Improved Identification and Calculation of Horizontal Curves with Geographic Information System Road Layers
Total Page:16
File Type:pdf, Size:1020Kb
Improved Identification and Calculation of Horizontal Curves with Geographic Information System Road Layers Hao Xu and Dali Wei Horizontal curve data are collected or calculated by transportation the length of each particular curve class (A through F) of a seg- agencies for different purposes because there is always a demand to ment. For example, the total length of Level B curves on an HPMS generate new curve data either to improve the data quality or to extend record segment is reported; tangents and different levels of curves the data coverage. There are different approaches and technologies for are included. The reported data could not give detailed information curve data collection. The selection of methods takes into consideration on curve location, length, or radius. The curve data in HPMS also accuracy, cost, and available sources. There is, consequently, a wide cover only paved principal arterial and rural minor arterial sample interest in extracting curve information from available road network panel sections; it is optional for an HPMS database to include other data of the geographic information system (GIS). When the cost of sections beyond the limits of the sample panel. The HPMS curve GIS data extraction is much lower than the cost of other approaches, data cannot meet new requirements in traffic engineering and high- it is significant to improve the accuracy of GIS curve identification way safety. For example, accurate radius information is a mandatory and calculation. This paper analyzes errors related to existing GIS curve input for state systemic safety improvements and the highway safety cal culation methods and introduces a new method to identify and to cal software of SafetyAnalyst. Transportation agencies have been col- culate horizontal curves with improved accuracy. The new method uses lecting or calculating horizontal curve data either to improve the data regression analysis of road vertex direction–location profiles to calcu quality or extend the data coverage. late curve radii. A freeway segment on Interstate 80 in Nevada, includ There are different approaches and technologies for curve data col- ing the eastbound and westbound directions, was selected to evaluate lection. Research activities were performed to extract horizontal curve the accuracy of the new method. The evaluation compared the curve data from satellite imagery (2), a GPS survey (3, 4), laser scanning (5), calculation results and the accurate curve information from project and GIS (6). The satellite imagery approach loads satellite imagery contract plans. The evaluation results proved the effect of the new pro into MicroStation Inroads or AutoCAD software, then the operator cedure on improving the accuracy of horizontal curve identification manually draws curves along highway centerlines, and finally the and estimation. The new method can be implemented as a GIS tool software calculates the radius of the drawn-out curves. Research was to scan GIS road networks automatically and create horizontal curve also done to develop an automatic process for curve extraction with data layers. This new method can also be used to generate curve data satellite images (7). However, the automatic approach is not widely from GPS survey data. used because of the lack of commercial tools and the uncertainty of the accuracy with differing image quality. Laser scanning can provide highly accurate survey data, but its cost is much higher Horizontal curve data are important in the design, operation, and than that of other approaches. When laser scanning is applied for performance evaluation of highway facilities, and curve data are especially significant for highway safety improvement. Curve-related data collection on a large road network, data management can also crashes, such as head-on crashes and run-off-road crashes, have be very challenging. The selection of curve data generation methods been identified as one of the emphasis areas that evolved from the normally takes into account the accuracy, cost, and existing data AASHTO Strategic Highway Safety Plan (1). Highway curve infor- sources owned by transportation agencies. While road GPS survey mation is an important component of the highway performance moni- data exist in only some transportation agencies, GIS data are avail- toring systems (HPMS) data set and has been collected and maintained able in almost all agencies. There is no data collection cost for using in each state. Curves in HPMS data sets are classified into levels of GIS to calculate curve data. GIS data are maintained by transportation A through F to describe the sharpness of a curve. Level A means that agencies and improved along with the data use, so most GIS road a curve is smooth with a large radius, while Level F means sharp networks are of good quality with relatively high accuracy. There is curves with