Precipitation characteristics of CAM5 physics at mesoscale resolution during MC3E and the impact of convective timescale choices

The CAM5 atmospheric physics components have recently been ported into the regional WRF model where their behavior can now be investigated under various scenarios that would be more difficult if attempted with their original global implementation. For example, a regional configuration can be used with known large-scale synoptic conditions such that direct comparisons with observations can be made in both time and space. The differences in the simulated fields can then be assumed to come from local behavior within the domain and not from biases in large-scale wave patterns that could be present in global domains.

This study takes advantage of the new capabilities by investigating the precipitation characteristics of the CAM5 physics suite at a mesoscale resolution with 32 km grid spacing, which is close to where CAM5 will be regularly run in the next few years. The Midlatitude Continental Convective Clouds Experiment (MC3E) has been chosen as a test period representing springtime weather conditions in the central United States. A series of simulations have been performed and the precipitation characteristics have been evaluated to quantify general precipitation behavior. With the default configuration, which in WRF uses a convective timescale parameter in the Zhang-McFarlane deep convective parameterization of 600 s, the model generates the total precipitation accurately in a regional total sense, but the day-to-day variability is too tied to the diurnal cycle. By using a longer convective timescale of 3600 s, which is the default for 2-degree grid spacing in the global CAM, the link to the diurnal cycle weakens and the diurnal cycle is more accurately predicted. However, this comes at the cost of accurately predicting the quantity of precipitation. Using an even longer convective timescale exacerbates this problem. This trade-off between accurate timing and amount is similar to what has been documented at the coarser scales where CAM5 has been run in the past. The fact that the Zhang-McFarlane parameterization acts similarly at mesoscale resolutions as at the coarser GCM resolutions is useful knowledge that helps one appropriately choose the resolution dependent parameters. However, it emphasizes the deficiencies in the ability of the CAM5 suite as a whole to accurately predict all aspects of the precipitation budget.