Assessment of the North American monsoon as represented in present day AOGCMs

The North American monsoon is a pronounced aspect of the warm-season hydrologic cycle over northern Mexico and the southwestern United States, and continental-scale teleconnections from the monsoon core have been described as a "robust" feature of the warm-season hydroclimate over the central U.S. Ability to predict the effects of climate change on the North American monsoon therefore is essential to water resource planning in much of the U.S. Motivated by this concern, we have assessed simulations of the North American monsoon from 17 coupled atmosphere-ocean general circulation models (AOGCMs) that have submitted results in support of the IPCC Fourth Assessment Report. We evaluated the ability of the models to represent the monsoon in the current climate by analyzing the period 1980-1999 from "Climate of the 20th century" runs from each available model. All models produced the main signal of the North American monsoon, i.e., an increase in precipitation over northwestern Mexico from June through August, though the magnitude of the increase varied from about half to about double the observed magnitude. Their ability to reproduce the observed teleconnections of the monsoon was poor. Most models showed a lag in the development of the monsoon when compared to observations. Changes to the North American monsoon in future climate were evaluated by comparing projected monthly precipitation for 2070-2099 to the 1960-1999 "Climate of the 20th Century" precipitation. Changes were mostly small and were not consistent among the models. Preliminary conclusions are that present-generation AOGCMs are seriously deficient in their representation of the North American monsoon, giving little confidence in their projections of future changes to the monsoon. It is recommended that the influence of various aspects of AOGCM configuration (especially horizontal resolution and the parameterization of deep convection) on model results for the North American monsoon be explored in a systematic way, and that regional climate models be used to dynamically downscale the AOGCM predictions.