Analysis of dynamics of road weather information system data

Abstract

Road and Weather Information System consists of a network of roadside Environmental Sensor Stations (EES) collecting meteorological data. Equipped with a variety of sensors, these stations gather data including ambient temperature, subsurface temperature, precipitation level and type, brightness, and more. Consequently, RWIS systems have been critical for increasing road safety over the years by providing valuable weather information helpful in anticipating and preparing for adverse weather conditions and reducing traffic collisions. The advancement of Artificial Intelligence (AI) and Machine Learning (ML) methodologies enhances our ability to leverage these systems through the analysis of temporal datasets, unraveling dynamic behavioral patterns over time, and constructing more accurate and dependable predictive models. The research study reported in this thesis focuses on understanding the dynamics of weather data to enhance road safety and the utilization of RWIS, especially during winter exhibiting hazardous road surface conditions. Utilizing ambient and subsurface temperature data from RWIS stations along Interstate 35, managed by the Oklahoma Department of Transportation (ODOT), the study methodically investigates the climatological patterns across different times of day and seasons. Parametric regression modeling is conducted to characterize the behavioral patterns of weather data for diurnal and nocturnal periods through seasonal progression. A notable aspect of this research is the demonstration of how incorporating data from subsurface temperature probes enhances the performance of machine learning models for the classification of road surface conditions and weather events. The study explored the influence of road clearance time on traffic speed during snowstorms employing statistical data-driven techniques of change in speed pre- and post-clearance. Machine learning classification techniques are employed to automate the detection of hazardous road surface conditions like ice and snow based on weather and speed data.