In previous literature, much attention has been given to the occurrence frequency of those accidents by the
police and incident frequency is far beyond full study. Incident data obtained by CCTV (Closed Circuit
Television) monitoring system on urban expressways in Shanghai City are applied in this study to guarantee the
completeness of indents. These incident data are quite different from those traditional incident (including
accident) reports by the police or incident management personnel: the CCTV-monitoring incident dataset
contains large quantities of minor crashes, which are always neglected in the field of safety analysis, and
numerous short-duration vehicle disabilities, which does not need the help of rescue facilities before moving and
thus left unrecorded. In this study, the average duration of an incident is only 9 min and 50% of incidents last
records, but must be considered in terms of congestion alleviation. It should be noted that distinction is made in
this paper between accidents and crash incidents: accidents are relatively severe crashes traditionally recorded by
the police, while crash incidents include those minor ones usually beyond the concern of traffic safety.