Evaluation of the AnnAGNPS model for predicting runoff and sediment yield in a small Mediterranean agricultural watershed in Navarre (Spain)

• AnnAGNPS was evaluated for predicting runoff and sediment in agricultural watershed.
• AnnAGNPS showed good ability to simulate runoff after calibration.
• Sediment load was satisfactory predicted at annual scale but not monthly and seasonally.
• Findings showed a suitable applicability of AnnAGNPS at annual scale.

AnnAGNPS (Annualized Agricultural Non-Point Source Pollution Model) is a computer model developed to predict non-point source pollutant loadings within agricultural watersheds. It contains a daily time step distributed parameter continuous simulation surface runoff model designed to assist with determining Best Management Practices (BMPs), the setting of Total Maximum Daily Loads (TMDLs), and for risk and cost/benefit analyses. The purpose of this study was to evaluate the capabilities of the model to simulate runoff and sediment loads in a small Mediterranean agricultural watershed (207 ha) located in the region of Navarre (Spain) using nine years of continuous data monitoring. Data for the first five years (2003–2007) were used for calibration and the remaining four years (2008–2011) for validation. No significant channel erosion nor in-stream structures were identified within this small and homogeneous watershed. Thus the set of models for analysing stream networks and corridors were not considered in this study. Firstly, AnnAGNPS was calibrated for runoff by modifying Curve Number values for different stages of the main crops. Results showed that the model satisfactory simulated surface runoff at monthly, seasonal and annual scales both in calibration and in validation processes. A Differential Sensitivity Analysis (DSA) was carried out to evaluate the sensitivity of eight input parameters for sediment load prediction. Based on DSA results, the AnnAGNPS model was calibrated for sediment load simulation. The model was capable of simulating the sediment load at an annual scale with a difference of less than 1% for calibration and 7% for validation. However, the results at monthly and seasonal scale were less precise. The use of stream networks and corridors models, already inserted in the system of computer models that make up AnnAGNPS, seem to be necessary for a more precise explanation of sediment yields and loads at monthly and seasonal scales, even for small watersheds.