Modelling water chemistry for a major Scottish river from catchment attributes

1. The prediction of water quality is increasingly required in river catchm ent management, but methods are still developmental. We therefore derived e mpirical models to predict the concentrations of base cations, H+ and alkal inity at any point in a complex Scottish river system, and under diverse di scharge conditions. Input data were readily available from geological and t opographic maps, whole rock composition data and catchment land use invento ries in geographical information systems (GIS). 2. A key and novel feature of the model was prediction using geological dat a for the riparian zone within 50 m of the river. The discharge contributio ns that passed through soil derived from each parent rock were estimated an d used as weightings in predicting final run-off quality. 3. Typical equations fbr upland catchments for mean, maximum or minimum riv er water Ca concentration, [Ca], were of the form [Ca] = a + b root{Ca-rz}, where {Ca-rz} denotes flow routing-weighted rock CaO concentration of the riparian zone rock types present. These equations were significant at P < 0 .0001. Similar approaches were applicable to alkalinity and to other base c ations. 4. Predictions of [Ca] in catchments with mixed land use were improved by i ncluding model terms for catchment riparian zone cover of agriculturally im proved (intensified) grassland and arable land. These results indicated ant hropogenic effects on base cation flux that would not be represented by geo logical data alone. 5. Similarly, concentrations of Na, Mg and K were correlated with Cl concen tration in the river water, primarily as a consequence of marine-derived Cl . Including distance from the east coast as a predictive variable in place of [Cl] obviated the need for direct [Cl] measurement in model operation. 6. We advocate further work to assess whether similar models can be develop ed and applied in other geographical locations, where features such as land use, geology and sea salt inputs will all vary.