A generalized methodology for ranking climate models based on climate indices for sector-specific studies: An application to the Mekong sub-basin.

This study proposes a methodological framework to evaluate and rank climate models based on extreme climate indices of precipitation and temperature for impact studies in seven sectors: Cryosphere, Energy, Forestry/GHGs, Health, Agriculture & Food Security, Disaster Risk Reduction (flood and drought), and Water Resources & Hydrology. The ranking of the climate models is based on their performance in sector-relevant extreme climate indices. Extreme climate indices for observed and climate models' datasets for a historical period and overall performance statistics were used to create a payoff matrix. The payoff matrix then served as an input to a multi-criteria decision-making process to rank the climate models for each of the climate indices. The final sector-specific ranking was achieved by averaging the ranks obtained in the sector-relevant indices. The developed methodology is demonstrated with an application to the Songkhram River Basin (Thailand), a sub-basin of the Mekong. Eighteen CMIP6 GCMs are used for the proposed evaluation and ranking processes and four performance statistics were used. Weights to each of the four performance statistics were determined using the entropy method. Compromise programming was applied to rank the GCMs based on the distance technique. The results indicate that the six best performing models are different for different sectors, with the GFDL_CM4 model common in all the seven sectors considered in the study. KACE1_0_G, GFDL_ESM4, GFDL_CM4, MRI_ESM2_0, and ACCESS_ESM1_5 models are the five top (ranked 1 to 5 respectively) performing models for the Water Resources & Hydrology sector. The developed framework is generic and can be applied to any region or basin; at the same time, it can also provide researchers and policymakers with specific information on best-performing models for particular sectors.