Selecting CMIP6 precipitation models by integrating relative importance metrics, compromise programming index, and Jenks optimized classification

Accurate projections of future climate are key to sustainable long-term water management and planning, and rely on the rigorous assessment of the performance of General Circulation Models (GCMs). While various methods have been employed to rank GCMs, further exploration of advanced ranking approaches is necessary to enhance the reliability of the GCM selection for specific applications and regions. This nationwide study introduces a novel framework that combines three ranking approaches-Relative Importance Metrics (RIMs), Compromise Programming Index (CPI), and Jenks Optimized Classification (JOC)-to evaluate 14 different CMIP6 GCMs and identify the most suitable models for projecting the spatiotemporal distribution of precipitation across 154 grid points in Iran. The performance of GCMs was evaluated using gridded climate data from ERA5 and CHIRPS for the period 1985-2014. MPI-ESM1-2-LR emerged as the top-ranked model for both datasets, with a CPI value of 0, and was subsequently bias-corrected using the Delta Change method. The selected model was then used to project the spatiotemporal distribution of precipitation for the near future (2025-2054) and far future (2055-2084) under three scenarios representing SSP1-2.6, SSP2-4.5, and SSP5-8.5. The findings from both datasets indicated a mix of positive and negative precipitation changes in the near future, whereas the far future showed more declines, particularly under SSP2-4.5 and SSP5-8.5, highlighting risks of water scarcity across more than 90 % of Iran's grid points. Precipitation increases were observed between October and January, while from February to September, nearly all scenarios projected substantial reductions, reaching up to -60 % in the far future.