1 Introduction

The thermal bioclimate affects various aspects of human well-being, encompassing environmental, physical, social and economic perspectives. The favorable outdoor thermal environments promote outdoor living (Nikolopoulou and Lykoudis 2007; Lin et al. 2012) diminishing the reliance on indoor air-conditioning and thereby alleviating the strain on energy efficiency. Such settings encourage the adoption of eco-friendly modes of transportation, such as walking, biking, and public transit. This contributes to the improvement of air quality by curbing pollutant emissions and enhances physical activity reducing the prevalence of chronic diseases (Mueller et al. 2016). Moreover, the presence of individuals in outdoor spaces promotes neighborhood security and social networking. Beyond this, the favorable thermal environments are associated with reduced risk of mental and behavioral disorders (Li et al. 2021), enhance occupational productivity (Bueno et al. 2021) and stimulate the tourism industry (Mora et al. 2018). Conversely, unfavorable thermal environments can pose direct health risks, potentially resulting in heat exhaustion, heat stress, hypothermia and the exacerbation of pre-existing health issues (WHO 2018).

Within a climate context, various configurations can create diverse bioclimatic conditions. Several factors, including topography, geography, urban development, building design, and human activities, interact to modify temperature, humidity, wind patterns, and solar radiation (Charalampopoulos et al. 2013; Feidas et al. 2014; Wang et al. 2021; Qian et al. 2022). Local climate can be significantly influenced by elevation, proximity to water bodies, terrain, the presence or absence of vegetation and the land use patterns. Urban areas can exhibit urban heat islands, with higher temperatures due to heat absorption and radiation from buildings, concrete surfaces, and asphalt (Kim and Brown 2021). Tall buildings can alter wind patterns and shading, impacting ground-level temperature and comfort (Deilami et al. 2018). Human activities, such as agriculture, irrigation, and industrial processes, can modify local climate through heat, moisture, and pollutant release, further changing bioclimatic conditions (Dale 1997).

The assessment of thermal bioclimatic conditions can be accomplished using thermal indices that consider multiple meteorological factors including air temperature, humidity, wind speed and radiation fluxes (de Freitas and Grigorieva 2015). There is a great variety of thermal indices, each developed based on different methodologies considering the number of variables they incorporate, the underlying principles, and their application (de Freitas and Grigorieva 2015, 2017). Thermo-physiological indices such as the Physiologically Equivalent Temperature (PET) (Hoppe 1993; Höppe 1999) and Universal Thermal Climate Index (UTCI) (Fiala et al. 2001, 2012; Bröde et al. 2012), are particularly suitable for evaluating the thermal environment, as they use human thermal energy models in their estimations and incorporate physiological variables (in addition to meteorological ones), such as age, height, weight, activity level, and clothing insulation (de Freitas and Grigorieva 2017; Matzarakis 2021). Both, PET and UTCI are commonly employed for assessing the thermal bioclimatic conditions (Giannaros et al. 2018; Potchter et al. 2018; Di Napoli et al. 2021a). However, especially in the Mediterranean climate of Greece, UTCI has shown greater effectiveness in evaluating pedestrians’ thermal perception (Pantavou et al. 2013) and applicability for operational purposes (Pantavou et al. 2023a) compared to other thermal indices, including PET.

The bioclimate of Greece is examined in studies of large-scale investigations, such as global or Mediterranean-scale, using data at a relatively low spatial resolution (Di Napoli et al. 2021a) or focusing on bioclimatic extremes (Giannaros et al. 2018). The few studies specifically centered on Greece are concentrated to specific regions using data from the network of surface meteorological stations (Bleta et al. 2014; Katavoutas and Flocas 2018). In previous nationwide studies (Matzarakis and Mayer 1997; Matzarakis et al. 1999), data from a low number of surface stations and for one-hour daily values (at 12 UTC) were used, incorporating Predicted Mean Vote and PET for assessing the Greece bioclimate, while their findings require update.

