Najafi et al. (2025) Assessment of C3S monthly to seasonal climate forecast models for mean and extreme precipitation over Iran

Identification

• Journal: Natural Hazards
• Year: 2025
• Date: 2025-12-26
• Authors: Mohammad Saeed Najafi, S. M. Sani, Razieh Noroozian, Samin Danandeh Saribaglou
• DOI: 10.1007/s11069-025-07786-z

Research Groups

• Department of Water Resources Study and Research (WRR), Water Research Institute (WRI), Tehran, Iran
• Department of Meteorology and Climatology, Doctoral School of Exact and Natural Sciences, Nicolaus Copernicus University, Toruń, Poland
• Department of water and hydraulic structures, civil engineering, K. N. Toosi University of Technology, Tehran, Iran

Short Summary

This study evaluates the performance of seven Copernicus Climate Change Service (C3S) models and their multi-model ensemble (MME) in predicting mean and extreme precipitation over Iran. It finds that MME consistently outperforms individual models, with ECMWF and UKMO showing the highest skill, especially in western and northeastern Iran and at shorter lead times.

Objective

• To evaluate the performance of seven Copernicus Climate Change Service (C3S) models and their multi-model ensemble (MME) in predicting mean and extreme precipitation indices across Iran.
• To identify the most reliable models for seasonal precipitation forecasting to support water resource management and disaster preparedness in Iran.
• To understand the intrinsic strengths and limitations of raw model outputs without statistical post-processing.

Study Configuration

• Spatial Scale: Iran, covering diverse climatic zones, evaluated at 148 synoptic stations.
• Temporal Scale: Historical period of 1997–2016 (20 years) for observations and forecasts, with lead times of one, two, and three months. Monthly and annual precipitation totals, and annual extreme precipitation indices.

Methodology and Data

• Models used: Seven C3S seasonal forecast models (CFSv2, CMCC, DWD, ECMWF, JMA, Météo-France, UKMO) and a Multi-Model Ensemble (MME) generated using a multivariate regression (MRMM) technique.
• Data sources:
  • Observational data: Daily and monthly precipitation from 148 synoptic stations across Iran (1997–2016), obtained from Iran’s Meteorological Office (IRIMO).
  • Seasonal forecast data: Daily global precipitation forecasts from C3S models (1997–2016) at 1.0° spatial resolution, initialized monthly, accessed via C3S Climate Data Store.
  • Teleconnection index data: Eight Sea Surface Temperature-driven Teleconnection (SSTDT) indices (AMO, IOD, ONI, PDO, Niño 1+2, 3, 3.4, 4) from the NOAA website.
• Evaluation Metrics: Nash–Sutcliffe Efficiency (NSE) and Pearson correlation coefficient (Corr).
• Extreme Precipitation Indices: Six ETCCDI indices: Consecutive Dry Days (CDD), Consecutive Wet Days (CWD), Annual maximum 1-day precipitation (Rx1Day), Annual maximum 5-day precipitation (Rx5Day), Number of days with precipitation exceeding 10 mm (R10mm), and Annual total precipitation when daily precipitation > 95th percentile (R95p).
• Evaluation Technique: Point-pixel evaluation, comparing station observations with the nearest grid cell value, with an altitude-based regression correction for predicted values.

Main Results

• The Multi-Model Ensemble (MME) consistently outperformed individual C3S models across all climate zones and lead times for both mean and extreme precipitation.
• Among individual models, ECMWF and UKMO systems achieved the highest skill, especially for shorter lead times (1-month), while models like Météo-France and CFSv2 generally performed worse.
• Forecast skill was notably higher in western and northeastern Iran, regions influenced by Mediterranean and Sudanese low-pressure systems, and declined as lead time increased.
• Forecasts were weakest along the humid Caspian coast and in southeastern Iran, where precipitation is influenced by small-scale convective systems or monsoons, which are challenging for coarse-resolution models.
• Three SSTDT indices (Nino1+2, AMO, PDO) showed a statistically significant relationship with monthly precipitation in Iran. UKMO and ECMWF models were particularly effective at simulating this relationship at shorter lead times.
• For extreme precipitation indices, MME generally performed best. ECMWF and UKMO were the most reliable individual models, particularly for western and northeastern regions, while CFSv2, CMCC, and DWD were less effective.

Contributions

• Provides a comprehensive, station-by-station assessment of all seven raw C3S seasonal forecast models and their MME for both mean and extreme precipitation over Iran, addressing a gap in existing literature.
• Quantifies the intrinsic strengths and limitations of raw dynamical model outputs without statistical corrections, which is crucial for understanding their core capabilities in a region with high climatic variability and complex topography.
• Identifies the most effective C3S models and the MME for different regions and lead times in Iran, offering operationally relevant guidance for water resource management and disaster preparedness.
• Highlights the varying skill of models across Iran's diverse climatic zones and for different extreme precipitation indices, providing detailed regional insights.

Funding

• No specific funding projects, programs, or reference codes were explicitly listed in the paper.

Citation

@article{Najafi2025Assessment,
  author = {Najafi, Mohammad Saeed and Sani, S. M. and Noroozian, Razieh and Saribaglou, Samin Danandeh},
  title = {Assessment of C3S monthly to seasonal climate forecast models for mean and extreme precipitation over Iran},
  journal = {Natural Hazards},
  year = {2025},
  doi = {10.1007/s11069-025-07786-z},
  url = {https://doi.org/10.1007/s11069-025-07786-z}
}