Traffic Volume Analysis from GPS / Probe Data

TL;DR

Instead of (or alongside) roadside counters, traffic can be estimated from “probe data” — GPS traces from vehicles, smartphones, and ETC2.0 units. On 7/21 at Ishikawa College you’ll learn the fundamentals and applications. The maintenance link: traffic loading drives deterioration, so traffic data drives prioritization.

What is probe / GPS data?

• Probe data = position/speed traces from moving “probes”: GPS-equipped vehicles, fleet telematics, smartphone apps, and Japan’s ETC2.0 roadside-communication units.
• Yields travel time, speed, route choice, congestion, and — with modeling — traffic volume estimates.
• Big advantage: wide spatial/temporal coverage without installing counters everywhere.

From traces to volume

The core challenge

Probe vehicles are only a sample of all traffic. Estimating true volume needs a penetration/expansion factor (what fraction of vehicles are probes), plus map-matching of noisy GPS to the road network and statistical correction.

graph TB
    P[Raw GPS traces] --> M[Map-matching to road network]
    M --> F[Filter / clean]
    F --> E[Expand by penetration rate]
    E --> V[Estimated volume, speed, OD flows]
    V --> U[Maintenance prioritization & planning]

Why traffic data matters for maintenance

• Loading: heavy/frequent traffic accelerates pavement and bridge deterioration → predicts where to inspect/repair first.
• Prioritization: limited budget should target high-volume, high-consequence links.
• Detours & resilience: after a disaster or closure, probe data shows how traffic reroutes (very relevant to Noto).
• Low-cost for developing agencies: smartphone/GPS data can substitute for expensive sensor networks — a strong comparison point.

Why it matters for this workshop

This session widens “maintenance” from the structure itself to the demand on it. A good forum angle: could probe/GPS data give a low-budget agency the prioritization power that rich agencies get from dense sensors? Ask the Ishikawa College hosts what data sources and penetration rates they use.

Sources

• Probe/travel-time data collection handbook: FHWA
• ETC2.0 probe-data applications (MLIT/NILIM literature).
• Host: National Institute of Technology, Ishikawa College.