Operational meteorologists have limited tools available when making nowcasts or short-term forecasts for the 0 to 6 hour timeframe. Large-scale operational numerical models like the RUC, ETA, WRF, and GSM are available to forecasters as well as trends in surface observations, satellite, and radar. With continued advances in the speed of computers, it is becoming feasible for each National Weather Service Forecast Office to economically run their own local-domain, high-resolution numerical model as an additional in-house nowcasting tool.

Such models can help operational meteorologists forecast a wide variety of situations, including the initiation or evolution of severe or flood-producing thunderstorms. To assess the usefulness of a high-resolution, local-domain operational model, a case study of Tropical Storm Allison was conducted. Allison produced catastrophic flooding over Houston, Texas over the period of 5-10 June 2001. The main goal was to examine how well the high-resolution model predicted the short-term evolution of rainbands associated with the flooding rainfall of Allison.

Allison was well-suited for this study since it dumped extreme rainfall totals in a short time period over the fourth largest city in the U.S. The maximum storm total precipitation over the 6 days Allison affected southeastern Texas was 38.8 inches (980 mm). The maximum rainfall over a 12-hour period was 28.3 inches (719 mm), and over a 6-hour period was 21.5 inches (546 mm). Tropical Storm Allison was billed as the costliest and deadliest tropical storm in U.S. history, causing nearly $5 billion in damage and 41 direct deaths.

The Advanced Regional Prediction System (ARPS), available free of charge from the Center for Analysis and Prediction of Storms (CAPS) at the University of Oklahoma, was used in this experiment. In addition to conventional forms of data, the ARPS is able to assimilate radar data by modifying the initial wind field, hydrometeors, and relative humidity in the atmosphere, which helps produce the most accurate analysis possible.

Hourly analyses of meteorological fields were produced while the storm affected southeastern Texas. Even though Allison was officially classified as either a tropical storm or depression during that time, there is evidence that it was really a hybrid or subtropical system. Allison structure was examined as it stayed organized over land and remained classified as a tropical depression for four days before reemerging into the Gulf of Mexico.

The Advanced Regional Prediction System (ARPS) was run for 9-hour periods on a one-way triply-nested 27 km to 9 km to 3 km grid for the evening of 8 June and morning of 9 June, when Houston received its largest rainfall totals from Allison. Sensitivity studies were conducted for this time period to study the changes in the model forecasts that occurred by varying the model’s resolution, terrain, background field, microphysics scheme, inputted observations, soil model, radiation model, microphysics scheme, convective parameterization, and use of data nudging. Based on these results, a comparison of the computational expense of each advanced model option with the gain in model forecast accuracy was discussed.

For the latest information on this research, visit http://weather.ou.edu/~bobwein/Allison.html