Watershed acidification models using the knowledge-based systems approach

Abstract This paper presents a novel approach to prediction of lake acidification. We use for illustration the Cation Denudation Rate (CDR) and the Trickle Down (TD) Acidification Models. Instead of selecting one model and discarding the other, we utilize each model in those cases where it is most applicable, an approach which requires both a quantitative and qualitative judgement or rules to choose the proper model. This model has been implemented in a workstation environment — RAISON Micro — which has been designed to facilitate automated model selection and analysis. We present the results of a preliminary test using the water chemistry data from 53 southern Quebec watersheds in Canada with 364 sampling stations. Statistical comparison with observed data was found to be more favourable than that for the individual models. The results are improved by the system's built-in facility to correct anomalous behavior in circumstances for which an incorrect model choice has been made in the absence of definitive knowledge. The uncertainties of the individual models and the combined model were found to be greater for higher SO 2 inputs but became smaller for reduced loads.