1 3.13 ON USING DATA ASSIMILATION IN DISPERSION MODELING Anke Beyer*, George S. Young, Sue Ellen Haupt The Pennsylvania State University, University Park, Pennsylvania 1. INTRODUCTION Transport and dispersion models are important tools for addressing the issue of a chemical, biological, radiological, or nuclear (CBRN) release. Since the 1950’s transport and dispersion modeling with application to CBRN emergencies is a topic of intense research (Fast et al. 1995). The Chernobyl nuclear accident, for example, promoted the development of new long- range transport models and applications of existing air quality models to radioactivity in many research facilities (Ishikawa 1994). A number of field experiments have been conducted in the last 40 years, for example the 1980 Great Plains Mesoscale Tracer Field experiment (Moran and Pielke 1995) and the Dipole Pride experiments (Watson et al. 1998). The dispersion of hazardous puffs or plumes is the result of three factors: the transport by the wind field, dispersion by turbulence, and chemical reactions. The basic data requirements for a dispersion modeling system are hazard source characterization, surface topography, and meteorological data. Using these three elements, and the equations of transport, dispersion, and chemical reaction, it becomes possible to accurately forecast the spread and evolution of hazardous materials. Currently available modeling systems range from relatively simple to highly complex (Arya 1999). A simple example is the Gaussian plume model, which has been used for almost a century to predict dispersion from continuous point sources in air quality applications (Weil et al. 1992). 2. DATA Observations play a key role in numerical weather prediction (NWP) as well as in transport and dispersion modeling. In operational NWP over 11,000 land surface observations, over 7,000 marine surface observations and approximately 900 upper air soundings are available each day * Corresponding author address: Anke Beyer, Dept. of Meteorology, 503 Walker Building, Pennsylvania State Univ., University Park, PA 16802-5013; e-mail: aub166@psu.edu. (WMO 2005). This conventional observational network is augmented by satellites, radar and aircraft. With the average scale of synoptic eddies being approximately 2000 km and their evolution measured in days, the data coverage and the time resolution for NWP forecasts is generally sufficient. In contrast; the characteristic length scale of a fatal hazardous release is much smaller than that of a synoptic disturbance: instead it will interact most strongly with the boundary layer eddies. These eddies scale with the boundary layer depth (Kaimal et al. 1976) so their typical length scales range from 1 to 5 km and their evolution takes less than an hour. Thus, much higher spatial and temporal sampling rates are required to map a hazardous release and its interaction with the dominant eddies. As with NWP, it is expected that several sensors will be needed per eddy to successfully initialize a model of the transport and dispersion of a plume. Unfortunately, the current State and Local Monitoring Network for air quality in the USA contains only approximately 4,000 fixed monitoring stations that measure criteria pollutants (particulate matter, sulfur dioxide, carbon monoxide, nitrogen dioxide, ozone, and lead) (EPA). An additional 500 air-sampling devices have been deployed in 31 US cities in the framework of BioWatch, a program of the Department of Homeland Security (Bridis 2003). The BioWatch sensor network detects biological agents by combining state-of-the-art laboratory testing and traditional filter sampling methods (Bridis 2003). But, even with multiple mobile instruments being rapidly deployed to the site of events there is no guaranteeing that all relevant turbulent eddies are observed. Also, the traditional types of data collection used in the EPA and BioWatch network have the disadvantage of a long sampling time. In an immediate emergency situation real-time data is needed. An alternative to gridded ground based measurements are Unmanned Aerial Vehicles (UAVs). UAVs share the advantage of an aircraft in rapid deployment to arbitrary locations and heights, but reduce the cost and safety restrictions a piloted plane would present (Watai et al. 2006).