Abstract

Sustainable environmental management is only possible within an ecosystem-based approach. It requires long-term decision making and, given the degree of ecosystem uncertainty and dynamics, it is hardly achievable without simulation modelling. Real-scale frameworks consistst of simulators with different temporal and spacial resolution, and an obvious problem is to unify the interaction of inputs and outputs of different models in the framework. There is also a well-known dilemma of large models versus small models. Sophisticated models involve a large number of state variables and even larger number of parameters. They are characterized by long development cycles, high costs and difficulties with communicating to and understanding by decision and policy makers, general public and other stakeholders of the environmental tools. For these reasons, very often a preference is given to small and simple ecological models over large and comprehensive models. It is, however, important to keep descriptive and forecasting power of the model and to provide practitioners with a tool sufficient for the purposes of environmental decision making and management. The paper presents an approach permitting to overcome both methodological problems using the methods of experiment design as a means of both model aggregation and building a simple model from existing large simulation model developed for a particular environmental problem. The proposed approach is illustrated by a case study on sustainable management of a forested watershed.