In this paper we focus on the performance of the FSL special model runs for these two convective types. The models run during IHOP were a 10 km version of the RUC, and a 12 km and nested 4 km version of MM5. The MM5 runs included an initialization using LAPS and a special "hotstart" method to capture convection ongoing at the model start. The MM5 model utilized the Kain-Fritsch convective parameterization for the 12-km grid, whereas the 4-km grid did not employ any convective parameterization. For both grids, microphysical processes were represented using the Schultz scheme. Since the conclusion of IHOP there have been some updates to the hotstart package to address overprediction of pcpn sometimes observed durg the first few hours of the fcst; these were incorporated into reruns for the entire IHOP period of the MM5/12 km model as well as a 12 km version of the WRF (no reruns were made of the RUC). These updates better matched the LAPS cloud initialization to the model microphysics. Further, the convective parameterization was turned off in the reruns, in part because the LAPS initialization of cloud updrafts partially makes up for the known tendency for delayed convective initiation that results at these resolutions when no parameterization of convection is employed.
In this paper we include the reruns in our assessment, in an attempt to determine whether models of this type have an ability to forecast if supercell storms and storm organization (upscale growth) can occur, even at the somewhat large grid resolution of 10 and 12 km. We will also use, where possible, the IHOP MM5/4 km runs for comparison to examine the effect of much better resolution, and also hope to rerun the post-IHOP versions of the MM5 and in particular the WRF at 4 km resolution for a few selected cases.
Supplementary URL: http://laps.fsl.noaa.gov/szoke/IHOP/ihopmodeleval.html