Modeling free-flow speed variations under adverse road-weather conditions: Case of cold region highways

Modelling speed behaviour in rural highways: Safety analysis of driving under adverse road-weather conditions

“…Thus, understanding the speeding behaviour under adverse road-weather conditions is essential for successful implementation of a weather-responsive VSL program. Desired speed, which is the speed chosen at driver’s discretion at specific traffic and roadway conditions directly reflects driver behaviour and is extensively used in studying driver behaviour [ 9 , 10 ]. Desired speed is manifested through the free-flow speed of vehicles [ 11 ] and is referred to as “speed” hereafter in.…”

“…In other words, desired speed explains the speed at which the drivers are comfortable under the prevalent driving conditions. Adverse road-weather conditions such as icy road surface conditions and poor visibility are proven to affect the desired speed significantly [ 9 , 12 ]. Accordingly, speed behavior, which is defined as the choice between several possible speeds [ 13 ], is widely chosen as a measure to study the impacts of adverse road-weather conditions on driver behaviour.…”

“…Past literature explored several approaches to investigate the impacts of adverse road-weather conditions on speed behaviour including statistical modelling and machine learning techniques [ 9 , 14 , 15 ]. For instance, [ 9 ] developed a group of statistical models to enumerate the speed variations resulted by adverse road-weather conditions, where the set of multiple linear regression models with 20-minute aggregate speeds as the dependent variable (Group II models) was concluded as the best performing model. Hoogendoorn [ 16 ] developed a new unified approach to estimate the free speed distributions based on the composite headway distribution model and the method of censored observations.…”

“…While some studies confirmed significant changes in speed behaviour under adverse road-weather conditions [ 12 , 14 , 17 ], other studies contradict these observations, mainly in terms of the intensity of speed dispersion [ 9 , 18 ]. Further, the performance of conventional approaches such as regression modelling to study speed behaviour has been criticized [ 9 ]. The underlying reasons for discrepancies in the literature are mainly threefold.…”

“…The underlying reasons for discrepancies in the literature are mainly threefold. The main prospective reason discussed is limited sampling to represent specific road-weather conditions [ 9 , 19 ]. On one hand, adverse road-weather conditions are intrinsically scarce, restricting the number of samples, which can be collected under specific inclement road-weather conditions.…”