Cerato et al. (2025) Summer Drought Predictability in the Euro-Mediterranean Region in Seasonal Forecasts

Identification

• Journal: Journal of Hydrometeorology
• Year: 2025
• Date: 2025-11-07
• Authors: Giada Cerato, Katinka Bellomo, Jost von Hardenberg
• DOI: 10.1175/jhm-d-25-0039.1

Research Groups

• Politecnico di Torino, Department of Environment, Land, and Infrastructure Engineering, Turin, Italy
• Department of Geosciences, University of Padova, Padova, Italy
• National Research Council, Institute of Atmospheric Sciences and Climate, Turin, Italy
• Copernicus Climate Change Service (C3S) contributing modeling institutions (ECMWF, Météo-France, U.K. Met Office, CMCC, DWD, ECCC)

Short Summary

This study evaluates the ability of state-of-the-art seasonal forecast systems to predict summer drought in Europe, finding that the Standardized Precipitation Evapotranspiration Index (SPEI-3) offers more spatially coherent and higher forecast skill than the Standardized Precipitation Index (SPI-3), particularly in southern Europe. The multimodel ensemble (MME) provides the most robust solution for early summer drought detection, demonstrating widespread significant skill up to a 1-month lead time.

Objective

• To identify the most predictable drought index for forecasting summer drought conditions across Europe.
• To rigorously assess the probabilistic skill of current Copernicus Climate Change Service (C3S) systems based on the identified index.

Study Configuration

• Spatial Scale: Europe, specifically the Euro-Mediterranean region (11°W–33°E, 34°N–70°N), with analysis distinguishing between the Mediterranean (south of 49°N) and Northern Europe (north of 49°N).
• Temporal Scale: Boreal summer (June–August, JJA) drought conditions. Seasonal forecasts were evaluated for lead times of 0, 1, 2, and 3 months (initialized on 1 June, 1 May, 1 April, and 1 March, respectively). The verification period spans 1993–2023.

Methodology and Data

• Models used: State-of-the-art seasonal prediction systems from the Copernicus Climate Change Service (C3S):
  • European Centre for Medium-Range Weather Forecasts (ECMWF) SEAS5
  • Météo-France (MF) System 8
  • U.K. Met Office (UKMO) GloSea6-GC3.2
  • Centro Euro-Mediterraneo sui Cambiamenti Climatici (CMCC) CMCC-SPS3.5
  • Deutscher Wetterdienst (DWD) GCFS 2.1
  • Environment and Climate Change Canada (ECCC) GEM5.2-NEMO
  • A Multimodel Ensemble (MME) constructed by combining 10 randomly sampled members from each individual system.
• Data sources:
  • E-OBS gridded observational dataset (Cornes et al. 2018) for precipitation, minimum, and maximum temperature data at 0.25° spatial resolution and daily frequency (1993–2023), remapped to 1° spatial resolution.
  • ERA5 reanalysis data (Hersbach et al. 2020) for comparison.
  • Drought indices: Standardized Precipitation Evapotranspiration Index (SPEI-3) and Standardized Precipitation Index (SPI-3), calculated over a 3-month time scale (JJA). Potential evapotranspiration (PET) for SPEI was estimated using the Hargreaves and Samani (1985) method.
  • Forecast skill evaluation metrics: Anomaly Correlation Coefficient (ACC), Brier Score (BS), Area Under the Receiver Operating Characteristic curve (AUC), Fair Continuous Ranked Probability Score (FCRPS), and Rank Histograms (RH).

Main Results

• SPEI-3 consistently exhibits higher predictability and more spatially coherent skill for summer drought conditions across Europe compared to SPI-3, particularly in southern Europe. This is attributed to the better predictability of temperature-related variables (e.g., daily temperature range) than precipitation.
• When initialized at the onset of the summer season (0-month lead time), all individual forecast models demonstrate good quality in terms of correlation, accuracy, reliability, and discrimination skills, with performance generally better in southern Europe than in northern Europe.
• The Multimodel Ensemble (MME) generally outperforms individual models, showing the highest percentage of grid points with statistically significant skill across various metrics. It provides the most robust solution for continental-scale applications, with significant skill for SPEI-3 forecasts up to a 1-month lead time.
• Forecast skill declines with increasing lead time across all models and the MME, and is generally lower and more spatially fragmented in northern Europe due to higher synoptic variability and poorer precipitation predictability.
• CMCC and UKMO models consistently achieve high rankings across different skill metrics, while DWD exhibits the most pronounced bias and underdispersion in its ensemble forecasts.

Contributions

• Provides a comprehensive, multimetric evaluation of state-of-the-art seasonal forecast systems from the Copernicus Climate Change Service (C3S) for summer drought conditions across the Euro-Mediterranean region.
• Demonstrates the superior predictability of SPEI-3 over SPI-3 for summer drought, explicitly linking this advantage to the higher predictability of temperature-related variables.
• Highlights the robustness and widespread skill of the multimodel ensemble (MME) as the most reliable option for early summer drought detection and risk management at a continental scale.
• Offers actionable insights into the spatial and temporal utility of seasonal drought predictions, informing climate-sensitive sectors on where and when forecasts are most reliable.

Funding

• European Union Next-GenerationEU [National Recovery and Resilience Plan (NRRP), Mission 4, Component 2, Investment 1.3 D.D. 1243 2/8/2022, PE0000005] within the RETURN Extended Partnership.
• "The Geosciences for Sustainable Development" project (Budget Ministero dell’Università e della Ricerca–Dipartimenti di Eccellenza 2023–27 C93C23002690001) for K.B.

Citation

@article{Cerato2025Summer,
  author = {Cerato, Giada and Bellomo, Katinka and Hardenberg, Jost von},
  title = {Summer Drought Predictability in the Euro-Mediterranean Region in Seasonal Forecasts},
  journal = {Journal of Hydrometeorology},
  year = {2025},
  doi = {10.1175/jhm-d-25-0039.1},
  url = {https://doi.org/10.1175/jhm-d-25-0039.1}
}