Regional and seasonal variability in human thermal stress in Poland

The study objective was to determine the spatial and temporal variability of Poland’s bioclimatic conditions and to designate heat-stress regions with the Universal Thermal Climate Index (UTCI) application. The study was based on daily data from the multiannual period 1966–2021 obtained for 37 stations in Poland, provided from the resources of the Institute of Meteorology and Water Management–National Research Institute (IMGW-PIB). The aforementioned data provided the basis for the calculation of the Universal Thermal Climate Index (UTCI). The study revealed high variability of bioclimatic conditions in Poland, both in temporal and spatial terms. Bioclimatic regions characterised by the different occurrence of heat stress were distinguished and characterised. Regions in the south-west and west of Poland proved the most favourable in bioclimatic terms, with the highest number of days with no thermal stress. In these regions, the highest UTCI values were observed, while the lowest were recorded in the northeast of Poland and at the east coast of the Baltic Sea. Among unfavourable biometeorological conditions, the ones causing hypothermia have so far occurred more frequently than the ones causing overheating of the human organism. However, UTCI has increased during the study period; therefore, under proceeding global warming, an increase in the frequency of occurrence of heat stress can be expected. So far, the most intensive increase in UTCI values in Poland was recorded in spring and the weakest in winter.


Introduction
Contemporary climate warming raises no doubts and has been intensifying since the early twenty-first century (IPCC 2021). The changes are manifested in the increasingly frequent occurrence of thermal extremes in the form of heat waves (Shevchenko et al. 2014;Spinoni et al. 2015;Tomczyk et al. 2022). A characteristic feature of recent years has also been recording high mean annual temperatures globally. Since 1880, ten warmest years occurred in the twenty-first century, and the eight warmest years have been the last 8 years (2014)(2015)(2016)(2017)(2018)(2019)(2020)(2021) (NOAA 2022). The occurring changes are also related to a change in bioclimatic conditions. Research from recent years was conducted in Poland points to increasingly frequent summer occurrence of conditions associated with heat stress, including extremely strenuous conditions (Tomczyk and Owczarek 2020;Krzyżewska et al. 2021). In winter, increasingly seldom occurrence of conditions related to cold stress has been observed. Such changes have particularly been evident in the case of days with strong and very strong cold stress (Wereski et al. 2020;Kuchcik 2021a, b;Owczarek and Tomczyk 2022).
The most appropriate way to characterise atmospheric conditions from the point of view of the functioning of the human organism is to consider many meteorological elements in the form of biometeorological indices. This type of research has been conducted in Poland at a local (Nidzgorska-Lencewicz 2015; Rozbicka and Rozbicki 2018; Mąkosza 2021), regional (Kolendowicz et al. 2018;Miszuk 2021), and national scale (Tomczyk and Owczarek 2020;Wereski et al. 2020;Krzyżewska et al. 2021;Kuchcik et al. 2021a, b;Owczarek and Tomczyk 2022). The cited studies determined bioclimatic conditions based on the Universal Thermal Climate Index. A somewhat different approach was adopted by Błażejczyk and Matzarakis (2007), who conducted research based on other indices, e.g. physiological subjective temperature and physiological equivalent temperature. Similar research, based on different biometeorological indices, has been conducted in many regions of Europe (Matzarakis et al. 2014;di Napoli et al. 2018;Pecelj et al. 2020) and around the globe (Abdel-Ghany et al. 2013;Bal and Matzarakis 2022).
The cited studies showed considerable variability of bioclimatic conditions in Poland. The situation encouraged an attempt of biometeorological regionalisation of Poland. To date, Polish literature has provided two regional divisions performed in the second half of the twentieth century. In the 1980s, a bioclimatic regionalisation of Poland was proposed by Kozłowska-Szczęsna (1986, 1991. It was based on the frequency of occurrence of selected characteristic days. The author designated six regions, and in three of them, subregions were designated additionally. Based on the analysis of biothermal conditions, Błażejczyk (2003 modified the proposed regionalisation of Poland. The new division covered eight regions with no designation of subregions. In the times of progressing climate change, it is justified to perform valid characteristics of bioclimatic conditions in Poland and verify previous Poland's division into regions. Therefore, the study objective is to determine the spatial and temporal variability of bioclimatic conditions in Poland and designate regions of heat stress with the application of the Universal Thermal Climate Index (UTCI).

