How well do the GCMs/RCMs capture the multi-scale temporal variability of precipitation in the Southwestern United States?

Multi-scale temporal variability of precipitation has an established relationship with floods and droughts. In this paper, we present the diagnostics on the ability of 16 General Circulation Models (GCMs) from Bias Corrected and Downscaled (BCSD) World Climate Research Program’s (WCRP’s) Coupled Model Inter-comparison Project Phase 3 (CMIP3) projections and 10 Regional Climate Models (RCMs) that participated in the North American Regional Climate Change Assessment Program (NARCCAP) to represent multi-scale temporal variability determined from the observed station data. Four regions (Los Angeles, Las Vegas, Tucson, and Cimarron) in the Southwest United States are selected as they represent four different precipitation regions classified by clustering method. We investigate how storm properties and seasonal, inter-annual, and decadal precipitation variabilities differed between GCMs/RCMs and observed records in these regions. We find that current GCMs/RCMs tend to simulate longer storm duration and lower storm intensity compared to those from observed records. Most GCMs/RCMs fail to produce the high-intensity summer storms caused by local convective heat transport associated with the summer monsoon. Both inter-annual and decadal bands are present in the GCM/RCM-simulated precipitation time series; however, these do not line up to the patterns of large-scale ocean oscillations such as El Nino/La Nina Southern Oscillation (ENSO) and Pacific Decadal Oscillation (PDO). Our results show that the studied GCMs/RCMs can capture long-term monthly mean as the examined data is bias-corrected and downscaled, but fail to simulate the multi-scale precipitation variability including flood generating extreme events, which suggests their inadequacy for studies on floods and droughts that are strongly associated with multi-scale temporal precipitation variability.