_____________________________________________________________________________________________
All Public Road Geospatial Representation Study ARNOLD Reference Manual
Technical Appendix Page 90 DOT Contract #GS-35F-0001P
September 2014
Routing and navigation applications
typically require a node at every
legitimate intersection in order to model
items such as vehicle turn restrictions
and directionality, and for geocoding.
rules are stored as errors, but exceptions can be marked accordingly. Network topology can be encoded
and its properties measured using graph theory, which is incorporated at a fundamental level in most
GIS products. Network analysis based on graph theory is commonly supported in GIS data models. LRS
are an extension to GIS, adding dynamic segmentation and allowing State DOTs to measure the location
of features and events on the road network, thus handling linear measurements along a road segment.
In this GIS-T case, the node geometry of a road network might be dissolved along logically continuous
routes. Conversely, in a non-GIS-T case, the road network might be subdivided into a multitude of
segments for graphic purposes to support cartographic representation, or geocoding. These approaches,
while valid for their own purposes, can adversely impact the topology needed for traffic flow and
routing applications. In other words, a dataset that is optimized for linear measurements, or
cartographic representation, or geocoding, is less than optimal for traffic flow modeling and routing.
Topology, particularly for connectivity, forms the basis for network analysis and routing. A routable
network is key to many GIS applications, such as fleet
management, drive-time analysis, and most notably
emergency response. Routing allows first responders to
determine the quickest route from their location to the
dispatch destination. Typically, routing applications
require a node at every legitimate intersection where a
turn can be made, which may differ from the data
structure of the LRS. For example, to achieve processing efficiencies for non-routing applications, State
DOTs have traditionally dissolved segmented networks into non-segmented networks, to reduce the
number of nodes and to define numbered routes with the minimum number of essential segments.
However, for routing purposes, nodes and related intersections are very important for modeling vehicle
restrictions (turns, heights, weights, speeds, direction, stop signs, traffic signals, congestion,
construction, detours, etc.); and for geocoding purposes for matching addresses with a location along a
street.
Routing applications require a network topology in which the relationships of each node and link with
other nodes and links are modeled in terms of constraints and possibilities. Fundamentally, two tables
are required to geospatially represent a network data model: 1) a node table, including unique IDs and X
and Y coordinates for each node; and 2) a link table, also with unique IDs, as well as the “from” node
(the origin or starting point), and the “to” node (the destination or ending point). When these tables are
relationally linked, the topological structure of the network can be built, and network analysis based on
graph theory can be applied.83F
Rodrigue, Comtois, and Slack. The Geography of Transportation Systems. Taylor & Francis Group: New York,
2009, pp. 76-77.