Skip to Main content Skip to Navigation

Appréciation à l'aide d'un modèle empirique des effets d'actions anthropiques sur la relation pluie-débit à l'échelle d'un bassin versant

Abstract : This report addresses the opportunities of improvement in the field of rainfall-runoff empirical modelling. Two aspects are considered: (i) modelling rainfall-runoff processes taking place in ephemeral river basins and (ii) modelling rainfall-runoff processes taking place in basins under land use change. The model used in the present research is an empirical lumped rainfall-runoff model developed in the CEMAGREF and the CERGRENE laboratories, called GR4J model. The empirical rainfall-runoff research program is first addressed, in order to investigate whether it may still be a fruitful approach, in the context of hydrology. A research program focuses on the theoretical basis, the hypotheses and guidelines, as well as the mathematical and experimental techniques, utilised by hydrologists while modelling hydrologjcal processes. The initial model employed in modelling ephemeral river basin rainfall-runoff processes, GR3J, is the original model from which GR4J has been developed. The main objective that motivates this research is to extend the domain of the model GR3J, which was successfully employed in perennial river basin rainfall-runoff simulation. The need for a change in the structure of the model, to include a fourth parameter, became evident after an in-depth analysis. This study takes into account an analysis of ephemeral phenomenon physical causes, as well as the reasons of GR3J inadequacy to simulate it. Systematic trials which change the model structure in order to improve its performance in this case are done for more than 40 model versions. The new model version adequately simulates rainfall-runoff process in ephemeral river basins (7 basins tested) as in perennial river basins (113 basins tested). Coping with an empirical rainfall-runoff model on account of basin nonstationarities requires a more demanding validation procedure than the classical split-sample test. We must verify at least (i) whether the model is already able to explain streamflow fluctuations induced by natural climate variability; (ii) whether parameter values of the model are stationary under basin stationary conditions and (iii) the sensibility of the model parameters to changes in basin physical characteristics. This study concerns the test of the model GR4J under basin stationary conditions as well as under man induced non-stationarities. Four stationary basins, a land drainage and a wildfire case studies are considered. Also an analysis about the performance of the GR4J local search calibration method on account of the risk of local maxima is done. As a conclusion, the local search method can be considered perfectly adapted to the model, showing good qualities of reliability and satisfactorily converging to the maximum maximorum. Both the land drainage basin and the paired basin used in this study could be considered stationary. Therefore, we may conclude that no identifiable changes in the rainfall-runoff relationship were observed in this case study. All of the other four basins exhibit non stationarities in the rainfall-runoff relationship, even the paired basins used in the wildfire case study. Therefore, in the particular wildfire case study it is not absolutely proved that the changes observed in the rainfall-runoff relationship are caused by the disappearing of the vegetal cover.
Complete list of metadata
Contributor : Ecole Des Ponts Paristech Connect in order to contact the contributor
Submitted on : Thursday, February 24, 2011 - 10:59:26 AM
Last modification on : Thursday, April 5, 2018 - 11:02:02 AM
Long-term archiving on: : Wednesday, May 25, 2011 - 2:50:46 AM


  • HAL Id : pastel-00569059, version 1


Nilo de Oliveira Nascimento. Appréciation à l'aide d'un modèle empirique des effets d'actions anthropiques sur la relation pluie-débit à l'échelle d'un bassin versant. Hydrologie. Ecole Nationale des Ponts et Chaussées, 1995. Français. ⟨pastel-00569059⟩



Record views


Files downloads