This study aims to examine the thermal bioclimatic conditions in Greece using the ERA5-Land dataset (Muñoz-Sabater et al. 2021) for the estimation of UTCI in the last three decades (1991–2020) at a 0.1° spatial resolution (≈10 km) in order to identify high-risk areas. The UTCI is a universal and widely used thermo-physiological index (Potchter et al. 2018; Di Napoli et al. 2021b) facilitating comparisons across different regions and climates that shows high applicability in the Mediterranean climate (Pantavou et al. 2013, 2023a). Additionally, this study assessed the intensity of two recent heatwaves that occurred in 2021 and 2023. Heatwaves are periods with exceptionally unfavorable thermal conditions exceeding the highest thresholds of heat stress. Studying these conditions provides insights into the human tolerance to heat stress. In this study, the analysis at the national level includes urban areas, which are particularly vulnerable to heatwaves because of the urban heat island effect (Santamouris 2020). This phenomenon intensifies heat stress in densely populated regions, intensifying the challenges faced during extreme heat event. The research on thermal bioclimate analysis can increase awareness about the sustainability of outdoor environment and contribute to the formulation of adaptation strategies to address health risks associated with the thermal environment. Moreover, it can be useful for tourism, urban design and energy efficiency purposes.

2 Materials and methods

2.1 Study area

Greece is located in the eastern Mediterranean Sea (Latitude: 35°N to 42°N, Longitude: 19°E to 28.5°E), has a Mediterranean climate (Csa) (Peel et al. 2007), and is renowned for its complex and diverse topography. The country's topography varies significantly, encompassing mountains, plains, islands, and a long coastline (Fig. 1). The topography of mainland Greece is characterized by key mountain ranges, including the Pindus Mountains in the western part of the country, the Rhodope Mountains in the northern region, and the Taygetos and Parnon Mountains in the southern part (Peloponnese peninsula). The eastern region encompasses the main plains of Macedonia and Thessaly. Several smaller plains are scattered throughout the country including East Macedonia and Thrace (Evros, Serres), Epirus (Arta, Preveza), Central Greece, Western Greece (Mesolongi, Agrinio, Pirgos), and Peloponnese (Argos, Kalamata, Sparta).

Fig. 1
figure 1

Elevation map of Greece. The boundaries show the first level administrative entities and some major cities of Greece 

Full size image

As of the 2021 census, the population of Greece is approximately 10.5 million people distributed as follows: 36.4% in Attica prefecture, 17.1% in the prefecture of Central Macedonia and between 1.9% (North Aegean prefecture) and 6.6% (Thessaly prefecture) in the remaining 11 prefectures (Hellenic Statistical Authority (HSA) 2021) . The capital of Greece, Athens, is located in Attica prefecture in the southern part of the country near the eastern coast of mainland Greece and along the Aegean Sea.

In recent years, Greece experienced two prolonged and extremely intense heatwaves, spanning from July 26th to August 6th, 2021 (12 days duration), and from July 13th to 27th, 2023 (15 days duration). During these periods, the daily maximum air temperature exceeded the threshold of 37.0 °C (Founda and Santamouris 2017) for consecutive days at numerous surface weather stations across Greece (Meteo 2023). On the hottest days August 3rd, 2021 and July 23rd, 2023, the number of stations operated by the METEO unit of the National Observatory of Athens (Lagouvardos et al. 2017) that recorded air temperatures exceeding 40.0 °C was 205 and 180 respectively (Meteo 2023).

2.2 Data

The data were extracted from the ERA5-Land reanalysis dataset (Muñoz-Sabater et al. 2021). The ERA5-Land dataset is produced by reprocessing the land component of the European Centre for Medium-Range Weather Forecasts (ECMWF) ERA5 climate reanalysis. Reanalysis integrates model data with global observations, ensuring a globally comprehensive and coherent dataset grounded in the principles of physics. With its high spatial and temporal resolution, the ERA5-Land dataset proves valuable for a wide range of weather-related applications.

Specifically, data on air temperature (Tair) and dew point temperature at 2 m, 10 m components of wind speed (WS), and surface solar radiation downwards were retrieved at spatial resolution of 0.1 ◦ × 0.1 ◦ (≈10 km) covering the territory of Greece. The data were collected at a temporal resolution of one hour, spanning the period from 1991 to 2020. The retrieved hourly data were used to estimate the UTCI (◦C) (Fiala et al. 2001, 2012; Bröde et al. 2012) which was used to assess the thermal bioclimate. Additionally, gridded population data were used from the statistical office of the European Union (Eurostat 2023) which were adjusted to the ERA5-Land grid using the Climate Data Operators (CDO 2023). This approach allowed the impact assessment of unfavorably thermal conditions on population level.

