Information theory for evaluating environmental classification systems

Environmental pollution data are often ranked in rule-based classification systems. These environmental data are separated in predetermined classes of a classification system for a better and smarter characterization of the state of pollution. Often the measured values are transformed, e.g. in pseudocolor maps, and can then be presented in maps. For some environmental compartments different classification systems for evaluating environmental loadings are used. Because of the dissimilarity of the various classification systems direct visual comparison is difficult. However, by means of information theory an objective comparison of these various classification systems based on their information content enables a decision to be made about which system is the most informative for objective assessment of the state of pollution. By means of the new measure “multiple medium information content” (multiple entropy) objective and simultaneous comparison of all channels (in an environmental classification system: pollutants) of each classification system is now possible. Furthermore the development of the state of pollution over the whole investigation period can be detected by means of information theory. On the basis of the conditions of the established rule-based systems the use of information theory enables definition of new ranges of classes in order to reach the optimum of information during conversion into the environmental classification system.