Skip to main content

Table 2 Description of predictor variables and interaction terms included in models without (NO_MET) and with (MET) meteorological variables

From: Modeling hourly weather-related road traffic variations for different vehicle types in Germany

Variable name Variable type Variable description
Potential predictor variables for NO_MET
hour Categorical (24) Hour of the day
dow Categorical (7) Day of the week (public holidays are treated as Sundays)
mon Categorical (12) Month of the year
holiday Categorical (2) School holiday in the federal state of the traffic station
trend Continuous Linear trend in time
hour:dow Interaction Different diurnal cycles on different days of the week
break Categorical (n) \(n={2,3,4}\) segments determined by breakpoint detection (if \(n=1\) this term and its interactions are excluded)
break:trend Interaction Different trends in different segments
break:hour Interaction Different diurnal cycle in different segments
Potential predictor variables for MET
temp\(^k\) Continuous Daily maximum temperature at 2 m height at the ERA5 grid cell closest to the traffic station
cloud\(^k\) Continuous Daily mean total cloud cover at the ERA5 grid cell closest to the traffic station
wind\(^k\) Continuous Daily maximum wind gust at 10 m height at the ERA5 grid cell closest to the traffic station
precip\(^{1/k}\) Continuous Average hourly precipitation sum of all RADOLAN grid cells within a radius of 10 km around traffic station
weekend Categorical (3) Distinguish between working day, Saturday, and Sunday (only included in interaction terms below, but not as single variable)
weekend:temp\(^k\) Interaction Different relationships between temperature and traffic
weekend:precip\(^{1/k}\) Interaction Different relationships between precipitation and traffic
weekend:clt\(^k\) Interaction Different relationships between cloud cover and traffic
weekend:wind\(^k\) Interaction Different relationships between wind speed and traffic
  1. For categorical variables the number of categories is shown in brackets. Meteorological variables are introduced to the model selection process with different exponents k \(=\) {1,2,3,4} to allow non-linear functional relationships