A sensitivity and benchmark study of RAMS in the Eastern Range Dispersion Assessment System

This paper describes a component of the Applied Meteorology Unit’s (AMU) evaluation of the Regional Atmospheric Modeling System (RAMS) contained within the Eastern Range Dispersion Assessment System (ERDAS). ERDAS is designed to provide emergency response guidance for operations at the Cape Canaveral Air Force Station (CCAFS) and Kennedy Space Center (KSC), Florida, in the event of an accidental hazardous material release or an aborted vehicle launch. The evaluation protocol is based on the needs of Eastern Range safety and weather personnel, and designed to provide specific information about the capabilities, limitations, and daily use of RAMS in ERDAS for operations at KSC/CCAFS. The prognostic data from RAMS is available to ERDAS for display and input to the Hybrid Particle and Concentration Transport (HYPACT) model. The HYPACT model provides three-dimensional dispersion predictions using RAMS forecast grids. Thus, the accuracy and sensitivities of the HYPACT model are contingent upon the prognostic data from the RAMS model.

In ERDAS, RAMS is run in real-time on four nested grids with horizontal resolutions of 60, 15, 5, and 1.25 km using full microphysics on all grids. The model is initialized twice-daily at 0000 and 1200 UTC using operationally-available observational data and run for a 24-h forecast period on local workstations with multiple processors. The 12-h forecast from the National Centers for Environmental Prediction (NCEP) Eta model provides the background field for the initial condition and the Eta 12-36-h forecasts are used as boundary conditions.

As part of the RAMS evaluation protocol, the AMU conducted two experiments during the Florida warm-season months of May through August 1999, including a sensitivity and benchmark study. The sensitivity study compares the current RAMS 4-grid configuration to a 3-grid setup. For each model simulation during the sensitivity study, RAMS was run using only the first 3 model grids. A high-resolution network of KSC/CCAFS wind towers located entirely within the 1.25-km RAMS grid was used for model validation. An algorithm routinely interpolated the RAMS forecasts to the locations of the KSC/CCAFS wind tower sensors for both the 4-grid and 3-grid simulations. The AMU compared point error statistics of temperature, wind, and moisture in order to determine the impact of horizontal resolution on the resulting model errors over the KSC/CCAFS wind-tower network in east-central Florida.

The benchmark experiment consists of a comparison between NCEP Eta and RAMS point forecasts at 14 selected stations across the southeastern United States. Again, an algorithm extracted and interpolated RAMS forecasts to the locations of these 14 stations in order to generate surface and upper-air error statistics. The AMU computed point error statistics for temperature, wind, and moisture in order to compare the performance of RAMS versus the Eta model. The benchmark experiment is an important component of the evaluation since it quantifies the potential added value of running RAMS on local workstations at much finer spatial resolution than the current national-scale operational Eta model. This paper will summarize the most significant findings from the sensitivity and benchmark experiments described above.