Modelling landscape experience using “experions”

Highlights

• A new metric for quantifying landscape experience.

• The integration of space and time.

• Represents varying viewing distances.

• Novel application of GIS and visibility analysis.

Abstract

This research provides a method for modelling landscape experience that combines space, time, and landscape character. The visual landscape experience is particularly challenging to model because it involves perception that is specific to an individual, views that vary in extent, a period of time that the experience takes place, and a complex interaction of landform and land-cover, including cultural elements. Landscape experience is important to the quality of people's lives and is a substantial component of the tourism industry. To represent such experiences there is a need to develop techniques to model landscape experience. The method demonstrated is based on the landscape experience while walking a track and combines GIS viewshed analysis with a GIS generated landscape character classification. This method takes into account near and distant landscape views and weights the area that each landscape type is viewed by the length of time that it is viewed. The results produce a new metric that combines space and time and has the units of hectare-hours. This paper argues that there is a need for a unit that combines space and time so that landscape experience can be measured, compared and researched. It is proposed that this unit be called “experions” and this paper shows how such a unit can be measured using GIS. This metric measures landscape character experience rather than quality, but can be converted to quality when combined with landscape preference information.

Introduction

This paper outlines a study that provides a new method for representing, conceptualising, and measuring landscape character experience that includes space and time. The method does not attempt to evaluate the quality of landscapes; however, there is a logical step to extend the findings of this study to an assessment of landscape quality using the empirical findings from psychophysical research that shows the relationship between landscape character and landscape quality (Brown and Brabyn, 2012, Daniel and Vining, 1983). There are many studies that assess landscape character but this is generally based on a map providing an aerial top-down perspective (Wheatly & Gilling, 2000), or from a point perspective (Brabyn & Mark, 2011). Most people experience the landscape from the perspective of a traveller who is moving through the landscape along a route. This movement could be by car along a road or walking a path. Such an experience can be measured in time increments and can include a variety of views and landscapes. It is therefore important to research methods that capture the various dimensions of this landscape experience so that landscape is more accurately represented. The example used in this paper to develop this method is a hiker walking a track in New Zealand, but this could have easily been applied to other landscape experiences involving travel.

Landscapes add value to the daily lives of many residents and plays a substantial role in the tourism industry. In New Zealand (NZ) landscape/scenery is the number one reason international tourists visit NZ and in 2013 these visitors spend approximately NZ$10 billion and overall tourism employs approximately 5.7% of the workforce (Statistics NZ). Not surprisingly, landscape change resulting from development is often highly contentious and results in costly and prolonged disputes. To ensure wise land-use decisions are made that carefully consider the landscape values, it is important that landscapes are researched and represented. This is challenging because of the complexity associated with landscape perception and landscapes are experienced from a range of views and personal perspectives. Landscape is a multidimensional experience involving the physical layers of the landscape, and how that experience is interpreted, which can be referred to as the perceptual layers of the landscape (Brown & Brabyn, 2012).

The landscape experience while walking tracks has many different permutations. Consider a simple track that consists of views of two landscapes – flat pasture and forested mountains. This track is unlikely to consist of exactly equal portions of flat pasture and forested mountains. There may be mainly forested mountains views and the occasional view of flat pasture, or there could a whole range of different permutations of these two landscape views. The views of forested mountains may extend for hundreds of hectares and the views of flat pasture may only be a few hectares. If this track takes 60 min to walk, the landscape experience may be 100s of hectares of forested mountains for the first 30 min, and then both flat pasture and forested mountains for 15 min, and then finish with just forested mountains. Each minute of the track may have a different landscape view that includes different landscape character types and different degrees of spatial extent. Generally such experiences are represented with qualitative descriptions – “the track has extensive forested mountains views with the occasional flat pasture”. Or the track may be quantified in terms of the length of track that passes through different landscapes – “45 min of forested mountains and 15 min of flat pasture”, which ignores the view from the track. There are now GIS tools, such as visibility analysis, and GIS data layers, such as landscape character, that enable such a landscape view experience to be quantified, which will be demonstrated in this paper. This paper will introduce the concept of “experion” which is a unit of experience that consists of hectare-hours. One experion is one hectare-hour. In the above example, the track may be described as having 200 experions of forested mountains and 2 experions of flat pasture. In this example the 15 min of flat pasture views would have only consisted of 8 ha (8 ha times 0.25 h equals 2 experions). This provides a quantification of the landscape experience.