The UTCI is a thermo-physiological index that provides a comprehensive evaluation of how people perceive their surrounding thermal environment. UTCI is defined as the air temperature of a reference environment (still air, 50% relative humidity– vapor pressure not exceeding 2 kPa, and radiant temperature equal to Tair) under the following assumptions: i.e. walking at 4 km/h (with a metabolic rate of 135 W/m2) and wearing clothing adjusted to prevailing thermal conditions (Błazejczyk et al. 2013). UTCI is valid in all seasons and climates. Its estimations can be assigned to a category of an assessment scale including 10 categories of varying levels of thermal stress (i.e., one indifference category, and five for cold and four for heat stress) and ranging from ‘extreme cold stress’ to ‘extreme heat stress’ (Appendix Table 2).

2.3 Statistical analysis

Summary statistics, such as measures of central tendency, dispersion and extremes, were employed to provide an overview of the human thermal bioclimate based on UTCI during 1991–2020. Initially, the climatology of UTCI was investigated by analyzing the spatial distribution of mean UTCI equivalent temperature at annual, summer (June, July and August) and winter (December, January and February) timescales. This involved also the assessment of the mean daily maximum and minimum UTCI equivalent temperature in summer and in winter, respectively, as well as the monthly frequency of UTCI categories. At the second step, we evaluated the thermal burden across the country, and particularly in key regions encompassing major cities. This included calculating the frequency of UTCI categories to find the percentage (%) of hours categorized as unfavorably warm or cool; defined as hours falling at least within the category ‘moderate heat/cold stress’, i.e., falling within the category of ‘moderate heat/cold stress’ and above/below. This choice was motivated by the suggestion that fatalities in Mediterranean countries often occur in conditions of moderate and strong heat stress (Di Napoli et al. 2018).

Following this, the trend of mean annual UTCI equivalent temperature and of the percentage of hours characterized as unfavorably warm/cool annually and seasonally were examined. Linear regression was employed to determine trend’s slope and the Mann–Kendall non-parametric test was used to assess the statistical significance at the 95% confidence level (Sneyers 1990). At the last step, the two recent heatwaves, which occurred from July 26th to August 6th, 2021 (HW1) and from July 13th to 27th, 2023 (HW2) were assessed by comparing their intensity and persistence to the climatology of the period 1991–2020. For the comparative analysis of the two heatwaves we calculated the mean, mean daily maximum and absolute maximum UTCI equivalent temperature. Moreover, we calculated the accumulated number of hours categorized as unfavorably warm, i.e., with at least ‘moderate heat stress’, over the 15-days period encompassing each heatwave of 2021 and 2023. Additionally, we estimated the distribution of the accumulated number of hours with at least ‘moderate heat stress’ over a rolling 15-day period of the summers 1991–2020 in order to define the relevant climatological distribution. This perspective enabled us to determine the percentile in which the 15-day periods of HW1 and HW2 are positioned, facilitating the evaluation of HW1 and HW2 persistence compared to the climatology.

Furthermore, the UTCI estimations were used to determine the proportion of the affected geographical area and population in order to have an estimation of the thermal-related risk with respect to exposure. The exposure was defined by comparing the population and the UTCI estimations in each grid point. Changes of UTCI values with latitude were examined using linear regression.

3 Results

3.1 Climatology of UTCI

UTCI ranges between -49.8 °C and 51.9 °C [mean ± standard deviation (sd) = 13.3 ± 12.2 °C] encompassing all the ten categories of the UTCI assessment scale, i.e., from ‘extreme cold stress’ to ‘extreme heat stress’. The lowest UTCI equivalent temperature of -49.8 °C was found in the North Aegean region, and the highest UTCI of 51.9 °C in southern Peloponnese. Areas experiencing ‘extreme cold stress’ are identified in East Macedonia and Thrace (Evros), North Aegean, and on some islands in the Aegean Sea.

The mean UTCI for the period 1991–2020 ranges from 7.5 °C to 17.8 °C (mean ± sd = 13.3 ± 2.1 °C; Fig. 2a), classifying thermal stress into two categories of UTCI’s assessment scale i.e., from ‘slight cold stress’ to ‘no thermal stress’. Across the country, thermal stress predominantly falls into the ‘no thermal stress’ category, with the exception of specific areas such as northern mountainous areas of West Macedonia (Pindus) and East Macedonia and Thrace as well as a few islands in Aegean Sea where thermal conditions indicate ‘slight cold stress’. An analysis of UTCI variations across different locations in Greece revealed a decrease of 0.5 °C in mean UTCI equivalent temperature for every degree increase of latitude (p-value < 0.001).

