Author(s)

D. Vicente, A. Vellido, E. Martí, J. Comas & I. Rodriguez-Roda

Abstract

The STREAMES (STream REAch Management, an Expert System) European project is an attempt to develop a knowledge-based environmental decision support system (KB-EDSS) to help water managers in making decisions. Such a KB-EDSS involves the evaluation of the ecological status of a river. Particular emphasis is placed on Mediterranean locations. The data from the STREAMES project are scarce and incomplete. They include several types of variables, including physical, chemical and biological parameters. The following preliminary question must be addressed: are we allowed to assume that the same water quality model is valid over the variety of rivers under consideration, or should we assume a different model for each river? Between these two extreme assumptions, rivers could be grouped according to the available measurements. This is a clustering problem, and the current study attempts to find an answer to the posed question using the Generative Topographic Mapping (GTM), a probabilistic alternative to the Self-Organizing Map (SOM). Amongst the many advantages of the probabilistic setting of GTM, one is especially relevant to the problem at hand: its ability to reconstruct missing data in a principled way. Keywords: Generative Topographic Mapping (GTM), ecological status of a river, clustering, data visualization, missing data, decision support system, SOM.

Keywords

Generative Topographic Mapping (GTM), ecological status of a river, clustering, data visualization, missing data, decision support system, SOM.