International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B4. Istanbul 2004
realtime with a joystick or mouse. Animations may be derived
from either method, by compiling a sequence of stills or by
recording an interactive navigation.
Portability is another important issue for many visualisation
methods. Simple still images can easily be reproduced and
distributed in a variety of ways, but cannot be modified, for
example to remove the uncertainty information. Interactive
methods, and some animation techniques, sometimes involve
the use of particular applications or plug-ins (e.g. Audition for
OpenFlight realtime models, or Flash for multimedia content on
the internet), which may allow customisation and thereby the
addition of some degree of user control. Such methods,
however, limit the distribution and use of visualisations to those
with the appropriate computer software, hardware, and possibly
network connections, which may go against the desire to open
up decision-making to a wider audience.
5. POTENTIAL METHODS FOR REPRESENTING
UNCERTAINTY IN LANDSCAPE VISUALISATION
Many of the techniques in Section 3 do not immediately seem to
be directly applicable to the issue of uncertainty in landscape
visualisation, as they can conflict with the needs of the
visualisation itself. That is, changes in colour, sharpness or
position in an effort to show uncertainty may be interpreted as a
change in real life rather than an attempt to illustrate non-visual
information (Goodchild, 1991). For this reason, altering the
appearance of the landscape or objects within it (e.g. changes in
colour/size, or multiple instances of items) is potentially
unhelpful in any decision-making process, since responses are
usually requested based at least partially on the visual
appearance of a landscape.
The suggestion that complications may arise from the use of
highly detailed visualisations whereby the viewer equates visual
detail with certainty of prediction was brought up by planning
professionals in previous work (Appleton & Lovett, in press).
There, respondents put forward the idea that different elements
of the image could be visualised at different levels of detail
depending on their certainty. On the other hand, the results of a
different research project suggest that viewers find high levels
of detail helpful in imagining a visualised landscape (Appleton
& Lovett, 2003), implying that it could be unhelpful to alter the
level of detail in an effort to express additional information. It
was also evident that there may be a “lowest common
denominator” effect whereby an image is only as realistic
overall as its least detailed element, perhaps throwing doubt on
the advisability of varying the detail level to convey uncertainty.
This is an interesting disagreement which needs to be
investigated in more depth, but it does not mean that visual
modifications should be discounted as useful techniques.
Methods whereby the user can control the visibility of
uncertainty information (perhaps using a sliding scale of
transparency or some other transition), or display it in a parallel
window, are possible alternatives to the permanent alteration of
visualisations, with animation techniques offering the option of
similar displays where interactivity is not practical (e.g. with a
large audience).
The techniques which would seem to be most suitable for
representing uncertainty may be summarised as follows:
e Deliberately creating low-detail visualisations (or elements
within them)
e Giving a range of possibilities, if appropriate
428
e Altering colour, either by adding false colour or
desaturating (greying out)
e Blurring
e Written information, either on labels within the image or
accompanying text
e Sound
Some of the above techniques may increase the time taken to
draw or display a landscape image, by adding an extra
dimension of information, which is relatively unimportant for
single images, but has more impact on multi-frame animations
and could well affect the perceived smoothness of interactive
methods by reducing the frame rate. However, interactivity also
brings an important extra dimension to most of the above
options. Not only does it allow visual methods to be applied
without permanently distorting the image, but it could be used
to control the level of verbal information displayed, perhaps
allowing expert and non-expert viewers to be catered for with
the same visualisation. A range of visualisations could be
presented in response to the viewer's choice of scenario or
changing of one or more given parameters. Sound-based
techniques could respond to the pointing/navigation device
used, either representing uncertainty on its own, or being used
as an alert device to draw attention to the presence of other
uncertainty information. It therefore seems likely that some sort
of user interface would be helpful, to allow the viewer to choose
the information to be displayed, showing multiple sources at a
time either in separate windows or overlaid on one another.
It is important to consider non-interactive methods as well as
interactive ones for the reasons mentioned in Section 4. If the
decision-making process is to be widened to include a greater
cross-section of society, the visual information intended to
facilitate their participation must be made available in an
accessible way. This does not preclude the use of sophisticated
technologies but does mean that alternatives should be
considered and investigated.
The audience may also affect the choice of methods in other
ways. In reaching out to more non-experts, environmental
decision-making must still be useful to expert participants, and
their needs and perceptions must be considered. One interesting
question is whether environmental professionals might have a
greater inherent understanding of uncertainty when presented
with an environmental future, and therefore there is less need to
draw their attention to its existence and more need to explain.its
sources and magnitude. Such information may be overwhelming
to a non-expert viewer, but could be presented on an on-demand
basis as suggested above.
6. WORK SO FAR
Preliminary work is being carried out to assess the technical
feasibility of the various options for presenting uncertainty
information. This phase is ongoing.
The underlying image (Fig 1) was chosen from a previous
project within the research group (Lovett et al, 2003) and shows
a rural, agricultural English landscape in the south-east of the
country. It is based on Geographical Information Systems (GIS)
data, specifically UK Ordnance Survey elevation and land use
data, with the land use information being greatly enhanced
through fieldwork and other sources to reflect the varieties of
crops grown in the area. The GIS information was used with
Visual Nature Studio from 3DNature to create a landscape
Internati
model, 1
view fro
Figure 1
The ima
time unc
efforts t
changes
availabil
changes
hedges),
Uncertai
sources,
the type
and soc
farmers.
defined,
explorat
are give
fairly st
turbines
process,
possible
Figure .
colours
Figures
as applic
Figure
distingu
levels i:
providec
degrees
more sc
probably