a short radius. The reported HPMS curve data reflect wide interest in extracting curve information from GIS data. When the related cost is low, it is significant to improve the accuracy of GIS curve identification and calculation. H. Xu, Department of Civil and Environmental Engineering, College of Engineering, University of Nevada, Reno, 1664 North Virginia Street, MS 258, Reno, NV 89557. The common procedure for curve extraction from the GIS road D. Wei, Partners for Advanced Transportation Technology, University of California, network is to obtain location information of each line vertex, and Berkeley, 1357 South 46th Street, Richmond, CA 94804. Corresponding author: then to calculate road directions, to identify curve segments, and H. Xu, [email protected]. finally to calculate curve radii and lengths. One example of GIS Transportation Research Record: Journal of the Transportation Research Board, curve data extraction tools is Curvature Extension, which is a plug-in No. 2595, Transportation Research Board, Washington, D.C., 2016, pp. 50–58. tool of Esri ArcGIS software developed by the Florida Department DOI: 10.3141/2595-06 of Transportation (DOT) (8). The tool calculates curve radius and 50 Xu and Wei 51 length when a user manually identifies a curve’s start point and end the calculation are shown in Figure 1c. The curve radius can be point. The accuracy of curve identification and estimation depends converted to the degree of curve with Equation 2. highly on the user’s judgment and experience, so it is difficult to con- trol accuracy when the tool is used for curve data extraction for 180 RL= (1) a large road network. Automatic GIS curve extraction procedures π∆ and tools were developed to save data extraction effort and improve accuracy (9). The accuracy of curve data extracted by the automatic where tools is decided by the quality of the GIS data and by the calculation R = curve radius, methods used in the procedure. L = length of curve, and This paper analyzes errors related to existing GIS curve calculation Δ = central angle of curve in degrees (deflection angle), which methods and introduces a new approach to improve the accuracy of is the absolute value of PT direction in degrees minus PC curve identification and calculation. The new method uses regression direction in degrees. analysis of direction–location profiles of road segments to calculate the curve radius. A freeway segment was selected on Interstate 80 180 (I-80) in Nevada, including the eastbound and westbound directions, 100 π 18,000 5,729.57 ∆ to evaluate the new method. The evaluation compared the curve D = = ==100 (2) calculation results and the accurate curve information from project RRπ RL contract plans, which were provided by the Nevada DOT. The evalu- ation results showed obvious improvements in the accuracy of curve where D is the degree of curvature (the angle subtended by a 100-ft radius calculation. Curve extraction from GPS survey and GIS data arc along the horizontal curve). uses similar calculation methods. When data formats of GPS points and GIS line vertices can be converted one to the other, methods and tools developed for one data type can also be used for the other. Error Analysis Therefore this new procedure can also be used to generate curve data When this traditional method is used, the accuracy of the calcu- from GPS survey. The new method does not take spiral curves into lated radius is sensitive to the accuracy of the identified PC and PT account, so it cannot be used to identify and calculate spiral curves. locations. When the error with curve length is assumed to be e, the The rest of this paper is organized as follows. Error analysis of calculated radius with error can be estimated by Equation 3; the existing GIS curve calculation methods is presented next, followed error is (57.3/Δ)e. by a description of the curve identification and calculation procedure, including the improved curve identification method and the regression 180 180 180 180 analysis approach for calculating curve radii. The evaluation results RLerrore= rror = ()Le+= Le+ of the new method are presented next, and then the paper concludes π∆ π∆ π∆ π∆ with a discussion of the research and findings. 180 57.3 =+Re=+Re (3) π∆ ∆ TRADITIONAL GIS CURVE ESTIMATION where AND ERROR ANALYSIS Rerror = curve radius calculated by Equation 1, which includes radius error caused by curve length error; For horizontal curve extraction from GIS, the point of curve (PC) e = curve length error; and and point of tangent (PT) mileposts (MPs), curve length, and curve Lerror = curve length affected by error. radius (or curve degree) need to be estimated with available GIS road layers (10). Curve radius and curve degree are particularly When the PC and PT direction difference is less than 57.3°, the important to highway safety compared with other attributes. The length error will be amplified and will be involved in the calculated general method of curve calculation from GIS data is introduced in radius.