Data-Driven Approach to Weather-Responsive Speed Management of Road Tunnel Access Zones

Abstract

In South Korea, one of the leading causes of motor vehicle crashes occurring at road tunnel entrances is speeding or failure to maintain a safe distance, especially during adverse weather conditions. The South Korean government released a strategic plan to address this issue through weather-responsive speed limit adjustments at tunnel entrances. This study aimed to assist the South Korean government’s plan with a data-driven safety approach. In particular, crash data from freeway tunnel access zones were augmented with real-time traffic speed and weather data, and standard binomial logit regression (BLR) with random forest (RF)-based variable selection was employed. The key variables contributing to crashes at freeway tunnel access zones during adverse weather conditions were identified as: speeding or failure to maintain a safe distance; no existence of a roadside fence; freeway sections with a distance of 100 to 250 m from the tunnel entrance; and interactions of traffic speed with wet pavement, average hourly rainfall of 6 h, visibility, and humidity preceding crash. The current standards for weather-responsive speed limit adjustment and appropriate deployment of variable message signs can be informed by the findings of this study.