Identifying multi-modal deserts: A multivariate outlier detection approach

Providing diverse modes of travel facilitates people's access to jobs, healthcare, critical activities, and other services. To assess the equity of access to transportation services, it is essential to consider different travel modes. In this study, we propose a concept called “multi-modal deserts” and develop an approach to identify them. Multi-modal deserts refer to areas with limited mobility options, which restrict people's access to essential services and opportunities. Based on the concept of multi-modality, our methodology integrates Mahalanobis distance for multivariate outlier detection to identify if an area's mobility services significantly deviate from other areas considering road network factors and travel modes. Downtown Tampa, Florida, was selected as an empirical case to demonstrate the proposed method, and 11 multi-modal deserts were identified among 182 Census Block Groups. In addition, spider charts were used to illustrate and compare the characteristics of these multi-modal deserts. The results identified several multi-modal deserts with different poverty levels and transportation constraints. The insights can assist local authorities in identifying mobility gaps, allocating resources more effectively, and improving equal access to opportunities for all residents.