Author(s): Vladan Babovic; Martin Baptist; Arthur Mynett
Keywords: Roughness coefficient; Evolutionary computing; Genetic programming; Knowledge discovery
Abstract: Proper modelling of flow in wetlands and vegetated floodplains is of great practical importance. Both analytical and experimental studies of vegetation-related resistance to flow and the equivalent resistance coefficients have shown that the resistance coefficients are water-depth dependent. Consequently, the traditional approach of using a single resistance coefficient fails to correctly describe the physics of the phenomenon. One way of improving upon this description is updating the equivalent resistance coefficient based on the computed water depth. In order to do so, a relationship between vegetation characteristics, bed resistance, water depth and equivalent resistance coefficient is needed. Two main approaches for creating such a relationship are contrasted. The first approach is the time-honoured method where a scientist uses whatever knowledge is available on the physics about the phenomenon and assembles an equation based on detailed understanding of the phenomena involved in the process. This understanding takes the form of many small models of subphenomena that are assembled to create an overall equation. The second approach employs a genetic programming technique to induce a set of relationships that are subsequently selected and improved by a scientist.