Extraction of basic roadway information for non-state roads in Florida.

PDFTech ReportBDV29-977-07Data collectionAccident reportsGeographic information systemsUnpaved roadsTraffic lanesMediansRoad shouldersSpeed limitsFloridaFlorida. Dept. of Transportation. Research CenterWu, WanyangGan, AlbertAlluri, PriyankaFlorida International University. Lehman Center for Transportation ResearchUS Transportation CollectionThe Florida Department of Transportation (FDOT) has continued to maintain a linear-referenced \ue2\u20ac\u153All-Roads\ue2\u20ac? mapthat includes both state and non-state local roads. The state portion of the map could be populated with select datafrom FDOT\ue2\u20ac\u2122s Roadway Characteristics Inventory (RCI). However, the RCI data are available for only a smallportion of the local roads in the All-Roads map, leaving a majority of the local roads in the map without the samedata. Given the large number of local roads in the map, it is clearly not cost feasible to collect the data in the field.Methods that make use of existing data as alternatives to field data collection are thus needed.One potential source of existing data is police crash reports. For every reported crash in Florida, the law enforcementofficers record information on more than 300 variables to describe the site and time of the crash, the geometricconditions, the traffic control, and drivers\ue2\u20ac\u2122 and pedestrian\ue2\u20ac\u2122s characteristics. Accordingly, this project aims to developmethods to extract roadway data recorded in crash reports as a means to both acquiring and continually updating theAll-Roads map for local roads in Florida. To the extent possible, the project attempted to extract data for thefollowing four variables that are included in Florida\ue2\u20ac\u2122s crash reports: number of through lanes, posted speed limit,shoulder type, and median type.The data extraction process to acquire data for the four variables in this project includes three steps. The first stepinvolves the extraction of data from crash records for as many road segments as possible. The second step covers thecase in which a road segment does not have any crashes. In this step, the values are derived from their adjacentsegments based on the assumption that roadway features are likely to be continuous. Finally, the third step focuseson the remaining segments for which data could not be extracted or derived in the first two steps. In this step, themissing data are manually collected using a web-based data collection application that is designed specially tofacilitate the process of observing and recording information from satellite images in Google Maps