Lindersson et al. (2025) SHEDIS-Temperature: linking temperature-related disaster impacts to subnational data on meteorology and human exposure

Identification

• Journal: Earth system science data
• Year: 2025
• Date: 2025-11-24
• Authors: Sara Lindersson, Gabriele Messori
• DOI: 10.5194/essd-17-6379-2025

Research Groups

• Department of Earth Sciences, Uppsala University, Sweden
• Swedish Centre for Impacts of Climate Extremes (climes), Uppsala, Sweden
• Centre of Natural Hazards and Disaster Science (CNDS), Uppsala, Sweden

Short Summary

This paper introduces SHEDIS-Temperature, an open-access dataset linking temperature-related disaster impacts from EM-DAT with subnational meteorological and population exposure data to enable comprehensive analysis of disaster drivers and outcomes. It reveals that more populated subnational areas tend to experience higher temperatures, highlighting critical risk hotspots.

Objective

• To bridge the knowledge gap in disaster impact databases by providing subnational hazard and exposure information for temperature-related disaster impact records, thereby enabling more comprehensive analysis of drivers of severe disaster outcomes.

Study Configuration

• Spatial Scale: Subnational administrative units (level-1: province/equivalent, level-2: county/district/equivalent) for 2835 locations in 71 countries, with underlying data at 0.1° grid resolution.
• Temporal Scale: 1979–2018, with meteorological data at 3-hourly intervals and population data annually interpolated from 5-year estimates.

Methodology and Data

• Models used:
  • Apparent temperature calculation (Steadman, 1994) using the "apparentTemp" function in the R package "HeatStress".
  • Percentile-based threshold analysis for event detection (31-day moving window, 1981-2010 reference period).
• Data sources:
  • Emergency Events Database (EM-DAT): International disaster impact records.
  • Geocoded Disasters (GDIS): Geocoded EM-DAT records to subnational locations.
  • Global database of Administrative boundaries (GADM v3.6): Administrative subdivision polygons.
  • Multi-Source Weather (MSWX-Past): High-resolution (0.1°, 3-hourly) meteorological data (2 m air temperature, 2 m relative humidity, 10 m wind speed), bias-corrected and downscaled from ERA5 reanalysis.
  • Global Human Settlement Population grids (GHS-POP): Global population maps (30 arcsec resolution, 5-year time series 1975-2020).
  • E-OBS dataset (v31.0e): Europe-wide station-based daily ensemble mean of maximum and minimum temperatures for validation.

Main Results

• SHEDIS-Temperature provides hazard and exposure data for 2835 subnational locations linked to 382 disaster records (243 cold waves, 139 heat waves) from 1979–2018 in 71 countries.
• The dataset reveals distinct patterns of human exposure, with North America, Europe, and northern Asia experiencing very cold extremes in highly populated areas, while India, Pakistan, and Bangladesh are notably affected by high temperatures and large populations.
• A global pattern shows that warmer administrative subdivisions tend to be more populated, especially in heat wave-impacted areas.
• Percentile-exceeding events were detected for a vast majority of disnos (all 139 heat wave disnos had at least one pct90 event; 233 of 243 cold wave disnos had pct10 events), supporting the reliability of EM-DAT reports.
• India consistently ranks highest in person-days exposed to both extreme heat (pct95) and cold (pct05) events.
• Technical validation against E-OBS showed high consistency (Pearson correlation coefficients ≥0.9, mean absolute errors ≤2 °C) for extreme daily temperatures, with minor systematic biases (slight overestimation of minimums and underestimation of maximums by MSWX).
• Comparison with EM-DAT reported temperatures showed reasonable consistency for heat waves (MAE = 2.6 °C) but greater variability for cold waves (MAE = 8.3 °C), highlighting challenges in capturing localized extremes and potential reporting discrepancies in EM-DAT.

Contributions

• Presents SHEDIS-Temperature, an open-access dataset linking subnational temperature-related disaster impacts with detailed hazard and exposure information, bridging a critical knowledge gap in existing disaster impact databases.
• Provides comprehensive hazard metrics, including absolute indicators (e.g., apparent temperature accounting for humidity and wind speed) and percentile-based indicators of temperature exceedances, derived from high-resolution 3-hourly data.
• Offers detailed population exposure data, including annual population figures for impacted units and person-days of exposure to threshold-exceeding temperatures.
• Delivers outputs at three levels (grid point, subdivision, and EM-DAT record) to support diverse interdisciplinary research needs.
• Enhances the reliability of disaster impact analysis by systematically integrating and cross-verifying data from multiple sources, addressing limitations of existing databases like EM-DAT regarding spatiotemporal detail and hazard magnitude.

Funding

• Swedish Centre for Impacts of Climate Extremes (climes)
• Centre of Natural Hazards and Disaster Science (CNDS)
• Swedish Research Council (Vetenskapsrådet; grant no. 2022-06599)
• Formas (grant no. 2023-01774)

Citation

@article{Lindersson2025SHEDISTemperature,
  author = {Lindersson, Sara and Messori, Gabriele},
  title = {SHEDIS-Temperature: linking temperature-related disaster impacts to subnational data on meteorology and human exposure},
  journal = {Earth system science data},
  year = {2025},
  doi = {10.5194/essd-17-6379-2025},
  url = {https://doi.org/10.5194/essd-17-6379-2025}
}