Evaluation of infiltration using the green-ampt model and rainfall-runoff data for lagan and sambret catchments, Kericho, Kenya

Abstract

The Green-Ampt model was used to evaluate the infiltration parameters by testing it on Sambret and Lagan catchments of Kericho district, Kenya. The comparison of observed and predicted surface runoff formed the basis for evaluating the parameters in the Green-Ampt infiltration model. The model parameters were obtained from texture based nomographs emanating from United States Department of Agriculture-Soil Conservation Service (USDA-SCS). Values of antecedent soil moisture were obtained from available records. The nomographs have not been previously examined in Kenya. The storm hyetographs and hydrographs provided the necessary hydrologic information in the analysis. For the selected events In all catchments, surface runoff was over estimated by the model both in its original form, for ponded conditions and the modified form, for predicting rainfall infiltration. This was regardless of the influence of interception which was estimated as a percentage of storm rainfall for each event considered. Poor prediction by the original Green-Ampt model was attributed to its poor representation of infiltration during rai~fall events. It assumes immediate ponding at the onset of rain so chat; surface runoff starts at the moment rainfall comnences. The result is an ove~rediction of surface runoff. Overorediction of sur:::acer..moffby the modified Gree!1-Fmptmodel was at~ributed to the low values of saturated hydrauiic conductivity (Ko) and x effective porosity (85) predicted from the nomographs. A low value of ~ implies that less water is allowed to infiltrate resulting in high surface runoff volume. A low effective porosity (85) means a reduced moisture deficit due to the small difference between the porosity and antecedent soil moisture. Attempts were made to establish a correct value of ~ for each catchment through an optimization and validation exercise. K, lS the most variable and unpredictable parameter. 85 is a function of soil texture and varies only slightly with land use changes. The measured value of 85 (85=0.7crn3/crn3)obtained from records was used In the procedure of optimization and validation. Saturated hydraulic conductivity (~) for Sambret, Sambret sub-catchrnent , and Lagan catchments were found to be 0.03, 0.04, and 0.05crn/h respectively. Since these optimized values were those obtained with interception taken into consideration, the results therefore show that interception has a significant effect on the rainfallinfiltration- runoff process in all the catchments studied. These results also show that the nomographs require modifications before they can be used on Kenyan catcr.ments. The infiltration rate curve derived from the original Green-~npt model was compared to the curve obtained from double ring infiltrometer measurements. The results indicaced that the concencric cylinder measurements yield highly variaele and UDxeliable results which must be used cautiously.