Geography Reference
In-Depth Information
Data and data models
2.2
A model is simply a means of representing 'reality' and spatial data models provide
abstractions of spatially referenced features in the real world. h e focus of this topic is
on analysis of spatial data rather than the ways in which spatial data are structured.
However, a very brief introduction to data models was considered useful as knowledge
of analysis of spatial data requires at least a basic understanding of data structures.
Most topics on GIS stress the division of data models into the well-known raster and
vector representations. h e key properties of the two data models, which will be useful
in making sense of the rest of this topic, are outlined here. h e data model available
determines the choice of method for spatial analysis as do the characteristics of the
particular data set.
Representations of real-world features are ot en divided into (1) entities and (2)
i elds (Burrough and McDonnell, 1998). Entities are conceptually distinct objects like
point locations, roads, or administrative boundaries. Fields convey the idea of values
of some property at all locations. For example, elevation can be measured or estimated
at all places and elevation does not usually have distinct edges, in contrast with, for
example, buildings. Objects that are well described as distinct entities are sensibly
represented using the vector data model. Properties that tend to vary quite smoothly
from place to place (i.e. they are spatially continuous and their values do not tend to
change abruptly from place to place) are frequently represented using the raster data
model. h ere are notable exceptions and these include isolines and contours, which
are vector-based representations of continuous phenomena such as temperature or
elevation (there are, of course, exceptions—a clif edge represents an abrupt change in
elevation and so temperature is perhaps a more conceptually straightforward exam-
ple). h e raster and vector data models are briel y dei ned in turn. A more in-depth
account of the way in which information is stored using the two data models is given
by Wise (2002).
2.2.1
Raster data
While it is assumed that readers are familiar with raster grids, some key issues are
addressed here. Raster grids are conceptually simple structures, comprising square
cells with numeric values or classes attached to each cell. A simple example of a raster
grid is given in Figure 2.1; in this case the value represents elevations in metres. Where
the cells contain categorical or integer (i.e. whole number) values the number of
instances of each class may be stored in a table. In cases where values with decimal
places are used, all information is conventionally stored in the raster itself. h ere are
huge amounts of data available in raster grid format—remotely sensed imagery (see
Section 2.8.3) comprises a major component of such data sources. h e spatial resolu-
tion of a raster refers to the area in the real world covered by a cell. For example, a grid
with a spatial resolution of 5 m covers an area of 5 by 5 (= 25) square metres. Remotely
sensed images with very i ne spatial resolutions (e.g. 1 m) have been generated for
many parts of the world, although ease of access (cost, etc.) varies geographically.
Search WWH ::
Custom Search