Spatial statistical calibration on linear networks: an application to the analysis of traffic volumes

The estimation of traffic volumes on a street network represents a fundamental step to improve transport planning protocols and develop effective road safety interventions. The traditional ways to derive traffic figures involve manual counts or fine-tuned automatic tools (e.g. cameras or inductive loops). Unfortunately, the manual counts are extremely time consuming, whereas the fixed instruments are typically very expensive and geographically sparse. However, given the increasing availability of mobile sensors (e.g. smartphones and GPS sat-nav), in the last years we observed a surge of methods to infer traffic counts from geo-referenced mobile devices. This paper proposes a spatial statistical calibration technique to combine accurate fixed counts and extensive GPS mobile data for the estimation of traffic flows, re-adapting the statistical methods to the spatial network context. The suggested methodology is exemplified using data collected in the City of Leeds (UK).