Hydrological Modeling of the Niger River (West Africa) for Flow Prediction in Ungauged Basins.

English Abstract:

In West Africa, many drainage basins are ungauged or poorly gauged, and in some cases existing measurement networks are declining. Prediction of discharge and other hydrological characteristics of ungauged basins is therefore an important priority for water resources planning and management and for ecological studies. The ultimate goal of our study is flow estimation in ungauged catchments of the Niger basin through hydrological modeling of the Bani River using SWAT and its parameters transfer to the upper Niger. The calibration task presented in this article, is the foundation for regionalization of model parameters, which will be addressed in the companion paper. In the present article the modelling framework: model description, study area and data, calibration strategy and results are presented. The Bani is located in the upper part of the Niger River and drains an area of about 101, 000 km² at the outlet of Douna. The climate is tropical, humid to semi-arid from the South to the North with an average annual rainfall of 1050 mm. Global datasets used are from USGS hydrosheds DEM, USGS LCI GlobCov2009 and the FAO Digital Soil Map of the World. Daily measured rainfall from nine rain gauges and maximum and minimum temperature from five weather stations covering the period 1981-1997 were used for model setup. Sensitivity analysis, calibration and validation were performed within SWATCUP using GLUE procedure at the catchment’s outlet (Douna) and at two additional internal stations (Pankourou and Bougouni). Model parameters were calibrated at daily time step on ten years (1983-1992) then validated on 5 years (1993-1997). A warm-up period of two years (1981-1982) was considered. Calibration and validation results are good at the catchment’s outlet (Nash and R² equal to 0.76 and 0.79 for calibration, 0.84 and 0.87 for validation). Model performance was slightly lower for the additional gauging stations (Nash equal to 0.73 and 0.58) but again always higher for the validation period (0.77 and 0.65). These statistics suggest that the model performance can be judged as satisfactory, especially considering limited data condition and a high climatic gradient characterizing the Bani catchment.