Spatial pattern recognition of the structure of urban land uses useful for transportation and land use modelling

Transportation and land use planners and modelers often use zone systems to take advantage of readily available socio-economic data in their modelling exercises. The most important issue is that analysis zone size is usually large, and therefore, homogeneous or single-type land uses cannot be achieved in many cases. Subsequently, intra-zone travel and mixed activities distribution cannot be captured and modeling accuracy has to be compromised. This study analyzes the form and spatial structure/pattern of different LUs within zone and find correlations between LU types and several morphological properties (building height, building area, building perimeter, building compactness, and parcel area) of parcels and their spatial arrangement indexes (Gabriel Line and Gabriel Length). Binary logistic model is then applied and fitted to the morphological properties and spatial indexes in order to extract residential and commercial LUs. The final Result demonstrates that residential and commercial LUs are extracted with an overall accuracy of 98.4% and 69.0% respectively.