Monte et al. (2025) Skilful seasonal predictions of droughts in the Mediterranean region

Identification

Journal: Climate Dynamics
Year: 2025
Date: 2025-10-28
Authors: Thomas Dal Monte, Andrea Alessandri, Annalisa Cherchi, Marco Gaetani
DOI: 10.1007/s00382-025-07872-9

Research Groups

Department of Science, Technology and Society, University School for Advanced Studies IUSS, Pavia, Italy
Institute of the Atmospheric Sciences and Climate (CNR-ISAC), National Research Council of Italy, Bologna, Italy

Short Summary

This study investigates the skill of seasonal prediction systems (SPSs) in forecasting meteorological drought in the Mediterranean region using SPI3 and SPEI3 indices. It demonstrates that optimized multi-model ensembles (MME) significantly enhance drought prediction skill, outperforming individual systems and climatology across most of the region.

Objective

Assess the skill of state-of-the-art Seasonal Prediction Systems (SPSs) in forecasting meteorological drought (Standardised Precipitation Index (SPI3) and Standardised Precipitation Evapotranspiration Index (SPEI3)) in the Mediterranean region across four seasons.
Identify the most effective Multi-Model Ensemble (MME) strategies for maximizing drought prediction skill.

Study Configuration

Spatial Scale: Mediterranean region, defined within the domain [10° W-40° E, 30° N-50° N], with a spatial resolution of 1° x 1°.
Temporal Scale: Hindcast period from 1993 to 2016 (24 years, 23 years for winter season), analyzing 3-month drought indices (SPI3, SPEI3) at a 1-month lead time, across four seasons (DJF, MAM, JJA, SON).

Methodology and Data

Models used: Eight Seasonal Prediction Systems (SPSs) from the Copernicus Climate Data Store (CDS): CMCC, ECMWF System 5, Météo France System 8, UK Met Office GloSea5-GC2, Deutscher Wetterdienst GCFS 2.0, US National Weather Service’s NCEP, Japan Meteorological Agency CPS3, and Environment and Climate Change Canada GEM5-NEMO. Multi-model ensemble (MME) strategies included combinations of the best-performing SPSs (B1, B2, B3, B4), the maximum skill from these (MaxB), and a full MME (B8).
Data sources:
- SPSs: 2-meter temperature and precipitation data from the Copernicus Climate Data Store (CDS) archive.
- Observations: Gridded global total precipitation and potential evapotranspiration data from the Climate Research Unit CRU4.07 (upscaled to 1° x 1° resolution).
- Sensitivity analysis: GPCC monthly dataset from the Global Precipitation Climatology Centre.
- Drought indices: Standardised Precipitation Index (SPI3) and Standardised Precipitation Evapotranspiration Index (SPEI3), with potential evapotranspiration calculated using the Hargreaves equation. Drought events were defined when the index fell below -1 (or the 16.67th percentile).
- Evaluation metric: Brier Skill Score (BSS) with climatology as the reference, calculated using a leave-one-out cross-validation (LOO-CV) procedure, and significance assessed at a 5% level using a bootstrap technique (1000 iterations).

Main Results

Individual SPSs showed limited skill, with median BSS values generally around zero, indicating performance often worse than climatology for 50% of the grid points.
SPEI3 demonstrated higher predictability than SPI3, particularly during summer and spring seasons, with distinct spatial patterns of skill.
Optimized MME strategies (MaxB) significantly improved prediction skill, achieving positive BSS values (outperforming climatology) for 80–90% of grid points across all seasons.
The median BSS for MaxB was approximately 0.1, with upper quartiles exceeding 0.2, especially for summer SPEI3. Peak skill values (BSS > 0.3) were observed for autumn SPI3.
The "best single SPS" (B1) strategy contributed the most to areas with high skill (BSS > 0.2), accounting for 60–80% of such grid points. MME combinations (B2, B3, B4) provided additional benefits in regions where individual model performance was moderate (0 < BSS < 0.2).
Regions exhibiting high and significant prediction skill (BSS > 0.1-0.2) included the Iberian Peninsula, North Africa, the Balkans, Anatolia, and the Middle East, particularly during summer and spring for SPEI3, and summer (Spain, North Africa) and autumn (Middle East) for SPI3.
Winter prediction skill was generally lower for both indices compared to other seasons.
High Pearson Correlation Coefficients (0.5 to 0.75) between observed and predicted SPI3 in the Middle East during autumn were identified as a key factor for the high skill in that region.

Contributions

Provides a comprehensive, multi-system evaluation of seasonal meteorological drought prediction skill across the entire Mediterranean region, extending beyond previous assessments focused on individual models or limited areas.
Demonstrates that optimized multi-model ensemble (MME) strategies significantly enhance drought prediction skill, offering robust improvements over individual models and climatological forecasts.
Identifies the most effective MME combination strategies for maximizing skill, showing that tailored optimization can yield better results than a full MME, especially for high-skill areas.
Highlights regional and seasonal variations in predictability, revealing that SPEI3 often outperforms SPI3 in drier regions and seasons, suggesting the importance of temperature-dependent processes in drought predictability.
Offers a foundation for developing more effective early warning systems for drought-sensitive sectors in the Mediterranean by leveraging the complementary strengths of multiple forecasting systems.

Funding

PhD programme in Sustainable Development And Climate Change at the University School for Advanced Studies IUSS Pavia, Cycle XXXVIII, supported by a scholarship financed by Ministerial Decree no. 351 of 9th April 2022, based on the NRRP - funded by the European Union - NextGenerationEU - Investment 4.1 “Extension of the number of research doctorates and innovative doctorates for public administration and cultural heritage”.
European Union’s Horizon Europe Research and Innovation Program under Grant Agreement No. 101081193 (OptimESM project).
ICSC—Centro Nazionale di Ricerca in High Performance Computing, Big Data and Quantum Computing, funded by European Union—NextGenerationEU (Concession Decree N. 1031 of 17/06/2022 adopted by the Italian Ministry of University and Research).
Project “Dipartimenti di Eccellenza 2023–2027” (funded by the Italian Ministry of Education, University and Research at IUSS Pavia).

Citation

@article{Monte2025Skilful,
  author = {Monte, Thomas Dal and Alessandri, Andrea and Cherchi, Annalisa and Gaetani, Marco},
  title = {Skilful seasonal predictions of droughts in the Mediterranean region},
  journal = {Climate Dynamics},
  year = {2025},
  doi = {10.1007/s00382-025-07872-9},
  url = {https://doi.org/10.1007/s00382-025-07872-9}
}

Generated by BiblioAssistant using gemini-2.5-flash (Google API)

Original Source: https://doi.org/10.1007/s00382-025-07872-9