Data and methods
The study was based on daily data from the multiannual period 1966-2021 obtained for 37 stations in Poland, provided from the resources of the Institute of Meteorology and Water Management-National Research Institute (IMGW-PIB) (Fig. 1). The study employed data from 12:00 UTC such as air temperature (°C), relative humidity (%), wind speed (m•s −1 ), and total cloudiness (okta).
The aforementioned data provided the basis for the calculation of the Universal Thermal Climate Index (UTCI). The index is defined as equivalent air temperature at which in reference conditions, the basic physiological parameters of the organism would adopt the same values as in real conditions. It is a one-dimensional value reflecting the response of the organism to multidimensionally described meteorological and physiological information (Błażejczyk et al. 2012;Bröde et al. 2012). It is assumed that heat exchange between man and the surroundings depends only on air temperature (Ta) at a constant level of the remaining meteorological parameters (Błażejczyk et al. 2012;  . The index is based on the multimodal thermoregulation model of the human organism by Fiala (Fiala et al. 1999(Fiala et al. , 2001, covering two heat exchange regulation subsystems: passive and active. UTCI values are a measure of heat stress of the organism and are expressed in °C. The calculation of UTCI values employed the BioKlima 2.6 package  with implemented computation algorithms with the application of the sixth-degree exponential function (Błażejczyk et al. 2013a).
The obtained data provided the basis for the calculation of mean UTCI values for a year, seasons of the year (i.e. spring: March-May, summer: June-August, autumn: September-November, winter: December-February), and for the multiannual period 1966-2020 (in the case of winter 1966/1967-2020/2021). Then, changes in the calculated values in the multiannual period were analysed. The rate and trend of changes were assessed using linear regression, and the statistical significance of trends was verified with a Student t-test. Moreover, absolute minimum and maximum UTCI in particular stations were determined for the entire multiannual period.
Next, based on UTCI values, classification of days was performed following the scale of assessment of heat stress of the organism prepared by the authors of the index (Table 1).
A division of Poland into regions of occurrence of heat stress was then performed. The division was based on a 55-year course of daily UTCI values in the analysed stations. The stations were grouped employing the Ward minimum variance method, one of the hierarchical grouping of multivariable objects most frequently applied in atmospheric sciences (in this case, the variables were UTCI values at 12:00 UTC in repetition 55 × 365 = 20,075) (Wilks 2011). As a hierarchical method, it begins with single-member groups and merges two groups at each step, until all objects are in a single group. The criterion for choosing which pair of groups to merge at each step is minimising the sum of squared distances between the points and the centroids of their respective groups, summed over the resulting groups. The graphical results of the Ward method is a tree diagram, which illustrates every stage of hierarchical clustering. Problem appears to decide which stage of clustering should be chosen as the final solution. Theoretically, the principal goal is to find the level of clustering that maximises similarity within clusters and minimises similarity between clusters; however, in practice, the best number of clusters is usually not obvious. Wilks (2011) postulates that determining the number of groups requires a subjective choice that depends on the goals of the analysis.
The next stage involved the characteristics of the designated regions. First, the analysis covered the average annual course of UTCI in particular regions. Moreover, the regions were compared in terms of parameters such as average, maximum, and minimum UTCI value in a year and particular seasons of the year, as well as the average annual number of cases in particular classes of heat stress.

Characteristics of bioclimatic conditions in Poland
In the period 1966-2020, mean annual UTCI in Poland was 6.0 °C (Fig. 2, Table A). The index value, however, was spatially variable, i.e. it decreased from the southwest and west towards the north and northeast. In particular stations, mean UTCI varied from 1.8 °C in Łeba to 9.0 °C in Opole.
