100m climate and heat stress data up to 2100 for 142 cities around the globe [PREPRINT]

ARTICLE INFORMATION 1 Article title 2 100m climate and heat stress data up to 2100 for 142 cities around the globe 3 Authors 4 Niels Souverijnsa,* 5 Dirk Lauwaeta 6 Quentin Lejeuneb 7 Chahan M. Kropfc,d 8 Kam Lam Yeungc 9 Shruti Nathe 10 Carl F. Schleussnerf 11 Affiliations 12 a Environmental Intelligence Unit, Flemish Institute of Technological Research (VITO), Mol, Belgium 13 b Vrije Universiteit Brussels (VUB), Brussels, Belgium 14 c Institute for Environmental Decisions, ETH Zurich, Zurich, Switzerland 15 d Federal Office of Meteorology and Climatology MeteoSwiss, Zurich, Switzerland 16 e Department of Physics, University of Oxford, Oxford, United Kingdom 17 f Integrated Climate Impacts Research Group, International Institute for Applied Systems Analysis 18 (IIASA), Laxenburg, Austria 19 Corresponding author’s email address and Twitter handle 20 [email protected]; https://x.com/nielssouverijns 21 Keywords 22 Urban climate; Urban heat; Projections; High-resolution 23 Abstract 24 Cities worldwide are increasingly facing the challenges of heat stress, a problem expected to worsen 25 with ongoing climate change. The lack of detailed, city-specific data hinders effective response 26 measures and limits the adaptive capacity of urban populations. In this data descriptor, we introduce 27 a comprehensive database providing climate and heat stress information for 142 cities globally, 28 covering the present and extending projections up to 2100 across three distinct climate scenarios, 29 including two overshoot scenarios. This dataset includes 34 heat stress indicators at a spatial resolution 30 of 100 meters, offering a unique database to identify vulnerable areas and deepen the understanding 31 of urban heat risks. The data is presented through an accessible, user-friendly dashboard, enabling 32 policymakers, researchers, and city planners, as well as non-experts, to easily visualise and interpret 33 the findings, supporting more informed decision-making and urban adaptation strategies. 34 SPECIFICATIONS TABLE 35 Subject Earth & Environmental Sciences Specific subject area Microscale climate information for cities worldwide Type of data Spatially explicit 100m climate information at decadal timesteps for the period 2010-2100 in NetCDF format Data collection The UrbClim urban boundary layer model is used to dynamically downscale largescale climate information to the extent of individual cities and their rural surroundings at very high resolution (100m). This hourly information is translated to decadal heat stress indicators for the period 2010-2100. Impact indicators are calculated using the CLIMADA model. Data source location 142 cities around the world Data accessibility Repository name: Zenodo Data identification number: https://doi.org/10.5281/zenodo.13361538 Direct URL to data: https://zenodo.org/records/13361538 Related research article None 36 37 1. VALUE OF THE DATA 38 • The dataset presented in this paper provides a first-of-its kind archive of 142 cities covering all 39 continents (excluding Antarctica) with detailed climate, heat stress and impact information for 40 both present and future time scales (until 2100) at 100m spatial resolution. 41 • The 100m spatial resolution allows to identify the most vulnerable areas within the city and 42 the long-term availability of the data permits to calculate heat stress impacts towards the end 43 of the century under different emission pathways, including overshoot scenarios and their 44 uncertainty. This provides invaluable support for urban planning, enhancing public health 45 responses, emergency response planning and climate impact and adaptation strategies. 46 • The data is presented in an easy-to-access dashboard (https://climate-risk47 dashboard.iiasa.ac.at/impacts/explore), allowing not only researchers, but also non-experts 48 and policy makers to easily access, visualise and interpret the heat stress data. Lastly, a toolbox 49 is presented to obtain similar data for cities that are currently not represented in this dataset. 50 • The model tools (UrbClim & CLIMADA) are validated both on temperature and humidity, 51 important components for calculating heat stress. 