, through its extensions, Spatial Analyst and Geostatistical Analyst.The spatial dependence of the THI in the State of Minas Gerais was analyzed by means of the semivariogram ad justment using the OLS method, adjusting the spherical model, and the interpolation was carried out using ordinary kriging (Ferraz et al., 2015).
The zones of thermal comfort and discomfort for rural workers were defined considering that the rural workers exert moderate activities, with metabolic rate of 175 W, working stand-up, with movement of arms and legs and the use of light clothes, that is, with thermal resistance of 0.09°C W m -2 (Oliveira et al., 2006).
The thermal environ ments were classified as comfo rt (THI <74), hot (74≤THI <79), very hot (79≤THI<84) and extremely hot (THI≥84) (Lamberts et al., 1997;Oliveira et al., 2006).Thermal discomfort begins in the hot environment and can cause health problems and reduce rural workers' performance.The very hot environment indicates danger and can cause serious effects to health and the extremely hot environment can cause very serious health risks.

RES ULTS AND DISCUSS ION
The spatial distribution of the average historical values  of THI min and THI mean indicate that the thermal environ ment in almost all seasons was classified as a comfort situation (THI <74), except for spring.At this season, the THI mean in some parts of the mesoregion of the Triângulo Mineiro/Alto Paranaíba and part of the boarder of the mesoregions of the Northwest and North of Minas Gerais presented values in the interval between 74 and 79, characterizing the region as hot (Figure 2).Oliveira et al. (2006).
Thus, periods with thermal conditions classified as hot (74 ≤ THI <79) and very hot (79<THI≤ 84) were observed at all seasons in the State of Minas Gerais for THI max , corroborating with Oliveira et al. (2006).Therefore, minor or serious health problems, in addition to the reduction in the performance of rural workers are expected.It is emphasized that, hourly variations of the THI can cause situations of thermal discomfort (Buriol et al., 2015), intensifying the discomfort.
The comfort classification verified for THI mín in the four seasons in the historical period (1976-2014) d id not change in 2024 (Figure 3), although tendencies of increase and reduction of these values were observed (Figures 2A,  2D, 2G, 2J).
The reductions in the occurrence frequency of the condition classified as comfort for the THI mean in the Engenharia Agrícola, Jaboticabal, v.38, n.2, p.173-179, mar./apr. 2018 summer, fall, winter and spring seasons were 6.25%, 0.00%, 0.00% and 2.08%, respectively.These percentages of reduction were consequently added to the condition of the thermal environment classified as hot (Figure 3).Reduction on 6.25% of the situation classified as hot and increase of very hot situation was verified for THI max (Figure 3A) in the summer season.
In the fall, it was observed the decrease of occurrences of THI max values for co mfort conditions of 2.09 (Figure 3B).Consequently, it was observed a 2.09% increase in the ext remely hot classificat ion for THI max when comparing the 2024 scenario with the historical period .

A.
B.
100,00 100,00 0,00 0,00 0,00 0,00 0,00 0,00 100,00 100,00 0,00 0,00 0,00 0,00 0,00 0,00 100,00 100,00 0,00 0,00 0,00 0,00 0,00 0,00  The trend analysis indicates, in general, the increase of the occurrence frequency of the hot classification in summer and spring for THI mean and THI ma x , when comparing the 2024 scenario with the historical period .So, it was verified the increase of the extremely hot condition in autumn and spring.This profile due to climatic changes increases human discomfort especially in summer, fall and spring, in which the sums of THI max occurrences classified as very hot and extremely hot are 85.42%, 54.16% and 58.33 %, (Figure 3), respectively.The increase in human discomfort due to climate change was also verified by Potchter & Ben -Shalo m (2013).Zhang et al. (2014) reco mmended the upper limits of t db and RH of 29.2°C and 50% (THI = 77.1)and 28.0°C and 70% (THI = 77.4)so that 90% of people can be satisfied in circulation spaces in a hot and humid area of China.The satisfaction level of 73% was obtained for t db and RH of 31.0°C and 50% (THI = 79.5) and of 29.5°C and 70% (THI = 79.4).
Changes in the thermal environ ment are related to declining work capacity (Dunne et al., 2013) and to the increased incidence of cardiovascular diseases (Ezekowit z et al., 2013) and mental disorders (Vaneckova & Bambrick, 2013).The absence of actions to mit igate the thermal environ ment on humans can cause reduction of performance and increase of diseases, a situation that can be aggravated by climate changes.
Given the current climatic conditions and trends to the future, strategies to mit igate the thermal effects of the environment on humans must be analyzed, such as the selection of surrounding vegetation (Lee et al., 2016), materials selection (Ragheb et al., 2016), the energy demand for heating or cooling (Van Hove et al., 2015), the design of the building, among others.This knowledge is essential for studies related to the reduction and adaptation to climate change (Hjort et al., 2016).

FIGURE 1 .
FIGURE 1. Location of the weather stations used in the present study.

FIGURE 3 .
FIGURE 3. Variation of the occurrence frequency of THI min , THI med and THI ma x in the seasons of (A) su mmer, (B) fall, (C) winter and (D) spring, for the period fro m 1976 to 2014 and for the year 2024, in the evaluated municipalit ies of the state o f Minas Gerais

TABLE 1 .
Conventional weather stations and their respective elevation.