In the analysed multiannual period, high year-to-year fluctuations of UTCI were recorded, although the variability was similar over the majority of the area, as suggested by the value of the standard deviation (in approximately 81% of the stations below 2.0 °C). The range of fluctuations of the standard deviation varied from 1.1 °C (Zakopane) to 2.8 °C (Suwałki). The lowest mean annual UTCI value in Poland was recorded in 1980. It reached 3.0 °C (Fig. 2, Table A). In that year, the lowest value in the analysed multiannual period was recorded in the highest number of stations. In the said year, the lowest mean values were recorded in the northeast of Poland (Suwałki − 3.0 °C) and the highest in the southwest, in the Oder River valley (Racibórz 6.8 °C). Equally low UTCI was recorded in 1970UTCI was recorded in , 1978UTCI was recorded in , 1985UTCI was recorded in , and 1987 with an average of 3.4 °C, 3.7 °C, 3.8 °C, and 3.5 °C, respectively.
The highest mean annual UTCI values for the entire area were recorded in 2018. It reached 8.9 °C (Fig. 2, Table A). Like in the case of 1980, high spatial variability of bioclimatic conditions was also observed. The lowest values were recorded on the east coast (Łeba and Ustka 4.7 °C) and in the northeast of Poland and the highest in the southwest, along the Oder River valley (Opole 11.3 °C). In that year, the highest index value in the analysed multiannual period was recorded in the highest number of stations. Equally high  2000, 2014, 2015, 2019, and 2020. It is worth emphasising that in 81% of stations, the highest mean annual index value occurred in the twenty-first century. The last years of the analysed period are usually years in which UTCI exceeded the mean value from the multiannual period. Except for two stations (Ustka, Racibórz), an increase in UTCI was recorded ( Table A). The most intensive changes occurred in the northeast of Poland. The increase in UTCI varied from 0.04 °C/10 in Kalisz to 1.41 °C/10 years in Suwałki. Changes exceeding 1.0 °C/10 years were also recorded in Mława and Świnoujście. In 89% of stations, the recorded changes were statistically significant.
In spring in the years 1966-2020, mean UTCI for the entire area reached 6.1 °C (Fig. 3, Table A). The calculated value showed spatial variability and decreased from the southwest and west towards the north and northeast of the country. In particular stations, mean seasonal UTCI value ranged from only 0.7 in Łeba to as much as 9.2 °C in Opole. In spring, higher year-to-year variability of UTCI was observed than in the case of the mean annual value, although the variability was similar over the majority of the area, as suggested by the values of the standard deviation, the highest in the northeast of Poland (Suwałki 3.3 °C). In the analysed period, the lowest mean UTCI for the entire area was recorded in 1970, reaching 1.4 °C. In that season, the lowest value in the analysed period was recorded in the highest number of stations. In particular stations, it ranged from − 2.6 in Suwałki to 8.1 °C in Racibórz. In the said season, values below 0.0 °C covered north and partially central Poland. Considerably higher values were recorded in the south of the country, particularly along the Oder River valley. Equally low values (below 2.0 °C) were recorded in 1980 and 1987. In only three stations (Kłodzko, Racibórz, Ustka), the lowest UTCI value in the multiannual period was recorded in the twenty-first century. Like in the case of the entire year, the highest mean value was recorded in 2018. It reached 9.9 °C. In that year, the highest index value in the analysed multiannual period was recorded in the highest number of stations. Apart from the Baltic coast, bioclimatic conditions were approximate and ranged between 10.0 °C and more than 12.4 °C in Toruń. In the analysed period, an increase in UTCI was observed in spring (except for Racibórz and Ustka). The most intensive changes were observed in the north of Poland. The maximum increase in UTCI was recorded in Suwałki (1.51 °C/10 years). In 86% of stations, the observed changes were statistically significant.
