Geography Reference
In-Depth Information
Linear
Referencing
OVERVIEW
Managers of artifacts with long linear structures (e.g., highways, pipelines, railroads) are generally
unconcerned with the precise two-dimensional geographic coordinates of the entity they deal with. A
locomotive is pretty well locked onto the railroad's one-dimensional linear structure—unlike, say, an
airplane, which can operate in three dimensions. As a result, many of the coordinate systems that have
grown up around linear structures are different from those we have been talking about up to now. Terms
like road miles, river miles, and rail miles are in common use in those industries. The average citizen who
uses the Interstate Highway System will probably be familiar with the small green signs with numbers on
them that indicate the number of miles from some origin—frequently a state border.
We are in the habit of representing linear structures with lines drawn between junction points (e.g.,
nodes). (Of course, these structures are portrayed on a two-dimensional field, but the smallest part of
each such line is one-dimensional [a vector], so we call them one-dimensional, or linear.) The difficulty
with simply representing such a structure—let's take a highway, for example—from intersection to
intersection is related to attributes. The requirement, so far in your studies, is that any GIS feature
(whether point, line, or polygon), must be homogeneous in all its attributes. For example, all of a given
cadastral polygon is owned by one entity, the taxes are paid on all of it or none of it, and so on. If different
attributes apply to two different parts of the feature, then you need two features (e.g., if different people
own different parts of a piece of land, then you need more polygons).
Imagine that you have a stretch of highway that is 2 miles long between its intersections A and B with
other roadways. Among the attributes you want to store for this length of road are speed limit, pavement
type, pavement quality, political jurisdiction, and number of lanes. The difficulty is that the speed limit
changes four times and the road goes from four lanes down to two and then back to four. It crosses a
county boundary. Part of it is blacktop and part concrete. Also, repairs have taken place on different
segments of the road at different times. To represent this “traditionally” in ArcGIS, we would have to
have a plethora of features. Every time an attribute changed (e.g., the speed limit changed from 55 to 45),
a new feature would have to be declared. So, the single-line feature that represented the 2 miles between
intersections might have to turn into dozens of short features. The complications created then—for
example, how would you find the distance from A to B?—are considerable.
The invention that turned out to make attribute representation tractable for such linear features is called
“linear referencing.” The fundamental idea is that you can have several sets of attribution information
accompany a single linear feature. What makes this possible is the concept described earlier in this
section: The thing that fixes the point at which a change takes place is a number—a distance—that is
related to the origin of the feature, in the style of, say, road miles.
Search WWH ::
Custom Search