ABSTRACT
ABSTRACT: Predicting algal blooms is an ambitious and difficult topic due to the complexity of aquatic ecosystem behaviour, insufficient knowledge of underlying processes, and shortage of high quality data. The same holds for the proper modelling of overland flows in wetlands and vegetated floodplains, which is of great practical importance in both flood early warning and in river restoration. Even though in both applications the purely hydrodynamic behaviour is conveniently formulated in terms of classical mathematical equations, some important physical, chemical or biological processes are often ‘hidden’ in the coefficients contained in these equations. It is demonstrated in this paper that Artificial Intelligence techniques can prove extremely valuable for establishing these coefficients or for providing complementary formulations of processes that can not (yet) be incorporated in these equations. Results are presented for the two cases of algal bloom prediction and vegetation flow resistance. It is shown that knowledge can be inferred from data directly, but that the techniques are applied most effectively when involving expert knowledge.
