A Framework for Map Agnostic Conflation: Challenges, Application, and Suitability in the Context of Mobility-Based Performance Measures

Map conflation has been an unfastidious yet important step in conflating both spatial and numerical roadway attributes. Oftentimes, a dataset that is pertinent to a roadway Network A is contrasted with another Network B. The process of spatially matching this pair of A and B poses certain logical and computational challenges. This paper illustrates these challenges and presents an algorithm that conflates proprietary third-party datasets (Network B) with the South Carolina Department of Transportation’s linear referencing system-based roadway network (Network A). This paper investigates the distortion of data caused by the map conflation and discusses its applicability in the context of several mobility-based performance measures. The study tested conflation pairs between the 73,899 Network A and 74,706 Network B segments, spanning 30,787 directional miles. Paired segments result in 99.23% spatial accuracy at a precision of 0.001 miles. The calculated statewide delay for the calendar year 2022 was 75,966.4 and 75,967.11 vehicle-hours for Networks A and B, respectively, which is a close match. In addition, the conflated maximum level of travel time reliability between the Network A and B segments showed high resemblance, with only about 0.5% mismatches. Eliminating semantic attribute considerations from the conflation process presented in this study created a method that is efficient, highly accurate, agnostic of input networks, and is applicable to any network with a clearly defined relationship between the segment topology and direction of travel.