In summer in the analysed multiannual period, mean UTCI for the entire area was 20.5 °C (Fig. 4, Table A). The lowest index values were recorded in the northern regions (Łeba 16.1 °C) of the study area and the highest in the southern ones, particularly in the south-western ones (Racibórz 22.7 °C). It is worth emphasising that except for the northern regions, bioclimatic conditions were approximate. The year-to-year variability of conditions was similar throughout the area, and the values of the standard deviation in a large majority of stations were the lowest among all seasons of the year. In most stations, the lowest mean index value for summer was recorded in 1980, reaching 16.5 °C. In particular stations, it varied from 13.5 in Suwałki to 19.1 °C in Racibórz. Equally low values were observed in 1974 and 1978. The highest mean value in the multiannual period for the entire area was recorded in 2019. It reached 24.2 °C. The year in which the maximum seasonal value in the multiannual period was recorded in the highest number of stations was 1992, with an average of 24.1 °C. In particular stations, the value ranged from 18.0 in Łeba to 27.2 °C in Racibórz. In the analysed multiannual period, except for the station in Ustka, an increase in UTCI in summer was recorded, with the greatest intensity in Suwałki (1.06 °C/10 years). The study showed that in 89% of stations, the changes were statistically significant.
In autumn, bioclimatic conditions were similar to the conditions occurring in spring, and mean seasonal UTCI for the entire area reached 6.4 °C (Fig. 5, Table A). An increase in the values occurred from the northeast and north towards the southwest and west of Poland. The lowest mean seasonal UTCI values were recorded in Suwałki (2.1 °C) and the highest in Słubice (9.4 °C). The yearto-year variability of the index over the major area was approximate, particularly in the western regions. The highest values of the standard deviation were recorded in the east of Poland, with a maximum in the north-eastern regions (3.2 °C). In the entire multiannual period, the lowest mean value for the study area occurred in 1973. It reached 2.7 °C. The lowest values were determined in the north-east of Poland (Suwałki − 5.6 °C) and at the east coast (Łeba − 0.9 °C) and the highest in the south-western regions (Racibórz 7.9 °C). In the aforementioned season, the lowest values throughout the multiannual period were recorded in the highest number of stations. Similar conditions were also observed in 1971, 1972, 1974, 1980, and 1998. The highest UTCI was recorded in 2006, with an average value for the study area of 10.5 °C. The highest index values occurred in the west and south-west of Poland. In particular stations, the mean value in the season varied from 5.8 in Suwałki to 14.2 °C in Świnoujście. Equally high index values were observed in 1982, 2000, 2014, and 2018. In the studied multiannual period (except for Racibórz), an increase in UTCI values in autumn was determined, the most intensive in the north-east, north-west, and south-east of Poland, with a maximum in Suwałki (1.45 °C/10 years). In 84% of stations, the changes were statistically significant.
During the study period, mean winter UTCI for the entire area was − 9.3 °C (Fig. 6, Table A). The lowest values were recorded in regions east of the Vistula River and at the east coast, and the minimum was observed in Suwałki (− 15.0 °C). The southern and western regions showed the highest index values with a maximum in Zakopane (− 3.3 °C). In the studied multiannual period, the lowest mean value for the season in the study area was recorded in winter 1969/1970, reaching − 13.9 °C. Particularly, low values were recorded in the north-east of Poland and at the boundary of the Mazurian Lake District and Mazowiecka Lowland with a minimum in Suwałki (− 21.0 °C). Higher winter values were recorded in the southern and western regions with a maximum in Zakopane (− 5.8 °C). The highest UTCI value for the entire area was observed in winter 2000/2001 (− 5.4 °C). The maximum value in the multiannual period in the highest number of stations, however, was determined in the season 2019/2020, with an average for the entire area lower by 0.1 °C, reaching − 5.5 °C. In the said season, UTCI values in particular stations varied from − 11.6 in Łeba to 1.0 °C in Świnoujście. Low index values were recorded at the east coast and in the south-east of Poland and higher in the belt of south and west Poland. In the studied multiannual period, except for three stations (Kalisz, Racibórz, Ustka), an increase in UTCI in winter was observed. Like in the remaining seasons of the year, the most intensive increase occurred in the south-east and north-west of Poland, with a maximum in Suwałki (1.65 °C/10 years). The study evidenced that in 68% of stations, the changes were statistically significant.

