Rough set approach to the evaluation of stormwater pollution
Luca Rossi, Roman Słowiński, Robert Susmaga · International Journal of Environment and Pollution · 1999
This paper addresses the problem of urban stormwater quality and its impacts on the receiving water. In order to describe this phenomenon, research aimed at creating a mathematical model of pollution has been conducted. Although various deterministic and stochastic models exist already, the quality of stormwater phenomenon deserves a very careful and thoughtful validation. In order to take advantage of some empirical observations available, we introduce an inductive learning method to discover some regularities in form of "if...then..." rules. We suggest the use of the rough set theory for this purpose. This approach is able to process both quantitative and qualitative data and also accepts inconsistency in the dataset. The rules generated from the data explain the phenomenon in terms of relevant attributes, and can be used to predict the level of pollution from future events. We present the results of the rough set analysis of a dataset concerning five catchments in the urban area of Lausanne and Geneva, Switzerland. The application of the rules to events from other catchments, particularly in France, proves the good prediction ability of our model.