Land-use change models that accurately replicate the complex dynamics of land development provide vital information for urban planning and policy. These models require both detailed data and advanced statistical methods. Many factors influence land-use change decisions, such as parcel characteristics, accessibility to activities, and current and historical neighborhood conditions. Therefore, spatial and temporal components must be incorporated in a model at the highest possible disaggregation level in order to achieve robust results. A spatio-temporal multinomial autologistic model, incorporating space and time and their interactions, is introduced to investigate land-use dynamics at the parcel-level, and is applied to Delaware County, Ohio. It is able to capture the impacts of the existing and historical neighborhood conditions of parcels with high accuracy. Advanced computational methods are used to deal with the computational challenges of parameter estimation. The model is validated, estimating 91.4% of all observations correctly for the period 2005–2010, and is applied to land-use forecasting.
Read full abstract