|
Author |
Thread:
Issue: Is GML GIS-centric? |
|
Paul Scarponcini
|
Issue: Is GML GIS-centric?
Posted:
Saturday, January 29, 2005 4:04 PM (EST)
|
|
(I have separated out what are emerging as key issues regarding GML into separate threads to encourage focused discussion on these topics.)
Actually no. It is feature-centric. As it derives from the ISO 191xx family of standards, GML breaks with the traditional GIS approach. That approach says that you draw a line or polygon and then attach feature attributes to it as if the line were the feature itself. In 191xx (and therefore GML), you create a feature (like a road) first, and then attach attributes to it (like number of lanes) and, by the way, one possible attribute that you can attach to the feature is its geometry/location. And, you can have multiple geometries associated with the same feature. So you can have a 1:24,000 (GIS) scale geometry as a (straight) linestring to support GIS type applications. And you can, at the same time, have a 1”=50’ (engineering) scale geometric representation complete with true circular curvature. With GML, you can have both geometries for the same road feature. That way you do not have to have a separate GIS and CAD feature for the same road, each one having to know, independent of the other, the number of lanes, pavement type, etc.
GML3 already has geometries defined to support geospatial representations. These go beyond the traditional GIS straight line geometries supported by GML2. Agreed, GML3 does not go far enough to support all needed engineering geometries. But the standards groups are working on this. OGC, for example, has established a CAD/GIS Working Group to focus on these issues. The Linear Referencing System standard adopted by ISO as one of the 191xx standards is general enough to support engineering stationing as a Linear Referencing Method.
The long term vision is for TransXML to support all phases of the transportation facility life cycle. It therefore needs to simultaneously support multiple types and instances of (CAD, GIS, and other) geometry. A feature-centric approach will allow this to happen.
|
|
|
|