Egli et al. (2026) Detecting anthropogenically induced changes in extreme and seasonal evapotranspiration observations

Identification

• Journal: Nature Communications
• Year: 2026
• Date: 2026-01-23
• Authors: Marius Egli, Sebastian Sippel, Reto Knutti, Vincent Humphrey
• DOI: 10.1038/s41467-025-67748-8

Research Groups

• Institute for Atmospheric and Climate Science, ETH Zurich, Zürich, Switzerland
• Institute for Meteorology, Leipzig University, Leipzig, Germany
• Federal Office of Meteorology and Climate MeteoSwiss, Zurich, Switzerland

Short Summary

This study investigates anthropogenically induced changes in extreme and seasonal evapotranspiration (ET) using climate models and observational data. It robustly detects increases in high ET extremes and seasonal mean ET in observational products, particularly in the Northern Hemisphere, indicating an increased risk of flash droughts.

Objective

• To examine whether increasing evaporative demand, driven by anthropogenic climate change, has led to detectable increases in high evapotranspiration (ET) extremes and seasonal mean ET in observational products.
• To test whether projected increases in potential ET manifest in actual ET, especially when focusing on high ET extremes (defined as the seven consecutive days with maximum ET, or ETx7d).
• To identify regions that have experienced robust changes in extreme or seasonal ET and assess the consistency of these changes with climate model simulations.

Study Configuration

• Spatial Scale: Global, hemispheric (Northern and Southern), and regional (IPCC AR6 SREX regions over land).
• Temporal Scale:
  • Observational data periods: 1950–2023 (ERA5 Land), 1980–2023 (GLEAM), 2001–2021 (X-BASE).
  • Primary analysis period for trends: 1980–2023.
  • Climate model simulations: Historical (from 1850), SSP5-8.5 (future), and PiControl (pre-industrial, continuous simulations split into 44-year chunks for trend estimation).

Methodology and Data

• Models used:
  • CMIP6 climate models (23 models for historical/SSP5-8.5 scenarios, 15 models for PiControl simulations).
  • Regularized linear regression (ridge regression) for detection and attribution of global and hemispheric changes, reducing internal variability and accounting for climate model disagreement.
  • Theil-Sen trends for regional analysis across SREX regions.
• Data sources:
  • Observational products:
    • ERA5 Land reanalysis (global, 0.25° spatial resolution, 1950–2023).
    • Global Land Evaporation Amsterdam Model Version 4.2a (GLEAM) (global, 0.1° spatial resolution, 1980–2023).
    • X-BASE (FLUXCOM product, global, 0.25° spatial resolution, 2001–2021).
  • IPCC AR6 SREX regions for subcontinental analysis.
  • Meteorological variables from ERA5 Land for composite analysis (e.g., precipitation, net shortwave radiation, surface air temperature, relative humidity, surface pressure, volumetric soil water).

Main Results

• High ET extremes (ETx7d) are primarily driven by warm spells characterized by substantial atmospheric moisture demand and high incoming radiation, with peak values reaching approximately 1.16 x 10^-7 m/s (10 mm/day). These events are intensified by anthropogenic climate change.
• Robust increases in extreme (ETx7d) and seasonal mean ET were detected in ERA5 Land and GLEAM observational datasets for the period 1980–2023, particularly in the Northern Hemisphere.
• The observed trends in ETx7d and Northern Hemisphere June-July-August (JJA) ET are statistically very unlikely (<10%) to exceptionally unlikely (<1%) to occur without anthropogenically induced climate change.
• Regionally, seasonal mean ET shows mixed increases and decreases from 1980 to 2023, while ETx7d universally increases or shows no significant change.
• In regions where both ETx7d and JJA mean ET increased, ETx7d trends were generally larger (e.g., up to 8.68 x 10^-11 m s^-1 year^-1), indicating an additional increase in ET variability beyond a shift in the mean.
• Higher latitude regions (e.g., Northwestern North America, Northern Europe, Russian Arctic, Russian Far East) consistently show increases in both ETx7d and JJA ET.
• Western Central Europe (WCE) exhibits strong and robust increases in both ETx7d and JJA mean ET, likely driven by increased surface radiation (global brightening) and temperature trends.
• In regions with decreasing JJA mean ET (e.g., Western North America, Southeastern South America, Southeastern Africa), no robust change in ETx7d was found, suggesting water limitation. Climate models generally failed to capture these observed decreases.
• Observational products and climate models agree on a tendency towards more recent record-setting ETx7d events in Europe.
• Regions experiencing strong ETx7d trends are at an increased risk of flash droughts due to a more rapid transition from wet to dry conditions.

Contributions

• Provides robust detection and attribution of anthropogenically induced changes in both extreme (ETx7d) and seasonal mean evapotranspiration using a sophisticated regularized regression model, effectively reducing the influence of internal climate variability.
• Highlights a critical divergence between mean and extreme ET trends, demonstrating that ETx7d trends are often larger than seasonal mean trends, which signifies an increase in ET variability and an amplified seasonal range.
• Systematically investigates regional changes across IPCC AR6 SREX regions, identifying specific areas globally with robust increases in extreme ET and assessing their consistency with CMIP6 climate models.
• Connects observed ET changes to an increased risk of flash droughts, providing further evidence for the alteration of the hydrological cycle as a fundamental consequence of climate change.
• Compares multiple observational products with a large ensemble of CMIP6 models, revealing areas of agreement and disagreement, particularly where models may not capture observed regional decreases in seasonal ET.

Funding

• Project ‘Constraints on near-term warming projections via distributionally robust statistical and machine learning’ (COPE; grant agreement C22-02), funded by the Swiss Data Science Center.
• Open access funding provided by Swiss Federal Institute of Technology Zurich.

Citation

@article{Egli2026Detecting,
  author = {Egli, Marius and Sippel, Sebastian and Knutti, Reto and Humphrey, Vincent},
  title = {Detecting anthropogenically induced changes in extreme and seasonal evapotranspiration observations},
  journal = {Nature Communications},
  year = {2026},
  doi = {10.1038/s41467-025-67748-8},
  url = {https://doi.org/10.1038/s41467-025-67748-8}
}