Modeling Multi-Scale Precipitation Variability in the Southwest US
Jiang, P., Gautam, M.R., Zhu, J. and Yu, Z. 2013. How well do the GCMs/RCMs capture the multi-scale temporal variability of precipitation in the southwestern United States. Journal of Hydrology 479: 75-85.
In an attempt to determine if today's climate models do indeed have that capability (or not), Jiang et al. assessed "the ability of 16 GCMs from the Bias Corrected and Downscaled (BCSD) World Climate Research Program's (WCRP's) Coupled Model Intercomparison 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 observed station data," focusing on four regions in the Southwest United States (Los Angeles, Las Vegas, Tucson and Cimarron), since these places "represent four different precipitation regions classified by clustering method." And in so doing, they specifically investigated "how storm properties and seasonal, inter-annual, and decadal precipitation variabilities differed between GCMs/RCMs and observed records in these regions."
The four U.S. researchers report that "RCMs tend to  simulate longer duration,  shorter inter-storm periods, and  lower storm intensity than observed." Moreover, they say that  "RCMs fail to simulate high average storm intensity during the summer period as seen in observed precipitation records." They also say that  bias-corrected and downscaled GCMs "lack the ability to reproduce observed monthly precipitation patterns." In addition, they note that "observed precipitation tends to be above average during the PDO warm phase, while precipitation during the PDO cold phase is below average," and that  "most of the considered GCMs failed to reproduce similar variability." And, last of all, they say their wavelet analysis revealed that  "even the successful GCMs on reproducing the low-frequency variability associated with ENSO and PDO, showed inconsistency in the occurrence or timing of 2-8-year bands."
Jiang et al. conclude that their "comparative analyses suggest that current GCMs/RCMs do not adequately capture multi-scale temporal variability of precipitation," and, therefore, they say that "using GCM/RCM output to conduct future flood projections is not creditable."