Role of Solar Radiation and Roadway Features on Crash Severity along Arterial Roads in Kentucky

Abstract: Weather factors, such as rain, fog, wind, snow, and solar radiation significantly affect crashes. This study seeks to evaluate crash scenarios that result in fatal (K), suspected serious injury (A), severe injury (KA), and non-severe injury (BCO) crash outcomes on Kentucky’s arterial roads under three solar radiation conditions: low solar radiation (0-100 watts/m2, i.e., early morning/evening/night), medium solar radiation (101-500 watts/m2, i.e., daytime), and high solar radiation (above 500 watts/m2, i.e., strong midday sun). Five-year crashes (January 1, 2019 through December 31, 2023) along arterial roads in Kentucky were extracted from the Kentucky Transportation Cabinet (KYTC) database. The crash database offers specifics about crash locations, encompassing latitude, longitude, date, time, and details about drivers, vehicles, and environmental context. The crash dataset was further merged with the roadway information and integrated roadway features, such as functional classification, median width, speed limit, and annual average daily traffic (AADT). Furthermore, additional effort was made to merge real-time weather information from the High-Resolution Rapid Refresh “HRRR” model within the National Oceanic & Atmospheric Administration “NOAA” database (during 1-hour of the crash) with the integrated crash-roadway data. The real-time weather variables have included: air temperature (in 0Fahrenheit “0F”), precipitation (in inches), humidity (in percentage), solar radiation (in watts/m2), visibility (in miles), and wind speed (in miles per hour).

This study used an association rule-based data mining technique to analyze different crash types (e.g., K, A, KA, and BCO) along Kentucky’s arterial roads at varying solar radiation levels (i.e., low, medium, and high). Additionally, the study developed and presented the top 10 rules for various model combinations using the “apriori” rule-based algorithm. The dataset included 5,260 crashes on Kentucky's arterial roads. The “apriori” association rules mining technique revealed that non-severe (BCO) crashes under low and medium solar radiation levels predominantly occurred on straight and level roads with narrow left shoulders ( < 2 feet). On the other hand, non-severe (BCO) crashes under high solar radiation levels were associated with increased traffic volumes (>1280 vehicles/day) and lower speed limits on asphalt roads. For severe (KA) crashes, regardless of the solar radiation level, asphalt roads and angle collisions were prevalent, with high solar radiation further introducing distractions on roads with median widths of moderate size (17-32 feet). Based on the association rules mining analysis, mitigation measures were suggested, including tackling driver distractions and refining road design in areas with poor visibility and higher air humidity (> 65%).