Identifying and assessing uncertainty in hydrological pathways: a novel approach to end member mixing in a Scottish agricultural catchment

A hydrograph separation based upon end member mixing was carried out to assess the relative importance of the hydrological pathways providing the main sources of runoff during five storm events in a 14.5 km(2) agricultural catchment in north east Scotland. The method utilised event specific end member chemistries to differentiate three catchment-scale hydrological pathways on the basis of observed Si and NO3-N concentrations in sampled source waters. These were overland flow (OF) (low Si and intermediate NO3-N); subsurface storm flow (high Si and high NO3-N) and groundwater flow (high Si and intermediate NO3-N). The hydrograph separation explicitly accounted for uncertainty in the spatial and temporal variation in end member chemistry using Bayesian statistical methods which assumed that each end member arose from a bivariate normal distribution whose mean vectors and co-variance matrices could be estimated. Markov Chain-Monte Carlo methods were used to model the average and 95 percentile maximum and minimum contributions that each end member made to stream water samples during storm events. Although there is large uncertainty over the contributions of each end member to specific events, the analysis produced hydrograph separations that were broadly believable on the basis of hydrometric observations in the catchment. Moreover, by using event specific end member compositions, the method appeared sensitive to the unique combination of event characteristics, antecedent conditions and seasonality in terms of producing feasible separations for very different events. It is concluded that OF generally dominates the storm peak and provides the main flow path by which P is transferred to stream channels during storm events, whilst subsurface storm flows usually dominate the storm hydrograph volumetrically and route NO3-rich soil water into streams. Consequently, nutrient enrichment in such streams is largely mediated by event-based hydrological flow paths, a finding that will need to be considered in any future mitigation strategies. (C) 2003 Elsevier Science B.V. All rights reserved.