SOIL EROSION PREDICTION USING MODIFIED UNIVERSAL SOIL LOSS EQUATION (MUSLE) IN TUGEN HILLS, BARINGO, KENYA

Abstract

Soil erosion by water is one of the primary causes of land degradation and occurs throughout the world. Soil erosion is contributing negatively to the already declining agricultural productivity thereby negatively influencing people’s livelihoods and economic empowerment. Therefore, there is need to understand erosion processes, quantify sediment yield, identify and rank critical sources on spatial domain of sediment. This will help in formulation of prioritized catchment conservation strategies. This study focused on estimation of sediment yield from Tugen Hills particularly Saimo catchment in Baringo County using Modified Universal Soil Loss Equation (MUSLE) model with a view to develop an understanding of inter-relationships between soil erosion and sediment yield. The input model parameters of runoff volume (Q) and peak flow rate (qp) were determined from runoff plots of dimensions 4.8m by 2m set up in the catchment with average slope of 2%. Soil erodibility factor (K) was calculated mathematically based on soil samples collected. Cover management (C) was obtained by percentage cover and support practice (P) factor was determined through observation and use of conversion table. Apart from determination of model parameters, the study calibrated and validated MUSLE for use in future studies within Saimo catchment and other catchments with similar characteristics. The mean bulk densities for top soil and bottom soil are 1.05g/cm2 and 1.07 g/cm3. The total value for fine sand and silt gives 37.1%. The saturated hydraulic conductivity varied from 8.0 μm/s to 41.3 μm/s with a mean value of 24.1 μm/s. There were only two classes high and moderately high translating to code 2 and 3, respectively. The analysis of variance (ANOVA) of the observed data showed that rainfall intensity affected the sediment yield production in the runoff plots and that there was no evidence to suggest that the soil homogeneity in the runoff plots affected the sediment yields. The observed and simulated MUSLE model values for calibration were PBIAS (0.83), R2 (0.75), r (0.87) and KGE (-0.20) and those for validation were NSE (0.96), PBIAS (-0.44), R2 (0.60), r (0.78) and KGE (0.46). Hence it can be concluded that the MUSLE model can be used successfully as an effective tool in soil conservation management. Future work for several seasons is however needed in order to capture different slopes and the varying climatic conditions for the model to be robust and to be used widely.