Natural landscape scenic preference: techniques for evaluation and simulation.
Abstract
The aesthetic beauty of a landscape is a very subjective issue: every person has their own
opinions and their own idea of what beauty is. However, all people have a common
evolutionary history, and, according to the Biophilia hypothesis, a genetic predisposition to
liking certain types of landscapes. It is possible that this common inheritance allows us to
attempt to model scenic preference for natural landscapes.
The ideal type of model for such predictions is the psychophysical preference model, integrating
psychological responses to landscapes with objective measurements of quantitative and
qualitative landscape variables. Such models commonly predict two thirds of the variance in the
predications of the general public for natural landscapes.
In order to create such a model three sets of data were required: landscape photographs
(surrogates of the actual landscape), landscape preference data and landscape component variable
measurements. The Internet was used to run a questionnaire survey; a novel, yet flexible,
environmentally friendly and simple method of data gathering, resulting in one hundred and
eighty responses. A geographic information system was used to digitise ninety landscape
photographs and measure their landforms (based on elevation) in terms of areas and perimeters,
their colours and proxies for their complexity and coherence.
Landscape preference models were created by running multiple linear regressions using
normalised preference data and the landscape component variables, including mathematical
transformations of these variables. The eight models created predicted over sixty percent of
variance in the responses and had moderate to high correlations with a second set of landscape
preference data. A common base to the models were the variables of complexity, water and
mountain landform, in particular the presence or absence of water and mountains was noted as
being significant in determining landscape scenic preference.
In order to fully establish the utility of these models, they were further tested against: changes in
weather and season; the addition of cultural structures; different photographers; alternate film
types; different focal lengths; and composition. Results showed that weather and season were
not significant in determining landscape preference; cultural structures increased preferences for
landscapes; and photographs taken by different people did not produce consistent results from
the predictive models. It was also found that film type was not significant and that changes in
focal length altered preferences for landscapes.