Flood estimation for roads, bridges and dams.

Abstract

Flood estimation can be classified into two categories, i.e. flood prediction and flood forecasting. Flood prediction is used for the estimation of design floods, which are floods associated with a degree of risk of being equalled or exceeded. Predictions are needed for the design and construction of infrastructure that are at risk to flowing water. Flood forecasting is used for the estimation of flood flows from an impending and/or occurring rainfall event (i.e. the estimation of the magnitude of future flood flows with reference to a specific time in the future). These are needed by catchment and disaster managers for the mitigation of flood damage. The estimation of flood magnitudes for flood forecasting requires the specific knowledge of prevailing surface conditions which are associated with the processes of rainfall conversion into flood runoff. In order to best achieve this, a distributed model (in order to exploit remotely sensed data and capture the spatial scale of the phenomenon) is used to continuously update the surface conditions that are important in this conversion process. This dissertation focuses on both flood estimation categories. In the first part of the dissertation, attention is given to the improvement of two simple event-based design flood prediction methods currently in use by design practitioners, namely the regional maximum flood (RMF) and the rational formula (RF) by comparison with statistically modelled historical flood data. The second part of the dissertation lays the theoretical and practical foundation for the implementation of a fully distributed physically-based rainfall-runoff model for real-time flood forecasting in South Africa. The TOPKAPI model was chosen for this purpose. This aspect of the research involved assimilating the literature on the model, testing the model and gathering and preparing of the input data required by the model for its eventual application in the Liebenbergsvlei catchment. The practical application of the model is left for a follow-up study.