A fast reliable algorithm for point source localization: Application to a new kitfox data set
Résumé
In case of dispersion of hazardous CBRN species in the atmosphere, Source Term Estimation (STE) algorithms can be employed to provide an estimation of the release parameters (localization, intensity,⋯) by making use of concentrations and meteorological measurements. These algorithms have to be fast, reliable and accurate. One of the fastest methods for solving the source estimation inverse problem is the renormalization technique. It gives a source estimate linear with respect to the observations. This estimate corresponds to a minimum weighted-norm solution which can be expressed by making use of the generalized inverse concept. In this paper, a method for an efficient computation of this inverse solution is proposed. This method is illustrated using continuous release trials conducted during the Kitfox field experiment from September 11th to September 15th 1995, just after the 52 well known releases of August 1995. From the author's knowledge, these smooth desert experiments have never been used for validation purposes and thus can be considerate as a new data set.