Validation of a Sensor-Driven Modeling Paradigm for Multiple Source Reconstruction with FFT-07 Data
Abstract
A Bayesian probabilistic inferential framework provides a natural and logically consistent method for source reconstruction that fully utilizes the information provided by a limited number of noisy concentration data obtained from a network (or, array) of detectors. This report addresses the application of this framework to the difficult problem of estimating the parameters of an a priori unknown number of sources, using an array of detectors. To this purpose, Bayesian probability theory is used to formulate the full joint posterior probability density function for the number of sources and the parameters (e.g., location, emission rate, activation and deactivation times) that describe each source. A simulated annealing algorithm, applied in conjunction with a reversible jump Markov chain Monte Carlo technique, is used to draw random samples from the posterior probability density function. By calculating the marginal posterior probability distribution of the number of sources from these samples, a maximum a posteriori estimate Ns for the number of sources can be obtained, and all samples of source distribution models with exactly Ns discrete sources can be used to provide best estimates for the source parameters (along with their associated uncertainties). The method is validated against a real dispersion experiment involving various combinations of multiple source releases conducted under a multinational cooperative Fusing Sensor Information from Observing Networks (FUSION) Field Trial 2007 (FFT-07) undertaken at US Army Dugway Proving Ground (DPG) in September 2007.
Document Details
- Document Type
- Technical Report
- Publication Date
- May 01, 2009
- Accession Number
- ADA512965
Entities
People
- E. Yee