Data-Driven screening of low-density urban freight for sustainable mobility planning

Cities need practical ways to identify freight activity that generates high local burden but remains weakly visible in municipal data. Low-density freight is one such case: sparse, infrequent or weakly utilised carrier activity can impose disproportionate emissions, congestion, noise and street-level pressure, yet local authorities rarely observe tour structure or load factors. This paper develops a data-driven screening framework for sustainable urban mobility planning in Pardubice, Czech Republic. The framework combines 74,022 delivery records, 3.25 million cordon movements from 13 cameras and 65 high-confidence carrier-profile observations used in the reproducible Pathway-A screen. Pathway A applies camera-based vehicle mix and observation frequency to flag candidate light-sparse carriers; one-sighting verdicts are retained as audit priorities. Pathway B uses administrative records to identify a heavy-sparse forwarder that camera-only screening cannot detect. A 10,000-resample stability test exposes the sampling fragility of sparse camera verdicts. A separate stops-per-tour sensitivity analysis shows that the forwarder-to-parcel external-cost ratio is 4.6 at the nine-stop baseline, 3.2 at 15 stops, 2.1 at 30 stops and 1.6 at 60 stops. External costs remain model-attributed and are not normalised by consignment weight, volume or value. The framework is a locally calibratable planning tool, not a universal classifier or enforcement rule.