Fig. 2
figure 2

Spatial distribution of mean Universal Thermal Climate Index (UTCI) equivalent temperature across Greece from 1991 to 2020 (a) throughout the year, (b) in summer (June, July and August), and (c) in winter (December, January, and February)

Full size image

In summer, the mean UTCI ranges between 18.1 °C and 29.5 °C (mean ± sd = 24.3 °C ± 2.1 °C) classifying thermal stress into two categories, ‘no thermal stress’ and ‘moderate heat stress’ (Fig. 2b). The highest value was found in the region of Central Macedonia (Serres) and the lowest in the northern mountainous area of West Macedonia. The spatial distribution of the mean UTCI in summer indicates that approximately 24.2% of Greece's land areas, where 54.4% of the population lives, experiences on average ‘moderate heat stress’. These areas are primarily the inland plains in Central Macedonia, Thessaly, Central Greece, Attica, Western Greece, few regions in eastern and southern Peloponnese, and a part of southern Crete.

In winter, the daily mean UTCI varied between -5.1 °C and 8.5 °C (mean ± sd = 2.1 °C ± 2.5 °C), falling within two categories of thermal stress, ‘slight’ and ‘moderate cold stress’ (Fig. 2c). The regions experiencing ‘moderate cold stress’ represent the 22.5% of the total area and the 7.7% of the population.

In summer, the average daily maximum UTCI ranges from 27.3 °C up to 39.8 °C (mean ± sd = 34.5 °C ± 2.1 °C) spanning across three categories of thermal stress, ‘moderate’, ‘strong’ and ‘very strong heat stress’ (Fig. 3a). About 4.1% of the country experiences ‘very strong heat stress’ affecting 5.7% of the population. Areas experiencing the most adverse warm conditions include the plains of Serres, Central Macedonia, Thessaly, Central Greece and southern Peloponnese.

Fig. 3
figure 3

Spatial distribution of mean daily (a) maximum in summer (June, July and August), and (b) minimum in winter (December, January, and February) of the Universal Thermal Climate Index (UTCI) equivalent temperature across Greece from 1991 to 2020

Full size image

In winter, the average daily minimum UTCI varies between -13.0 °C and 2.4 °C (mean ± sd = -5.3 ± 2.7 °C), falling within two categories of thermal stress, ‘moderate’ and ‘slight cold stress’ (Fig. 3b). The ‘moderate cold stress’ dominate across most areas of the country. The ‘slight cold stress’ is experienced in 2.0% of the country (areas Central Greece and Crete) impacting 3.5% of the population.

Figure 4 shows the distribution of monthly UTCI estimations in percentages per stress category. The category ‘no thermal stress’ occurs throughout the entire year. From May to October, the prevailing stress category is that of ‘no thermal stress’ with a monthly percentage of hours with UTCI equivalent temperatures falling into this category ranging from 54.9% (July) to 69.9% (September). Unfavorably warm conditions occur mainly from May (38.8%) to September (23.8%) with July and August registering the highest percentage of heat stress hours within a month reaching 45.2% and 42.4%, respectively. Unfavorably cool conditions occur mainly from December (38.9%) to March (25.4%). The highest percentage of hours with at least ‘moderate cold stress’ occur in January (47.5%) followed by February (41.5%).

Fig. 4
figure 4

Frequency of occurrences of Universal Thermal Climate Index (UTCI) stress categories in each month across Greece from 1991 to 2020

Full size image

At a yearly basis (Fig. 4), unfavorably warm/cool hours prevail during 32.0% of the total hours in a year. Among these, 17.2% are identified as unfavorably warm while 14.8% as unfavorably cool. During summer and at country scale the percentage of hours with unfavorably warm conditions is 42.2% (20.3% falling into the category of ‘moderate heat stress’). In winter and at country scale, the percentage of hours with unfavorably cool conditions is 42.7% (38.6% falling into the category of ‘moderate cold stress’).

3.2 Thermal burden

The analysis of the spatial distribution of the percentage of hours with unfavorably warm conditions, at annual timescale, shows that thermal burden occurs at a mean (± sd) percentage of 17.2% (± 3.3%). This burden is more pronounced in the plains, with the highest percentage reaching 25.9% in the plains of Central Macedonia Fig. 5a). In mountainous regions, the percentage of hours with unfavorably warm conditions is low, ranging from 7.0% to 12.0% of the hours. The unfavorably cool conditions (Fig. 5b) occur at a mean (± sd) percentage of 14.8% (± 5.4%) with the highest frequency found in West Macedonia (30.2%).