The combination of time and space are fundamental to landscape experience. However, the inclusion of time within landscape experience models has not yet gained traction. This is not surprising given the complex nature of landscape and the difficulty of providing spatial models of landscape experience that include perception. The modelling of space and time in relation to representing the movement of people and wildlife is a rapidly growing research frontier in geography, GIS, and GIScience. The widespread use of integrated GPS/GIS technologies and the associated collection of large datasets (over time and space) has provided geography communities opportunities to integrate sophisticated space-time analysis and models. This has led to the study of complex environmental and social systems, ranging from climate change to infectious disease transmission. Most of the focus has been on developing geovisualisation tools such as the space – time prism that pinpoint locations and also show the time (Kwan, 1998, Miller, 1991, Weber & Kwan, 2002). Most studies are associated with the movement of people or wildlife, to identify congestion, or connectivity in the case of wildlife. In relation to landscapes there has virtually been no research. Kwan (1998) used three-dimensional GIS visualization to show activity density of travellers, but the emphasis is on visualisation rather than quantification.

Mapping the environment based on human experience poses interesting challenges for geographical information science because it requires expertise in both GIS analysis techniques and an understanding of the psychology of how people experience landscapes. This paper will first discuss the complex nature of landscape and then present a new method for representing this complexity.

Landscape is a complex and ambiguous concept, but it can be defined simply as the appearance of the land and water from a distance. This definition is consistent with Appleton's (1980) definition; “the environment perceived, especially visually perceived” (p. 14), and Palka (1995) definition; “the assemblage of human and natural phenomena contained within one's field of view out-of-doors” (p. 71). The fact that landscape involves perception means that landscape is not only what is objectively “out there”, but also how people subjectively interpret what is “out there”, which can vary significantly between people.

Language provides an important insight into how the general public conceptualise landscape, and is the basis for a branch of landscape research known as cognitive categorization. Tversky and Hemenway (1983) published an early paper on the content of outdoor scenes. Mark et al. (1999) summarised this research and also conducted experiments for identifying category norms for landform, landscover, water, human settlements and human made infrastructure. It is clear that basic common language terms, such as hills, mountains, lakes and ocean, are the categories that normal people are using for conceptualising many different landscape components.

Information on landscape is a complex and growing body of knowledge developed by a wide range of disciplines and research approaches. Zube, Sell, and Taylor's (1982) classification of landscape research into four components (ecological/expert, psychophysical, phenomenological/experiential, and cognitive) is still an accurate representation of the range of topics covered. In terms of practical solutions for assessing landscape values for land-use planning, the psychophysical approach is regarded as the most theoretically robust method (Daniel and Vining, 1983) given that landscape “beauty is in the eye of the beholder”. Unfortunately, the cheaper “expert approach” is still widely used. However, recent studies using Public Participatory GIS have shown how psychophysical landscape evaluation is now affordable (e.g. Brown & Brabyn, 2012). The research described in this paper is not directly evaluating landscapes. Instead, the purpose is to represent landscape character as an experience, which could lead to more robust evaluation.

Central to all the landscape research approaches is the need to have a classification/ontology of landscape character so that there is a frame of reference for communication. The use of classification is critical in the physical science (soil, biology and geology) but a reluctant method in the social sciences, despite classification being inherent in all language and cognition. Several studies have shown the benefit of a landscape character classification and there are major landscape classification projects based around GIS being developed. These classifications tend to use a static map approach that provides an aerial top-down perspective. Given that landscapes are often experienced while travelling, research methods that capture the time and space components of travel, will enhance the characterisation and classification of landscape experience. A landscape character classification based on experience could potentially be used in psychophysical assessment. For example; people could be surveyed for their preferred landscape viewed from a walking track. These survey results could then be analysed in relation to the landscape character of the tracks and inferences made about people's preferred landscapes.

Visibility analysis is widely used in landscape assessment, which is not surprising given that landscape is first and foremost the appearance of the land. The generation of landscape viewsheds and visible areas using GIS was one of the earliest applications of digital elevation models (Amidon & Elsner, 1968), and is now widely available in GIS packages. Viewsheds are usually calculated from point locations using line of sight calculations. Points can be used to represent linear features (e.g. road) by using a series of evenly spaced points that follow the linear feature, or area features (e.g. lake) by using a regular grid of points. The analysis is dependent on having accurate surface elevation information, as poor quality will result in viewshed uncertainty (Fisher, 1991, Fisher, 1993). The NZ Landscape Character Classification used visibility analysis to identify water views. The technique has also been used to characterise the landscape seen in a photograph (Brabyn & Mark, 2011) for the purpose of automatically tagging photos. There is ongoing research to improve the accuracy of viewshed calculations. In particular, there is the effect of the distance between the observer and the perceived object. This effects the perceived size of an object and has been compensated using distance decay functions (Kumsap, Borne, & Moss, 2005). Other methods include the use of distance bands (Higuchi & Terry, 1983). Slope and aspect also effect visibility, as a surface that is facing the observer is more visible than a surface that is side on (Bartie, Reitsma, Kingham, & Mills, 2011). Viewshed analysis tends to provide a binary result showing whether a pixel can be viewed or not.