1.2.

The real world and representations of it 52

1.2.4 Spatial databases
Spatial databases are a specific type of database. They store representations of
geographic phenomena in the real world to be used in a GIS. They are special
in the sense that they use other techniques than tables to store these represen-
tations. This is because it is not easy to represent geographic phenomena using
tables. We will not discuss these more appropriate techniques in this book.
A spatial database is not the same thing as a GIS, though they have a num-
ber of common characteristics. A spatial database focuses on the functions we
listed above for databases in general: concurrency, storage, integrity, and query-
ing, especially, but not only, spatial data. A GIS, on the other hand, focuses
on operating on spatial data with what we might call a ‘deeper understanding’
of geographic space. It knows about spatial reference systems, and functional-
ity like distance and area computations, spatial interpolations, digital elevation
models et cetera. Obviously, a GIS must also store its data, and for this it pro-
vided relatively rudimentary facilities. More and more, we see GIS applications
that use the GIS for the spatial analysis, and a separate spatial database for the
data storage.
The assumption for the design of a spatial database schema is that the rel-
evant spatial phenomena exist in a two- or three-dimensional Euclidean space.
Euclidean space can be informally defined as a model of space in which loca-
tions are represented as coordinates—(x, y) in 2D; (x, y, z) in 3D—and notions
like distance and direction have been defined, with the usual formulas. In 2D, we
also talk about the Euclidean plane.
The phenomena that we want to store representations for in a spatial data-
base may have point, line, area or image characteristics. Different storage tech-
niques exist for each of them. An important choice in the design of a spatial
database application is whether some geographic phenomenon is better repre-

first previous next last back exit zoom contents index about
1.2. The real world and representations of it 53
sented as a point, as a line or as an area. Currently, the support for image data
exists but is not impressive. Some GIS applications may even be more demand-
ing and require point representations in certain cases, and area representation in
other cases. Cities on a map may have to be represented as points or as areas,
depending on the scale of the map.
To support this, the database must store representations of geographic phe-
nomena (spatial features) in a scaleless and seamless manner. Scaleless means
that all coordinates are world coordinates given in units that are normally used
to reference features in the real world (using a spatial reference system). From
such values, calculations can be easily performed and any (useful) scale can be
chosen for visualization. A seamless database does not show map sheet bound-
aries or other partitions of the geographic space other than imposed by the spa-
tial features themselves. This may seem a trivial remark, but early GIS appli-
cations had map production as their prime purpose, and considered map sheet
boundaries as important spatial features.
All geographic phenomena have various relationships among each other and
possess spatial (geometric), thematic and temporal attributes (they exist in space
and time). Phenomena are classified into thematic data layers depending on
the purpose of the database. This is usually described by a qualification of the
database as, for example, a cadastral, topographic, land use, or soil database. A
spatial database not only serves to store the data and manipulate it, as it should
also allow the users to carry out simple forms of spatial analysis.
Spatial analysis involves questions about the data that relate topological and
other relationships. Such questions may involve neighbourhood, distance, direc-
tion, incidence, disjointness and a few more characteristics that may exist among
geographic phenomena. In the El Niño case, for example, we may want to find
out where is epicentre of warm water or where is the steepest gradient in water

first previous next last back exit zoom contents index about
1.2. The real world and representations of it 54
temperature.

first previous next last back exit zoom contents index about
1.2. The real world and representations of it 55
GIS and databases
A database, like a GIS, is a software package capable of storing and manipulating
data. This begs the question when to use which, or possibly when to use both.
Historically, these systems have different strengths, and the distinction remains
until this day.
Databases are good at storing large quantities of data, they can deal with
multiple users at the same time, they support data integrity and system crash
recovery, and they have a high-level, easy to use data manipulation language.
GISs are not very good at any of this.
GIS, however, is tailored to operate on spatial data, and allows all sorts of
analysis that are inherently geographic in nature. This is probably GIS’s main
stronghold: combining in various ways the representations of geographic phe-
nomena. GIS packages, moreover, nowadays have wonderful, highly flexible
tools for map production, of the paper and the digital type. GIS have an em-
bedded ‘understanding’ of geographic space. Databases mostly lack this type of
understanding.
The two, however, are growing towards each other. All good GIS pack-
ages allow to store the base data in a database, and to extract it from there
when needed for GIS operation. This can be achieved with some simple settings
and/or program statements inside the GIS. Databases, likewise, have moved
towards GIS and many of them nowadays allow to store spatial data also in dif-
ferent ways. Previously, they in principle were capable of storing such data, but
the techniques were fairly inefficient.
In summary, one might conclude that small research projects can probably
be carried out without the use of a real database. GIS have rudimentary data-
base facilities on board; the user should be aware they are really rudimentary.
Mid-sized projects use a database/GIS tandem for data storage and manipula-

first previous next last back exit zoom contents index about
1.2. The real world and representations of it 56
tion. Larger projects, long-term projects and institutional projects organize their
spatial data processing around a spatial database, not around a GIS. They use
the GIS mostly for spatial analysis and output presentation. We will look more
closely at these data processing systems in Chapter 3.

first previous next last back exit zoom contents index about