Linking models for land use analysis: experiences from the SENSOR project

In order to quantify the effects of a comprehensive set of policies on land use, interaction between sectors needs to be accounted for, while maintaining a high level of detail for each sector. This calls for a combination of sector specific and sector wide models. This paper describes such a modelling system, with emphasis on the linking of the models to a coherent system. Five sectors of significant importance for land use are modelled individually: Forestry, agriculture, urban land use, transport infrastructure, and tourism. All models are connected as sub-modules to an economy-wide partial econometric model. In addition, a land cover model is used to disaggregate land use down to 1km grid resolution. The linking of such a diverse set of models in a consistent way poses conceptual as well as practical issues. The conceptual issues concern questions such as which items of the models to link, how to obtain a stable joint baseline scenario, and how to obtain a joint equilibrium solution for all models simultaneously in simulation. Practical issues concern the actual implementation of the conceptually sound linkages and provision of a workable technical solution. The linked system allows us to introduce a shock in either of the models, and the set of results will provide a joint solution for all sectors modelled in SENSOR. In this manner, the models take a complex policy scenario as argument and compute a comprehensive set of variables involving all five land use sectors on regional level, which in turn forms a basis for distilling out the impact on sustainability in the form of indicators. Without the extensive automation and technical linkages, it would not have been possible to obtain a joint equilibrium, or it would have required exorbitant amounts of working time.