Weynants et al. (2025) Dheed: an ERA5 based global database of compound dry and hot extreme events from 1950 to 2023
Identification
- Journal: Earth system science data
- Year: 2025
- Date: 2025-11-28
- Authors: Mélanie Weynants, Chaonan Ji, Nora Linscheid, Ulrich Weber, Miguel D. Mahecha, Fabian Gans
- DOI: 10.5194/essd-17-6621-2025
Research Groups
- Max Planck Institute for Biogeochemistry, Jena, Germany
- Institute for Earth System Research and Remote Sensing, Leipzig University, Leipzig, Germany
- Helmholtz-Centre for Environmental Research, UFZ, Leipzig, Germany
Short Summary
This study introduces Dheed, a novel global database of compound dry and hot (CDH) extreme events from 1950 to 2023, derived from ERA5 reanalysis data, and confirms a significant increase in the frequency and spatial extent of these events over recent decades.
Objective
- To develop and present Dheed, a global, gridded, daily-scale database of compound dry and hot (CDH) extreme events from 1950 to 2023, based on ERA5 reanalysis data.
- To characterize the spatiotemporal patterns and trends of CDH events and validate the database against reported historical events.
Study Configuration
- Spatial Scale: Global, with a spatial resolution of 0.25° in both latitude and longitude (1440 x 721 grid cells). Analysis focused on land areas.
- Temporal Scale: 1 January 1950 to 31 December 2023 (74 years). Hourly ERA5 data aggregated to daily time steps. Drought indicators (PEI) calculated over 30, 90, and 180 antecedent days. CDH events are defined as lasting at least three consecutive days.
Methodology and Data
- Models used:
- FAO's Penman-Monteith equation for calculating hourly reference evapotranspiration (ET0).
- Empirical quantile-based thresholds (1st percentile) for detecting extreme daily maximum temperature (Tmax) and drought indicators (PEI).
- Spatio-temporal connected component analysis for grouping discrete extreme occurrences (DEOs) into unique CDH events.
- Theil-Sen approximation for trend analysis.
- Julia programming language with YAXArrays.jl, ImageFiltering.jl, and ImageMorphology.jl packages.
- Data sources:
- ERA5 climate reanalysis data (Hersbach et al., 2020, 2023) from the European Centre for Medium-Range Weather Forecasts (ECMWF). Variables include 2 m temperature (T2), 10 m zonal and meridional wind speed (u10, v10), atmospheric surface pressure (Ps), surface net solar and thermal radiation (S, L), saturation water vapour pressure (es), vapour pressure (ea), and total precipitation (PT).
- Validation against a priori compiled list of 40 historical extreme events reported in scientific and grey literature.
- Comparison of drought detection with EOBS-based daily Standardized Precipitation Evapotranspiration Index (SPEI) (Pohl et al., 2023) and ERA5-based daily SPEI (Liu et al., 2024) at ICOS sites.
Main Results
- The Dheed database contains 26,351 unique labeled CDH events for the period 1970–2023.
- A highly significant positive trend was observed in the global annual percentage of land area and days affected by compound dry and hot extreme events from 1970 to 2023 (Theil-Sen estimator: 0.0066 % per year).
- This increasing trend in CDH events was an order of magnitude larger in recent years (1998–2023) compared to earlier years (1970–1997).
- The year 2023 recorded the highest percentage of extremely dry and hot annual days and land area.
- Europe is the continent most affected by CDH extremes, followed by Africa and South America. Africa experienced the steepest increase in annual cumulative area subject to CDH days. Antarctica and Oceania showed non-significant trends.
- While most events are short (median duration of 4 days) and spatially small, the distribution of spatiotemporal volume has consistently shifted towards larger events in recent years.
- All ten largest and ten longest labeled CDH events identified in Dheed occurred after the year 2000 and could be linked to documented droughts and/or heatwaves in the scientific or grey literature (e.g., 2010 Russian heatwave, 2012 USA drought, 2023 Amazon drought).
- Validation against 40 independently reported historical extreme events showed that 38 could be associated with labeled CDH events in Dheed. Two reported droughts not associated with heatwaves were not captured.
- Comparison of Dheed's extremely dry days (1% threshold) with SPEI < -2 from two other datasets showed partial agreement (29% of Dheed's dry days were not classified as extremely dry by both other SPEI datasets), with agreement increasing for longer accumulation periods.
Contributions
- Provides the first global, gridded database of compound dry and hot (CDH) extreme events at a daily scale, offering explicit spatio-temporal delineation of individual events.
- Offers a unified, analysis-ready dataset (Dheed) that reconciles the differing timescales of drought and heat by using multi-scalar daily drought indicators (30, 90, 180 days).
- Facilitates comprehensive studies on the impacts of CDH events on ecosystems, specific species, and society, serving as a basis for sampling high-resolution satellite imagery and assessing ecological monitoring networks.
- Establishes a generic event detection pipeline that can be adapted for investigating other types of meteorological extreme events and their combinations.
Funding
- European Space Agency (AI4Science project "Multi-Hazards, Compounds and Cascade events: DeepExtremes")
- European Space Agency (AI4Science project "Climate Adaptation, Extremes, Multi-Hazards and Geo-Hazards Science: ARCEME")
- European Space Agency (AI4Science project "The DeepESDL AI-Ready Earth System Data Lab")
- EU Horizon 2020 Framework Programme, H2020 Societal Challenges (grant no. 101003469)
- Max Planck Society (for article processing charges)
Citation
@article{Weynants2025Dheed,
author = {Weynants, Mélanie and Ji, Chaonan and Linscheid, Nora and Weber, Ulrich and Mahecha, Miguel D. and Gans, Fabian},
title = {Dheed: an ERA5 based global database of compound dry and hot extreme events from 1950 to 2023},
journal = {Earth system science data},
year = {2025},
doi = {10.5194/essd-17-6621-2025},
url = {https://doi.org/10.5194/essd-17-6621-2025}
}
Original Source: https://doi.org/10.5194/essd-17-6621-2025