Comparing physically based and empirical models in predicting the sediment yield from urban watersheds Case study: Lubigi Catchment

Abstract

The performance of many urban drainage systems in Uganda has been affected by sediment deposition in drainage channels. These sediments reduce their carrying capacity leading to flooding that causes destruction of property and poor standards of living. Models have been used to predict the sediment yield from bigger river catchments but very little modelling has been carried out for the sediment yield in these urban catchments. Therefore, using a case study of Lubigi Catchment, this study predicted the sediment yield from the watersheds of these urban channels using a physically based model i.e. Water Erosion Prediction Project(WEPP) and an empirical model i.e. Universal Soil Loss Equation(USLE). The model output results were then compared with the actual measured load in the channel so as to determine the prediction accuracy of these models. Water sampling from the channel was carried out thrice a month and the samples were taken to the laboratory to determine their sediment concentration. This sediment concentration was multiplied with the measured discharge to obtain the sediment load. The measured sediment yield was 387.78 ton/month in January, 611.01 ton/month in February, 559.12 ton/month in March and 454.61 ton/month in November The inputs for the WEPP model were the DEM climate file, land use file and a soil file. The sediment yield obtained was 557 ton/month in January, 770.53 ton/month in February, 733.38 ton/month in March and 683.39 ton/month in November. For USLE, the required files were the DEM, monthly rainfall data, satellite images and the soil map. The sediment yield obtained was 648.96 ton/month in January, 992.99 ton/month in February, 1106.56 ton/month in March and 948.48 ton/month in November. These models were used in conjunction with GIS software to enable the determination of both the spatial and total monthly sediment yield for Lubigi catchment. Sensitivity analysis was conducted to identify parameters that most influenced the predicted sediment yield. The Nash Sutcliffe Efficiency (NSE) was used to evaluate model performance with NSE values greater than 0.5 regarded as satisfactory performance. Calibration was done using the January to March dataset while validation was carried out using the November dataset. This study also involved design of a sediment detention pond to reduce the amount of sediment deposited in Lubigi drainage channel. The sediment pond was to be constructed at Kalerwe, Bwaise and Namugoona and was to be desilted every seven months. The pond volume was divided into the settling zone (1454.76m3) and the sediment storage zone (727.38m3).