Fig. 5
figure 5

Percentage of hours with (a and c) unfavorably warm and (b and d) cool conditions, (a and b) throughout the year, (c) in summer (June, July and August), and (d) in winter (December, January, and February) according to the Universal Thermal Climate Index (UTCI) from 1991 to 2020

Full size image

During summer and at country scale (Fig. 5c), the mean (± sd) percentage of hours with unfavorably warm conditions is 42.2% (± 5.8%) while the highest percentage, about 61%, is found in Central Macedonia (Serres), Central Greece and southern Peloponnese. Regions with local maxima exceeding a percentage of 45.0% of the hours experiencing unfavorably warm conditions are found in the plains of Central Macedonia, Thessaly, Central Greece, Attica, Western Greece and the Peloponnese. The relevant affected area is 34.0% of the total country while the relevant affected population is 66.8%. The maximum frequency of unfavorably warm conditions is found in Central Macedonia (42.5%).

In winter and at country scale (Fig. 5d), the mean (± sd) percentage of hours with unfavorably cool conditions is 42.7% (± 12.7%). The highest percentage is found in West Macedonia (72.5%). Regions with local maxima exceeding a percentage of 45.0% of the hours experiencing unfavorably cool conditions cover 43.5% of the country while the relevant affected population is 20.1%.

In Fig. 6 the percentage of hours at annual timescale which is ranked as unfavorably warm/cool for major cities in Greece, is presented. The cities are ranked based on decreasing latitude (from north to south). The percentage of hours with at least ‘moderate heat stress’ throughout the year ranges from 16.9% in Rhodes up to 23.2% in Serres and Argos with an average over all cities of 20.2% (Fig. 6a). Conversely, the percentage of hours experiencing at least ‘moderate cold stress’ varies from 3.5% in Iraklio to 14.9% in Ioannina, with an overall average across all cities of 9.5% (Fig. 6a). In summer, the percentage of hours with at least ‘moderate heat stress’ reaches the 54.3% in Serres and 51.6% in Argos (Fig. 6b). Areas experiencing increased hours of heat stress in summer include Agrinio (50.8%), Larissa (50.6%), Chalkida (50.2%), Trikala (49.1%) and Arta (48.9%). Specifically, among the three most populus cities in Greece, Athens, Thessaloniki and Patra, the percentage of hours with unfavorably warm conditions in summer is 48.0% in Athens, 46.0% in Thessaloniki, and 45.9% in Patra. In winter, areas with increased hours of cold stress encompass Thessaloniki (46.4%), Ioannina (43.7%), Volos (40.8%), Trikala (36.5%), and Preveza (33.8%) (Fig. 6c).

Fig. 6
figure 6

Frequency of hours with (a) unfavorably warm and cool conditions throughout the year, (b) unfavorably warm conditions in summer (June, July and August), and (c) unfavorably cool conditions in winter (December, January, and February) in big cities of Greece (sorted by decreasing latitude)

Full size image

3.3 Trend analysis of UTCI

Trend analysis showed an increase of annual mean UTCI over the years for the period 1991 to 2020. UTCI increase ranges between 0.02 to 0.09 °C/year (mean = 0.05 °C/year; Fig. 7a) suggesting that the annual mean UTCI at the end of 2020 was increased by 0.6 to 2.7 °C across Greece compared to 1991. The maximum increase was found on the islands of the eastern Aegean Sea (0.08 – 0.09 °C/year), in the mountainous areas of West Macedonia (0.07 °C/year), and in Evros (0.06 – 0.07 °C/year). The minimum increase was found in the mountainous areas of Central Greece (0.02 – 0.03 °C/year) and in the Peloponnese (0.03 – 0.04 °C/year). The UTCI trend was found to increase with latitude across Greece with a rate of 0.0032 °C per degree of latitude (p-value < 0.001) indicating a higher increase of annual mean UTCI equivalent temperature in higher latitudes.

Fig. 7
figure 7

Spatial distribution of the trend in mean Universal Thermal Climate Index (UTCI) (°C/year): (a) annual trend for the period 1991–2020, and seasonal trends for (b) winter, (c) spring, (d) summer, and (e) autumn. Dots indicate statistically significant trends at the 95% level

Full size image

Investigation of the seasonal trends (Fig. 7b-e) showed that spring, summer and autumn exhibit a uniform maximum increase of the annual mean UTCI by about 0.09 °C/year on the islands of the eastern Aegean Sea. However, variations are noted in East Macedonia and Thrace, Central Greece, Attica, and western Peloponnese where the increasing trend is more pronounced in spring compared to summer. The highest percentage of the area with an increase exceeding 0.09 °C/year was identified in spring, encompassing regions such as the islands of central Aegean Sea, East Macedonia and Thrace, regions of West and Central Macedonia, and regions of Thessaly. The positive trend in winter is non-significant in most of the country.

