||Mathematical models are playing increasingly important roles in science and engineering. Model-based design and optimization is nowadays the dominant engineering paradigm to systematic design and maintenance of engineering systems. Also, driven by the "grand" challenges society are facing, e.g. energy and environmental considerations, new model-based control applications are emerging, as for power grids, medical or environmental systems. These new applications represent highly complex systems with high demands on autonomy, and adaptation, calling for data-driven models.
This is true for water systems which include complicated ingredients, like non-linearities, time-varying parameters and time-delays. The phenomena are often stiff, i.e. the ratio between the fastest and the slowest modes are of several orders of magnitude and the data are most of the time irregularly sampled. All these features make the system identification very challenging.
Surface water pollution is a serious issue in areas with intensive agriculture. Policy makers of the European Union and elsewhere in the world aim at improving water quality in receiving surface water bodies. This study aims at improving load estimates from concentration measurements by reconstructing the responses of contaminant concentrations to rainfall events using commonly available quantitative hydrological data.
Recent new tools developed for nonlinear system identification will be investigated and extended to improve the understanding of the fate and transport of contaminants in drinking and source water. As most mathematical models of hydrological systems are formulated on the basis of natural laws, such as dynamic conservation equations, often expressed terms of continuous-
time, linear or nonlinear dierential equations, the data-driven modeling approach will focus on the direct identification of continuous-time models from sampled data.