Sequential filter for river pollution monitoring

A sequential least squares filter is dovelopcd to estimate the pollution in a multi-soctional river. The Nash game approach proposed by Özgüner and Perkins (1978) is applied. The sections are defined according to mouths of larger tributaries. Streeter-Phelps hyperbolic equations model the contents of dissolved oxygen and its demand. The problem is defined as simultaneous minimization of quadratic estimation functionals defined on suecessive sections. The resulting decentralized filter looks like a cascade of local filters, each of which, besides measurements of its own section, uses those of preceding ones. Thus the filters are informed in advance if pollution occurs upstream. An orthogonal collocation approximation is applied to develop a numerical solution.