Abstract
The Highway Capacity Manual (HCM) provides weather Speed Adjustment Factors (SAFs) to estimate expected speed reductions under adverse conditions such as rain and snow. For example, at a free-flow speed of 65 mph, the heavy-rain SAF = 0.93, representing a 7% reduction. However, these default adjustments are based on limited calibration datasets and may not reflect site-specific precipitation conditions, roadway environments, or regional driver behavior. To address this limitation, this thesis applies a data-driven, site-specific analysis integrating 15-minute INRIX probe speeds, Continuous Count Station volumes, and Road Weather Information System (RWIS) data across four Virginia freeway sites (two urban, two rural).
Precipitation was categorized by intensity (light, medium, heavy rain) and aligned with traffic density states (free-flow, transition, congested) and ambient lighting (day/night). Median speeds were computed across conditions, and reductions relative to no-precipitation baselines were compared with HCM SAFs. Statistical testing using Analysis of Variance (ANOVA) and Tukey’s Honestly Significant Difference (HSD) identified significant differences in mean speeds across precipitation categories and traffic phases.
To further assess distributional impacts, speed-density boxplots and interquartile ranges (IQR) were introduced. These measures captured the variability in driver responses across conditions, complementing the median-based analysis. Results indicated that precipitation effects on variability were site and time-dependent. At night, IQRs were consistently wider under precipitation than under dry conditions, reflecting greater divergence in driver behavior under reduced visibility. Snow conditions were consistently associated with greater variability than freezing precipitation, though limited sample sizes in some bins suggest these findings should be interpreted with caution.
Observed speed reductions routinely diverged from HCM guidance. Light rain reduced free-flow speeds by 1 to 5 mph (~1-7%), despite the HCM providing no adjustment. Heavy rain reduced median free-flow speed by up to 11 mph (~18%) at one urban site, over twice the HCM adjustment. A winter weather case study further showed that pavement surface condition, not considered in the HCM, played a critical role.
Overall, findings indicate that weather-related speed changes vary significantly by site, precipitation intensity, lighting, and surface state. Incorporating both central tendency and variability measures provides a more comprehensive understanding of weather impacts on traffic, and results highlight the need to revisit HCM guidance using broader categories and more representative datasets.
