Effect of loss model on evaluation of Manning roughness coefficient of experimental concrete catchment

To simulate a runoff hydrograph from a concrete catchment that is subject to natural rainfall, the use of computer model is essential. To develop such a model, the modeller is required to specify the rainfall data, the physical characteristics of the catchment, the Manning roughness coefficient (Manning n) of the concrete surface, and the loss model. To evaluate the Manning n for simulating runoff hydrographs, it is common to use the hydrograph fitting technique. With this technique, the simulated hydrographs are actually dependent on the loss model. Hence, the Manning n that is being evaluated is also dependent on the loss model. In view of this, the effect of loss model on the Manning n has been examined. Based on the rainfall and runoff data on a 25 m2 experimental concrete catchment comprising two overland planes and one rectangular channel, the examination shows that: (1) During the initial portion of the events, as the concrete was dry, the actual loss was higher than those assigned in the four loss models (the proportional, the initial and proportional, the upperbound, and the initial and upperbound). Hence, the evaluation of Manning n should omit this portion and only consider the subsequent wet portion of the events. (2) Based on the wet portions of the events, the effect of loss model on the optimum Manning n is small. (3) For the four loss models, the overall optimum Manning n for concrete are 0.013–0.015. These values are within the recommended Manning n for steady, uniform flow in concrete channel. Hence, the Manning n for steady, uniform flow is applicable to runoff simulation in which the flow is unsteady and non-uniform. (4) A comparison of the simulated hydrographs by the upperbound loss models and the proportional loss models shows that the upperbound loss models produce simulated hydrographs that are closer to the observed. (5) By adding one parameter to account for the initial loss in the loss model, it produces simulated hydrographs that are marginally closer to the observed. (6) The loss model that produces simulated hydrographs that are closest to the observed is the initial and upperbound loss model.