Capturing Landscape Visual Character Using Indicators: Touching Base with Landscape Aesthetic Theory

Landscape Research ISSN: 0142-6397 (Print) 1469-9710 (Online) Journal homepage: https://www.tandfonline.com/loi/clar20 Capturing Landscape Visual Character Using Indicators: Touching Base with Landscape Aesthetic Theory Åsa Ode, Mari S. Tveit & Gary Fry To cite this article: Åsa Ode, Mari S. Tveit & Gary Fry (2008) Capturing Landscape Visual Character Using Indicators: Touching Base with Landscape Aesthetic Theory, Landscape Research, 33:1, 89-117, DOI: 10.1080/01426390701773854 To link to this article: https://doi.org/10.1080/01426390701773854 Published online: 14 Jan 2008. Submit your article to this journal Article views: 7726 Citing articles: 118 View citing articles Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalinformation?journalcode=clar20

Landscape Research, Vol. 33, No. 1, 89 117, February 2008 Capturing Landscape Visual Character Using Indicators: Touching Base with Landscape Aesthetic Theory ÅSA ODE*, MARI S. TVEIT{ & GARY FRY{ *Swedish University of Agricultural Sciences, Sweden {Norwegian University of Life Sciences, Norway ABSTRACT This paper presents one way that landscape visual character can be captured using indicators derived from nine theory-based concepts related to landscape perception. The paper aims to establish links between landscape aesthetic theory and visual indicators, thus exploring what landscape indicators are really indicating. The steps from abstract visual concepts to measurable visual indicators are described, and links are made to theories of landscape preferences and perception. The focus of the paper is on the application of indicators, including a presentation of the possible data sources of the presented indicators. The paper includes a discussion on the selection of appropriate landscape indicators through a suggested filtering process. The relationships between the concepts and the ability of visual indicators to capture changes in landscape character and other issues related to interpretation are discussed. KEY WORDS: Visual character, landscape indicator, landscape analysis Introduction This paper describes a framework for assessment of landscape visual character using theory-based visual indicators. The landscape perspective in management, planning and policy is a current focus in Europe, as part of the European Landscape Convention (ELC) adopted in the year 2000 by the European Council. The ELC defines landscape as an area, as perceived by people, whose character is the result of the action and interaction of natural and/or human factors (Council of Europe, 2000). This definition puts the focus on the human experience of landscape, highlighting issues of the perception and character of a landscape. The framework presented in this paper relates closely to Landscape Character Assessment (Swanwick, 2002), but encompasses only the visual character of a landscape. A landscape assessment focuses on describing the landscape in contrast to a landscape evaluation which strives to identify what makes one landscape better or worse. Correspondence Address: Åsa Ode, Department of Landscape Architecture, Swedish University of Agricultural Sciences, PO Box 58, SE230 53 Alnarp, Sweden. Email: asa.ode@ltj.slu.se ISSN 0142-6397 Print/1469-9710 Online/08/010089-29 Ó 2008 Landscape Research Group Ltd DOI: 10.1080/01426390701773854

90 A. Ode et al. Landscape Character Assessment (LCA) has been developed as a tool to include issues of the experience of landscape (among others) within management, planning and monitoring (Wascher, 2005). One of the most widely applied schemes is the system developed by the Countryside Agency and Scottish Natural Heritage, and carried out for England and Scotland (Swanwick, 2002). Landscape character is defined here as a distinct, recognisable and consistent pattern of elements in the landscape that makes one landscape different from another, rather then better or worse (Swanwick, 2002). In England, the landscape character types identified form the basis for the monitoring scheme Countryside Quality Counts, where change in character is one important component. The method developed in England and Scotland has been used as the basis for the implementation of landscape character assessment across Europe (e.g. Denmark [Miljøministeriet, 2007], Sweden [La nsstyrelsen i Ska ne, 2006]). The LCA as developed for England and Scotland has a holistic approach towards landscape character, integrating all the aspects contributing to character for defining character areas. This contrasts with systems such as LANDMAP in Wales where the different aspects (e.g. visual and sensory, landscape form, historical landscape, cultural landscapes) contributing towards character are kept as separate information layers that can be combined for different purposes (Countryside Council for Wales, 2006). It has been argued that identifying character is, to a large extent, built upon human perception and therefore landscape character assessment can be questioned with regards to its scientific rigour and hence its role as an analytical tool for landscape planning (Wascher, 2005). This paper describes an approach to the development of visual landscape indicators linked to landscape character. We believe that such an approach can make a valuable contribution to the development and application of landscape characterization (Table 1). Capturing Landscape Visual Character Using Indicators Landscape indicators provide possibilities for a more objective basis for identifying landscape character through dividing the totality of our visual perception of the physical landscape into quantifiable characteristics. Visual landscape indicators are less well developed than for those of other landscape functions (Dramstad & Sogge, Table 1. Definitions used in the paper Definitions Landscape Landscape visual character Visual landscape assessment Landscape analysis An area, as perceived by people, whose character is the result of the action and interaction of natural and/or human factors (Council of Europe, 2000) The visual expression of the spatial elements, structure and pattern in the landscape A process that aims at analysing visual landscape character A systematic process of describing landscape attributes, their spatial pattern and their importance to people

