A method for predicting the extremes of stream acidity and other water quality variables

A hydbrid deterministic - statistical approach is proposed for modelling the extremes of water quality in catchments subjected to long-term acidification. The approach is based on process models describing the long-term variations in mean chemistry. Superimposed on these mean projections are distributions providing information on the extremes of water quality which are fitted to catchment data using maximum likelihood techniques. The approach is general and can be applied to the prediction of other water quality variables where samples can be regarded as belonging to a parametric probability distribution. A simple implementation of the approach using chemical data and a calibrated deterministic model for the Allt a' Mharcaidh is used as an illustrative method.