In the last few years Scientific Visualisation (ViSc) (as defined in the
landmark report to the US National Science Foundation; McCormick et al.,
1987) has become one of the standard investigative tools of all the physical
and natural sciences (Hall, 1994). Visualisation in the sense used here is not
a set of techniques to communicate the known as is done with traditional
graphics and maps, but a means, a mechanism, for exploring data sets and
structures, to generate ideas and explore alternatives. This has been defined
as 'exploring data and information graphically, as a means of gaining
understanding and insight into the data' (Earnshaw and Wiseman, 1992:
Brodlie et al., 1993), and is often referred to as visual data
Many factors, both practical and theoretical, have driven this change from
traditional static displays or presentations to visualization:
- Developments in sensor technology and automated data capture have provided
data at rates faster than can be easily converted into knowledge. These
streams, or firehoses, of data are being added to the warehouses
full of old data (McCormick et al., 1987). Such developments are
appropriate to much social science data at, for example, the Essex Data Archive
such as the Census and Opinion Polls.
- Some of the most exciting discoveries in science have been associated with
non-linear dynamics where apparently simple equations like the finite
difference form of the logistic Xt+1 = aXt ( 1 -
Xt ) conceal enormously complex, but real-world like behaviour that
can only be appreciated when displayed graphically (Lorenz, 1993).
- As science has progressed to produce ever more complex simulation models,
so it has become necessary to use visualisation as the only practicable way to
assimilate all the model outputs.
- Improvements in computing mean that we now have networks of very fast
workstation computers available whose primary output is to high resolution,
colour screens. In looking at these effects of technology it is easy to lose
sight of what is fast becoming commonplace. An obvious example is resolution
and colour. In 1980 the Census Research Unit at Durham University published an
Atlas of the 1971 UK Census of Population (Census Research Unit, 1980) which
used the then new laser technology to produce, at great difficulty and expense,
maps with individual colour symbolism for each and every kilometre grid square
over Britain. At the time, these were the most detailed population maps at this
scale and resolution ever produced, but using modern hardware similar displays
are relatively easy to create at even higher spatial resolution using a full
gamut of colours (Martin and Bracken, 1991).
This newer technology
means that we can create entirely new forms of display which extend the number
of available graphic variables beyond Bertin's original seven (plan, size,
shape, value, orientation, hue, texture). These new visual variables include
projection (Dorling, 1992, 1994), animation (Tobler, 1970; Moellering, H.,
1976) and sound (Fisher, 1994). A major problem confronting the use of these
new graphic variables is that, unlike conventional maps we know almost nothing
about good design using them (Monmonier, 1991; Krygier, 1996). This leads to a
potential for contrasting styles of use (Table 1).
Comparison of Traditional and Computer Displays
GRAPHICS and MAPS
aim for realism (VR)
to use as much data as possible
aids to understanding
for many viewers
by one person
ViSc is clearly in the tradition of exploratory data analysis in statistics
(Tukey, 1977) in as much as it emphasises the use of graphics in the
development of ideas, not, as in traditional graphics, in their presentation.
Indeed, ViSc often turns the traditional process on its head by developing
ideas graphically and then presenting them using non-graphic means.
McCormick et al. (1987) talk about visualisation as a method of analysis
which 'transforms the symbolic into the geometric, enabling researchers to
observe their simulations and computations' (p3) and 'a tool for both
interpreting image data fed into a computer, and for generating images for
complex multidimensional data sets'.