Author(s): Adam P. Piotrowski; Pawel M. Rowinski; Jaroslaw J. Napiorkowski
Linked Author(s): Pawel M. Rowinski
Keywords: No Keywords
Abstract: Different approaches allowing the prediction of pollutant transport in rivers have attracted a great deal of attention among scientists throughout the world. Unfortunately, the prerequisite for running most of the proposed models is the detailed information about the hydraulic and morphometric conditions of the considered river. Even when the simplest advection-diffusion model is planned to be used, identification of its parameters is usually based on specially designed experimental studies performed in the reach under consideration. In practice however, an unexpected spillage of pollutant may occur in an ungauged river in which no tracer tests had been performed in the past and the model is expected to provide some evaluation of the pollution fate. Managers and decision makers, who have just a modelling tool at their disposal and the basic information about the stream, are supposed to derive some conclusions about the admixture pattern in the stream after its release at some location. In the present paper the outline of methods based on artificial neural networks for estimation of parameters of two pollutant transport models is presented. As the key issue is the question of availability of hydraulic and morphometric data for particular site, three different possible cases are considered, which differ in a number of information at the user's disposal. If the concerned river reach is sufficiently well-recognized, the artificial neural network based method of estimation of transient storage zone model parameters is suggested. When little information is available, only parameters of advection-diffusion equation may be estimated. In such a case two versions – the simple one and the requiring more information are discussed. Clearly the pollutant transport prediction error increases significantly with the decrease of available information about the river reach. The significance of proper estimation of water velocity is indicated as crucial for the correct prediction in every case.
Year: 2009