An increase over time in summer was also identified in the annual percentage of hours with unfavorably warm conditions. This increasing trend ranged from 0.06 to 0.52%/year (mean = 0.14%/year; Fig. 8) suggesting that the percentage of hours with unfavorably warm conditions was increased at the end of 2020 by 1.8% to 15.6% across Greece compared to 1991. The areas with the highest increasing trend were found in the region of Central Macedonia (0.40%/year) and Western Greece (0.30%/year), on Ionian islands (0.30%/year) and on a few islands in the Southern Aegean (0.50%/year).

Fig. 8
figure 8

Spatial distribution of the trend in the percentage of hours with unfavorably warm conditions according to the Universal Thermal Climate Index (UTCI) in summer. Dots indicate significant trends at the 95% level

Full size image

3.4 Heatwaves 2021 and 2023

During the two most recent and severe heatwaves which occurred in Greece in the periods of July 26th—August 6th, 2021 (HW1) and 13th – 27th July, 2023 (HW2), unfavorably warm conditions were identified mainly in the plains of the mainland. The mean UTCI equivalent temperature over the whole country ranged from 22.9 °C to 35.5 °C in HW1, and from 23.6 °C to 35.6 °C in HW2 (Table 1), exceeding the respective mean summertime daily values of the period 1991–2020 by about 5.0 °C (18.1 °C to 29.5 °C; Fig. 2b). Notably, during both heatwaves the percentage of the Greek areas which experienced unfavorably warm conditions is two-fold than that of the reference period (44.6% in HW1 and 43.5% in HW2 versus 24.2% in the reference period as derived from Fig. 2b) and the population affected was about 1.5 times higher (78.1% in HW1 and 76.6% in HW2 shown in Table 1 versus 54.4% in the reference period shown in Fig. 2b). The mean maximum UTCI equivalent temperature during the heatwaves reached up to the ‘extreme heat stress’ category (Table 1). In the HW2, the absolute maximum UTCI reached 52.1 °C, surpassing the absolute maximum value of the period 1991–2020 which was 51.9 °C.

Table 1 UTCI equivalent temperatures and heat stress categories in two heatwave periods, July 26th– August 6th, 2021 (HW1) and July 13th – 27th, 2023 (HW2), along with the percentage (%) of area extent and population affected for each heat stress category
Full size table

Both heatwaves HW1 and HW2 were very intense. There was approximately a two-fold increase (41.5% in HW1 and 42.9% in HW2 versus 21.9%) in the percentage of hours categorized at least as of 'strong heat stress' compared to the mean summer value of the period 1991–2020. Focusing on the percentage of hours classified as of ‘very strong heat stress’, there was a five-fold increase accounting for 22.5% in HW1 and 22.2% in HW2 compared to the reference period that was found 3.9%.

In order to further evaluate the intensity of HW1 and HW2, we calculated the accumulated hours that the population experienced unfavorably warm conditions, i.e., at least ‘moderate heat stress’ (or worse), over the 15-days period encompassing each heatwave. This calculation was compared with the distribution of the accumulated number of hours with at least ‘moderate heat stress’ over the rolling 15-day period of summers from 1991 to 2020. Figure 9 supports the fact that both heatwaves were extremely intense, with HW2 being the worst. During the HW1 in about 34% of Greece the total number of hours with unfavorably warm conditions fell within the 99.9% percentile of the distribution of the total hours over the rolling 15-day period of summers from 1991 to 2020, setting new records of hours with unfavorably warm conditions. This areal percentage expanded to 85.0% for HW2. The affected areas included all major cities of Greece during HW2, whereas in HW1 there were regions including big cities that did not break records such as Thessaloniki, Mesolongi, a part of Athens, Pirgos, Patra, Iraklio, Rhodes. This distinction becomes particularly evident when assessing the estimated population affected by the accumulated hours exceeding the 99.9%, which was 44.4% for HW1 but increased to 93.2% for HW2. Both heatwaves were less extreme over the islands of Aegean Sea as the total number of hours with unfavorably warm conditions closely approached the 95% percentile.

