Use of genetic algorithm in new approach to modeling of flood routing

Journal of Oceanology and Limnology - The hydrological models and simplified methods of Saint-venant equations are used extensively in hydrological modeling, in particular for the simulation of the...     

The rainfall-runoff model GR4J optimization of parameter by genetic algorithms and Gauss-Newton method: application for the watershed Ourika (High Atlas,Morocco)

Hydrological modeling aims to place the disposal of water actors, very simple tools for hydrological simulation to start with, the basic observations made on rainfall until the reproduction of the response the flow of a basin. The aims of this work consist, first, of testing the model GR4J (Génie Rural à 4 paramètres Journalier) performances of the rain-flow in semi-arid environment watershed and secondly, of highlighting the contribution of new optimization techniques (genetic algorithm, Gauss-Newton) in determining the optimal parameters of the model. The results obtained show that those methods of optimization are robust and reliable in comparison with the “solver” optimization used in Excel that usually applies for GR4J model. In terms of the model performances, we obtained a very satisfying results. Also, the simulations clearly show the influence of the melting snow on the Ourika’s basin flows; indeed, taking into account this component in the model allowed a consecutive improvement of the simulations.     

Combination of GIS and mathematical modeling to predict floods in semiarid areas: case of Rheraya watershed (Western High Atlas,Morocco)

The Tinitine sub-watershed of Rheraya is renowned for its semiarid climate, irregular supply of water flow, and its historical floods; for these reasons, it is classified as vulnerable areas during a rainfall event. We conducted this study to propose flood prediction models adapted to this risky zone. For this, a hydrological method of flood forecasting rainfall-runoff used to determine a model conforms to the semiarid basin. This model is based on the articulation of the series production and transfer function. The goal of the production function is to determine the portion of the rainfall runoff, which is performed by two approaches that differ in their structures: (1) the first approach is based on the mathematical model of Horton; (2) the second approach is based on the determination of the part of the rain contributes to runoff and obtained by a spatial map of runoff coefficient (GIS). The transfer function is based on the two models of Clark and Nash, rainy efficient routes to the catchment outlet from a unit hydrograph. The characteristic parameters of these models are obtained through the application of mathematical optimization methods such as genetic algorithms. Thereby, the coupling function producing Horton (identified by the parameters: initial infiltration?=?15.03, final infiltration?=?0.3, and phi?=?0.45) and Clark transfer function (identified by the parameters C_A?=?0.21 and C_B?=?0.79) was given very satisfying results, mainly the low difference between observed and simulated hydrograph and Nash coefficient which is about 85%. This shows the interest of this coupling model in flood forecasting.