Forced versus Unforced Variability in the North American Monsoon

The North American Monsoon (NAM) system, in Southwestern US and Northwestern Mexico, is a critical component of the region's hydrology, with an average of 60% of the annual precipitation occurring during the peak monsoon months of June through September. Therefore it is important to understand what controls the year-to-year and long-term variability of monsoon precipitation. Previous analyses of global climate model simulations of NAM indicate that it has dynamical aspects that are challenging for the current class of models used in recent major intercomparisons, such as the 3rd and 5th Climate Model Intercomparison Projects (CMIP3 and CMIP5). While there have been improvements in the model representation of monsoon timing from CMIP3 to CMIP5 (although only two thirds of the CMIP5 models have a phase lag of zero months), a significant wet bias persists in CMIP5. Dynamical and statistical downscaling studies of NAM suggest that resolution plays an important role in representing the NAM system due to the interaction between multiscale circulation and the complex (small-scale) topography in the region. Toward investigating the hypothesis that resolution improves the representation of moisture transport and precipitation in simulations of NAM, we compare the climatology of the NAM region between a low and high resolution global climate model (100 km and 25 km respectively). Further, we investigate the role of ocean variability in forcing interannual variability in the NAM by comparing high resolution GCM simulations with and without interannually varying sea surface temperatures. Our results suggest that high resolution indeed improves the dynamics of NAM and that a surprisingly large fraction of NAM interannual variability is driven by intrinsic land-atmosphere variability rather than by ocean variability.