Modelling Off-Street Parking Supply at the Parcel Level: A Regression-Based Case Study in two German Cities

Kurzfassung

Data on off-street parking supply is poorly documented, although highly needed for effective parking policies. To address this gap, this study models off-street parking supply in the German cities Hamburg and Munich. Parking capacities are derived using aerial imagery-based surface parking segments, cadastral data, and OpenStreetMap features, yielding over 1.3 million off-street parking spaces in Hamburg and over 600,000 in Munich. Negative binomial count regression is conducted to model parking supply and its determinants on parcel level. Ground floor area and open space ratio show strong positive relationships with the residential parking supply density, while the latter is decreased with rising standard land values and higher household purchasing power. Non-residential coefficients are less stable throughout the three investigated samples. Parking supply prediction using cross-validation reveals moderate errors for residential models driven by outliers and substantially higher errors for non-residential uses. Transferring the Hamburg model to Munich data suggests that model specification is a greater challenge than spatial disparities between cities. This study provides a replicable approach for off-street parking supply modelling that future research can build upon, with particular need for further investigation into non-residential parking provision and the spatial transferability of model results across diverse urban contexts.