Future Projections of Heat Waves and Associated Mortality Risk in a Coastal Mediterranean City
[ad_1]
1. Introduction
Climate change is widely considered as one of the greatest global threats and health hazards of the 21st century. The latest IPCC report notes that in the last decade (2011–2020), global surface temperature was approximately 1.09 °C higher than the reference period of 1850–1900. Meanwhile, future projections suggest a total warming of 1.4–4.4 °C, depending on the emission scenario [1]. Rising temperature rates are even more prominent in the Mediterranean region, which has been characterized as a climate change ‘hotspot’ because of the particular responsiveness and vulnerability of the region [2,3,4,5,6,7]. Mediterranean warming has exceeded global average rates since the 1980s, with present annual mean temperatures being 1.5 °C above the late 19th century levels, and with future annual and summer warming rates projected to be 20% and 50% larger than the annual global average, respectively [7,8]. Further increases in the summer temperatures of the region are of major concern, since future projections indicate a warming of up to 7 °C in daily maximum temperatures compared to the late 20th century, under high emission scenarios [9,10].
An especially alarming feature of climate change is the increase in the frequency of hot extremes [11]. Climate change may impact both the means and variances of temperature distributions, variously affecting the likelihood of extreme temperatures [12]. Specifically for the Mediterranean region, there have been numerous studies reporting increases in the frequency, duration and intensity of heat waves during the past few decades [7,13,14]. For example, [14] found that the heat wave intensity, length and number increased by a factor of 7.6, 7.5 and 6.2, respectively, in the period 1960–2006 in the eastern Mediterranean. In addition, the number of tropical nights ( > 20 °C) also presents a significant increase in various studies across the region [7]. Looking at the future, significant rises in hot extremes are expected across the Mediterranean, especially in the summer season, with regard to both daytime and nighttime temperatures [13,15]. Defining warm nights and cold days using 90th percentile and 10th percentile of the daily values of the calendar day, respectively, [16] showed that under a global warming of 4 °C, the region will experience predominantly warm nights and almost no cold days during the year. Model projections also indicate a lengthening in the season of hot extremes (by ~1 month until the middle of and by more than two months until the end of the century) as well as major increases in the frequency, duration and intensity of heat waves, with some studies demonstrating increases of the order of 20 times more heat wave days in southern Europe by the end of the century [13,15,17,18,19].
The growing concern over the impact of excessive heat on human health under climate change is further emphasized in urban areas because of two specific features of the urban environment. The first is related to the large concentration of the population in urban areas (~55% globally), which is projected to further increase in the future (~68% by mid-century) [20,21]. The second is related to the well-documented Urban Heat Island phenomenon, under which higher temperatures are observed in the urban environment compared to the suburban and rural surrounding areas, especially during the nighttime [22,23,24,25]. Meanwhile, station measurements suggest a slightly faster warming trend in urban centers compared to rural areas in the Mediterranean and the Middle-East–North Africa region [26]. These urban features raise major implications regarding the safety of the Mediterranean populations and necessitate comprehensive investigation of the relationship between climate change, heat exposure and the urban environment.
When modeling urban areas, using small horizontal grid cell sizes is crucial for the correct representation of their microclimatic features [27]. In the absence of small enough grid cells, features from urban areas are averaged together with the surrounding suburban and rural areas, leading to obscuration of the distinctive features of each category. High-resolution modeling allows for the proper distinction of the above-mentioned areas and even for the discrimination between different intra-urban levels of heat exposure (e.g., due to different elevations, proximity to sea or land use categories), ensuring more reliable and representative projections [28]. Nevertheless, increasing the modeling resolution significantly increases the computational resources and time necessary for the simulations, and it often requires compromises in other modeling parameters (e.g., larger time steps or less simulated time). As a consequence, studies evaluating the future evolution of hot extremes in urban environments commonly employ large model grid cells, restricting the analysis to a few (or even one) grid cells. For instance, studies by [29,30] focus on future heat waves in various cities of Europe and the Iberian Peninsula by analyzing the model grid cell nearest to each city. Similarly, [31,32] assess future heat wave conditions in multiple European cities by examining a few grid cells closest to the city centers.
