Najafi et al. (2025) The skill of RegCM4 in forecasting Iran’s precipitation: a basin-scale intra-seasonal to seasonal analysis

Identification

• Journal: Theoretical and Applied Climatology
• Year: 2025
• Date: 2025-09-12
• Authors: Mohammad Saeed Najafi, Omid Alizadeh
• DOI: 10.1007/s00704-025-05731-5

Research Groups

• Department of Water Resources Study and Research (WRR), Water Research Institute (WRI), Tehran, Iran
• Geography Department, Humboldt-Universität zu Berlin, Berlin, Germany

Short Summary

This study evaluates the skill of the RegCM4-CFSv2 model in forecasting intra-seasonal to seasonal precipitation over seven basins in Iran. It finds that the model exhibits moderate skill, particularly at shorter lead times and for extreme events, but performance declines with increasing lead time and shows significant regional variability.

Objective

• To investigate the impact of dynamically downscaling CFSv2 using an ensemble approach with RegCM4 to improve intra-seasonal to seasonal precipitation predictions at the regional scale over Iran.
• To evaluate the predictive performance of the RegCM-CFSv2 ensemble in simulating monthly and seasonal precipitation patterns across Iran using a regionalized approach.

Study Configuration

• Spatial Scale: Seven hydrological basins across Iran, covering an area of 1,648,000 square kilometers. RegCM4 was configured with a 20 kilometer horizontal resolution.
• Temporal Scale: 2000–2019 (20 years), focusing on the October-April wet season. Forecast lead times of one, two, and three months (L1, L2, L3).

Methodology and Data

• Models used: Regional Climate Model v4.7 (RegCM4.7) for dynamical downscaling, forced by the Climate Forecast System version 2 (CFSv2). An ensemble system with 24 members was used, combining multiple initial conditions, model parameters, and four sets of physical parameterization schemes (e.g., Grell, Emanuel, Kain-Fritsch, Tiedtke for cumulus convection; BATS, CLM4.5 for land surface).
• Data sources: CFSv2 model data for initial and boundary conditions. Monthly mean precipitation data from 157 synoptic stations across Iran (provided by the Iran Meteorological Organization) for verification.

Main Results

• The RegCM4-CFSv2 model demonstrated moderate skill in forecasting seasonal precipitation over Iran, with performance consistently declining as lead time increased.
• At a 1-month lead time (L1), the average Kling-Gupta efficiency (KGE) and correlation coefficient (CC) were approximately 0.31 and 0.37, respectively, with a relative root mean square error (RMSE) of about 32%.
• At a 3-month lead time (L3), performance significantly weakened, with average CC and KGE dropping to approximately 0.18 and 0.15, and average RMSE increasing to about 36%.
• Model performance exhibited significant regional variability, with better skill observed in basins along the southern Caspian Sea coastal plains (R2, R3) and western/northwestern regions (R1, R2), and lower skill in central and southeastern Iran (R7).
• Categorical evaluation showed moderate accuracy, particularly for ‘below-normal’ (BN) and ‘above-normal’ (AN) precipitation categories, suggesting utility for flood and drought management.
• The model was prone to false alarms, especially in the ‘normal’ (N) precipitation category, and showed a bias towards under-predicting BN events and over-predicting N events, which became more pronounced at longer lead times.
• A comparative analysis indicated that WRF-CFSv2 outputs generally exhibited greater accuracy than RegCM4-CFSv2 and the coarser CFSv2 for October-April seasonal precipitation forecasts in the studied basins.

Contributions

• This study provides a comprehensive, basin-scale evaluation of the RegCM4-CFSv2 ensemble's skill in intra-seasonal to seasonal precipitation forecasting over Iran, addressing a gap in the existing literature.
• It quantifies the benefits of dynamical downscaling for improving precipitation forecast accuracy at regional scales in a complex topographical domain like Iran.
• The research highlights the model's strengths in detecting extreme precipitation events, offering valuable insights for water resource management and disaster risk reduction strategies (flood/drought management).
• It identifies specific regional and temporal limitations of the RegCM4-CFSv2 model, guiding future research towards targeted improvements through multi-model ensembles or statistical post-processing.

Funding

• Water Research Institute (WRI), Ministry of Energy (server, hardware, and other supports)
• Iran National Science Foundation (INSF) (Grant number 99031973)

Citation

@article{Najafi2025skill,
  author = {Najafi, Mohammad Saeed and Alizadeh, Omid},
  title = {The skill of RegCM4 in forecasting Iran’s precipitation: a basin-scale intra-seasonal to seasonal analysis},
  journal = {Theoretical and Applied Climatology},
  year = {2025},
  doi = {10.1007/s00704-025-05731-5},
  url = {https://doi.org/10.1007/s00704-025-05731-5}
}