A CRITICAL APPRAISAL OF FLOOD FORECASTING MODELS

Abstract

Flood forecasting, a non-structural measures of flood management, is one of the most important aspects of applied hydrology. River flood forecasting methods usually incorporate conventional form of routing techniques. Routing using conventional hydrological approach usually uses a storage relationship between the reach storage and the inflow and outflow and their derivatives in addition to the lumped continuity equation for developing a routing model. Any form of storage relationship can be developed from the Kulandaiswamy's generalized storage equation. Even with the advancement of computational capabilities, the linear flood forecasting models are still used in the field, due to their simplicity for easy understanding and less data intensive nature. However, these models would serve a larger purpose if physical relationship behind most of such models are brought out. The present study, aims to investigate_ the performance of four linear routing models by developing them from the generalized storage equation (Kulandaiswamy, 1964) investigated by Kulandaiswamy et al. (1967). Hypothetical data were used first to study the actual behaviour of these models. Performance of the models are evaluated based on Nash-Sutcliffe criterion, percentage errors in attenuation of routed peaks and errors in time to peak of routed flow. These linear models were further investigated to take into account the non-linearity effect of the flood propagation process. Here the non-linearity refers to the variation of parameters in a physically based manner, but maintaining the linear structure of the model throughout the routing process. The capability of the lower order variable parameter model, in the form of variable parameter Muskingum method, was extended to a higher order variable parameter routing model. But study reveals that such an extension model performs poorly in comparison with the performance of lower order model, the Variable Parameter Muskingum method, in reproducing the actual characteristics of flood propagation process.