Freeway management systems are becoming increasingly important, serving as a core of many intelligent transportation system (ITS) applications. Advancement in various areas of traffic monitoring system technology has provided planners and engineers with rich sets of traffic data, both macroscopically and microscopically, that can be used to achieve even better freeway management. In this paper, we propose a new methodology specifically designed to process traffic data that vary in time according to their state. This methodology consists of two algorithms, i.e. a scale space smoothing algorithm and a time segmentation algorithm, which can be used either separately or sequentially to process the data. The scale space algorithm smoothes time series traffic data to eliminate the embedded random effect while preserving the sharp transition state of the data. The time series segmentation algorithm divides the time series data into different "semantically" meaningful segments belonging to different states. This paper also discusses potential applications of this methodology to a variety of ITS implementations. Several experiments have been conducted on selected applications and their results verify the effectiveness of the proposed methodology.