Automated source parameter and low level wind estimation for atmospheric transport and dispersion applications

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 4 February 2014: 9:00 AM
Room C206 (The Georgia World Congress Center )
Jake Zaragoza, UCAR, Boulder, CO; and L. M. Rodriguez, A. J. Annunzio, and P. Bieringer

The National Center for Atmospheric Research (NCAR) developed a source term estimation (STE) algorithm to determine the source location and quantify the release amount of a contaminant if observations of the contaminant are available. The algorithm, called the Variational Iterative Refinement STE Algorithm (VIRSA), combines methods from inverse modeling and nonlinear optimization to provide an accurate and computationally efficient solution. Due to VIRSA's combination of accuracy and computational tractability, it has been selected for use in chemical and biological hazard assessment models including The Hazard Prediction and Assessment Capability (HPAC) and the Joint Effects Model (JEM). These hazard assessment models are designed to provide downwind hazard area information for chemical or biological agent releases. When the source parameters are not fully known, this requires an accurate STE technique to accurately calculate the STE parameters and corresponding downwind hazard area.

This VIRSA STE algorithm has been tested on field trials including FUSION Field Trial 2007 (FFT07), a field program funded by the Defense Threat Reduction Agency (DTRA) to test STE algorithms. VIRSA testing with FFT07 yielded positive results; however, the field trial provides a limited number of environmental and operational scenarios that the algorithm may face. To extend the STE test space we developed a number of synthetic, physically realistic transport and dispersion realizations using the Large Eddy Simulation capability. Using these synthetic data sets in conjunction with FFT07 we have conducted a more rigorous examination that includes a broader variety of plume dispersion scenarios with varying travel distances, sensor configurations, and plume detection methods. In addition, the limits of model performance were investigated.