Estimation of Road Surface Temperature Using NOAA Gridded Forecast Weather Data for Snowplow Operations Management

Abstract

Monitoring road surface temperatures is crucial to establishing winter maintenance strategies by the State Departments of Transportation (State DOTs) in the United States. Traditionally, transportation agencies rely on the information provided by Road Weather Information Systems (RWIS) for road surface temperatures along roadways. However, these systems are costly and only provide estimates at specific locations, resulting in distant areas being under-represented. In recent years, some interpolation techniques have been considered to address this gap by estimating the road surface temperatures between the RWIS stations. Nevertheless, these techniques are only valid when the RWIS data are available. This study aims to estimate the road surface temperatures using forecast weather data which are available at high spatial resolution in the National Weather Service Database maintained by the National Oceanic and Atmospheric Administration (NOAA). To this end, road surface temperature data were collected from roadways using a vehicle-mounted infrared temperature sensor. Furthermore, the associated forecast weather parameters from the National Weather Service database were used to develop relationships between the publicly available weather forecast data and the actual road surface temperatures using multiple linear regression. The authors developed two estimation models for dark and light groups and leveraged the gridded forecast weather data from the national weather service database to visualize the estimated road surface temperatures along roadways using a GIS approach. The results showed that the ambient temperature, relative humidity, wind speed, average temperature of the previous day, and road surface conditions (wet/dry) are statistically significant in estimating the road surface temperatures using gridded forecast weather data. The performance of the models was validated, and satisfactory accuracy metrics (i.e., mean absolute error) of approximately 1°C and 2°C were achieved for the dark and light groups, respectively. The proposed method was implemented in the TxDOT Wichita Falls district as a part of a Snowplow Operations Management System to provide information about the estimated road surface temperatures to transportation managers for the 2021–2022 winter season. This information facilitates establishing proactive anti-icing measures in locations where possible low surface temperatures are expected. The findings of this research contribute to a better understanding of the influence of publicly available weather forecast parameters on road surface temperatures.