Ershadfath et al. (2025) Selecting CMIP6 precipitation models by integrating relative importance metrics, compromise programming index, and Jenks optimized classification

Identification

• Journal: The Science of The Total Environment
• Year: 2025
• Date: 2025-11-14
• Authors: Farnaz Ershadfath, Rouhollah Davarpanah, Zulfaqar Sa’adi, Mikołaj Piniewski, Dennis Trolle, Jørgen E. Olesen
• DOI: 10.1016/j.scitotenv.2025.180935

Research Groups

• Water Engineering Department, Faculty of Agriculture, Bu-Ali Sina University, Hamedan, Iran
• Department of Hydrology, Meteorology and Water Management, Institute of Environmental Engineering, Warsaw University of Life Sciences, Warsaw, Poland
• Centre for Environmental Sustainability and Water Security (IPASA), Research Institute for Sustainable Environment (RISE), Universiti Teknologi Malaysia (UTM), Johor Bahru, Malaysia
• WaterITech ApS, Skanderborg, Denmark
• Department of Agroecology, Aarhus University, Tjele, Denmark

Short Summary

This study introduces a novel framework combining Relative Importance Metrics (RIMs), Compromise Programming Index (CPI), and Jenks Optimized Classification (JOC) to evaluate 14 CMIP6 General Circulation Models (GCMs) for projecting precipitation across Iran. MPI-ESM1–2-LR was identified as the top-ranked model, projecting significant precipitation declines from February to September, especially in the far future under high emission scenarios.

Objective

• To introduce a novel framework integrating Relative Importance Metrics (RIMs), Compromise Programming Index (CPI), and Jenks Optimized Classification (JOC) to evaluate 14 CMIP6 GCMs for projecting precipitation in Iran.
• To identify the most suitable CMIP6 GCM for spatiotemporal precipitation projections in Iran, addressing predictor collinearity in GCM evaluation.
• To project future spatiotemporal precipitation distribution for the near (2025–2054) and far (2055–2084) future under Shared Socioeconomic Pathways (SSPs) 1–2.6, 2–4.5, and 5–8.5 scenarios.

Study Configuration

• Spatial Scale: Nationwide study across Iran, utilizing 154 grid points. GCM data were re-gridded to a 1° spatial resolution.
• Temporal Scale:
  • Historical/Evaluation period: 1985–2014 (30 years)
  • Near future: 2025–2054 (30 years)
  • Far future: 2055–2084 (30 years)
  • Monthly and annual precipitation projections.

Methodology and Data

• Models used:
  • 14 CMIP6 General Circulation Models (GCMs): ACCESS-CM2, ACCESS-ESM1–5, CESM2-WACCM, CMCC-CM2-SR5, CMCC-ESM2, FGOALS-g3, GFDL-ESM4, INM-CM4–8, INM-CM5–0, MIROC6, MPI-ESM1–2-LR, NESM3, NOR-ESM2-MM, NorESM2-LM.
  • Ranking methods: Relative Importance Metrics (RIMs) (betasq, pratt, genizi, last, lmg, first, car), Compromise Programming Index (CPI), Jenks Optimized Classification (JOC).
  • Bias correction method: Delta Change (DC).
• Data sources:
  • Gridded climate data: ERA5 (reanalysis product, 0.25° resolution), CHIRPS (gauge-satellite-based product, 0.25° resolution).
  • Observational data: Monthly precipitation from 31 synoptic stations across Iran (for validation of gridded datasets).
  • Future scenarios: Shared Socioeconomic Pathways (SSPs) 1–2.6, 2–4.5, and 5–8.5.

Main Results

• MPI-ESM1–2-LR consistently ranked as the top-performing CMIP6 GCM for precipitation simulation across Iran, based on both ERA5 and CHIRPS datasets, achieving a CPI value of 0.
• Both ERA5 and CHIRPS datasets showed reasonable agreement with observations, with mean correlation coefficients of 0.71 for ERA5 and 0.61 for CHIRPS. ERA5 generally outperformed CHIRPS in capturing precipitation variability.
• Future annual precipitation projections (bias-corrected with MPI-ESM1–2-LR) indicated a mix of positive and negative changes in the near future (2025–2054).
• The far future (2055–2084) showed more pronounced annual precipitation declines, particularly under SSP2–4.5 and SSP5–8.5. Under SSP5–8.5 in the far future, mean annual precipitation decreased by -12.4 % (CHIRPS) and -15.5 % (ERA5).
• Over 90 % of grid points showed reduced annual precipitation under SSP2–4.5 and SSP5–8.5 in the far future (e.g., 93 % for CHIRPS under SSP5-8.5, 96 % for ERA5 under SSP5-8.5).
• Monthly projections indicated significant precipitation reductions from February to September across nearly all scenarios, reaching up to -18.7 mm (ERA5) and -14.2 mm (CHIRPS) in the far future.
• Increased monthly precipitation was generally observed from October to January, with January and December projected to experience the most significant increases under SSP5–8.5 (approximately 61-69 % increase).

Contributions

• Introduces a novel and robust framework for GCM selection by integrating Relative Importance Metrics (RIMs), Compromise Programming Index (CPI), and Jenks Optimized Classification (JOC) at a national scale, which is a methodological innovation for country-scale climate change studies.
• Provides the first comprehensive, nationwide evaluation of 14 CMIP6 GCMs for precipitation projection over Iran, addressing a regional data gap.
• Highlights the value of employing multiple gridded datasets (ERA5 and CHIRPS) for GCM evaluation, revealing complementary insights and dataset-specific biases.
• Offers a reliable and adaptable approach for identifying top-performing climate models, applicable to diverse geographical regions and other climatic variables.
• Delivers detailed spatiotemporal precipitation projections for Iran under various SSPs, providing crucial evidence for water resource management and adaptation strategies in a water-scarce region.

Funding

Not explicitly stated in the paper.

Citation

@article{Ershadfath2025Selecting,
  author = {Ershadfath, Farnaz and Davarpanah, Rouhollah and Sa’adi, Zulfaqar and Piniewski, Mikołaj and Trolle, Dennis and Olesen, Jørgen E.},
  title = {Selecting CMIP6 precipitation models by integrating relative importance metrics, compromise programming index, and Jenks optimized classification},
  journal = {The Science of The Total Environment},
  year = {2025},
  doi = {10.1016/j.scitotenv.2025.180935},
  url = {https://doi.org/10.1016/j.scitotenv.2025.180935}
}