Skip to main content
ICTD 2025 Main banner
Final Program


Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.
Back

ICTD 2025

Artificial Intelligence (AI) in Transportation

POSTER RECEPTION (View list of posters here)

Enhanced Road Surface Temperature Prediction Using Random Forest Model and NWS Weather Forecast Data

Tuesday, June 10, 2025
4:45 PM – 6:00 PM MT
Abstract: Effective winter road maintenance depends on obtaining precise information about the temperature of the road surface. Transportation agencies typically utilize Road Weather Information Systems (RWIS) to monitor Road Surface Temperature (RST). However, due to the high costs associated with RWIS deployment, their availability is limited, especially in remote areas. The National Weather Service (NWS) provides high-resolution gridded weather forecasts, presenting an opportunity to estimate RST in areas without RWIS coverage by correlating these forecasts with RST using predictive models. However, current studies have not explored the application of Random Forest (RF) as a robust predictive model for estimating RST. This study evaluates the potential of the Random Forest model to enhance RST predictions based on NWS forecasts data. The RF model was trained using a dataset of actual RST measurements from North Texas roadways and corresponding NWS weather forecasts. The model’s performance was evaluated against other conventional predictive models, including Linear Regression, Deep Learning, Support Vector Machine, and K-nearest Neighbor. The Random Forest model demonstrated the highest accuracy, with an R-squared value of 0.96, a root mean square error of 1.37°C, and a mean absolute error of 1.03°C. This research advances road safety by offering more precise RST prediction models utilizing publicly available weather data. These models can be integrated into intelligent winter operations management systems, assisting highway agencies in obtaining reliable RST estimates in regions without RWIS.

Learning Objectives:

Attendees can expect to learn the following from this session:
  • Upon completion, participants will learn how the weather forecasts from the National Weather Service could be utilized to estimate the road surface temperature.
  • Upon completion, participants will learn how the random forests model could be used to correlate the NWS weather forecasts to road surface temperatures.
  • Upon completion, participants will understand how the accuracy of common predictive models compares to that of the random forest model.