Skip to Main content Skip to Navigation

Réflexions sur l'utilité des modèles mathématiques dans la gestion de la pollution diffuse d'origine agricole

Abstract : This work has two main objectives. First, it evaluates the ability of empirical and conceptual models to simulate the evolution of loads and concentrations of suspended solids, nitrates and phosphates at the outlet of a small rural watershed. A model (based on CREAMS and SWRRB) is used as an example of empirical models and HSPF as an example of conceptual models. Second, the usefulness of such models is analyzed towards the management of non-point agricultural pollution through the detailed study of five potential operational uses of models: (1) computation of pollutant load, (2) analysis of high concentration risk, (3) simulation and representation of pollutant transport mechanisms, (4) identification of pollutant sources, and (5) prediction of the impact of modifying agricultural practices on water quality. The first part of this work displays the importance of rural non-point source pollution. The different steps of water management procedures in agricultural watersheds are analyzed, pointing out the need for various management tools at each step. Then, the potential use of models as management tools is presented. This first part ends with the description of the normal structure of a modelling approach to a problem, and with the analysis of extrinsic reasons which can explain why models are scarcely used in France. The second part describes the physical, chemical and biological processes and subsequent sub-models interacting in the global cycle of water, erosion, nitrogen and phosphorus. Next, a comprehensive analysis is given of the most famous models, specifying their abilities and their applicability. The third part is devoted to the conducted modelling work. The results obtained after validation for both selected models on the Mélarchez watershed (7 km2) are analyzed. Finally, a conclusion is suggested as to die usefulness of models as management tools, on the basis of the answers that they are able to bring to real problems. The main conclusions to be drawn are as follows: - These models are too complex to simulate loads and concentrations at the ouüet of rural watersheds. The available information in the data is not sufficient to calibrate the model, hence parameters are under-determined. Moreover, simulations are generally rather poor, specifically in the case of the highest concentrations. As management tools, the models are not more efficient than simpler models to compute pollutant loads (for instance models considering a constant concentration). - In principle, these models should be able to simulate other state variables (e.g., soil nitrate storage) or to simulate the impact on water quality of modifying agricultural practices such as changing cultivated areas or crop rotations, or changing the type or amount of fertilisers. This presupposes that the realism of the representation of processes inside the watershed (not only at the outlet) should be verified. Unfortunately this is not the case, and furthermore the necessary data generally do not exist. Hence, at present, these models should not be trusted for decision making.
Complete list of metadatas
Contributor : Ecole Des Ponts Paristech <>
Submitted on : Wednesday, October 27, 2010 - 10:40:23 AM
Last modification on : Monday, October 19, 2020 - 10:54:49 AM
Long-term archiving on: : Friday, January 28, 2011 - 2:43:41 AM


  • HAL Id : tel-00529963, version 1


Luiz Augusto Kauark Leite. Réflexions sur l'utilité des modèles mathématiques dans la gestion de la pollution diffuse d'origine agricole. Modélisation et simulation. Ecole Nationale des Ponts et Chaussées, 1990. Français. ⟨tel-00529963⟩



Record views


Files downloads