Title:Evaluation of Crowdsourced Data on Unplowed Roads

Abstract:Transportation agencies routinely collect weather data to support maintenance activities. With the proliferation of smartphones, many agencies have begun using crowdsourced data in operations. This study evaluates a novel unplowed roads dataset from the largest crowdsourced transportation data provider Waze. User-reported unplowed roads in Virginia were compared to national and state weather data for accuracy, and found 81% of reports were near known snow events, with false positives occurring at a regular rate of approximately 10 per day statewide. Reports were largely located on primary roads, limiting the usefulness for transportation agencies who may be most concerned with poorly monitored secondary roads. An effort to encourage unplowed road reporting in Waze through targeted messages on social media did not increase participation. Low reporting may be due to the feature's novelty, recent mild winters, or COVID-19 school and business closures.