Modelling streamwater quality under varying hydrological conditions at different spatial scales

A simple and pragmatic technique for modelling streamwater acidity in a major Scottish catchment, the River Dee in Aberdeenshire, is presented. Based on End Member Mixing Analysis (EMMA), the technique is used to predict daily streamwater concentrations of Gran alkalinity and calcium at a range of spatial scales (1-690 km(2)). For a given site anywhere within the upland region of the catchment, the streamwater end member compositions are estimated from geological, soil and land cover data using a Geographical Information System (GIS). The utility of the approach is demonstrated by the predicted streamwater Gran alkalinity and calcium concentrations matching with a reasonable degree of accuracy those observed in the field. There are two assumptions: (i) the streamwater is composed of two end members and (ii) the end members are temporally invariant. The accuracy of the model predictions suggests that these two assumptions are reasonable, at least to a first approximation in large (> 50 km(2)) catchments. To model environmental change the technique must be linked to process based models, which predict how end members will change over time. (C) 1999 Elsevier Science B.V. All rights reserved.