Mathematical Modelling of Water Soil Erosion and Sediment Yield in Large Catchments
Thesis (MScEng (Civil Engineering))--University of Stellenbosch, 2006.
In many part of the world, but especially in Africa, land degradation leads to severe soil erosion and high sediment yields. Mathematical models and empirical methods can be used to simulate the sediment yields. In many cases spatial and temporal data are however limited in the large catchments often found in Africa. A model should be able to simulate the long-term hydrology and sediment yields for sub-catchments and should be physically based as far as possible. In this thesis several models were evaluated and the agrohydrological model of the University of Kwa- Zulu-Natal (ACRU) was applied on two large catchments with limited data in Kenya. The key aim of the thesis was to assess the applicability of the ACRU modelling system for sediment yield prediction in large catchments under conditions of limited data availability. Two catchments in Kenya which drain into Lake Victoria were selected for this research: Nyando (3562 km2) and Nzoia River (13692 km2). Lake Victoria, with a surface area of 68000 km2 and an adjoining catchment of around 184200 km2, is the second largest fresh water lake in the world and the largest in the tropics. The Lake Victoria Basin area is increasingly being used for domestic, agricultural and industrial purposes by the three riparian countries Kenya, Tanzania and Uganda. About 21 million people (year 2000) rely primarily on subsistence agricultural and pastoral production for their livelihoods. But pervasive poverty has hindered sustainable use of the land resources and there has already been considerable land degradation. There has also been expansion of the increasing on-site erosion (overland flow) and reducing buffering capacity of the natural vegetation in wetlands and in the riparian zones (Hansen, Walsh, 2000). A regional assessment identified the Nyando River Basin and Nzoia River Basin as major sources of sediment flow into Lake Victoria on the Kenyan side of the Lake. Accelerated run off sheet erosion over much of the Nyando catchment area has led to severe rill, gully and stream bank erosion in lower parts of the river basin (Swallow, 2000). The ACRU model is a hydrological model using daily time steps with the Modified Universal Soil Loss Equation (MUSLE, Williams, 1975) module to simulate soil erosion. The MUSLE sediment yield module uses factors that characterize physical conditions on the surface of a catchment as input information. Data required for the model include: sub-catchment daily rainfall, historical flow records, general catchment topographical information, meteorological information, land use and cover, soil characteristics, sediment yield data, etc. The model used daily time steps for a 55 years record for the period 1950 to 2004. During calibration it was found that the sediment yield is overestimated which was expected since the model is a soil erosion model (based on MUSLE). The model was calibrated in each catchment against observed sediment load data, but this data were limited. Verification of the model was carried out by using satellite images and independent sediment load data when available. Scenario analysis was carried out by changing land use in the model to investigate how soil erosion could be reduced. Grassland to replace subsistence farming was found most effective, but irrigated sugarcane was also investigated. The model was found to be very effective in indicating which sub-catchments contribute most of the sediment yield. Under limiting data conditions it was found that it is very important to calibrate the model against field data. The most sensitive parameters affecting the sediment yield were found to be: a) Hydrological: • Daily rainfall spatial distribution of rain gauge • Time of concentration • Mean annual precipitation • Minimum and maximum temperature • Monthly evaporation b) Soil and catchment characteristics: • Number of sub-catchments making up catchment in model • Catchment slope and slope length, steepness factor • Land cover • Crop coefficient • Soil texture class and depths • Soil erodibility factor