Application of the Modified Universal Soil Loss Equation (MUSLE) for the prediction of sediment yield in Agewmariam experimental watershed, Tekeze River basin, Northern Ethiopia.

The study utilized the Modified Universal Soil Loss Equation (MUSLE) to predict sediment loss and evaluate the model's performance in the Agewmariam experimental watershed in order to support the planning, management, and appropriate use of the soil and water resources in the watershed. The natural resources conservation service (NRCS) curve number method was used to model runoff energy factor. By overlaying maps of runoff energy, soil erodibility, slope length and steepness, cover management, and support practice factors with assigned values, the cumulative effect of these parameters for the suspended sediment yield was calculated using the ArcGIS raster calculator. The runoff energy factor was the most sensitive parameter, followed by slope length and steepness factor. To improve the model's fit to the local conditions, the initial abstraction to storage ratio (λ) of the runoff energy factor was reduced to 0.023, and the MUSLE model coefficient and exponent were adjusted to 1 and 0.59, respectively. During calibration, the mean observed and estimated suspended sediment yields were 0.2 and 0.23 ton/ha, respectively, while during validation, they were 0.7 and 0.53 ton/ha, respectively. The model evaluation showed that the MUSLE model, without calibration, was not appropriate for estimating runoff and sediment yield. However, with appropriate calibration, the model showed good performance with a coefficient of determination (R), coefficient of efficiency (E), and index of agreement (d) of 0.85, 0.85, and 0.96 respectively, during calibration and 0.84, 0.65, and 0.83 respectively, during validation. Based on these findings, this study suggests that the calibrated MUSLE model can be used to prioritize soil and water conservation interventions within the watershed or can be extrapolated to neighboring similar watersheds. Further refinement of model input parameters using more data from the watershed is recommended to increase the prediction accuracy of the model.