"COMPARISON OF THE PERFORMANCE OF tAINFALL-RUNOFF HYDROLOGICAL MODELS IN RUVUBU RIVER BASIN" A CASE STUDY OF RUVYIRONZA-NYABIRABA CATCHMENT

Abstract

Rainfall-runoff models have become accepted as important tools in operational hydrology for estimating information required for water resources planning, design, and operation. Specifically, rainfall-runoff models are normally useful tools where data are insufficient by simulating and by extending the time series. This thesis work presents an appraisal study to compare the performance of four hydrological models in RVZ- Nyabiraba catchment for Ruvubu river basin and to select the best candidate model for the catchment response prediction. In this appraisal study, to achieve our objective 2 empirical models: Simple Linear Model (SLM), Linear Perturbation Model (LPM), and 2 conceptual models: (HBV) and Soil Moisture Accounting and Routing (SMAR) were tested in Ruvyironza-Nyabiraba catchment. Parameter optimization is carried out by trial and error, ordinary least squares, Rosenbrok, Simplex and generic algorithm. The parameter set that gave the best objective function value over the calibration period in the ranges of the parameters was used for validation. T'he visual comparisons were also made for the low and high flow fit of the hydrographs. The comparison was also made on the basis of the relative error of peak (RL) criteria and the index of volumetric fit (IVF). From the models comparison performance criteria, it is shown that the Simple linear model (SLM) and HBV are not adequate in modelling the rainfall runoff transformation. However, the RVZ-Nyabiraba catchments exhibit marked seasonal behaviour and good results was also obtained with Linear Perturbation model (LPM) which involves the assumption of linearity between the departures from seasonal expectations in input and output series. Within the range (0.5-0.9) of the tested models performance, in the RVZNyabiraba catchment, out of the four models, SMAR was found to be the best candidate

model that can simulate the flows. Hence, SMAR is adequate in modelling the rainfall

runoff transformation. Further investigation should be made to generalize the

applicability of this model to all Ruvubu river basins.