In the context of climate change and highly congested cities, moving towards low-carbon transport systems becomes of utmost importance. One approach to reduce mobility in urban areas assumes that changing the urban form can influence individual car travel. In particular, strategic changes to the different features of compact development, also referred to as the 5D’s (density, diversity, design, destination accessibility, distance to transit), have long been considered as an effective lever influencing individual mobility.
However, the precise influence of the 5D’s on driving distance is still unknown. So far, this influence has mostly been studied through small data samples derived from questionnaires, making the obtained results highly case and context specific. Furthermore, existing studies only focussed on a low number of urban form features and hardly analysed the spatial scale of the impact of the 5D’s.
This study aims to close this gap by analysing more than 25 million trips over one year in Berlin, as well as very detailed urban form information, including LiDAR-derived 3D buildings and street space allocation data. We adopt a deductive research design and use a flexible machine learning approach with gradient boosting. We create various features for each of the 5D variables, and test if and how they individually and collectively impact driving distance in Berlin. On top, we examine how the impact of the features vary over different spatial scales to map out areas where infrastructure change might be most effective.
The results of this study contribute to a more nuanced understanding for the impact of urban form on individual driving distance. This is important, not only for academia to advance the discussion about the effects of the urban form on individual mobility behaviour but also for governments and city planners to create more effective, low-carbon urban planning strategies.