52 2. BACKGROUND 53 Heat stress is a natural disaster that is responsible for approximately 500.000 excess deaths per year 54 worldwide (1). In urban environments, temperatures are generally higher compared to rural 55 environments caused by the lower amount of vegetation and abundance of sealed surfaces. Towards 56 the future, one expects an increase in the number of heatwaves in cities (2) and their inhabitants are 57 prospected to experience twice as much heat stress compared to rural populations (3). Taking into 58 account that 68 % of the global population is projected to live in urban areas by 2050 (4), heat stress 59 in cities is a key priority to consider by policy makers, city planners and authorities. Despite the 60 acknowledgement of the increased vulnerability of city populations to heat stress (5,6), current 61 globally available datasets lack the spatial and temporal resolution to represent this additional heat 62 burden (7,8). The dataset presented here provides a first-of-its kind archive of 142 cities covering all 63 continents (excluding Antarctica) with detailed climate, heat stress and impact information for both 64 present and future time scales (until 2100) at 100m spatial resolution, building on the work that was 65 executed over Europe by Lauwaet et al. (2024) (9). 66 67 3. DATA DESCRIPTION 68 Indicators for each decade are available for a selection of 142 cities for the period 2010-2100 for 69 different future climate model scenarios and uncertainties (Figure 1; Supplementary Table 1). The 70 database of indicators is provided in both Geotiff and NetCDF format and is available in a local 71 projection (which can be retrieved from the metadata of the files) and in EPSG:4326. Individual files 72 and quick visualisations of indicator maps can be retrieved from the Climate Risk Dashboard 73 ((https://climate-risk-dashboard.iiasa.ac.at/impacts/explore), which allows to download the Geotiffs, 74 NetCDFs and visualisations in PNG format (Figure 2). A bulk data download option that allows to 75 download all indicators, time periods, scenarios at once is also provided via 76 https://doi.org/10.5281/zenodo.13361538. As all data is georeferenced, users can visualise, analyse 77 and manipulate the maps in GIS software tools and python. 78 79 Figure 1: Overview map indicating the 142 cities for which present and future climate and heat stress data is made available. 80 81 Figure 2: Snippet visualisations of the annual number of days with moderate heat stress (Wet Bulb Globe Temperatures above 82 25°C) over Berlin in the 2020 climate policies scenario from a (left)) spatial and (right) temporal perspective. Visualisations 83 obtained from the Climate Risk Dashboard. 84 An overview of the list of indicators that is calculated for each of the 142 cities is listed below. 85 - Temperature Indicators 86 o Average daily maximum temperature: Average daily maximum 2 meter temperature 87 over the full decade 88 o Average daily minimum temperature: Average daily minimum 2 meter temperature 89 over the full decade 90 o Average daily temperature: Average daily 2 meter temperature for the full decade 91 o Maximum temperature of the warmest month (10): The average maximum monthly 92 temperature of the warmest month throughout the year 93 o Maximum temperature of the coolest month (10): The average minimum monthly 94 temperature of the coolest month throughout the year 95 o Daytime Urban Heat Island: The average difference in daily maximum temperatures 96 between each pixel and the rural (non-water) spatial 10th percentile temperature 97 value. Temperatures are height-corrected by rescaling them to the average city height, 98 applying a lapse rate of 6.5 K*km-1. It captures the difference in temperatures due to 99 human activities and the modification of land surfaces. 100 o Nighttime Urban Heat Island: The average difference in daily minimum temperatures 101 between each pixel and the rural (non-water) spatial 10th percentile temperature 102 value. Temperatures are height-corrected by rescaling them to the average city height, 103 applying a lapse rate of 6.5 K*km-1. It captures the difference in temperatures due to 104 human activities and the modification of land surfaces. 105 - Temperature-based heat stress indicators 106 o Annual heatwave days: A heatwave is defined as a minimum of three days in which 107 both the daily maximum and minimum temperature exceed the 90th percentile 108 threshold of a base period (taken as the period 2011-2020). The 90th percentile 109 threshold is calculated over the full simulation domain (the city and its rural 110 surroundings) based on the definition in Romanello et al. (2022) (11). The indicator is 111 depicted as the average number of heatwave days per year. 112 o Annual heat-wave magnitude index daily (HWMId): The HWMId was defined by Russo 113 et al. (2015) (12) and allows quantifying the magnitude of heatwaves by accounting 114 for both their severity and duration, which makes it more suitable to compare extreme 115 temperature events across the world as well as past, present and future heatwaves. 