Rivoire et al. (2026) Identification of hydro-meteorological drivers for forest low greenness events in Europe

Identification

• Journal: Natural hazards and earth system sciences
• Year: 2026
• Date: 2026-03-09
• Authors: Pauline Rivoire, Sonia Dupuis, Antoine Guisan, Pascal Vittoz, Daniela I. V. Domeisen
• DOI: 10.5194/nhess-26-1183-2026

Research Groups

• Institute of Earth Surface Dynamics, Faculty of Geosciences and Environment, University of Lausanne, Lausanne, Switzerland
• Laboratory of Cryospheric Sciences (CRYOS), School of Architecture, Civil and Environmental Engineering (ENAC), École Polytechnique Fédérale de Lausanne (EPFL), Lausanne, Switzerland
• Institute of Geography, University of Bern, Bern, Switzerland
• Oeschger Centre for Climate Change Research, University of Bern, Bern, Switzerland
• Institute for Atmospheric and Climate Science, ETH Zurich, Zurich, Switzerland

Short Summary

This study identifies hydro-meteorological drivers of forest low greenness events across Europe using a random forest model and satellite Normalized Difference Vegetation Index (NDVI) data. It reveals that warm and dry conditions in spring and early summer, along with multi-year influences, are critical predictors for forest browning, with regional and forest-type specific variations.

Objective

• To identify and quantify the most critical hydro-meteorological drivers (available from operational sub-seasonal to seasonal forecast systems) for forest low greenness events (browning) across Europe at monthly to annual timescales.

Study Configuration

• Spatial Scale: Europe-wide analysis, with independent random forest models run at each 0.5° × 0.5° grid point. Input NDVI data at 0.01° resolution (upscaled to 0.1° for binarization), and hydro-meteorological data at 0.1° (temperature, soil moisture) and 0.5° (precipitation, latent heat flux).
• Temporal Scale: Data period from 1980/1981 to 2022. Analysis focuses on summer (July–August) low greenness events, with hydro-meteorological predictors considered at monthly to seasonal timescales up to 18 months prior to the observed browning.

Methodology and Data

• Models used: Random Forest (RF) classification model for prediction and driver importance. LASSO logistic regression used as a benchmark for predictive skill.
• Data sources:
  • Vegetation Greenness: Normalized Difference Vegetation Index (NDVI) 10-day composite dataset from Advanced Very High Resolution Radiometers (AVHRR) LAC data (1981–2022).
  • Hydro-meteorological Variables: ERA5 and ERA5-Land reanalysis datasets (1980–2022) for:
    • Maximum 2 meter air temperature (Max T2m)
    • Soil moisture at 28–100 centimeter depth (soil moist.)
    • Total precipitation (total precip.)
    • Surface latent heat flux (s.l. heat flux)
  • Forest Classification: CORINE Land Cover dataset (2006, 2012, 2018) for broad-leaved, coniferous, and mixed forest types.

Main Results

• The random forest model exhibits high predictive performance for forest browning across Europe, with a critical success index (CSI) greater than 0.75 for 65% of grid points and an Area Under the Curve (AUC) greater than 0.8 for 98% of grid points.
• Warm and dry conditions in spring and early summer are identified as the most essential predictors for forest browning. Maximum 2 meter temperature in June and July, and soil moisture in July, are among the most frequently selected top predictors.
• Multi-year influences are significant, with soil moisture and temperature anomalies from the preceding year playing a role, particularly in Scandinavia and for coniferous forests.
• Non-linear relationships are uncovered, such as both positive and negative precipitation anomalies at different lags contributing to browning risk for coniferous forests.
• For broad-leaved forests, hot and dry conditions in spring and early summer (current or preceding year) are strongly associated with browning.
• For coniferous forests, the influence of past conditions is more heterogeneous, with spring, autumn, and winter of the previous year contributing to browning risk, alongside current year summer temperatures.

Contributions

• Presents a novel, large-scale, spatially explicit analysis of forest browning drivers across Europe using a homogeneous and automated random forest modeling framework.
• Enables a region-specific comparison of hydro-meteorological drivers, capturing the diversity of forest responses across the continent.
• Identifies critical hydro-meteorological drivers at monthly to annual timescales using variables available from operational sub-seasonal to seasonal forecast systems, providing actionable insights for proactive forest management.
• Demonstrates high predictive skill of the random forest model for forest browning across Europe, outperforming a LASSO regression benchmark.
• Uncovers significant multi-year influences and non-linear relationships between hydro-meteorological variables and forest browning.

Funding

• Early Career Postdoctoral Fellowship from the Faculty of Geosciences and the Environment at the University of Lausanne.
• European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement no. 847456).
• Swiss National Science Foundation through project PP00P2_198896.
• Dr. Alfred Bretscher Stipendium for climate and air pollution research of the University of Bern.

Citation

@article{Rivoire2026Identification,
  author = {Rivoire, Pauline and Dupuis, Sonia and Guisan, Antoine and Vittoz, Pascal and Domeisen, Daniela I. V.},
  title = {Identification of hydro-meteorological drivers for forest low greenness events in Europe},
  journal = {Natural hazards and earth system sciences},
  year = {2026},
  doi = {10.5194/nhess-26-1183-2026},
  url = {https://doi.org/10.5194/nhess-26-1183-2026}
}