The Bani basin was classified into 4 clusters of similar catchments (Figures 2-4),  The topographic variables (Elev, ElevMin, ElevMax, Slo1), precipitation and the geographical position of the sub-catchment (Lat) were demonstrated to be the most important causes of similarity between catchments belonging to Cluster 2 and Cluster 4 (Table 2),  This study permitted to propose the two nomenclature: Group of northerly flat and semi-arid catchments, and group of southerly hilly and humid catchments.

Results showed that the model performance can be judged as very good (Moriasi et al., 2007) especially considering limited data condition and high climate, land use and soil type variabilities in the studied basin (Figure 1).  Prediction uncertainty is acceptable: most of the observed data (around 80& ) are bracketed by the 95PPU within an acceptable width (R-factor < 1). However, model is characterized by more prediction uncertainties during high flows (Figure 2).  The most sensitive parameters are mostly related to surface runoff reflecting the dominance of this process on the streamflow generation (Table 1).