Capturing Landscape Visual Character Using Indicators 91 2003). However, there has been a recent increase in developing the application of visual indicators in response to the demand to incorporate aspects of human perception of landscape (e.g. Jessel, 2006; Wascher, 2005; Palmer, 2004; Germino et al., 2001; Weinstoerffer & Girardin, 2000; van Mansvelt & Kuiper, 1999; Gulinck et al., 1999). Our aim is to present a framework which is able to capture the visual character of a landscape and describe landscape change over time. The approach builds on the conceptual framework established by Tveit et al. (2006), and will take the framework towards the application of visual indicators. The output is meant to be descriptive rather than normative, thus referred to as landscape visual character rather than visual quality. An objective landscape visual character assessment could however form a useful basis for subsequent evaluation of landscape visual quality, for example, in a management or policy setting. The framework consists of nine concepts of visual landscape character considered important in landscape aesthetic literature (see later), and the provision of indicators related to these different aspects of the visual landscape helps us identify which aspects are affected by landscape change. Thus, the framework makes it possible to identify the nature of landscape change, and thereby the impact of changes on the visual qualities of the landscape. We believe this could be very useful in assessment and monitoring of both particular aspects as well as the totality of a landscape s visual character. As in Tveit et al. (2006), this theoretical framework consists of four levels of abstraction linking indicators to landscape aesthetic theory; concepts, dimensions, landscape attributes and indicators. The concepts should be seen as an umbrella term under which different dimensions and synonyms of the concept are found. Both the concept and dimension levels are abstractions of the landscape s physical attributes. The indicators represent the level at which the landscape attributes could be measured and quantified. Visual Landscape Indicators and Their Support in Theory This paper is based on a literature review covering papers on landscape aesthetics, visual concepts and landscape preferences, including both suggested and empirically tested visual indicators. From the literature, nine visual concepts were identified which together characterize the visual landscape. These were: complexity, coherence, disturbance, stewardship, imageability, visual scale, naturalness, historicity, and ephemera. The nine concepts are supported by different theories explaining people s experience of landscape and their landscape preferences. While this paper focuses on landscape character, theories developed for explaining and predicting preference provide a basis for explaining what is important for our experience of landscape. These theories could thereby aid in identifying what characteristics of the visual landscape are important to describe. Table 2 presents the nine visual concepts and outlines their relationship to different landscape aesthetic theories. Through the use of the theoretical framework developed by Tveit et al. (2006) visual indicators could be linked to specific visual concepts that are contributing to landscape character. This section will present the visual indicators identified for the

92 A. Ode et al. Table 2. Concepts describing landscape character relationships to theories of landscape preference and experience Concept Theory References Complexity Biophilia Kellert & Wilson (1993) Coherence Information Processing Theory Kaplan & Kaplan (1982, 1989) Disturbance Biophilia Kellert & Wilson (1993) Stewardship Aesthetic of care Nassauer (1995) Imageability Spirit of place/genius loci/vividness Lynch (1960); Litton (1972); Bell (1999) Topophilia Tuan (1974) Visual scale Prospect-refuge theory Appleton (1975) Information Processing Theory Kaplan & Kaplan (1982, 1989) Naturalness Restorative landscapes Kaplan & Kaplan (1989); Ulrich (1979, 1984) Biophilia hypothesis Kellert & Wilson (1993) Historicity Topophilia Tuan (1974) Landscape heritage/ historic landscapes Lowenthal (1979, 1985); Fairclough et al. (1999) Ephemera Restorative landscapes Kaplan & Kaplan (1989); Ulrich (1979, 1984) nine concepts. Based on the literature we suggest ways to apply the indicators using different data sources: landscape photos, land cover data, orthophotos and field observation. Indicators of Complexity Complexity refers to the diversity and richness of landscape elements and features and the interspersion of patterns in the landscape. Complexity is a factor in the Kaplan s Informational Processing Theory, where complexity provides content and things to think about (Kaplan & Kaplan, 1989). An overly complex landscape is also likely to affect the legibility of the landscape as a consequence of offering too much information. The Biophilia hypothesis presented by Kellert and Wilson (1993) states the importance of diversity in relation to nature, both with regards to species and landscape types. The indicators of complexity describe the complexity of landscape both with regards to content and spatial configuration. In the literature three groups of indicators could be distinguished: 1. Distribution of Landscape Attributes, which focuses on the number of landscape elements:. Density of landscape elements (de la Fuente de Val et al., 2006; Gulinck et al., 2001; Schu pbach, 2002). Diversity of landscape attributes (de la Fuente de Val et al., 2006; Germino et al., 2001; Giles & Trani, 1999; Gulinck et al., 2001; Hunziker & Kienast, 1999; Palmer, 2004; van Mansvelt & Kupier, 1999)