116 o Annual number of days exceeding [25°C; 30°C; 35°C]: Annual number of days in which 117 the maximum temperature exceeds [25°C; 30°C; 35°C]. 118 o Annual number of nights exceeding [20°C; 25°C; 28°C]: Annual number of nights in 119 which the minimum temperature does not drop below [20°C; 25°C; 28°C] 120 o Annual cooling degree hours: Cooling degree hours is an international standard to 121 estimate energy usage for cooling dwellings using air conditioning. It is calculated as 122 the number of hours during which the temperatures rises over 25°C, multiplied by the 123 number of degrees the temperature rises above 25°C. The annual average value for 124 the decade is shown. 125 - Wet Bulb Globe Temperature based heat stress indicators 126 o Annual number of days WBGT > [25°C; 28°C; 29.5°C; 31°C]: Annual number of days in 127 which the WBGT exceeds [25°C; 28°C; 29.5°C; 31°C] for at least one hour. 128 o Annual number of nights WBGT > [25°C; 28°C]: Annual number of nights in which the 129 WBGT does not drop below [25°C; 28°C] 130 o Annual number of hours WBGT > [25°C; 28°C; 29.5°C; 31°C]: Annual number of hours 131 in which the WBGT exceeds [25°C; 28°C; 29.5°C; 31°C]. 132 o Lost working hours (LWH) for intense activities: Depending on the WBGT, workers lose 133 productivity or must take mandatory breaks. For intense activities (415W; see 134 ISO:7243 for examples) the following equation was constructed to calculate the lost 135 productivity in one hour (13): 136 𝐿𝑊𝐻=1 − {1 −0.165∗𝑊𝐵𝐺𝑇+5.3982 0{𝑖𝑓 𝑊𝐵𝐺𝑇 < 26.55918 𝑖𝑓 26.55918 ≤ 𝑊𝐵𝐺𝑇 < 32.59783 𝑖𝑓 𝑊𝐵𝐺𝑇 ≥ 32.59783 137 o LWH for moderate activities: Depending on the WBGT, workers lose productivity or 138 must take mandatory breaks. For moderate activities (300W; see ISO:7243 for 139 examples) the following equation was constructed to calculate the lost productivity in 140 one hour (13): 141 𝐿𝑊𝐻=1 − {1 −0.2195∗𝑊𝐵𝐺𝑇+7.2043 0{𝑖𝑓 𝑊𝐵𝐺𝑇 < 28.2656 𝑖𝑓 28.2656 ≤ 𝑊𝐵𝐺𝑇 < 32.82141 𝑖𝑓 𝑊𝐵𝐺𝑇 ≥ 32.82141 142 o LWH for light activities: Depending on the WBGT, workers lose productivity or must 143 take mandatory breaks. For light activities (180W; see ISO:7243 for examples) the 144 following equation was constructed to calculate the lost productivity in one hour (13): 145 𝐿𝑊𝐻=1 − {1 −0.5∗𝑊𝐵𝐺𝑇+16.5 0{𝑖𝑓 𝑊𝐵𝐺𝑇 < 31.0 𝑖𝑓 31.0 ≤ 𝑊𝐵𝐺𝑇 < 33.0 𝑖𝑓 𝑊𝐵𝐺𝑇 ≥ 33.0 146 - Impact indicators 147 o Population exposed to heatwave warning days 148 o Population exposed to heat stress days 149 4. EXPERIMENTAL DESIGN, MATERIALS AND METHODS 150 The UrbClim model (14) is used to derive the hourly meteorological output to calculate the indicators. 151 UrbClim is an urban boundary layer climate model, which is designed to dynamically downscale large152 scale climate information to the extent of individual cities and their rural surroundings at very high 153 resolution (up to 100m). The UrbClim model consists of a land surface scheme containing simplified 154 urban physics, coupled to a 3-D atmospheric boundary layer module, taking into account the 155 conservation equations of momentum, temperature, humidity and mass, while also specifically 156 accounting for turbulent fluxes and the mixing layer. The atmospheric boundary layer is tied to 157 synoptic-scale meteorological fields through the lateral and top boundary conditions, to ensure that 158 the synoptic forcing is properly considered. 159 The present-day simulation period spans a time period of 10 years from 2008-2017. For this period, 160 the UrbClim model is forced at its top and lateral boundaries by large-scale synoptic information from 161 the ERA-5 reanalysis product (an overview of the variables that are used can be found in (14). Apart 162 from meteorological input data, the main strength of the UrbClim model lies in a detailed 163 representation of the land surface properties. Depending on the region, different data sources have 164 been used to characterise the urban surroundings (Supplementary Table 2). These data sources are 165 resampled to the city modelling domains at 100m spatial resolution and allow to obtain spatial 166 heterogeneity within the urban canopy. Details on the approach can be found in (9,14). 167 The UrbClim model produces hourly output for meteorological variables such as temperature, 168 humidity, wind speed, but also soil properties and energy fluxes at 100m spatial resolution. Next to 169 these basic meteorological variables, heat stress (Wet Bulb Globe Temperature) is calculated based on 170 the model of Liljegren et al. (2008) (15). This metric accounts for temperature, humidity and radiation 171 and serves as a proxy for perceived temperature. It is calculated following ISO:7243 using the 172 meteorological output of UrbClim and solar radiation information from the reanalysis dataset ERA-5. 