In the methodology used, errors may be inherent in several steps performed. For instance, the original observed flow data were converted from stage height to discharge, assuming steady-state flow conditions. Additionally, the slope of the rivers, extracted from a digital elevation model, was averaged and does not account for sinks and waterfalls along the river course. Furthermore, the calculation of travel time using hydrographs, the approximation of hydraulic radius, the roughness coefficient, and slope extraction from the digital elevation model, as stated in the methodology section of the study, might also introduce further errors. The assumptions made in estimating variables and the subjectivity while interpreting and collecting data in the field were other factors that affected the routing procedure.
It was assumed that the reaches used in this study were uniform. However, the three reaches used (Klip and the Mooi Rivers) have varying widths and cross-sections, and they do meander as well as having different vegetation cover within the reach. The discharge is related to the channel dimensions using empirical formulae that were developed for different river conditions. Furthermore, the linearity assumption of the Muskingum-Cunge method along the river reach might also introduce errors in the computed hydrographs. Considering all the above assumptions made in this study to route floods in ungauged catchments, the following recommendations are made:
The empirical formulae and base roughness coefficients should be checked at a regional level and related to historical flow records to verify the empirical formulae,
The slope of a reach and roughness coefficient require practical experience to determine realistically. Hence, the estimated slope and roughness should be checked against practical scenarios,
The computed hydrographs do not fit the observed hydrographs as well in longer reaches (i.e., Reach-II and Reach-III) compared to a short reach (Reach-I). This may be explained by the inadequate estimation of lateral inflows to the main stream. Hence, lateral flow to the main stream requires a regional-level study to estimate the coefficients, such as β,
Overbank flow cannot be simulated by the Muskingum-Cunge method. Therefore, these cases need further studies or modifications to consider floods that spill over the banks, and
The methods are not applicable to steeply rising hydrographs. Hence, further research needs to be conducted to apply the Muskingum method to flows with high velocities in ungauged catchments.