Repository logo

An assessment of radar, gauge and kriged gauge rainfall data in Free State, South Africa.

Thumbnail Image



Journal Title

Journal ISSN

Volume Title



In the last decades, flooding has caused significant damages and death in South Africa. Flooding is associated with heavy rainfall events which vary largely in space and time. The forecast of such phenomena requires quality rainfall data to generate output with a reasonably high degree of confidence. This study aims to document the difference of rainfall estimates derived from radar, rain gauge and a kriged rain gauge network. A review on the use of weather radar in hydrological studies shows that there is an opportunity of using radar estimates in near-real time flood forecasting and warning applications. The quality of radar rainfall estimates is assessed using pluviometer rain gauges and a daily kriged rain gauge surface in the Vaalsub-catchment in the Free State, South Africa. The study uses data from the MRL5 S-band radar located at Bethlehem and from rain gauges within weather stations operated by South African Weather Services (SAWS). The analysis assesses the effect of rainfall seasonality, radar range dependencies and storm variation on the quality of estimated rainfall accumulations. In addition,gauge density is analysed to determine the effect it has on the performance of kriging estimation.During the research notable spatial rainfall variation and areas with quality radar estimates have been identified. The results show that there exists a seasonal bias between radar and rain gauge estimates with the radar underestimating low intensity gauge rainfall pronouncedly during the winter rainfall events by an average of 31%. The underestimation of rainfall by the MRL5 radar increases as we move away from the radar tower. During the summer rainfall events the radar estimates are almost similar to point gauge estimates especially during late summer (February and March) . Radar underestimation of winter rainfall is probably due to overshooting of the tops of stratiform rainfall by the radar beam. Correlations are high between MRL5 radar and rain gauges during summer rainfall events ranging between 0.7 and 1. Results from this study provide information to guide on the application and selection of rainfall estimation techniques.


Masters Degree. University of KwaZulu-Natal, Durban.