173 To accurately downscale radiation to 100m resolution, a detailed representation of the building 174 footprints and trees within the modelling domain is necessary. Their effect is two-fold. On the one 175 hand, they cast shade, while on the other hand, buildings also absorb and emit radiation, adding an 176 extra source of radiation. Detailed building footprint information is obtained from Open Street Map, 177 Google Africa Buildings and Microsoft Building Footprints, while the fraction of trees in each 100m 178 pixel is defined depending on the land use. 179 The future climate forcing data is obtained from the MESMER-FaIR ensemble. Both FaIR and MESMER 180 are climate model emulators that, with limited computational effort, can provide a large ensemble of 181 climate model realisations. The FaIR emulator (16) is used to translate greenhouse gas emissions to 182 the total strength of the forcing imposed on the climate system. This allows to calculate the (change 183 in) Global Mean Temperature (GMT), constrained by both historic warming and expected future 184 changes set out by the Intergovernmental Panel on Climate Change (IPCC). GMT is used by MESMER 185 (17) to emulate the evolution of key climate variables over land for each of the given Earth temperature 186 trajectories obtained from FaIR. In this work, the monthly downscaled module of MESMER is used, 187 MESMER-M. 188 For each of the 142 cities, the following scenarios have been considered: 189 - 2020 climate policies (IPCC AR6 scenario): This scenario assumes that no further climate action 190 is taken beyond the climate policies that were in place in 2020. Global warming reaches 2.9°C 191 in 2100 (best estimate), and would continue climbing into the new century. 192 - Delayed climate action (Gradual strengthening scenario in IPCC AR6): This scenario assumes 193 that decarbonisation is delayed to the 2030s, but then takes place in earnest. Fossil fuel use 194 never ends but is instead compensated for with high amounts of carbon dioxide removal. 195 Global warming in 2100 reaches 1.7°C (best estimate). 196 - Shifting pathway (IMP-SP scenario in IPCC AR6): This scenario explores how a broader shift 197 towards sustainable development can be combined with stringent climate policies. Global 198 warming peaks at 1.6°C in 2060 and goes back to 1.3°C in 2100 (best estimate). 199 The low spatial resolution and monthly temporal resolution prevents us from performing accurate 200 future simulations of UrbClim dynamically driven by the MESMER-M ensemble. For example, changes 201 in monthly average temperature might underestimate changes in the highest temperature quantiles 202 (i.e. extreme temperatures generally change with higher amounts than average temperatures, which 203 are of most interest in our study. To address this, we apply the quantile mapping bias algorithm (18). 204 Average monthly changes in temperature for each decade in the 2.5°x2.5° grid cell in which each city 205 is located are obtained from MESMER-M, while changes in different temperature quantiles (10 in total) 206 for different changes in monthly temperature are obtained from the CMIP6 archive, which has a higher 207 time resolution (daily). It is used to calculate changes in quantiles of daily temperature for different 208 levels of monthly temperature change within the city. These perturbations are added to the historical 209 data simulated by UrbClim, leading to a time series of the same length and time scale as the historical 210 time series but representative of future climate conditions. The approach above is applied for the three 211 future climate scenarios forcing it with data from the mean, 5th and 95th percentiles of the ensemble 212 of MESMER-M realisations until 2100. This provides an estimate of the uncertainty around the mean 213 changes in the calculated indicators. 214 Apart from meteorological information that is directly obtained from the UrbClim model, impacts are 215 computed using the open-sourced and open-access natural hazard risk model, CLIMADA (CLIMate 216 ADAptation) (19). The impact is calculated based on three components: hazard, exposure, and 217 vulnerability. The hazard data is obtained from the city-scale meteorological modelling with the 218 UrbClim model. The exposure is defined as the population and is obtained at 100m spatial resolution 219 from WorldPop for each of the 142 cities. The WorldPop Constrained Individual countries 2020 UN 220 adjusted data provides the top-down constrained gridded population data which is adjusted to match 221 the United Nations national estimate. Thus, the population data is validated for official reference. The 222 original WorldPop dataset is on a country scale. To match the hazard data on a city scale, the original 223 country-level gridded population data is trimmed into the specific city-level gridded population data. 224 Vulnerability, such as age group, is not considered. Each person in the exposure has an equal weighting 225 to the hazard. 