Calibrating the Rainfall-Runoff Model GR4J and GR2M on the Koulountou River Basin, a Tributary of the Gambia River

Abstract

Rainfall runoff modelling is the first step in water resources management. It is the only way to simulate the hydrological behavior of the basin for a good evaluation of the potentiality of this in term of water production. Many approaches are actually in use. In physically distributed models, deterministic relations issued from conservation laws of physics (mass conservation, moment momentum conservation) are solved to describe the hydrological processes generating flow and their interaction. A DEM that should be as complete as possible is associed. Complexity of the equations to be solved and the huge amount of required data, uncertainty in these data make these models of limited use. Conceptual rainfall-runoff models are often preferred by hydrologists. These models are based on equations relating in a realistic manner the different terms of the hydrological cycle. They are simpler than determistic models and more flexible, Conceptual models are generally global. According to the way hydrological cycle terms are taken into account, conceptual model can be classified as empirical or not. The aim of this paper is to evaluate the availability of water in the Koulountou river basin, a tributary of Gambia River. This river basin should reinforce the water resource in a neighboring Kayanga river basin. Two empirical models at daily and monthly scale, the GR4J and GR2M have been used to describe the hydrological behavior of this basin. These models have been realized by the CEMAGREF, a French research Office. They use as inputs daily or monthly rainfall and potential evapotranspiration and river basin area, and give as output daily or monthly runoff. The first step before applying a hydrological model is to calibrate it that is to estimate the best parameters that fit the outputs in a given period. The Nash criterion has been used as goodness-of-fit criterion. Model performs satisfactory when this criterion is greater than 0.70 according to available data. A period from 1971 to 1994 has been selected. This period have been divided into three parts: one for calibration (1971-1978), one for validation (1978-1986), and the last for application (1987-1994). The results we obtain shows that GR4J and GR2M performs well in the Koulountou river basin since the Nash criterion is greater than 0.8.