摘要
Integrated assessments that aim to support sustainable natural resources management require analysing how biophysical systems are impacted by human actions. These analyses are often performed by modelling the physical system, while human actions are prescribed as scenarios and introduced into the physical models by varying the model input. To achieve a more thorough analysis of the human system component in participatory integrated assessments, we developed a semi-quantitative approach for actor-based modelling which focuses on modelling actions of societal actors based on their problem perceptions but also computes the resulting changes of physical system variables. Our approach is intended to support transdisciplinary research and identification of sustainable development options in problem fields with high levels of uncertainty or ignorance, together with the actors that are being modelled. Actor-based modelling is done using an enhanced version of the DANA software. DANA allows modelling the actors in a specific problem field by representing the perception of each actor by directed graphs, and by computing optimal actions from the perspective of each actor. These perception graphs are semi-quantitative causal maps, which can easily be discussed among stakeholders in a participatory process. DANA was extended to support, in addition to actor modelling, the other two steps in actor-based modelling, modelling of actions and modelling of factors. Modelling of actions refers to determining the actions of each actor under certain scenario assumptions, taking into account the diverse problem perceptions of the individual actors in the problem field, the action of the other actors and exogenous changes. Modelling of factors refers to calculating, in a semi-quantitative way, the resulting changes of physical variables (e.g. pollutant emissions), which may serve as input to physical models. We applied actor-based modelling in an integrated assessment of mobile organic xenobiotics in rivers. Our study shows that actor-based modelling allows generating scientifically better founded and more transparent scenarios of the drivers of change in integrated assessments, in particular because they are based on a structured analysis of the actors' problem perceptions.