Evaluating Global Climate Models for Hydrological Studies of the Upper Colorado River Basin

Abstract Three generations of global climate models (GCMs), Coupled Model Intercomparison Project version 3 (CMIP3), CMIP5, and CMIP6, are evaluated for performance simulating seasonal mean and annual‐to‐decadal variability of temperature and precipitation in the Upper Colorado River Basin. Low‐frequency precipitation variability associated with drought is a particular focus and found to be a significant model shortcoming. The evaluation includes remote teleconnected atmospheric responses to the Pacific Ocean, including the El Niño/Southern Oscillation and Pacific Decadal Oscillation. GCMs have improved their simulation of the Upper Basin over model generations, but primarily in atmospheric circulation metrics. Persistent winter precipitation biases have changed little, including in multiyear precipitation variability. Users generally bias‐corrected GCM data before use; evaluation using a simple spatially and temporally averaged bias correction shows that the CMIP6 models outperform earlier generations after the bias correction, although more complex precipitation biases remain even after the simple bias correction. These model rankings will be useful when selecting GCMs for a variety of hydrological and ecological climate studies in the Upper Basin.