Present and future humid heat extremes and population exposure in Türkiye

Abstract

Literature communicates that humid heat extremes will intensify more than dry heat extremes. This favors studies that elaborate on the characteristics of humid heat, given that it is physiologically more relevant than dry heat in terms of understanding the physical interaction between human thermoregulation and heat stress, promoting the use of metrics that express heat stress through a combination of temperature and humidity. Global projections show that the frequency and intensity of humid heat extremes will surge in the land area already subject to them and expand toward the higher latitudes, making the mid-latitudes hotspots for these extremes. Therefore, a thorough explanation of their regional characteristics becomes crucial, given that the changes in these extremes can potentially render a large proportion of the global population at risk of being exposed to humid heat extremes that can drive morbidity and mortality. At this point, the risk and vulnerability assessments inevitably include quantifying population exposure to heat stress. Studies show that there has already been a dramatic increase in global exposure to humid heat extremes. Türkiye is located in the Eastern Mediterranean near the Middle East, a region susceptible to humid heat stress. Here, we perform the first analysis of present and projected changes in the intensity and frequency of humid heat extremes and quantify the population exposure to these extremes in Türkiye, using long-term simulations from the non-hydrostatic mesoscale model of Consortium for Small-scale Modeling (COSMO-CLM) under the Representative Concentration Pathways 8.5 (RCP8.5) emission scenario. We force the model with the global simulations provided by the coupled climate model of the Max Planck Institute for Meteorology (MPI-ESM-LR), a participant in the CMIP5 (Coupled Models Intercomparison Project Phase 5) collaboration. First, global simulations are downscaled to 0.44° horizontal resolution in a domain that includes the majority of the Eastern Mediterranean region. These simulations are then further downscaled to 0.11° horizontal resolution over Türkiye. The overall model configuration is the same between the 0.44° and 0.11° simulations. We use the microphysics scheme of the 2-category Ice Scheme; the convection scheme of Tiedtke, 1989; the radiation scheme of Ritter and Geleyn, 1992; the land surface scheme of TERRA-ML; and a turbulent kinetic energy-based surface transfer scheme. The simulations are run for the years 1979-2005 for the reference period and for the years 2024-2100 for the future period. Our analysis is based on the hourly wet-bulb temperature (WBT) as the humid heat stress metric, considering its robustness to climate change, calculated from hourly temperature and relative humidity. As part of preprocessing, we verify our long-term climate simulations using the ERA5-Land reanalysis data, and we bias-correct the hourly WBT values against the corresponding reanalysis data using an Empirical Quantile Mapping (EQM) procedure. The EQM is a distribution-free correction method that takes into account several predetermined empirical quantiles to map the model output quantiles to the observed quantiles. The empirical CDFs of the WBT from the ERA5-Land reanalysis and model output are here approximated using 16 manually defined quantiles, which are 0.03, 0.10, 0.16, 0.23, 0.30, 0.36, 0.43, 0.50, 0.56, 0.63, 0.70, 0.76, 0.83, 0.90, 0.96, and 1.00. The use of the EQM approach stems from the mismatch in the model behavior in different quantiles. Namely, while the humid heat extremes, characterized in our study with the WBT, are significantly overestimated over our domain, the median values are generally underestimated. As an alternative to the EQM approach, we utilize a simple linear scaling bias-correction approach. However, this approach did not produce good results in our case, as it applies a single daily correction factor for every hour of the day. We focus on three aspects of humid heat stress across the climatologically different regions of Türkiye. First, we define base (1985-2005) and projection periods (2081-2100) and quantify changes in the frequency and intensity of the humid heat extremes in each period. The intensity metrics we use include the average annual 99th quantile (q99) and maximum (max) values of the WBT, and we use three different frequency metrics based on an absolute and a relative heat stress threshold, which are the number of days above 27°C (NOD27), the number of consecutive hours above 27°C (NOCH27), and the number of consecutive hours above local q99 (NOCHQ99). And we reveal whether the trends in the average annual median and maximum WBT values are similar, and if not, we quantify to what extent they differ. For this, we employ the Mann-Kendall trend test, which determines the significance of trends by empirically ranking the values in a timeseries. Second, we document whether the change in the climate of Türkiye affects the diurnal climatology of humid heat stress, focusing on the hours at which it maximizes. Finally, using the commonly used person-days approach, which characterizes exposure by the multiplication of the population amount in a specific region by the number of days in a