Least Square Data Assimilation for Identification of the Point Source Emissions

Abstract

The identification of single and multiple-point emission sources from limited number of atmospheric concentration measurements is addressed using least square data assimilation technique. During the process, a new two-step algorithm is proposed for optimization, free from initialization and filtering singular regions in a natural way. Source intensities are expressed in terms of their locations reducing the degree of freedom of unknowns to be estimated. In addition, a strategy is suggested for reducing the computational time associated with the multiple-point source identification. The methodology is evaluated with the synthetic, pseudo-real and noisy set of measurements for two and three simultaneous point emissions. With the synthetic data, algorithm estimates the source parameters exactly same as the prescribed in all the cases. With the pseudo-real data, two and three point release locations are retrieved with an average error of 17 m and intensities are estimated on an average within a factor of 2. Finally, the advantages and limitations of the proposed methodology are discussed.