Evaluation of CMIP5 historical simulations in the Colorado River Basin

The Colorado River Basin (CRB) is the primary source of water in the

southwestern United States. A key step to reduce the uncertainty of future streamflow

projections in the CRB is to evaluate the performance of historical simulations of General

Circulation Models (GCMs). In this study, this challenge is addressed by evaluating the

ability of nineteen GCMs from the Coupled Model Intercomparison Project Phase Five

(CMIP5) and four nested Regional Climate Models (RCMs) in reproducing the statistical

properties of the hydrologic cycle and temperature in the CRB. To capture the transition

from snow-dominated to semiarid regions, analyses are conducted by spatially averaging

the climate variables in four nested sub-basins. Most models overestimate the mean

annual precipitation (P) and underestimate the mean annual temperature (T) at all

locations. While a group of models capture the mean annual runoff at all sub-basins with

different strengths of the hydrological cycle, another set of models overestimate the mean

annual runoff, due to a weak cycle in the evaporation channel. An abrupt increase in the

mean annual T in observed and most of the simulated time series (~0.8 °C) is detected at

all locations despite the lack of any statistically significant monotonic trends for both P

and T. While all models simulate the seasonality of T quite well, the phasing of the

seasonal cycle of P is fairly reproduced in just the upper, snow-dominated sub-basin.

Model performances degrade in the larger sub-basins that include semiarid areas, because

several GCMs are not able to capture the effect of the North American monsoon. Finally,

the relative performances of the climate models in reproducing the climatologies of P and

T are quantified to support future impact studies in the basin.