Regions of human heat stress
The final result of the study on the spatial and temporal variability of heat stress is the division of Poland into regions of occurrence of heat stress, based on a 55-year course of daily UTCI values in 37 stations. The division of stations into groups (by Ward method) proved spatially uniform, i.e. stations in the groups at all stages of grouping were neighbouring. Four basic regions were designated in which UTCI values are characterised by similar values and temporal course (Figs. 7 and 8): north-western region (NW) with the designated coastal subregion (dashed line in Fig. 7), north-eastern region (NE) with the coastal subregion, south-western region (SW) with the eastern and  UTCI value for summer in the SW region exceeds 21 °C, reaching approximately 24 °C in early August. The average for winter is − 6.7 °C, with values of approximately − 8 °C in the coldest January. The NW region proved the coldest. There, almost all UTCI characteristics show the lowest values, and the average in the coldest month is approximately − 12.5 °C. The greatest differences between the regions occur in the coldest season. Mean winter UTCI value for the NE region (− 11.5 °C) is almost twice higher than that for the SW region (− 6.7 °C), like the mean winter maximum value (12.9 °C in SW and 6.3 °C in NE). In spring and autumn, differences between regions are lower, and in summer, the average thermal impressions are similar in all regions, with differences of approximately 2 °C in favour of regions' warmest in summer, namely SW and NW (Fig. 8, Table 2).
Maximum UTCI values in Poland recorded in the period 1966-2021 exceeded 40 °C (absolute maximum 43.1 °C on 09 August in Słubice, NW region), and minimum values fell below − 50 °C (absolute minimum − 56.3 °C on 31 December 1978 in Suwałki, NE region). Only in the warmest SW region, no UTCI values lower than − 50 °C were recorded in the studied 55-year period. The absolute minimum UTCI value for the SW region was − 48.7 °C, recorded on 17 January 1972 in Kraków.
The statistics of occurrence of UTCI in classes of heat stress (Table 3) show that Poland usually features conditions with no thermal stress (from 142.6 days in the SW region to 126.2 days in the NE region). Very strong heat stress is sporadically recorded (from 38 to 46 °C), or extreme cold stress (< − 40 °C). Strong heat stress (32 to 38 °C) occurs on average on several days in a year (from approximately 7 days in the SW region to 3 days in the NE region), and extreme cold stress (− 27 to − 40 °C) is observed on average on approximately 5 days in a year in the NE region and on less than 2 days in the SW and NW regions. Strong or moderate cold stress usually occurs in the NE region (in total on approximately 136 days in a year) and most seldom in the SW region (approximately 111 days in a year), where mild cold stress is relatively frequently experienced. Moderate heat stress usually occurs in the SE and CSE region (on average on 28-29 days in a year).

Summary and discussion
The study revealed high variability of bioclimatic conditions in Poland, both in temporal and spatial terms. The lowest UTCI was recorded in the north-east of Poland and at the east coast of the Baltic Sea. The highest index values were observed in south-western and western regions of the country. Considering particular stations, the lowest values usually occurred in Łeba, Ustka, and Suwałki and the highest in Racibórz and Opole. The obtained results are in accordance with earlier research conducted in Poland (Mąkosza 2013;Krzyżewska et al. 2019;Błażejczyk et al. 2021;Kuchcik et al. 2021a, b). Low UTCI in the northeast of Poland is determined by low air temperature values. Except for mountain regions, the said area is the coolest in the country (Owczarek and Filipiak 2016;Ustrnul et al. 2021;Tomczyk 2022). Low index values at the east coast are related to higher than average wind speeds (Wibig 2021, Fig. 7 Bioclimatic regions in Poland based on daily UTCI values; solid lines mark the division into four main regions, and dashed lines indicate borders of subregions 2022). Regions with the highest UTCI are areas with high air temperature and lower wind speeds (Owczarek and Filipiak 2016;Ustrnul et al. 2021;Tomczyk 2022;Wibig 2021Wibig , 2022. Relatively high values in submontane and montane areas result from lower than average wind speed in exposed areas and those under the effect of foehn wind. The effect of wind on the intensification of convection heat losses from the organism is particularly manifested at very low air temperature (Tikuisis and Osczevski 2003). Even wind with low speed (approximately 3 m•s −1 ) increases the risk of hypothermia while being outdoors for a long time with no appropriate protection, and at a temperature below − 20 ℃, the risk of frostbite of exposed body parts within 10 to 30 min increases (Environment Canada 2020; Owczarek and Tomczyk 2022).
