Multi-entity field approximation for hazard origin estimation

In the event of an atmospheric contaminant (biological, nuclear, radiological, or chemical) release, it is crucial to ascertain the source information for the contaminant both for mitigation purposes and to predict subsequent transport and dispersion. Here, we propose a mixed Eulerian/Lagrangian approach to determine the source information for multiple instantaneous and continuous contaminant releases by approximating the observed concentration field with several entities in a Multi-Entity Field Approximation (MEFA). Here, the entity is the contaminant puff or plume, modeled as a Gaussian with three parameters: location, mass, and spread. The MEFA is a three-step process. First, we minimize the difference between the concentration observations and the modeled entity. We assume that each entity has the same Gaussian representation and optimize each functions' parameters. This approach requires nonlinear optimization, and thus, we use a robust optimization technique (a genetic algorithm) for this fit. Second, we must associate each entity with its observation. Such a process requires an expert system. Lastly, we estimate the source information, which is accomplished by inverting a simple set of dynamic equations. In addition, we establish guidelines to determine when a MEFA is appropriate, because in some scenarios a single entity field approximation may suffice. Such guidelines depend on the spacing of contaminant releases, the entity spread and the sensor network density. To investigate algorithm robustness, we test this method on the FFT07 dataset.