226 LIMITATIONS 227 An important limitation of the dataset is that it assumes static urban morphology. In reality, cities will 228 expand and transform in ways that can influence both local climate responses and the number of 229 people. By design, these factors are held constant to isolate the large-scale climate change signal at 230 high resolution for present-day cities. The dataset should therefore be interpreted as a climatological 231 baseline for assessing potential climate impacts, rather than as a projection of future urban 232 conditions. Future extensions could combine this framework with urban growth scenarios and 233 dynamic demographic projections to generate more application-oriented estimates of future urban 234 climate risk. 235 236 ETHICS STATEMENT 237 The authors have read and follow the ethical requirements for publication in Data Brief. We thereby 238 confirm that the current work does not involve human subjects, animal experiments, or any data 239 collected from social media platforms. 240 241 CRediT AUTHOR STATEMENT 242 Niels Souverijns: Conceptualization, Data Curation, Methodology, Formal Analysis, Writing. Dirk 243 Lauwaet: Conceptualization, Methodology, Formal Analysis. Quentin Lejeune: Conceptualization, 244 Methodology. Chahan M. Kropf: Data Curation, Formal Analysis. Kam Lam Yeung: Data Curation, 245 Formal Analysis. Shruti Nath: Data Curation, Formal Analysis. Carl F. Schleussner: Conceptualization, 246 Funding acquisition. All authors reviewed the manuscript. 247 248 ACKNOWLEDGEMENTS 249 This project has received funding from the European Union’s Horizon Europe research and 250 innovation programme under grant agreement No 101003687. 251 252 DECLARATION OF COMPETING INTERESTS 253 The authors declare that they have no known competing financial interests or personal relationships 254 that could have appeared to influence the work reported in this paper. 255 256 REFERENCES 257 1. Zhao Q, Guo Y, Ye T, Gasparrini A, Tong S, Overcenco A, et al. Global, regional, and national 258 burden of mortality associated with non-optimal ambient temperatures from 2000 to 2019: a three259 stage modelling study. The Lancet Planetary Health. 2021 July;5(7):e415–25. 260 2. Perkins-Kirkpatrick SE, Lewis SC. Increasing trends in regional heatwaves. Nat Commun. 2020 261 July 3;11(1):3357. 262 3. Marcotullio PJ, Keßler C, Fekete BM. The future urban heat-wave challenge in Africa: 263 Exploratory analysis. Global Environmental Change. 2021 Jan;66:102190. 264 4. United Nations. World Population Prospects 2022: Summary of Results [Internet]. United 265 Nations; 2022 [cited 2023 Dec 1]. (Statistical Papers - United Nations (Ser. A), Population and Vital 266 Statistics Report). Available from: https://www.un-ilibrary.org/content/books/9789210014380 267 5. Fischer EM, Oleson KW, Lawrence DM. Contrasting urban and rural heat stress responses to 268 climate change. Geophysical Research Letters. 2012 Feb;39(3):2011GL050576. 269 6. Wouters H, De Ridder K, Poelmans L, Willems P, Brouwers J, Hosseinzadehtalaei P, et al. Heat 270 stress increase under climate change twice as large in cities as in rural areas: A study for a densely 271 populated midlatitude maritime region. Geophysical Research Letters. 2017 Sept 16;44(17):8997– 272 9007. 273 7. Li D, Yuan J, Kopp RE. Escalating global exposure to compound heat-humidity extremes with 274 warming. Environ Res Lett. 2020 June 1;15(6):064003. 275 8. Saeed F, Schleussner C, Ashfaq M. Deadly Heat Stress to Become Commonplace Across South 276 Asia Already at 1.5°C of Global Warming. Geophysical Research Letters. 2021 Apr 277 16;48(7):e2020GL091191. 278 9. Lauwaet D, Berckmans J, Hooyberghs H, Wouters H, Driesen G, Lefebre F, et al. High 279 resolution modelling of the urban heat island of 100 European cities. Urban Climate. 2024 280 Mar;54:101850. 281 10. O’Donnell MS, Ignizio DA. Bioclimatic Predictors for Supporting Ecological Applications in the 282 Conterminous United States. U.S. Geological Survey; 2012 p. 10. (U.S. Geological Survey Data Series 283 691). 284 11. Romanello M, Di Napoli C, Drummond P, Green C, Kennard H, Lampard P, et al. The 2022 285 report of the Lancet Countdown on health and climate change: health at the mercy of fossil fuels. 286 The Lancet. 2022 Nov;400(10363):1619–54. 287