Sediment yield studies in the Mathare River catchment

Abstract

The Mathare catchment comprises an area of 26 Km2 and is located in the Western suburbs of Nairobi, Kenya. It is currently fitted with instruments for both rainfall measurement and runoff gauging at the Kabete field station. For the purpose of sediment gauging, a bridge was constructed for sediment sampling; two sets of staff gauges installed and a steel container erected for storage of equipment. Suspended sediment concentration in the river was estimated using four methods; the grab method, the equal discharge increment method, the equal transit rate method and the automatic single stage bottle sampling method. It was observed that there was no significant difference in statistical sense between the first three methods. The single stage automatic bottle sampler showed significant deviation in relation to the grab method. As a tool for the development of simple field gauging techniques, three methods of discharge measurement, the conventional multi-vertical current meter method, the single point velocity method and the equal transit rate sampling methods were investigated. It was found out that they all yield significantly different results and the last two required calibration to become useful for gauging. For this catchment, the instantaneous unit sediment graph (IUSG) was developed and tested for the prediction of sediment discharge. This involved development and analysis of the river discharge rating equation and the sediment graphs. The developed IUSG model produced a lower sediment yield in comparison to that derived from the temporal concentration graph and the flow duration curve analysis methods. The IUSG model based on the multireservoir cascading concept simulated the sediment graphs more accurately than that based on the time area histograms routed through a single reservoir concept. For the period 27/10/92 to 15/9/93, (324 days) the total estimated sediment yield of the Mathare catchment was 140, 150 and 165 tonnes by IUSG, temporal concentration analysis and flow duration curve analysis methods (uncorrected sediment rating curve) respectively. For the flow duration curve method, with the sediment rating equation adjusted by the correction factor (arising due to the lognormal distribution of the error component in the rating equation), a sediment yield estimate of 179.2 tonnes was obtained. The mean sediment production rate was found to be approximately 6.9 tonnes/km2/yr. Such a low estimate of sediment yield reflects the high level of conservation practices prevalent in the catchment. It is likely that the Mathare reservoir is still in healthy state in terms of siltation. The particle size distribution suggested that the total sediment load of the Mathare river comprises of suspended component with negligible bed load.