A Prototype Winter Road Condition System for Logistics Companies

Winter storms cause transit disruptions across logistics networks, resulting in delayed deliveries. We introduce a road condition forecast system built on publicly available weather forecasts and open source road condition models. Because existing fixed site in-situ road observations are sparse, we supplement these with road conditions derived from vehicle telemetry data. We present results that evaluate the METRo and SUMMA road models ability to predict road surface temperature and categorical condition during the winter of 2022-2023. We test how much forecasts degrade when each source of observations (road weather stations, traffic cameras, vehicle telemetry data) are removed from the data assimilation step. Finally, we discuss the challenges and opportunities of running an operational road model across the US to support a delivery network.