Report No SR3843



Jan 1995


A variety of classification schemes are used in Scotland and Northern Ireland to describe river water quality either on a national basis or for routine operational purposes. Operational staff with experience in the application of these schemes identified a number of opportunities for the improvement of the currently used schemes and SNIFFER commissioned a project with the following ultimate goal:

To develop a rapid, automated, objective classification system for flowing freshwaters in Scotland and Northern Ireland, driven by biological quality information, that can be used for operational and routine classification purposes.

Chemical and biological data from the 1990 quinquennial survey were collated from the RPBs in Scotland and the Environment Service in Northern Ireland to facilitate the investigation of options for a chemical classification scheme and the derivation of a biological classification scheme.

Three options for the chemical classification scheme were investigated. On balance, a scheme proposed by Forth RPB, including 5 classes and numerical class boundaries for 4 chemical determinands, was selected as the most appropriate.

Both bankside- and laboratory-generated biological data were to be included in the biological classification scheme. A statistically robust relationship between the two types of biological data could not be derived. This precluded the direct conversion of the bankside-generated data into an equivalent of laboratory-generated data. Consequently, separate biological classifications were derived for bankside- and laboratory-generated data using linear discriminant analysis. Biological classes from both of these classifications lay within one class of the chemical class for more than 70% of sites in all classes. The agreement of the classifications to within one chemical class allowed the investigation of a biological over-ride and, therefore, a unified classification. The application of the biological over-ride to the sites in the available data set resulted in a series of class changes, although these changes did not always produce a better description of water quality.

There is potential for the automation of a system ranging from a simple classification routine to a sophisticated water quality archive, classification and plotting package.

Further refinement of the analyses performed in this study is recommended if any of the approaches investigated are to be implemented operationally, with particular attention to be paid to the range of chemical components measured, the frequency of samp ling, the summary statistics used and the desirability of a unified system.


River water quality; classification scheme.

Copies of the report are available from FWR, price 30.00, less 20% to FWR Members.