10/01/94
While spatial features are typically located using planar (2-dimensional) referencing systems, many transportation features are located using a linear (1-dimensional) referencing system (LRS). These features include bridges and other structures, changes in pavement condition, or accident sites. Often, spatial and linear-referenced features must be combined for specific transportation applications. Examples include: (1) finding the minimum clearance height on a section of highway by locating all bridges that cross over the highway; or (2) using dynamic segmentation to highlight sections of a highway with poor pavement conditions to prioritize rehabilitation activities.
Although a LRS cannot entirely replace a planar referencing system for geographic display, it does represent the format in which most transportation infrastructure and incident data are currently reported. In the absence of extremely high accuracy base maps and locational measurements, using planar coordinates alone to match point data to the transportation network will yield anomalous cases where the resulting spatial object does not lie on the network. This cannot happen using a LRS.
It is the position of the FGDC Ground Transportation Subcommittee that a LRS is an essential component of transportation network spatial databases. The FGDC Ground Transportation Subcommittee therefore recommends that:
The remainder of this paper provides more specific recommendations for LRS requirements on network data structures and for the three major transportation networks -- roads, railroads, and navigable waterways.
For the purposes of this paper, the following definitions are used:
In a LRS, features are, by definition, restricted to lie on or along the segments that make up the base network. Their locations are described by a distance along a specified route from a known reference point. If the reference point and the route are explicitly identified in the base network, then the linear-referenced feature can be located by calculating the distance along the segments comprising the specified route. In order to carry out these calculations, each segment in the base network must include the following attributes:
Several popular Geographic Information System (GIS) software packages have implemented procedures for linking linear-referenced data to a spatially referenced transportation network. However, each vendor's software procedure operates somewhat differently, and each requires that key LRS attributes be in specific formats to facilitate the linking. For example, several of the leading transportation GIS software packages explicitly require that a route must begin and end at a node. Consequently, the network database must include nodes at every location where a new route begins and ends.
In order to provide a consistent framework for incorporating LRS within a network data structure, the FGDC Ground Transportation Subcommittee recommends that a standard LRS data structure be included as part of any approved SDTS Transportation Network Profile. Establishment of this standard LRS data structure will require consideration of the specific software requirements of key GIS software and development of a software-neutral data format. However, an initial set of key attributes and rules is proposed below:
Any network segment that is used in a LRS must include the following attribute fields:
There is currently no single national LRS for roads and highways. Any organization responsible for collecting data about a particular road network creates its own unique LRS. It is not unusual for a local transit agency to maintain a LRS for its bus routes, while the municipal government uses a different LRS for local streets, and the State DOT uses yet another LRS for state roads. Thus, a single road segment could be represented by three or more different LRS, and may require all of these to link attributes from each agency's databases. Standardization of agency- specific LRS occurs only if linear-referenced data must be shared across agencies and if the costs associated with converting to a common LRS are less than the costs associated with collecting and maintaining redundant data linked to each agency's own LRS.
The issues associated with coordinating diverse agency-specific LRS at the national level are currently being addressed by the Federal Highway Administration (FHWA) as part of its Highway Performance Monitoring System (HPMS). 2 All rural arterial, urban principal arterial and National Highway System (NHS) segments included in the HPMS database must be referenced using each State's own LRS. Furthermore, each State is required to submit to FHWA a digital highway network database and/or a set of maps showing the location of all HPMS inventory routes and their corresponding LRS attributes. FHWA will use the information from these submissions to develop a LRS for the FHWA National Highway Planning Network (NHPN) Version 2.0 database. The NHPN and its LRS attributes will be updated annually.
The HPMS submissions will establish a common LRS for each rural arterial, urban principal arterial and NHS segment in each State. The FHWA will accept different systems from different States. For example, some States use a county-based system, where milepost values are reset to zero each time a route crosses into a new county. This system requires a county identifier in addition to the route identifier to uniquely locate a point on the network. Other States use a statewide control section atlas where every inventory route has its own unique identifier, regardless of what county it is located in. Still other States may uniquely identify each segment in their network, creating what amounts to a unique route for each segment. FHWA requires that each road segment be associated with one and only one inventory route, and that each road segment be uniquely identified by the combination of State/county code, inventory route, inventory subroute, and beginning and ending mileposts (or kilometerposts).
The FHWA requires that States submit a LRS for those roads and streets functionally classified as Interstate (rural and urban), other Freeways and Expressways (urban), other Principal Arterials (rural and urban), and Minor Arterials (rural), plus any remaining roads designated as part of the NHS. FHWA does not require States to submit a LRS for other roads, such as collectors and local streets. Indeed, depending on the municipality, milepost-based LRS may not even exist for many of these roads. On the other hand, most local roads do have address ranges.
Address ranges are, in fact, simply another type of LRS. With address ranges, the route number is replaced by a road name, and the beginning and ending distance measurements are replaced by beginning and ending addresses on each road segment. An initial set of address ranges were included in the TIGER/Line databases, but these were limited to urban areas as defined in the 1980 GBF/DIME files. The address range coverage was expanded in TIGER/Line 1992 from 50 million to 80 million potential addresses, based on input from Census takers during the 1990 decennial Census.
The existence of duplicate street names within a county can cause locational ambiguities using address ranges. These ambiguities can be reduced significantly by including Zip Codes as a further qualifier to the State/county FIPS Codes. However, neither address ranges nor Zip Codes were designed specifically as map-based locational referencing systems. Consequently, even with Zip Codes, certain address locations, such as vanity addresses, centralized mail boxes, or rural postal routes still cannot be matched to a position on the road network.
The combination of a milepost-based LRS for higher level roads, and address ranges for lower level roads functionally classified as urban minor arterials, rural/urban collectors, and local streets should provide reasonably complete LRS coverage for a national road network. It may therefore be necessary to include attributes for both address ranges and the HPMS milepost-based LRS on any national road network database.
