Computer-Aided Civil and Infrastructure Engineering 17 (2002) 104–118 From Neural Networks to Qualitative Models in Environmental Engineering Dominik Wieland Technische Universität Wien, Institut für Informationssysteme, Favoritenstraße 9–11, A-1040 Wien, Austria Franz Wotawa* Graz University of Technology, Institut for Software Technology IST, Inffeldgasse 16b/II, A-8010 Graz, Austria & Gerhard Wotawa Universität für Bodenkultur, Institut für Meteorologie und Physik, Türkenschanzstraße 18, A-1180 Wien, Austria Abstract: As an alternative to physical models, artificial neural networks (ANNs) are a valuable forecast tool in environmental sciences. They can be used effectively due to their learning capabilities and their low computational costs. Once all relevant variables of the system are identi- fied and put into the network, it works quickly and accu- rately. However, one of the major shortcomings of neural networks is that they do not reveal causal relationships between major system components and thus are unable to improve the explicit knowledge of the user. Another problem is due to the fact that reasoning is only done from the inputs to the outputs. In cases where the opposite is requested (i.e., deriving inputs leading to a given output), neural net- works can hardly be used. To overcome these problems, we introduce a novel approach for deriving qualitative infor- mation out of neural networks. Some of the resulting rules can directly be used by a qualitative simulator for pro- ducing possible future scenarios. Because of the explicit representation of knowledge, the rules should be easier to understand and can be used as a starting point for cre- ating models wherever a physical model is not available. * To whom correspondence should be addressed. E-mail: wotawa@ist.tu-graz.ac.at. Moreover, the resulting rules are well adapted to be used in decision support systems. We illustrate our approach by introducing a network for predicting surface ozone con- centrations and show how rules can be derived from the network and how the approach can be naturally extended for use in decision support systems. 1 INTRODUCTION Artificial intelligence has been successfully applied to solve problems regarding physical and meteorological systems. Application areas include explanation of physical systems, forecasting, decision support, and modeling. The problem of giving meaningful explanations of the behavior of a physical system has been tackled by qualitative reason- ing (QR), for example, by Weld and de Kleer (1989) and Dague (1995). Several approaches for representing physi- cal systems in a qualitative way have been proposed. The most influencing techniques are qualitative simulation (see Kuipers, 1986), confluences (see De Kleer and Brown, 1984), and qualitative process theory (see Forbus, 1984). Although one of the goals of QR was to develop mod- els able to reason about physical systems and consequently to give a program some kind of physical problem-solving © 2002 Computer-Aided Civil and Infrastructure Engineering. Published by Blackwell Publishing, 350 Main Street, Malden, MA 02148, USA, and 108 Cowley Road, Oxford OX4 1JF, UK.