Geography Reference
In-Depth Information
high precision. h e possibility of errors being introduced at any stage of data process-
ing or analysis should be taken into account and the generation of high-quality graphic
outputs should not disguise the fact that any output is only equal in quality to the lowest
quality input. h is topic is discussed further below. All spatial data should be associ-
ated with metadata—that is, data about the data sources which indicate key informa-
tion on how and when the data were collected, as well as detailing any conversions or
modii cations undertaken. If detailed metadata are kept, these act as an invaluable
resource for future users of the data in that they provide a means of assessing factors
that may have an ef ect on applications which make use of these data.
2.9.1
Uncertainty in spatial data analysis
h e use of alternative procedures, or selecting dif erent options in the application of
one method, will ot en lead to dif erent results. It is essential in any use of spatial data
to take into account such potential problems. h e modelling of propagation of errors
from one processing stage to another, and of the degree of uncertainty in representa-
tions of features and their attributes, are signii cant areas of research. h e quality
of outputs from a spatial analysis is a function of (1) the quality of the data, (2) the
quality of the model, and (3) interactions between the data and the model (Burrough
and McDonnell, 1998). When data from dif erent sources are combined, the ef ects
of many dif erent kinds of uncertainties (e.g. measurement errors, scale dif erences,
temporal dif erences, and other factors) may also combine. Spatial data quality and
uncertainties in spatial data are among the subjects of the topic chapter by Brown and
Heuvelink (2008).
Visualizing spatial data
2.10
Visualization is the i rst stage of any spatial analysis. Simple viewing of a spatial data set
may seem conceptually straightforward. However, there may be a multitude of deci-
sions that have to be taken into account when visualizing data which may impact
strongly on interpretations of those data and on the ways in which any analysis might
proceed. Simple point patterns (i.e. point event locations with no attributes attached)
are ot en presented using points to represent each event location. In the case of objects
with categorical attributes (e.g. urban area or rural area), depiction may be based on
the selection of dif erent colours or shades to represent each category. In such cases,
selection of colours or shades that enable dif erentiation between categories is impor-
tant; a map with two classes depicted using similar shades or colours may be very
dii cult to use. Continuous variables (e.g. measurements of an airborne pollutant) are
usually represented using a range of colours or shades (e.g. white for small values,
shades of grey for intermediate values, and black for large values). Where the data model
is a grid, a continuous grey scale or colour scale may be used, as shown in Figure 2.1.
A common means of displaying areal data (e.g. population densities in administra-
tive zones) or values attached to other discrete objects such as points, in particular,
Search WWH ::
Custom Search