Fig. 9
figure 9

Spatial representation of percentiles where the total hours classified as unfavorably warm over a 15-day period including the (a) July 26th—August 6th, 2021 (HW1) and (b) July 13th – 27th, 2023 (HW2) stand in comparison to the respective hours over a rolling 15-day period during the summers of 1991 to 2020

Full size image

4 Discussion

This paper presents a climatological and trend analysis of the thermal bioclimate in Greece. The analysis employed the ERA5-Land reanalysis dataset with a resolution of 0.1° × 0.1° for the estimation of the thermo-physiological index UTCI in an hourly basis for the period from 1991 to 2020. The UTCI was also used to assess the intensity of the two most recent heatwaves which occurred in July 2021 and 2023.

The analysis shows that the mean UTCI equivalent temperature exhibits a spatial variability of about 10.0 °C across the country, consistently falling within the categories of ‘no thermal stress’ and ‘slight cold stress'. A latitude-dependent decrease in UTCI equivalent temperature was identified. In summer, areas experiencing ‘moderate heat stress’ are identified based on the mean UTCI estimations, whereas areas experiencing ‘very strong heat stress’ are found based on the mean daily maximum UTCI values. In winter, areas with moderate cold stress are found based on both the mean UTCI and mean daily minimum UTCI estimations. Unfavorably warm conditions are mainly met from May to September and cool from December to March. The highest percentage of hours with heat stress is recorded in July and August as these months are associated with the maximum number of heatwaves (Galanaki et al. 2023). Conversely, January and February experienced peak occurrences of cold stress.

Furthermore, the results suggest that areas characterized by low UTCI equivalent temperatures include mountainous regions. In contrast, the mainland plains display notably higher UTCI values. Heat stress is prevalent in inland plains, particularly in the eastern regions of the country in accordance to previous studies (Matzarakis and Mayer 1997). During summer, coastal areas within the mainland and the islands generally experience lower UTCI equivalent temperatures than the mainland plains (Matzarakis and Mayer 1997; Matzarakis et al. 1999; Matzarakis 2001), possibly due to the influence of the cool and dry Etesian winds (Matzarakis 2001). Similar outcomes were found in terms of another thermo-physiological index, the PET (Matzarakis 2001). Central Macedonia, Thessaly, and Central Greece have the highest frequency of hours categorized as unfavorably warm, while West Macedonia experiences the highest frequency of hours categorized as unfavorably cool. Thessaly and Central Greece were also identified as areas experiencing increased unfavorably warm conditions based on PET (Matzarakis 2001) while they have been characterized as heatwave hotspots in a recent study examining heatwave climatology and trends in Greece (Galanaki et al. 2023). At the city level, the regions of Serres, Argos, Larissa, Trikala, Agrinio, and Chalkida are some of the hottest regions of the country, exhibiting the highest percentage of hours with unfavorable warm conditions in summer. On the other hand, the regions of Ioannina and Thessaloniki are among the coolest, displaying the highest percentage of hours with unfavorably cool conditions in winter.

Regarding 2021 and 2023 heatwaves, the findings of this study indicate that both were exceptionally prolonged and intense, with a great part of the country experiencing unusually high number of hours under heat stress. During the heatwave of July 2023, about 85% of country’s regions experienced a new record of the 15-day period total number of heat stress hours compared to the period 1991–2020. In previous studies focusing on Athens, the heatwave in 2021 has been described as the most severe on record, specifically in terms of the total hours of unfavorable conditions (Founda et al. 2022; Giannaros et al. 2023). However, the heatwave in 2023 set a new record for both the highest absolute maximum UTCI value and the total number of hours with heat stress within the period 1991–2020. The most affected areas were primarily the plains, such as the plains of Central Macedonia, Thessaly, Peloponnese, Western Greece, and Crete. The islands, particularly in the Aegean Sea, were less affected due to the influence of the cooling sea breezes. Both 2021 and 2023 heatwaves were characterized by their extended duration (the 2023 heatwave being three days longer), as well as the absolute maximum air temperatures, with both events setting record high temperatures of the surface weather stations network operated by the METEO Unit at the National Observatory of Athens (NOA) (Lagouvardos et al. 2017; Meteo 2023).

