Surface runoff, soil erosion and water quality modelling in the Alpone watershed using AGNPS integrated with a Geographic Information System

Abstract

A modern approach to problems of non-point source water pollution and erosion in a watershed consists in linking distributed models to Geographic Information Systems (GIS). Water quality and quantity are affected by the inherent spatial and time variability of the hydrological attributes within a watershed. Spatially distributed models and data are more suitable than conventional lumped ones. They act directly on spatially distributed data to assess hydrological transport processes affecting water quality. The linkage of GIS is a powerful tool to handle and provide the large amounts of detailed input data that these models require, reducing uncertainty caused by spatial averaging. The AGNPS (Agricultural Non-Point Source) model was used to assess runoff, soil erosion, and associated non-point source pollution for events monitored in the Alpone watershed (Italy). This is an event-based model and operates on a cell basis so that the spatial variation in parameters of each cell can be accounted for in the analysis throughout the whole watershed. Runoff predictions were compared for 20 observed events that occurred in 1992 and 1993. Predictions of sediment yield and nutrient loading were compared with seven observed events. An integrated AGNPS/GIS system was developed to efficiently create the necessary large number of data inputs. Moreover, graphical displays of the results have proven to be a very effective and efficient way of interpreting the results and of decision-making during model calibration. The GIS output layers may be used in planning studies to determine locations, within a watershed, that are critical in sediment yield and in the contribution of pollutant nutrients. The early results of the study point out that the runoff volume is properly estimated by the model while peak runoff rates are poorly predicted, both at low and high values. Predicted nutrient load generally parallels predicted sediment load: in fact in the case of nutrients, overestimation or underestimation are found to extend over the same ranges as those predicted for sediment load by AGNPS. The results from the Alpone watershed have illustrated the capability of the model for estimating sediment and nutrient yields and their spatial distribution particularly in the case of large storm rainfall events, which produce most of the total annual load of sediment and nutrients in the watershed. Nevertheless, the need is felt both of checking the model more carefully and of conducting further studies in order to take into account the considerable space-time variability of rainfall data within the watershed.