01677210

Monograph

http://www.trb.org/Main/Blurbs/177916.aspx

The CurveFinder tool automatically identifies horizontal curves and extracts their geometric characteristics solely from geographic information system (GIS) roadway maps. Applying CurveFinder to datasets from multiple states has revealed the need to improve the accuracy of CurveFinder for low-quality datasets that contain errors or do not represent the actual roadway alignment accurately. States also expressed the need to obtain curve information compatible with Federal Highway Administration’s Model Inventory of Roadway Elements (MIRE). The goal of this IDEA project is to: incorporate the ability in CurveFinder to extract curve-related MIRE elements; improve the accuracy of CurveFinder for low-quality GIS data sources; and develop an online portal called CurvePortal where agencies can upload their GIS roadway maps for obtaining horizontal curve location and geometric information. MIRE has a specific subcategory for horizontal curve data, and includes eight elements. The updated CurveFinder extracts all MIRE elements from GIS roadway maps except for Element 111 Superelevation, which can only be obtained from field surveys. For each curve extracted by CurveFinder, information on curve type, curve degree, curve radius, curve length, whether there is a presence of transition section, the deflection angle if the curve is a horizontal angle point, and the curve direction is included as attributes in the GIS shapefile. CurveFinder preparation, calibration, and run for production of final results has been streamlined into a five step process. To improve the accuracy of CurveFinder for low-quality GIS datasets, two smoothing methods are used to preprocess the data before applying the CurveFinder algorithm. Calibration is performed within desktop GIS software by an iterative process by which the original road segments as well as smoothed road features by means of best fit and spline methods with a variety of offset and angle thresholds are passed to the curve finder algorithm. The preprocessing and calibration steps resulted in better identification of the start and end points of curves as demonstrated by the increased coverage of identified curves as compared to ground truth curves. CurvePortal was developed to serve as an intuitive web interface by which users can upload spatial road features as well as optionally define an attribute field on which distinct road segments participating in a given roadway can be joined. Additionally, the interface offers the option to upload ground truth features that can be used for calibration purposes and identification of optimal algorithm parameters prior to the final curve identification step.

This Highway IDEA project was conducted by University of Wisconsin-Madison.

NCHRP IDEA Project 185

English

37p

2018-3

Digital/other

Figures; References; Tables

Accuracy; Calibration; Curves (Geometry); Data quality; Geographic information systems; Geometric design; Highway curves; Software; User interfaces (Computer science)

Innovations Deserving Exploratory Analysis (IDEA) (Program); Model Inventory of Roadway Elements

Data and Information Technology; Design; Highways; Safety and Human Factors

TRIS, TRB, ATRI

Jul 31 2018 9:44AM