MULTICRITERIAL ANALYSIS AND CHOICE OF HYDROLOGICAL MODELS IN TROPICAL ZONE: APPLICATION TO THE SANAGA WATERSHED IN CAMEROON

The Sanaga watershed covers an area of 135,000 km2 and is drained by the longest river in Cameroon (918 km). Its natural flow varies from 7,000 m3 /s in flood periods to less than 150 m3 /s in periods of low waters. In order to regularise this flow, the river has been equipped with four basins tanks which are unfortunately far from hydroelectric production sites. One of the problem that results is the lack of a reliable method for predicting flow rates of the watershed. To face it, the operator of the electrical sector has set up a digital model. This is a rain model semi-distributed daily flow which has some weaknesses. First, the target flow in low water can only be guaranteed with a failure rate of 25%. Then, after the development of this tool, the production park was enriched by a fourth reservoir dams. Finally, a hydroelectric dam and a water catchment and treatment works are being implemented on this river at a place called Nachtigal, located 65 km from the city of Yaoundé. This article proposes globalized conceptual models to characterize the hydrological behaviour of Sanaga at this outlet. The approach used is that of rural engineering (GR) models that have the reputation of being robust models. The GR1A, GR2M, GR3M and GR4J models are compared using the ELECTRE methods using three criteria: accuracy, absence of bias and robustness. Using the local step-by-step method, four algorithms were encoded on MATLAB. At the end of the multi-criteria analysis, a core of actions consisting of models GR2M, GR3M and GR4J is retained. The GR2M model has Nash of 81.57%. The GR3M model has a 78.56% Nash. These three tools predict at different times the flow rates on the Sanaga with appreciable efficiency. As for the GR4J model Nash criteria is 62.61%