The potential impacts of prolonged exposure to extreme heat stress during heat waves have become clearer in recent decades [33,34,35,36,37]. Standout illustrations consist of the 2003 European and the 2010 Russian heat waves that caused tens of thousands of excess deaths [38,39,40,41]. The USA Center for Disease Control and Prevention reports that heat waves were the highest cause of death related to weather conditions during the decade 2000–2009 [42], while in Australia, heat waves are responsible for more deaths than all the other natural hazards combined [43]. Apart from extreme heat events, increases in moderate ambient temperatures have been shown to also impact mortality rates substantially [44,45]. Multiple studies demonstrate the adverse effects of heat on mortality in several European and Mediterranean cities [46,47,48,49]. In a study by [47], it was demonstrated that while the maximum temperature threshold for increased mortality rates was higher in Mediterranean cities (29.4 °C) compared to north-continental cities (23.3 °C), the overall increase in all-natural mortality per 1 °C was greater in the Mediterranean cities (3.12% versus 1.84%). In addition, [49] found that for the 50+ age group, the percentage increase in daily mortality during heat wave days was by 22% higher in Rome and 8% higher in Stockholm compared to normal summer days. Both studies also highlight the importance of locally conducted research, since the varying city population, topographical and climatic characteristics can result in important differences in heat mortality estimates.
In the context of heat-related mortality studies, thermal exposure metrics that employ more meteorological variables than temperature alone are often utilized in order to better capture the heat stress on the human body, such as the apparent temperature [44,45]. For example, refs. [50,51] found that a 5.5 °C increase in mean apparent temperature corresponds to 2.3% and 1.8% increases in mortality, respectively, while [52] found that an increase of 1 °C in the same metric resulted in a 2.1% and a 1.5% increase for Lisbon and Oporto, respectively. It is also noteworthy that most of the studies regarding heat-related mortality typically employ daily maximum or mean values of the heat stress indicator. There are studies, however, which indicate that the lack of relief and rest through sleep, attributed to high nighttime temperatures, are additional important contributions to heat mortality [53,54,55]. Ref. [53] used the Hot Night Excess (HNE) index to quantify the intensity of thermal stress during nighttime. They indicate that under the SSP2–4.5 scenario, the increased nocturnal thermal stress could lead to ~1% higher attributable fraction of mortality compared to increases in mean daily temperatures. Such findings further exacerbate future health impact concerns for urban areas. Meanwhile, future projections indicate a greater heat mortality burden under climate change [31,56,57,58,59].
The objective of this work is to leverage high-resolution modeling (2 km) in order to comprehensively assess the future impacts of climate change in the frequency, duration and intensity of heat waves and the associated health impacts in large Mediterranean urban environments. This study’s focus is Thessaloniki, Greece, which is a large city that combines vulnerability traits stemming both from its Mediterranean location and urban characteristics. At the same time, its complex topography, coastal location and land use heterogeneity make it particularly suitable for demonstrating the benefits of high-resolution modeling. Regarding the recent trends of hot extremes in the city, [60] demonstrate that the annual number of heat wave episodes has increased by 72% during the period 1990–2020 and that the duration of the longest heat wave episode per summer has been increasing at a rate of 0.4 days/decade since the mid-20th century. Meanwhile, [61] found that 2.34% of all-cause deaths during 2006–2016 in the city were attributable to heat stress, while mortality was estimated to increase by 1.95% per 1 °C above neutral heat conditions.
Six heat wave indices typically examined in the heat wave-related literature were computed for the middle and the end of the century and were compared to the present climate. Furthermore, exposure–response relationships derived by [56,61] for the area of Thessaloniki were used to investigate potential increases in mortality risk during heat wave days. For both the definition of heat waves and the derivation of the exposure–response relationships, the apparent temperature was employed, since it is a heat metric that also accounts for the effect of humidity in heat-related discomfort. Lastly, the effects of exacerbated urban nighttime heat conditions on health were also investigated. This was achieved by defining heat waves and establishing exposure–response relationships not solely based on the daily maximum but also the daily minimum values of apparent temperature.
Put simply, this study aims to achieve the following objectives: (1) to reinforce the literature concerning the impact of climate change on heat waves in large Mediterranean urban environments through very high spatial resolution modeling, (2) to investigate the use of a more physiologically oriented heat metric in the analysis of heat waves and the associated health risks in a Mediterranean city, (3) to investigate the potential exacerbation of health impacts due to the aggravated nighttime heat conditions in a Mediterranean urban environment, (4) to facilitate a better understanding of the potential heat-related health impacts due to climate change for the city of Thessaloniki and thus bolster the background for more well-informed and targeted mitigation and adaptation strategies.
4. Conclusions
This study explored the future impact of climate change in the frequency, duration and intensity of heat waves, as well as in the associated mortality risk, in the large Mediterranean city of Thessaloniki, Greece. For that purpose, high-resolution (2 km) climate simulations were conducted with the WRF model for three 5-year periods describing the present, mid-century and end of the century conditions, under the RCP8.5 scenario. Using small horizontal grid cells allowed not only for proper discrimination between urban and non-urban areas but also for the examination of differences within the urban environment itself, which is caused by elevation, proximity to sea or land-use differences. Nevertheless, while the allocation of computational resources to achieve a more realistic representation through high resolution is valuable, it comes with the trade-off of the inability to execute an ensemble of model projections. This limitation hinders the capacity to provide robust uncertainty quantification for the model results, and, therefore, the uncertainty of the model results in this study can only be characterized by the model evaluation performed in [24].
