Modeling land use decisions with Bayesian networks: Spatially explicit analysis of driving forces on land use change

Land use decisions result from complex deliberative processes and fundamentally influence the livelihoods of many. These decisions are made based on quantitatively measurable information like topography and on qualitative criteria such as personal preferences. Bayesian networks (BN) are able to integrate both quantitative and qualitative data and are thus suitable to approach such processes. We model land use decisions in a pre-Alpine area in Switzerland, integrating biophysical data and local actors' knowledge into a spatially explicit BN. A structured experts' process to elaborate three different BN including agriculture, forestry, and settlement provides the base for the modeling. A spatially explicit updating of the BN via questionnaires enables us to take local actors' characteristics into account. Results show which drivers are most important for land use decision-making in our case study region, and how an alteration of these drivers could change future land use. Furthermore, focusing on the probability of occurrence of various land uses in a spatially explicit manner gives insights into path-dependency of land use change. This knowledge can serve as information for planners and policy makers to design more effective policy instruments.