Estimating the Influence of the Neighbourhood in the Development of Residential Areas in the Netherlands

Abstract

Land-use models have been developed for predicting future land-use patterns. We describe a logistic regression model for residential development in the Netherlands that takes the neighbourhood effects of land use and infrastructure into account. Hereto, a new methodology was developed, single parameter impact estimation, for estimating the influence of land use and infrastructure in the neighbourhood on residential developments. The model also incorporates initial land use and a set of policy-related and suitability-related variables. In essence, the neighbourhood effects typically simulated with cellular automata-based land-use models have been integrated into a logistic regression model. The model was calibrated and validated on historic land-use developments in the Netherlands. On the basis of the model, residential areas develop preferably at the fringe of the city, on agricultural land in the neighbourhood of residential areas. The probability of residential development depends on the cluster size of the land-use categories. Large clusters of allotments, cemeteries, or sports fields are relatively attractive for residential development, while forest, nature, and industrial areas become more attractive when present in small clusters.