A related long-term effort by FHWA's Office of Research is investigating the development of a national (if not universal) Locational Referencing System, to be used for vehicle location, in- vehicle navigation, and routing under various Intelligent Vehicle Highway System (IVHS) projects. A national Locational Referencing System is necessary to unambiguously and efficiently communicate the locations of places and moving vehicles across different spatial network databases, using different display software and different communications media.
A preliminary assessment of the existing mix of address ranges and LRS 3 suggests that they will not be adequate for IVHS applications. One proposed solution calls for the development of a comprehensive system of national route identification numbers for principal roads and highways and a master street name list for local non-route numbered roads. Each route or named street would be uniquely and unambiguously identified according to some established hierarchical naming convention (e.g., roads that have both an Interstate and a State or county route number would be identified by their Interstate route number). Every route would have well defined start and end points, including geographic coordinates to some standard level of accuracy and unambiguous descriptions of intersections with other spatial features (e.g., a State boundary, another route, etc.). A national register of route numbers and street names will be much easier to maintain if States and localities adopted consistent route naming conventions for use in their own LRS.
Another long-term effort of the FHWA Office of Environment and Planning, USGS, and Bureau of the Census is to explore the feasibility of a methodology for the development of a unique identifier for each roadway section. The objective would be to create an identification system that is uniform across the country, will require minimum maintenance, and can track changes (e.g, new roads, realignments, or abandonments) over time.
The FGDC Ground Transportation Subcommittee recommends the establishment of a standardized, national LRS for U.S. roads, and supports the development of a unique road identification system that can be used for both current inventory applications and future vehicle location and navigation systems emerging from IVHS.
At this time, it is too early to recommend a specific development approach, but the FGDC and FHWA should closely monitor the various locational referencing initiatives that are now in progress, and should endeavor to coordinate these initiatives into a single, consensus strategy.
The Subcommittee recognizes that the development of a unique identification system for all U.S. roads is a long-term goal. In the interim, additional steps can, and should be taken to address current LRS requirements associated with road inventory data and to facilitate the transition to a national LRS. These steps are presented below.
a. Establish a well-defined hierarchy of route numbers (e.g., Interstate, U.S. Primary, State Route, County Route, etc.) such that a single route number can be associated with each network segment.
b. Eliminate the practice of using equations to tie reconstructed routes back historically to original reference points. All routes would be newly referenced after major realignment.
c. Change from a county-based to a State-based road inventory system. More specifically, State sign routes which cross county lines would be sequentially referenced from their beginning to end instead of being reset at county boundaries.
NOTE:Items 5.b and 5.c can be accomplished by electronically transforming existing State LRSs to statewide referencing systems without equations.
Unlike the road network, there is a single industry-wide approach for linear referencing of railroads. The approach is based on track ownership, and a particular line is uniquely identified by a combination of owner railroad and its internal operating designations. Although the definitions of operating segments (regions, divisions, subdivisions, districts, or other operating entity) may differ from railroad to railroad, in general each operating segment is delineated by a system of mileposts which orient the operating personnel as to the beginning and end of that geographical subsection -- starting at a specific base point and proceeding linearly along the rail line to the end of that segment. Virtually all main line and branch line tracks are mileposted in this manner; track segments within rail yards, sidings, or private spurs are not typically mileposted.
Most railroads were originally mileposted based on the actual distance along the track. However, over time, mergers, line abandonments, new construction or other factors such as changes in ownership and track configuration have altered the uniformity of the milepost numbering system. Mileposts are published in engineering department track charts and in employee timetables issued by each railroad, which contain detailed operating rules and instructions as well as other information such as signals, speeds, etc. relating to specific mileposts or sections.
Milepost markers are usually placed along the line every mile at the whole mile points. While intermediate points such as stations, turnouts, and other operating features are not usually marked in the field, they generally are referenced in the track charts and timetables in tenths (or even hundredths) of a mile between the markers. Junction points are often repeated for each connecting line and may be assigned different milepost numbers for each segment for which they are a part.
The major problem associated with appending this LRS to a national railroad network database is that most of the data will have to be input manually, using time tables obtained from individual railroads. This will be a time consuming and costly effort, and will require the cooperation of the railroads themselves.
The FGDC Ground Transportation Subcommittee recommends the following approach to developing a national LRS for U.S. railroads.
A national standard LRS has been established for navigable inland and intracoastal waterways by the Corps of Engineers. The standard is based on the waterway name and a milepost measured from a known reference point (e.g., a confluence, the mouth, or the termination of a navigable channel). The LRS is maintained by the Corps of Engineers and is being incorporated into the National Waterway Network.
The FGDC Ground Transportation Subcommittee recommends that current efforts by the Corps of Engineers to establish river mile referencing of all commercially navigable rivers and the Intracoastal waterway system be completed, and adopted as the national LRS for inland and intracoastal waterways.
1 It may be necessary to provide more than one attribute field in order to unambiguously identify a route. Typically, a second attribute field is used to differentiate among two or more subroutes which share the same primary route number, but do not satisfy the topological requirements of a route spatial object. For example, a divided highway may be represented by two parallel lines in a map database, and for inventory purposes, both lines would have the same route number. Alternatively, a State sign route may reset its reference points at county boundaries, and would require a subroute field to avoid ambiguities caused by having the same route milepost value (e.g., milepost 2.5) in two different counties.
2 Federal Highway Administration, Highway Performance Monitoring System Field Manual, August 1993, Chapter 5.
3 Viggen Corp., "Location Referencing Systems: Analysis of Current Methods Applied to IVHS User Services", Report E2 (draft), March 8, 1994.