Forest Ecosystem models are diagnostic tools to assess and understand ecosystem processes. Conflicting interests such as simplicity, observability and biological realism must be addressed to ensure a well balanced modelling approach. Because field observations are usually only available for short time periods or for a limited number of locations, models are important to extrapolate in space and time. The key to a successful modelling approach relies on finding the appropriate scale but is often limited by the availability of input data. In the practice of forest modelling, it is often necessary to produce meaningful outputs on the basis of rather weak information. In this paper we discuss spatial and temporal scaling issues within empirical modelling. Following the case studies the paper demonstrates how models address cross-scaling problems as they are relevant for the required data as well as the decision making and implementation process of specific end user needs.

Key words: forest modelling, scaling, end user