As alternative to physical models, neural networks are a valuable forecast
tool in environmental sciences. They can be used effectively due to their l
earning capabilities and their low computational costs. As far as the relev
ant variables of the system are measured and put into the network, it works
fast and accurately. However, one of the major shortcomings of neural netw
orks is that they do not reveal causal relationships between major system c
omponents and thus are unable to improve the explicit knowledge of the user
. To overcome this problem, we introduce an approach for deriving qualitati
ve informations out of neural networks. Some of the resulting rules can be
directly used by a qualitative simulator for producing possible future scen
arios. Because of the explicit representation of knowledge the rules should
be easier to understand and can be used as starting point for creating mod
els wherever a physical model is not available. We illustrate our approach
using a Network for predicting surface ozone concentrations and discuss ope
n problems and future research directions.