Given the Mediterranean climate of Greece, a higher occurrence of ‘extreme’ or ‘very strong heat stress’ conditions could be expected. This relative low frequency of the extreme categories could be due to the fact that UTCI was estimated using the ERA5-Land reanalysis dataset which although suitable and undoubtfully valuable, it may underestimate some weather variables compared to the surface meteorological stations (Araújo et al. 2022; Zou et al. 2022; Choudhury et al. 2023) particularly in warm conditions (Choudhury et al. 2023). Nevertheless, previous studies analyzing UTCI and using weather station data (Bleta et al. 2014; Katavoutas and Flocas 2018; Pantavou et al. 2023b), similarly reported either the absence or the extremely low frequency of the extreme UTCI categories, even though they considered regions in southern Greece (Bleta et al. 2014) and Athens (Katavoutas and Flocas 2018; Pantavou et al. 2023b). Different thermal indices may yield diverse estimations of thermal stress due to variations in their definitions and sensitivity to meteorological variables (Blazejczyk et al. 2012; Provençal et al. 2016; Pantavou et al. 2023a). The sensitivity of UTCI at low air temperatures to the wind speed (Bröde et al. 2012) may also contribute to the occurrence of some of the lowest UTCI equivalent temperatures observed in certain islands in the Northern Aegean region. In this study, UTCI was chosen because it is a universal index that facilitates comparisons across different climates and is widely used in both research and operational contexts (Krüger 2021). At the local level, UTCI showed high efficiency for operational purposes (Pantavou et al. 2023a) and superior applicability in estimating pedestrians’ thermal perception (Pantavou et al. 2013).

UTCI may not be directly used as a measure of climate change, although it can be influenced by various factors associated with global warming such as surface air temperature, local temperature increases to due to factors such as urbanization and changes in land use, changes in weather patterns leading to shifts in precipitation, wind patterns, and atmospheric circulation, as well as extreme weather events. Monitoring UTCI over time can provide valuable insights into how climate change is affecting outdoor thermal conditions in a region. In this study, the average annual UTCI equivalent temperature in Greece was found to increase over the years, aligning with the global air temperature increase (WMO 2023) and the air temperature increase found in other European countries (Tomczyk and Bednorz 2020). A recent study also suggested an increase in heatwave events between 1950 and 2020 in Greece (Galanaki et al. 2023). The increased trend found in our study is more pronounced in regions of higher compared to lower latitudes while it is particularly prominent in spring compared to other seasons.

The strengths of this study lie in the use of the recent 30-year period of the ERA5-Land dataset to examine the climatology and trends of the bioclimatic conditions in Greece. The ERA5-Land dataset offers a high temporal resolution of one hour and provides a comprehensive spatial dataset covering the entire country. The study employed UTCI, a contemporary thermo-physiological index widely embraced in both research and practical applications, to achieve these insights. A comprehensive statistical analysis was conducted covering the entire country, in all seasons, and examining the extent of the area and population affected by each UTCI category. On the other hand, this study has also certain limitations that should be considered. These mainly stem from the inherent difficulty in the ERA5-Land dataset, which, like any other reanalysis, might be unable to reproduce at the highest accuracy the local conditions especially the maxima/minima. The present study adopted the standard scale of UTCI assessment categories although these categories thresholds have been adjusted based on people’s perception (Pantavou et al. 2014). Nevertheless, the inclusion of ERA5-Land dataset and the original UTCI scale support global comparisons across different regions and climates.

5 Conclusions

This study outlined the overarching features of UTCI estimated by the ERA5-Land dataset across Greece. It provided bioclimatic maps and highlighted the inland plains of the country, especially in the eastern regions of the mainland as regions with increased heat stress levels. During summer, coastal areas within the mainland and islands typically register lower UTCI equivalent temperatures in contrast to the inland plains. A warming trend in the thermal bioclimate of Greece over the past three decades was identified.

This study is one of a few focusing on the thermal bioclimate of Greece based on the use of UTCI. UTCI already encompasses a broad range of applications and presents opportunities for further advancements in the future. Moreover, ERA5-Land's comprehensive coverage of meteorological variables, high spatial and temporal resolution, and availability make it a valuable resource for examining weather patterns, climate trends, land surface processes, and their impacts on various systems.

The analysis of the thermal bioclimate can contribute to the development of adaptation measures and enhance the capacity of national public health systems to respond to the thermally-induced health risks. Bioclimatic maps monitor changes of the thermal environment raising awareness over the sustainability of outdoor environment while their applicability extend to sectors such as tourism and recreation, urban design, and energy efficiency planning. They provide a foundation for future research and strategic planning in the face of climate variability and change.