The North American summer monsoon circulation exhibits substantial interannual variability. Establishing clear links between this variability and the slowly varying boundary forcing (sea surface temperatures - SSTs - and land surface conditions) has proven difficult. Recent observational and regional modeling studies have suggested that anomalies in snow cover/soil moisture in the Rocky Mountains/Great Basin have an impact on the monsoon. However, the scarcity of land surface observations has precluded a detailed assessment of this impact. Furthermore, our recent analysis of a century-long rainfall dataset shows that these relationships are evident only after 1962. Thus, the possibility remains that a substantial portion of the variability is due to dynamical processes internal to the atmosphere. Determining the relative roles of internal and lower boundary forcing processes in producing interannual variations is the major objective of our current research. Our methodology is based on analysis of simulations performed with the UCLA atmospheric general circulation model (AGCM). Two 20-year long simulations have been completed, one using climatological SSTs and the other using observed SSTs for the period 1979-1998 (CONTROL and AMIP, respectively).

We have compared the simulated and observed variability of monsoon precipitation in the southwest U.S. in CONTROL and found that its pattern and magnitude are similar to that observed. These results suggest that internal atmospheric processes alone can produce variability comparable in magnitude to that observed. In the AMIP simulation we do not find a systematic impact of SST anomalies on the monsoon. There is no increase in the overall interannual variability in the U.S. part of the monsoon. Furthermore, there is little correspondence between observed 'wet' and 'dry' monsoon years and the 'wet' and 'dry' years in the AMIP simulation. In fact, three of the four dry monsoon years in the simulation were observed to be wetter than normal.

We are examining three possible explanations for this result: 1) atmospheric circulation anomalies are forcing the extratropical SST anomalies rather than the SST anomalies forcing atmospheric anomalies, 2) the single realization of each year is not sufficient to distinguish the SST-forced signal from the 'natural variability' (noise), which according to our CONTROL simulation is substantial, 3) land surface processes are playing a significant mediating role in the relationship between SSTs and monsoon strength. To address 2) we have performed a twenty-member ensemble of integrations for the summer season (May-September) forced by the global SST anomaly field based on an observed composite of 'wet' minus 'dry' monsoon years. We again find that there is almost no systematic impact of the SST anomalies on the atmospheric circulation over North America. To investigate 1), we are currently performing an ensemble of simulations in which the AGCM is coupled to a mixed-layer ocean model. An analysis of this ensemble will be presented. To address 3), we have coupled a model of land surface processes (SSiB) to the AGCM. A brief overview of the SSiB land surface model will be presented along with preliminary results from AGCM integrations demonstrating the impact of including land surface processes.