Low-level clouds in the Southern California Bight: Sensitivity studies in mesoscale models

We investigate the ability of regional mesoscale models, specifically MM5 and WRF, to correctly simulate the low-level cloud cover in the Southern California Bight. Simulations are run for July 2003 and then compared with visible satellite images three times daily to assess goodness of cloud cover amount and location. The control run generally underestimates the amount of low cloud cover over the bight with a mean error of 72%. The control run improves to 49% error if only considering the morning hours. Tests of new boundary layer and shallow cloud parameterizations show only a slight improvement over the control run with mean error of 67% in low cloud amount. These new parameterizations have previously been shown to perform well at coarse horizontal resolution (~50 km), but still need improving at the high resolution used in this study, which is necessary to capture both the complicated flow patterns and heterogeneous cloud structures common in coastal zones (< 6 km). Vertical resolution was also found to be important with a 16% improvement in simulated low cloud amount when the number of vertical levels was increased from 24 to 38 in the control run. Possible avenues for model improvement will be discussed.