Flooding in KwaZulu-Natal : modelling, history and future aspects.

Abstract

The current state of flood modelling relies on statistical techniques revolving around either river or rainfall data that are used to produce an estimated flood return period (e.g. 1:100 year). These tend to ignore 1) the geological record as an archive of flood events; 2) the spatial distribution of flood producing weather systems; and 3) climatic cycles which may ultimately control episodes of flooding. This thesis developed a Flood Zone Model (FZM) model from existing Geographic Information Systems (GIS) datasets and GIS software models at a quaternary catchment (4th order basin) level. Model discharge estimates were derived from modified Regional Maximum Flood (QRMF) equations where it was found that flood elevations produced from QRMF estimated discharges could be directly related to the geological record of flood. This was achieved using Manning-derived calibration factors (CFs) based on reach slope. Comparison of the modelled flood elevation surfaces against the field data and available 1:100 year return period elevations showed R2 coefficients of 0.999 for all calibration factors. In one of the quaternary catchments investigated, geological evidence and discussion with local communities identified flood elevations attributed to flash flooding. On this basis, the Flood Zone Model was adapted to estimate peak discharges using the Rational Formula where it was found that the calibration factors were valid for flash flood modelling and that the flood elevations that resulted from flash flooding far exceeded the 1:100 year return period. To evaluate the spatial distribution of flood producing storms, daily rainfall data from KwaZulu-Natal (1890 - 2000) were gridded to produce regional storm event footprints. Storm events typically last between three to four days with the highest associated risk period from January to May. Flooding appears to be mostly influenced by migrating easterly waves. Compilation of all the storm event footprints defined five risk zones with a sixth zone at risk from tropical storms and cyclones. Comparisons between the annual regional storm event count and several climatic cycles show a significant correlation between the Pacific Decadal Oscillation (PDO) and regional storm event as a result of increased easterly wave activity. Assuming no change to the Pacific Decadal Oscillation cycle, increased periods of intense flooding will occur in the future.