Geoscience Reference
In-Depth Information
A
display
is composed of a number of primitive graphical objects that are combined together,
rather like the base types in a conventional programming language. Example objects might be a
point, a region or an arrow; these are sometimes termed
symbols
or
glyphs
. Symbols host visual
variables - such as colour, shape, transparency and position - to which data can be visually
encoded. Many visualisation environments allow multiple displays to be rendered together as a
scene
and support coordination of activities across these different displays, essentially providing
different views or perspectives onto the underlying data. For example, in GeoViz, it is common
to support a map display that is linked to more traditional statistical displays such as a scatterplot
and parallel coordinate plot and to provide interactivity between these displays so that user actions
in one display are mirrored onto the others (so-called linking and brushing or coordinated visual
displays: North and Schneiderman, 1999). In Figure 5.2, these connections between displays are
shown by
leader
lines
as well as by highlighting the region currently selected in the same manner
in the linked displays.
The visualisation environment typically allows a user to assign data to some or all of the visual
variables that a symbol supports, to determine its appearance. A simple example from cartography
involves using the size of a population to determine the area of a circle used to represent a city. More
complex examples are given later. In traditional cartography and in GeoViz based around maps,
the underlying geography determines position and shape* for the geospatial data, leaving visual
variables such as colour, transparency and hatching to encode any additional attributes (such as
land cover or soil type). By contrast, statistical graphs and charts often allow any arbitrary variable
to be encoded using position, such as in a scatterplot where the
x-
and
y
-axes are not restricted to
spatial coordinates (if they are, then the scatterplot essentially becomes an unprojected map). Any
visual variables that a user can choose to assign data to are known as
free variables
; the ones that
are reserved for some other system-defined role are known as
bound variables
. From GIS packages
to SciViz systems, we see a variety of charting and graphing tools that offer different arrangements
of free variables that can be used to encode data. For example, some systems might support symbols
that can move (be animated), or geometry that morphs, or the ability to change the viewpoint of the
observer to the scene, while others stick to a more rigid map-based paradigm where
x
and
y
must be
used for geographical position.
†
Some excellent examples of the range of possibilities for map-based
GeoViz are provided by Demaj and Field (2013).
A brief summary of some of the key terms introduced earlier is provided in Table 5.1. The termi-
nology used in visualisation can get rather confusing, not least because different communities use
different terms to mean the same thing.
5.2.3 g
eneSiS
of
g
eoViSualiSation
S
ySteMS
Visualisation is an evolving field of study, and it has its beginnings in several seemingly discon-
nected research communities. Perhaps the earliest focus on visualisation was seen in the engi-
neering and science communities during the 1970s and 1980s where rendering systems came to
the fore to help researchers understand complex engineering and scientific structures, such as
buildings, organic molecules and the ozone hole (Treinish, 1993). When visualisation started to
become more popular in the 1980s-1990s, several large, commercial visualisation systems arose,
such as
ENVI/IDL
,
NAG IRIS Explorer
and
IBM Data Explorer
. These provided comprehensive
development environments that greatly eased the task of creating a visualisation from scratch,
often by the use of visual programming techniques based on the flow of data. These systems were
*
Cartograms literally stretch this point by allowing some other variable such as population to warp the shape and position
of geographical regions.
†
Of course, most GIS software allows the map geometry to be morphed by the map projection and registration process,
but not by values that originate in the attribute data, such as the amount of rainfall received. The geometry variables are
bound
for all practical visualisation purposes.
