MODELING THE CLIMATE AND HYDROLOGY OF THE TONO BASIN IN GHANA, WEST AFRICA

Abstract

The study focused on the Tono basin in the Kessena municipality in the Upper East Region of Ghana. Due to poor data collection at the basin, there has not been any hydrological and climatic assessment of the basin. To address these challenges the study explored the potential of using a state of the art hydrological model (WRF-Hydro) in a coupled mode to assess these water resources, particularly the Tono basin in Ghana. A 2-domain configuration was chosen: an outer domain at 25 km horizontal resolution encompassing the West African Region and an inner domain at 5 km horizontal resolution centered on the Tono basin. The infiltration partition parameter (kdtref) and Manning’s roughness parameter (MannN) of the hydrological model were calibrated to fit the simulated discharge with the observed one. The simulations were done from 1999-2007, using 1999 as a spin-up period. The results were compared with TRMM precipitation, CRU temperature and available observed hydrological data. A standalone WRF model was run for the same period to assess whether the coupled model (WRF/WRF-Hydro) provides an improvement in estimating climate variables, e.g. precipitation and temperature. The forcing data used to drive the model runs were, ERA interim reanalysis and ECHAM6. The WRF/WRF-Hydro forced with ERA-I demonstrated a precipitation pattern of correlation with observed (TRMM) data of about 0.91 and its centered RMS error of 2.4 mm/day, whereas WRFonly forced with ERA-I produced a precipitation pattern of correlation of about 0.82 and a RMS error of 3.6mm/day. For temperature, WRF/WRF-Hydro produced a pattern of correlation with observed (CRU) data of about 0.94 and a RMS error of 0.6 0C, whereas WRF-only produced a pattern of correlation of about 0.87 and a RMS error of 1.2 0C. Similar characteristics were demonstrated by ECHAM6 model data; however, ECHAM6 produced the worst results especially for the coupled approach. These variations in model performance can be attributed to v the optimum physics option chosen, which may not be the optimum in the Tono basin (microscale effect) and also applying the same calibration for different models (coupled or uncoupled). The WRF-Hydro model demonstrated strong signal of streamflow estimate; with Nash-Sutcliff efficiency (NSE) of 0.78 and Pearson’s correlation of 0.89. Further validation of model results was based on driving the output from the WRF-Hydro to a water balance model to simulate the dam levels. The model-derived dam levels were in good agreement with the observed dam levels with a correlation (R2) of 0.81. The deficiency in the modelled dam levels could be attributed to the models’ over estimation of evaporation. Regarding climate change impact on the Tono dam, two climate change emission scenarios were applied (i.e. RCP4.5 and RCP8.5). Both scenarios did not agree on the signal of change with respect to precipitation but both indicated a warmer condition. RCP4.5 indicated an annual rainfall projection increase of +7%, which implies a future increase in flows of about 14% and RCP8.5 indicated rainfall projection decrease by -9.6%, implying about 20% reduction in flows. There is therefore the need to put in place adaptation measures to ensure the sustainability of the Tono dam in the face of climate change.