In the context of this study, potential intra-urban discrepancies were examined by constructing two different time series based on selected inland and coastal grid cells. Apparent temperature was the heat metric employed for defining heat waves and estimating heat-related mortality, since it offers a more physiologically oriented measure compared to temperature alone. Both the daily maximum and minimum values of apparent temperature were employed for defining heat waves and estimating heat-related mortality in order to also include the amplified health impacts due to exacerbated nighttime heat conditions in urban environments. Changes in six heat wave indices that describe heat wave frequency, duration and intensity were explored. The associated changes in heat-related mortality during heat waves was examined by applying heat–mortality exposure–response relationships that were derived for the city of Thessaloniki and were based on daily maximum and minimum apparent temperature.
An initial analysis of the future thermal environment of the area indicated an exacerbation of similar magnitude in both urban inland and coastal regions. Both and were found to increase by ~1 °C by the middle and by ~4–4.2 °C by the end of the century compared to the present climate. Examining instead of temperature, the hot season daytime thermal stress was found to be similar for the inland and coastal urban regions, while the coastal thermal environment was found to be slightly more aggravated during the nighttime. Looking at the greater area of Thessaloniki through high resolution, the increases in were found to be relatively homogeneous, while high areas were found to expand, changing the spatial pattern in the greater area by the end of the century.
While heat wave frequency indices present noteworthy increases already by the middle of the century, the picture changes drastically by 2100 with the HWF increasing by a factor of ×8 and HWN by a factor of ×5 in both inland and coastal urban regions. The results suggest that not only will the majority of future summer days adhere to heat wave conditions, but a few heat wave episodes will also occur in May and September. Indices related to duration and intensity showed relatively stable behavior until 2050. By 2100, the mean heat wave duration increases by ×1.7, while heat wave episodes that last about a month were detected in both inland and coastal urban regions. Finally, HWI presented an increase of ~1 °C, while HWA was found to increase by ~3.5 °C in the inland and by ~4.2 °C in the coastal areas. Using high modeling resolution also allowed an assessment of the frequency heat wave indices in the greater area of Thessaloniki. By 2050, all non-mountainous areas will present non-zero values in HWF and HWN, which is in contrast to the reference period. Results from the period 2096–2100 indicate that the majority of the area will experience more than 30 heat wave days and seven heat wave episodes annually, while a few areas inside the city will experience almost up to 70 heat wave days per year. These results further reinforce concerns for heat-related health impacts under climate change in Mediterranean urban environments while also highlighting the importance of high modeling resolution in differentiating between micro-climatic features that are averaged when coarser resolutions are used.
The relative risk induced by both daytime and nighttime exposure to heat wave conditions was found to remain stable by the middle of the century and similar in magnitude for both coastal and inland urban regions. Nevertheless, the corresponding increase in the annual number of heat wave days that the increased mortality risk persists for raises major concerns for a potentially considerable increase in annual heat-related mortality in urban environments already by the middle of the century. By 2100, the heat-related mortality regime was found to change drastically, with both – and -induced daily relative risk increasing to 1.17 (+5.4%) and 1.27 (+9.5%), respectively. At the same time, these conditions will persist for the majority of summer days, calling for rigorous and multi-sector measures for the mitigation of health impacts. The mean heat wave day RR() was found to be higher than the corresponding RR() in all study periods, while RR() distributions were found to be significantly less positively skewed than the corresponding RR() distributions. Meanwhile, the 95% empirical CIs for the RR() values were found smaller compared to those of RR(), indicating that by including effects of lack of relief during nighttime, can serve as a more reliable predictor for predicting daily heat-related mortality.
As already mentioned, the relative risk results presented in this work are rather conservative, as the exposure–response functions employed cannot account for any risk amplification due to exposure to consecutive days of heat wave conditions. The potential existence of such an amplification effect could further increase the relative risk during heat wave days estimated in this study and should be further investigated alongside the effect of lack of stress relief during nighttime. However, even without including any amplification effects, the results of this study incite considerable health alarms for Mediterranean urban environments. In addition to climate change mitigation policies, targeted adaptation strategies and proper public communication must be the focus of local authorities throughout the next few decades, as such measures have been shown to be effective in reducing heat-induced mortality in multiple countries [99].
[ad_2]
