The February 2000 floods on the Sabie River, South Africa: an examination of their magnitude and frequency

• Main Authors: G.L. Heritage, B.P. Moon, G.P. Jewitt, A.R.G. Large, M. Rountree
• Format: Article
• Language: English
• Published: AOSIS 2001-07-01
• Series: Koedoe: African Protected Area Conservation and Science
• Subjects: February 2000 flood, flood frequency, flood magnitude, Sabie River.
• Online Access: https://koedoe.co.za/index.php/koedoe/article/view/184

The floods that affected much of Southern Africa in February 2000 have been reported as the largest in living memory by many observers. However, the force of the floods damaged the majority of the gauging stations located on the affected rivers, many of which were not constructed to measure flows of such a magnitude. This paper presents an estimation of the peak flood discharge on 6 February 2000 for the bedrock influenced Sabie River in the Kruger National Park, by simulating the hydraulic and geometric characteristics of the peak flow and relating these to the roughness character of the channel. Peak water surface slope data in the form of strandline measurements at channel type breaks along the river were collected for six sites along the Sabie River within the Kruger National Park. Flood conditions within each channel type were considered to approximate to uniform flow. The cross-sections are located between major tributary inputs allowing for approximate sub-catchment flow contributions to be estimated. The results indicate that the flow peaked at around 3000 mVs at the Kruger Gate entrance to the Kruger National Park, increasing to approximately 5500mVs at Skukuza and 7000 mVs at Lower Sabie close to the Mozambique border following inputs from the Sand River sub-catchment. These estimates compare well with the simulated rainfall runoff total of 4300 mVs at Skukuza, however, precipitation inputs over the lowveld appear to indicate that the discharge only rises to 4950 mmVs at Lower Sabie. A flood flow of this magnitude has never been experienced based on the simulated flow data generated by the ACRU hydrological model calibrated against measured flows therefore suggesting a return period in excess of the 60 years of record.