Geography Reference
In-Depth Information
This goes on for about another 630 pages, which the publisher has, understandably, declined to include.
Our point is, it takes a lot of bits to represent even a small photo.
Continuous Nature of the Referencing Basis
In most databases, a particular, unique key points to a unique thing. For some examples, a given auto
license number identifies a particular car; a name or Social Security number tags an individual person;
a house number and street constitute a pointer to a residence. Spatial phenomena do not enjoy any
such autonomy, however—they are a mixture of discrete and continuous. That is, there is no natural
and completely satisfying one-for-one correspondence between spatial locators and the related data.
A virtually infinite amount of data is potentially available about even the smallest area of the real
world; we can store only a small part. Thus, by choosing a particular technique for organizing the
continuous into the discrete, we are screening out or “throwing away” an infinite amount of potential
information. Clearly, it takes some sophistication and forethought to select a technique to represent the
coordinates that apply to the continuous real world and have a database that will be useful in solving
problems.
Continuous Nature of Data
In addition to the continuum of two- and three-dimensional space just mentioned (i.e., the fact that
our basic referencing scheme potentially has infinitely many points in it), there are also problems with
the continuous nature of the data themselves. Soil type is probably a good example. Just as no two
snowflakes are alike, no two soils are exactly alike. Soils must be categorized into groups and a judgment
made about which group a particular soil belongs to. In naturally continuous variables, such as elevation,
the parallel issue of precision comes in: Do we measure (vertically) to the nearest meter? To the nearest
millimeter?
Abstraction of Entities
The simplest reference that can be made in a spatial database is to a point, but no material entity is
ever just a point. Many of the things we deal with are either linear features, areas, or volumes, so the
referencing scheme becomes more complicated. Where is a house? Well, it's a lot of places when you get
right down to it. Do you define it by its corners in plan view? Do you select a single point, a “centroid,”
and define the house to exist at that point? Do you simply say it exists in town “X,” with many other
houses? There are many fundamental variations in the way the “real world” is and can be referenced.
These varying methods can be incompatible, precluding any easy transfer of data or techniques for
manipulating data.
Multitude of Existing Spatial Coordinate Systems
Many spatial coordinate systems exist. Most of those used for planning and resource management
rely on the use of flat projections of curved surfaces. Many of the datasets that will be used to build a
multivariable spatial database will come from data recorded with distorted and dissimilar methods of
representation. Matters of units, datum, spheroid, and projection must be addressed. A single state may
use many coordinate systems in its various agencies. Examples are latitude and longitude (both NAD27
datum and NAD83 datum), UTM (both NAD27 datum and NAD83 datum), a state plane coordinate
system (one or more zones), road miles, river miles, a special coordinate system for particular features
(e.g., oil and gas wells), and so on.
Search WWH ::
Custom Search