High spatial variability of UTCI related to regional variability of climatic conditions in Poland permitted the designation of bioclimatic regions characterised by the different occurrence of heat stress, particularly in the cool season of the year. Regions in the south-west and west of Poland proved the most favourable in bioclimatic terms, with the highest number of days with no thermal stress. The division is largely determined by winter thermal conditions, in Poland characterised by the longitudinal arrangement of isotherms and a decrease in temperature from the west to the east (Tomczyk and Bednorz 2022). The least biometeorologically favourable NE region is relatively windy and the coldest in winter.
A previous division of Poland into bioclimatic regions (Kozłowska-Szczęsna 1991;Błażejczyk 2003) presents an image different from that proposed in this study because it is based on different criteria, namely the frequency of occurrence of selected characteristic days with stress conditions, and supplementarily, the biothermal conditions. The coldest north-east region has a smaller spatial range, and the extensive central region was combined with the central-western part of Poland. Kozłowska-Szczęsna (1991) and Błażejczyk (2003) designated coastal areas as a separate region. In this study, they constitute subregions NW and NE. In the analysed years in all regions, high year-to-year variability of mean annual and mean seasonal values was recorded. In each of the analysed time intervals, the greatest fluctuations, as suggested by the value of the standard deviation, occurred in the north-east of Poland. In most stations (in 68% of stations), the greatest UTCI variability was observed in winter and the lowest in summer (in 84% of stations).
Whereas conditions causing hypothermia of the human organism have so far occurred more frequently among unfavourable biometeorological conditions than those causing its overheating, the study period featured changes in bioclimatic conditions that in the vast majority of stations showed an increase in UTCI and therefore an increase in the frequency of occurrence of heat stress. The most intensive increase in UTCI values was recorded in spring and the weakest in winter. In the case of mean  Table 3 Mean number of days with heat stress categories in bioclimatic regions indicated in Fig. 7. The warmest conditions among the regions are marked red, and the coldest is purple. annual values and mean values for spring, only Racibórz and Ustka showed an inconsiderable decrease in UTCI. In summer, a decrease in the index value occurred in Ustka, in autumn in Racibórz, and in winter in Kalisz, Racibórz, and Ustka. Changes recorded for the mean annual and mean summer value in the highest number of stations were statistically significant (in 89% of stations). Changes in the winter average were statistically significant in the lowest number of stations (in 68% of stations). A similar trend of changes was observed in earlier studies from the territory of Poland (Kuchcik et al. 2021a, b). The cited research also showed that the most intensive changes in UTCI were recorded in the north-east of Poland, and a lower rate of increase may result from the analysed multiannual period not covering the recent years. Similar levels of change were also determined in research from west Poland (Mąkosza 2013), from the coast of the Baltic Sea (Owczarek et al. 2019), and from the Polish-Saxon border (Miszuk 2021). Further changes should be expected in the following years. According to Błażejczyk et al. (2013b), in the years 2000-2100 in Warsaw, the number of days with heat stress will increase (according to UTCI) at a rate of 0.9 day/10 years. In other regions, a similar trend of changes in the frequency of days with heat stress can be expected. As evidenced by Tomczyk et al. (2022), by the end of the twenty-first century, an increase in the frequency of hot days should be expected, related to considerable heat stress. The greatest changes are forecasted for south and central Poland.