Geography Reference
In-Depth Information
than five businesses are located at a single address. However, there are design problems with
this kind of visualization technique. First, it is only effective where the stacked symbols are
drawn with equal heights. (This is an example of the kind of conflict that frequently occurs
between visual variables and other visual effects in 3D data visualizations.) Second, where
the individual stack symbols are sorted by business attributes, as in this example, then the
kinds of businesses visible above the reference plane will not necessarily be representative of
the range of businesses within the entire stack. Despite these and other limitations, however,
reference planes afford numerous insights into value relationships across 3D scenes, such as
which symbols lie above or below a particular threshold value, or which land is lower/higher
than other land.
Divided symbol stacks
In 3D data visualizations which display stacks of locationally coincident point symbols,
there are usually no visual cues to indicate the vertical extent of each symbol. This prevents
the analyst from making comparative visual estimates of the number of items of partic-
ular colours, forms or sizes in each stack within the view. Where stacks contain several
symbols using the same set of visual variables (e.g. size, type, colour), it becomes even
more difficult to make comparative estimates of the number of symbols in stacks across
the entire scene. One way of providing an additional visual cue is to insert small gaps
between the symbols in each stack, automatically adjusted by the software to remain pro-
portional to the standard symbol height. An example is provided in Figure 10.3, which
uses stacks of coloured cylindrical solid symbols to indicate the genders of all individu-
als living in residential properties in part of the East End of London, using data from the
1881 census of population. Not only do the inter-symbol gaps enable the analyst to make
rough visual estimates of the number of individuals resident at each location, but the soft-
ware also assists the making of broader comparative judgements by automatically sorting
the symbols into common value groups (males vs females in this example) within each
symbol stack.
Non-perspective projections
Although non-perspective projections have been widely used in architecture and engineer-
ing, and the isometric grid became popular with computer games in the 1990s (e.g. SimCity,
Civilisation, A-Train and Theme Park), the perspective projection is almost universally used
for 3D data visualization. Wyeld (2005a, b) suggests that this is part of a visual tradition that
has dominated art and other visual media since the Renaissance, and that a deliberate effort
is needed to wean people away from this expected form of representation. For some data
interpretation tasks which involve the comparison of objects across a scene, a case can be
made for setting aside the perspective view, and using instead a non-perspective projection.
In Figure 10.4, for example, the view shown in Figure 10.3 is redisplayed using an orthogonal
projection, which makes it easier to make cross-scene comparisons of the sizes of symbol
stacks. (With this software, the viewer is able rapidly to toggle between the two projected
views with a single keystroke to assist their interpretation of the data.)
Search WWH ::
Custom Search