REPRESENTING UNCERTAINTY IN VISUALISATIONS OF FUTURE LANDSCAPES
K. Appleton*, A. Lovett, T. Dockerty, G. Sünnenberg
Zuckerman Institute for Connective Environmental Research, School of Environmental Sciences,
University of East Anglia, Norwich, NR4 7TK, UK
+44 1603 591095 k.appleton(@uea.ac.uk http://www.uea.ac.uk/zicer/ssevrel
KEY WORDS: GIS, visualization, understanding, landscape, decision support, photo-realism, agriculture
ABSTRACT:
The use of computer-based visualisations to depict landscape futures is increasingly common in environmental decision-making,
particularly where information must be communicated to a wide and non-expert audience. Continuing technological development
makes it possible to use real-world data as a base for such visualisations, with modifications made to the landscape data as
appropriate depending on the scenario to be illustrated. The effects of factors such as the level of realism in the images produced and
the capability for viewer interaction with the scene are beginning to be addressed by researchers, but relatively little attention has
been given to the representation of uncertainty. The complexity of environmental systems means that uncertainty always exists to
some degree when illustrating possible or proposed future states for any landscape, and it is essential that this be communicated to
stakeholders, decision-makers and other audiences. Cartographic techniques exist and have been used to illustrate error and other
quality issues relating to spatial data, but there has been no large-scale application to landscape visualisation. Options for
communicating levels of uncertainty are being explored using a study area in Norfolk, in the eastern part of the UK. This work first
considers sources of uncertainty; it then moves on to examine various illustration methods in terms of technical feasibility and data
needs as well as audience reaction and perception. In helping visualisation producers to illustrate uncertainty in the clearest and most
accessible way, the results of the study have the potential to enhance environmental decision-making processes.
KURZFASSUNG:
Computerbasierende Visualisierungen von zukünftigen Landschaftszuständen werden immer häufiger eingesetzt, um
Umweltentscheidungen zu treffen, insbesondere, wenn Informationen mit einem breiten Laienpublikum ausgetauscht werden
müssen. Die fortschreitende technologische Entwicklung macht es móglich Real-World-Daten für solche Visualisierungen
einzusetzen, wobei entsprechende Modifizierungen der Landschaftsdaten in Abhängigkeit von den darzustellenden Szenarien
vorgenommen werden. Wissenschaftler beginnen damit Auswirkungen von Faktoren wie Realitàtsnáhe der erzeugten Bilddaten und
die Fühigkeit der Interaktion der Betrachter mit diesen Bildern zu untersuchen. Wenig Aufmerksamkeit wurde dabei bisher der
Darstellung von Unsicherheiten gewidmet. Aufgrund der Komplexitát von Umweltsystemen bestehen bei der Darstellung von
móglichen oder vorgeschlagenen Landschaftszustánden immer zu einem gewissen Grad Unsicherheiten und es ist unabdingbar, dass
diese Interessenvertretern, Entscheidungsträgern und anderen Zielgruppen verdeutlicht werden. Kartografische Verfahren existieren
und werden genutzt, um Fehler oder andere Qualitätsgesichtspunkte in Bezug auf ráumliche Daten darzustellen, aber es erfolgte
bisher kein grofer Einsatz in Bereich der Landschaftsvisualisierungen. Móglichkeiten zur Verdeutlichung der bestehenden
Unsicherheiten werden bezogen auf einen Untersuchungsraum in Norfolk, im óstlichen Teil von England, erforscht. Diese Arbeit
berücksichtigt zunächst die Quellen von Unsicherheiten; darauf aufbauend werden verschiedene Illustrationsmethoden in
Abhängigkeit von technischer Umsetzbarkeit und erforderlichen Daten sowie die entsprechenden Reaktionen und Wahrnehmungen
der Zielgruppe untersucht. Durch den Einsatz der Ergebnisse der Studie kónnen Unsicherheiten in einer klaren und höchst
zugänglichen Form dargestellt werden und so zu einer Verbesserung von Umweltentscheidungsprozessen führen.
I. INTRODUCTION away from merely representing the visible landscape and
towards adding value to that representation by including more
The environmental decision-making process incrcasingly subtle and complex factors such as uncertainty.
involves the visualisation of spatial data in one form or another.
In particular, landscape visualisation is recognised as a useful
tool for communicating large amounts of environmental 2. UNCERTAINTY IN ENVIRONMENTAL
information, especially when working with the non-expert DECISION-MAKING
groups now beginning to participate in decision-making (Craig,
2002). Recent example applications of such work in the UK Environmental decision-making processes naturally deal with
include the presentation of coastal management options (Jude et predictions or proposals relating to future environmental states.
al, 2002) and the impacts of climate change on rural landscapes Some level of ambiguity is inevitable in any future scenario due
(Lovett et al, 2003). to the combination of various sources of uncertainty, including
natural environmental variability, data uncertainty, knowledge
However, as visualisation and display methods become more uncertainty, and model uncertainty (Willows & Connell, 2003:
sophisticated, users have control over an increasing number of 49). Our understanding of natural systems is imperfect; to this is
aspects of the output, such as lighting, the colour and texture of added the impossibility of completely predicting human actions
surfaces and objects, and the opportunity to add elements such and reactions on either an individual or collective scale, and
as labels and overlays. Attention is therefore beginning to turn issues of measurement including accuracy and precision.
Internatio
ae
Multiplic:
landscape
great nuir
Although
a end-use
is certain
be elimir
scientific
when de
decisions
overwhel
complica
visualisat
environm
done to a
Previous
in press)
considere
decision-
time is n
aspects «
specific.
scene W
comment
only inc
renewabl
which ap
by the au
explored
One pote
represent
define.
uncertair
within tl
outcome
conditior
the over:
for pref
landscap
different
what is
suitabilit
section.
It may e
enough :
informat
assess qi
the type
intended
It is wc
where v
detailed
the beli
Howeve
those th
question
that det
concern:
message
visualis:
audienc:
Someho
