Testing a Dynamically Computed Convective Time Scale in the Kain-Fritsch Scheme for Meso-Gamma-Scale Convection Parameterization in WRF

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 4 February 2014: 2:45 PM
Room C209 (The Georgia World Congress Center )
O. Russell Bullock Jr., EPA, Research Triangle Park, NC; and K. Alapaty, J. A. Herwehe, T. L. Otte, and C. G. Nolte

The U.S. Environmental Protection Agency is using the Weather Research and Forecasting (WRF) model to dynamically downscale global climate information to horizontal scales as fine as 12 kilometers. Specifically, we are using the Advanced Research WRF (WRF-ARW) employing sub-grid-scale convection schemes. We are finding a general positive bias in simulated precipitation, especially when the horizontal grid spacing encroaches into the meso-gamma scale. The positive bias at 12-km grid spacing is most severe in locations where sub-grid convective precipitation dominates over model-resolved precipitation. The Kain-Fritsch convective scheme was updated to allow sub-grid cloud effects in the RRTMG radiation scheme (Alapaty et al., Geophys. Res. Lett., 2012), which alleviates a portion of the positive precipitation bias at the 12-km scale. However, a significant positive bias persists. In this work we have further modified the Kain-Fritsch scheme to set the time scale for sub-grid-scale convection based on convection depth and updraft velocity and we find significant further improvement in the simulated precipitation.

The default method for estimating the time scale over which convection removes available buoyant energy in the Kain-Fritsch scheme is based on the model grid spacing and the horizontal wind speed at the lifted condensation level and the mid-troposphere, essentially forcing the sub-grid convection to dissipate at least 90% of buoyant energy based on a Courant condition. This convective time scale is restricted to a range of 30-60 minutes. Nonetheless, finer grid spacing tends to produce a shorter convective time scale for a given wind speed. This leads to more vigorous convection for our 12-km grid spacing which, one could argue, is where sub-grid parameterizations should be yielding to resolved model processes rather than becoming more vigorous.

This work will compare 12-km WRF simulations using the Kain-Fritsch scheme in its original form, with sub-grid radiation effects, and with both sub-grid radiation and the new convective time scale. Model simulations of precipitation are evaluated against monthly data from the Parameter-elevation Regressions on Independent Slopes Model (PRISM). The results show approximately one-third of the positive bias in precipitation is removed with sub-grid radiation effects and another third is removed by employing the new dynamically computed convective time scale.