Stephen J. Beskpalko
Sandia National Laboratories
PO Box 5800 MS 0977
Albuquerque NM 87185-0977
sjbespa@sandia.gov
Voice (505) 845-8847
Fax (505) 844-2057

Max M. Wyman, Ph.D.
Terra Genesis, Earth Modeling & Systems Restructuring Group 1268 East McNair Drive
Tempe AZ 85283
max.wyman@terragenesis.com
Voice (602) 345-0447
Fax (602) 345-8345

John C. Sutton, Ph.D.
GIS/Trans, Ltd.
2801 Business Center Drive, Suite 145
Irvine, CA 92715
jsutton@ca.gistrans.com
Voice (714) 222-0710
Fax (714) 222-0801

ACKNOWLEDGMENT AND DISCLAIMER

This work was partially supported by the United States Department of Energy and was performed at Sandia National Laboratories under Contract DE-AC04-94AL85000.
 
The opinions expressed in this document are those of the authors, and do not necessarily reflect the opinions or positions of their employers, other individuals, or other organizations. Mention of commercial products does not constitute endorsement by any person or organization.

POSITION

GIS for Transportation (GIS-T) is an interesting subset of the overall suite of technologies referred to as GIS. The requirements of GIS-T are intermediate between the purely cartographic applications and the more demanding topographic requirements of civil engineers. None-the-less, the GIS-T is an important component of the GIS community, and one where change is needed (Vander Veer, 1997).

Proposed herein is a new direction in GIS for Transportation (GIS-T) infrastructure modeling which is potentially free of the pathologies associated with current LRS data models. A model is proposed which builds towards a 3-D GIS based on location references provided by GPS and its World Geodetic Spheroid 1984 (WGS84).

 Current GIS-T is built up from three layers:

• A fundamental arc-node (point, line, and polygon) layer as derived from traditional cartographic systems (vector or raster). This model is frequently linked to tabular details through a relational database manager.
• A 1-D offset measurement technique known as a linear reference system (LRS).
• Dynamic segmentation as an enabling tool for assigning multiple attribute sets over a single linear event.

HISTORICAL PERSPECTIVE

Many of the current constraints found in GIS stem from decisions that made sense at the time they were made, but are no longer valid. The 2-D cartographic map was an acceptable representation when overpasses were rare and various modes of transportation were largely disjoint. During these earlier days, LRS was vital because few alternatives could record absolute field locations easily. A field accuracy of 0.5 km was acceptable because other than the distance measuring instrument (DMI) there was no alternative.

Dynamic segmentation was also the only alternative for representing more than one event along a linear feature. Because classical GIS-T techniques required the use of multiple local reference frames, data sharing between agencies was difficult or impossible. Problems multiplied as systems grew in complexity and temporary fixes were added to overcome the multiple datum requirement. Our current work leads us to conclude that even common national datum would not suffice to solve the problem.

Multiple datum LRS and color coding of complex infrastructure objects often result in misleading or incorrect computation, identified by Sutton (1996) as network pathologies. The incorrigible nature of these pathologies suggests the need for a radical change, rather than evolutionary changes to either traditional LRS data models or national datum (see Fletcher 1995, 1996, Vonderohe, 1995, 1996, Dueker and Butler, 1997).
 
The fundamental thesis of our position is that these convoluted historical constraints are no longer valid. There are clear indications that the layered LRS/multiple datum architecture is incapable of representing contemporary transportation features. Because GPS service is now ubiquitous, it is possible to build a fully three-dimensional and topologically correct model for transportation infrastructure.
 
The GPS provides the common origin needed to make data interchange fast, easy, and error-free, just as a geometrically correct 3-D GIS-T model would eliminate the pathological errors bound within the limits of LRS. Mutually incompatible datum and complex dynamic segmentation coding techniques would be replaced by a common language, and the gap between precision drafting methodologies and connectivity-bound topologies like GIS could be closed forever. The potential of 3-D GIS-T data storage should be viewed as profound and ready to meet the needs of local, state and federal agencies forced to accomplish more with less.

PROPOSED THREE DIMENSIONAL GIS MODEL

We propose to outline the basic elements of a new model based on a singular worldwide reference. Further, the model we propose must be lane-based for transportation, and likely something other than object-oriented (at least in terms of the current state of object technology). The implementation we propose would be based on a new form of database, which is constructed from loosely coupled schemas, as opposed to the traditional monolithic database architecture. This work on loosely coupled schemas funded by DARPA and is currently under way at the University of Colorado.

RESEARCH TOPICS FOR THE FUTURE

There is both short-term and long-term work that must be undertaken to transform the GIS from the link-node based technology of current systems to the accurate fundamental technology needed for the future.

1. The urban canyon problem must be addressed. There are promising mathematical techniques that need to be developed to reduce the likelihood of inadequate or reflected signals in some parts of the country.
2. A formal lane-based GIS model must be developed to insure interoperability between other GIS technologies and GIS-T.
3. Fundamental distance and reference algorithms must be developed and tested to demonstrate the appropriate accuracy is achieved with the new GPS technology.
4. Appropriate foundation work must be done to move the field away from the practice of incorporating ever more complicated patches on the link-node-datum models. The maximizing possibility of a single unified worldwide and highly accurate referencing system should be the goal of future spatial data research.
5. It must also be demonstrated that the three-dimensional lane-based and topologically correct model does indeed eliminate the possibility of pathological cases, as we suspect.

1. Brown, A. Extended differential GPS. Navigation 36(3): 265-285, 1989.
2. Dueker and Butler. GIS-T Enterprise Data Model with Suggested Implementation Choices. The Center for Urban Studies, Portland State University, 1996.
3. Fletcher, et al, Geographic Information Systems - Transportation ISTEA Management Systems Server-Net Prototype Pooled Funds Study. FHWA, 1995.
4. Kaplan, E., et al. Understanding GPS, Principles and Applications. Artech House Publishers, 1996.
5. Fletcher, et al, The Case for a Unified Linear Reference System. Alliance For Transportation Research, 1996.
6. Sutton, J. Network Pathologies Phase 1 Report. Sandia National Laboratories, Project AH-2266, November 1995.
7. Sutton, J. Network Pathologies Phase 2 Report. Sandia National Laboratories, Project AH-2266, March 1996.
8. Vander Veer and Bespalko. GIS: Reaching the Third Dimension. Society of Women's Engineers, 1997 National Convention, 1997. Also available as Sandia National Laboratories Technical Report SAND97-1616a.
9. Vonerohe, A. Results of a Workshop on A Generic Model for Linear Referencing Systems, University of Wisconsin - Madison, 1995.
10. Vonderohe, A. A Methodology for Design of a Linear Referencing System for Surface Transportation. Sandia National Laboratories, Project AT-4567, 1996.