As part of the International H₂O Project (IHOP) field program, which took place over the Southern Great Plains from 13 May to 25 June 2002, large amounts of high-resolution meteorological data were acquired from various platforms. These data, when assimilated into a mesoscale NWP model, could potentially provide for improvements in forecasts of convective events over those that might be made using more standard data sources. In this study, a southward propagating severe MCS is examined in this context, and the impact of the use of various mesoscale data sources and a complex cloud analysis scheme on analyses and forecasts of the event will be discussed.

The event in question took place on 15-16 June 2002. The MCS initially developed in the morning of the 15th over central Nebraska and propagated southward, reaching northern Oklahoma around 0Z on the 16th. The system then continued southward through central Texas, reaching the Gulf Coast around 12Z on the 16th. Its path was marked by several reports of wind damage. The MCS took on a distinct bow shape as it moved through Oklahoma.

High-resolution (9 km and smaller) analyses and simulations of this system were performed using the Advanced Regional Prediction System (ARPS), developed at the University of Oklahoma. As part of the analysis cycle NEXRAD Level II radar data and GOES satellite data were used in a complex cloud analysis routine. To examine the impact of the use of various data sources and the complex cloud analysis on the prediction of the MCS, five simulations (initialized at 12Z 15 June 2002 and run out to 24 hours) were performed for this study, denoted 9km_alldata_c, 9km_alldata, 9km_standard, 9km_eta, and 3km_alldata_c. 9km_alldata_c (the control run) included various surface mesonet data in its initial conditions as well as the NWS surface and upper air data networks, and also included a complex cloud analysis at the initial time. 9km_alldata was the same as 9km_alldata_c, except for the lack of a cloud analysis. 9km_standard used only the NWS surface and upper air data networks in its initial conditions, and the initial conditions for 9km_eta were interpolated from a 27 km ARPS grid that was in turn interpolated from the 12Z Eta analysis. The initial and boundary conditions for 3km_alldata_c were interpolated from 9km_alldata_c, and an additional cloud analysis was performed. Each of the 9 km simulations used a new version of the Kain-Fritsch convective parameterization scheme, while the 3 km simulation used explicit convection.

Results show that, for this particular case, the use of additional mesonet surface data in the initial conditions has a positive effect on the forecast of the MCS, which is seen by examining the position of the system at different forecast model times and comparing with observations. Three-hour accumulated precipitation spectra for the period starting at forecast hour 12 were also calculated for all grids and the results confirm that the use of more data in the initial conditions has a positive effect, especially on smaller scales.

The impact of the complex cloud analysis appears to be confined to the first few hours of the forecast, where the shape and position of the MCS is improved. Precipitation spectra for the accumulated precipitation over the first three hours show a clear improvement, again especially at the smaller scales, for simulations including the cloud analysis over those without. After this time period, the effect gradually becomes smaller and simulations 9km_alldata_c and 9km_alldata converge to nearly identical solutions.

The 3 km simulation (3km_alldata_c) is seen to be superior to any of the 9 km simulations, with much smaller position errors for the MCS throughout the forecast period. Since the only other difference between this simulation and 9km_alldata_c is an additional higher-resolution cloud analysis at the initial time, the increased resolution and the use of explicit convection in this simulation are apparently strong factors in the improved forecast of the MCS.

Future work will involve incorporating more high-resolution mesoscale data into the initial conditions of the model, and performing more 3 km simulations using initial and boundary conditions interpolated from the various 9 km simulations discussed previously. Higher resolution (1 km) simulations are also planned, and the use of intermittent data assimilation cycles of radar and mesonet data at various points during the forecast of the MCS will be employed. Also, forecast precipitation spectra will be compared with observed precipitation spectra, and spectra of other quantities will be computed and compared with observations. A more quantitative comparison of the strength of the cold pool and surface winds between the simulations and observations will also be performed.