Lane-to-lane urban traffic behaviors under different time scales

This paper examined the traffic time series data measured by flow and density. Traffic volumes and occupancies were collected from a loop detector station on a four-lane highway at an interval of 30 seconds. Two analysis time periods were selected: non-peak hour period 12 am to 4 am, and peak-hour period 4 am to 8 am. The traffic data were subsequently aggregated with different time scales; 1-min and 2-min for further analysis. The nonlinear dynamics of traffic time series data among each lane were investigated by using phase space reconstruction and Liapunov exponent methods. The results have shown that during the non-peak hours, traffic behaved stochastic on all lanes at all the three time intervals; however, for the peak-hour period, there is some discrepancy on flow and density results among different time intervals. The statistical analysis on 30-sec interval suggested traffic chaotic behaviors occurring on all four lanes except for being stochastic on lane 4 (next to exits) which was evident by flow time series analysis. For 1-min and 2-min intervals, traffic has exhibited some chaos on four lanes from both flow and density results.