Modal split and accessibility: forecasting transport use through the modal accessibility gap

Abstract

Accessibility is a central lens through which transport geographers understand potential mobility, yet it is rarely used to predict realised behaviour such as modal shares. This paper proposes a territory-based predictive modal split model in which the Modal Accessibility Gap (MAG). operationalised through multimodal Spatial Availability (mSPAV), serves as the primary predictor of car use in home-work trips across the Madrid metropolitan area. Three accessibility measures of increasing sophistication are evaluated: a Hansen-type gravity indicator, a unimodal Spatial Availability, and a population-group-neutral multimodal Spatial Availability. Results show mSPAVbased MAG significantly outperforms the others. A cubic regression model incorporating MAG alongside income and motorisation achieves R? = 0.658. The approach is theoretically grounded in random utility theory, data-efficient, and operationally simple, offering a spatially meaningful alternative to individual-based mode choice models in data-scarce or upstream planning contexts.