Regional water quality of the river Tweed

Results are presented from a regional analysis of water quality in the Tweed river catchment. The analysis is based on the extensive data records of the Tweed River Purification Board for the period 1985–1994. A wide range of inorganic, nutrient and micro-organic determinands have been measured for samples taken from rivers, sewage works, refuse sites and industrial effluents. A regional perspective is provided with river water quality being interpreted in relation to catchment geography and to the observed point source inputs to the catchment. For many determinands, both diffuse and point sources contribute to the riverine load. Point source inputs are important for phosphate, metals (especially copper and lead) and micro-organic pollutants. Diffuse agricultural sources are particularly important in lowland arable areas, e.g. for nitrate, phosphate and micro-organics. In addition, widespread background geological sources contribute to loads of metals such as iron and zinc. Sewage effluent is the most widespread and most significant point source of many pollutant chemicals in the Tweed basin. High biochemical oxygen demand, nitrate, total ammonia, chloride and phosphate result from domestic effluent and industrial inputs, e.g. from sectors such as food processing and textiles. Heavy metals are also high at a few sites, especially at Selkirk and Galashiels, owing to wastes from the electronics industry. Micro-organic chemicals in effluent tend to be very site specific because of their industrial sources. Direct industrial discharges are very limited, most industrial effluent being discharged to the sewer. Refuse sites appear to have little impact on water quality; the most significant effects are for suspended solids, iron and dichlorvos. Fish farms cause depletion in dissolved oxygen levels and increases in suspended solids and total ammonia but, on the whole, these changes do not cause problems. The paper considers ways in which regional models can be developed given the resource of extensive water quality records. A basic empirical regression model is investigated for nitrate and is found to be able to explain regional variations in terms of land use and soil type. This simple type of approach may provide a means of estimating how stream water quality will respond to future changes in land use.