year when extreme WBT circumstances occur in that region, we evaluate the long-term changes in the number of people annually exposed to specific WBT thresholds throughout the base and projection periods, allowing both the population and climate in Türkiye to evolve. To isolate the relative contribution of the extreme humid heat stress to the total exposure, we perform an alternative exposure analysis in which we hold either the population count or the climate constant throughout the analyzed period. Namely, while the climate component can be interpreted as the change in population exposure if the climate of Türkiye evolved into the future while the population kept at the year 2005, the population component of the total exposure physically refers to the change in population exposure if the climate kept at the average of the last five years in the base period while the population evolved into the future. We portray not only the nationwide changes in the humid heat extremes and population exposure but also their regional aspects by exploiting the K-means clustering algorithm. We use this algorithm to conceptualize seven climatologically different regions. The model outputs from the hourly 2-meter temperature, 2-meter RH, and total precipitation are aggregated to seasonal values for each season in a year and averaged over the base period for the clustering of these regions. Each of the K-means regions is named according to their geographical location and proximity to water bodies. For instance, the region restricted to the country's western coast is named RWC (i.e., Region West Coastal), and the region located along the southern peripheries of the nation is named RSC (i.e., Region South Coastal). We apply the conceptualized K-means regions to each analysis explained above and show regional changes in humid heat stress and exposure. First, our results suggest a remarkable increase in q99 and max values across most of the country, particularly along the coasts, relative to the base period. The average annual q99 of the WBT exceeds 29°C in some regions in the projection period, with the average annual max values approaching the dangerous heat stress threshold of 31°C during the same period. The country's all-time maximum WBT reaches 33.8°C, a nearly deadly condition. Regionally, the highest WBT values are seen in RSC and RWC, which account for 21.41% and 25.60% of Türkiye's total population, respectively, in both the base and projection periods. The frequency of humid heat stress shows similar changes as well, with the NOD27 and NOCH27 projected to increase as much as 45 days and 12 hours, respectively, in a geographically wide area, which includes the surroundings of Adana, Antalya, Izmir, Sakarya, Ordu, and Diyarbakir, most of which are coastal locations. Over most of these regions, nearly a month every year is anticipated to experience severe humid heat stress, and almost 56% of the country's land area will experience local upper tail heat stress conditions for at least an additional ten consecutive hours. This is in contrast to the NOD27 and NOCH27 in the base period, which are geographically constrained to the city of Adana. Our findings convey the higher likelihood of humid heat extremes reigning within these regions. Second, the projections show that the WBT values at every hour of the day will be at least two degrees Celsius greater than in the base period. However, this increase is more pronounced for the maximum (minimum) values of the WBT, which approximately occurs during the afternoon (nighttime). Overall, the response of the diurnal cycle of humid heat stress is consistent across regions with different climates. Finally, we explicate a significant rise in the number of people exposed to severe humid heat stress, concentrated along most coastal regions, by as much as 1.6 million person-days. The individuals living in the RSC, RWC, and RNC regions are expected to face the highest levels of exposure to humid heat stress, with regional mean population exposure to humid heat stress amounting to almost two orders of magnitude from the base period to the projection period. RWC, in particular, is subject to maximum mean exposure to 29°C in the projection period up to seven times the maximum mean exposure to 27°C in the base period. On the other hand, the exposure to the same conditions in the base period seldom exceeds 1 million person-days and is geographically limited to the southeastern city of Adana and east of the Marmara Sea. The projected rise in exposure is primarily due to climate change, as the contribution of the population component of the total exposure to the surge in exposure is minor compared to the contribution of the climate component. Further, more than 20% of Türkiye's population may confront severe humid heat stress for at least one hour, with that percentage falling to 4.15 percent for at least five consecutive hours, which indicates that people will not only endure more intense humid heat stress but also be exposed to these conditions consecutively over a period of many hours. Our findings highlight the increase in humid heat extremes and the resulting population exposure to these circumstances, especially along Türkiye's coast, and have important ramifications for society and numerous industries, including the economy and tourism.