Capturing Landscape Visual Character Using Indicators 93 2. Spatial Organization of Landscape Attributes, focusing on which degree this could be perceived as complex or simple. For this group the following indicators have been suggested in literature:. Edge density (Germino et al., 2001; Palmer, 2004). Heterogeneity (Dramstad et al., 2001; Fjellstad et al., 2001). Aggregation (de la Fuente de Val et al., 2006) 3. Variation and Contrast between landscape elements.. Degree of contrast (Hands & Brown, 2002; Arriaza et al., 2004). Shape variation (Giles & Trani, 1999; de la Fuente de Val et al., 2006; Gulinck et al., 2001; Hulshoff, 1995; Palmer, 2004; Weinstoerffer & Girardin, 2000). Size variation (Giles & Trani, 1999; de la Fuente de Val et al., 2006; Gulinck et al., 2001; Hulshoff, 1995; Palmer, 2004; Weinstoerffer & Girardin, 2000) Complexity is a concept that has been focused on in landscape ecological studies (e.g. Green et al., 2007) and hence several types of indicators have been developed for the application on land cover and orthophotos, as shown in Table 3. Most of these indicators could be applied using programs such as FRAGSTAT (McGarigal et al., 2002). Landscape photos and field observation can provide detailed information about landscape elements not covered by land cover data, for example, linear and point elements and the density of these in the landscape. Landscape photos and field observations will often be necessary to assess the contrast between different land covers. Indicators of Coherence Coherence relates to the unity of a scene, the degree of repeating patterns of colour and texture as well as a correspondence between land use and natural conditions. Coherence is one factor for predicting preference within the Information Processing Theory, and it refers to a more immediate understanding and readability of our environment (Kaplan & Kaplan, 1989). The indicators of coherence identified in the literature focus, to a large extent, on the spatial arrangement of landscape elements and can be broadly divided into: 1. The Spatial Arrangement of Water.. Presence of water (Kuiper, 2000; van Mansvelt & Kuiper, 1999). Correspondence of land form and location of water (Kuiper, 2000; van Mansvelt & Kuiper, 1999) 2. Spatial Arrangement of Vegetation.. Correspondence with expected natural conditions (van Mansvelt & Kuiper, 1999). Fragmentation (Litton, 1972; Palmer, 2004). Repetition of pattern across the landscape (Kaplan & Kaplan, 1989; Pearson, 2002)

94 A. Ode et al. Table 3. Complexity suggested indicators and application using different data sources Concept Data source Complexity Landscape photos Orthophotos Land cover data Field observations 1. Distribution of landscape attributes. Richness of landscape elements Number of landscape elements per view. Diversity of land cover Number of different land covers per view Number of landscape elements per area Number of landscape elements per area Diversity and Diversity and evenness indices a 2. Spatial organization of landscape attributes. Edge density Edge density a Edge density a. Heterogeneity Heterogeneity Index b Heterogeneity Index b. Aggregation of land cover/patches Aggregation indices a Aggregation indices a 3. Variation and contrast. Contrast Degree of contrast between land covers in view. Shape variation Degree of variation between shapes in view. Size variation Degree of variation between size in view Number of landscape elements per area Number of evenness indices a different land covers per area Degree of contrast between land covers Shape indices a Shape indices a Degree of variation between shapes Size distribution Size distribution Degree of variation indices a indices a between size a A range of diversity, evenness, edge density, aggregation, shape and size distribution indices are found within landscape metric software such as FRAGSTAT (McGarigal et al., 2002) and IAN (DeZonia & Mladenoff, 2004) developed within landscape ecology. b The heterogeneity index is the proportion of points on different land types and is calculated using a grid of points for which land types are recorded (see Fjellstad et al., 2001, for full detail of how to calculate the index).

Capturing Landscape Visual Character Using Indicators 95 The indicators represented in the literature are limited for coherence (see Table 4). Spatial arrangement of water could be estimated for all four types of data. The correspondence between land form and water location does, however, require information on elevation in order to estimate the degree of low lying areas with water. The spatial arrangement of vegetation could be quantified for all three types of data using measures of fragmentation and repetition of pattern. Estimation of the correspondence with natural conditions requires detailed information. When information on what could have naturally occurred is available, the degree to which the landscape is in agreement with this could be estimated using land cover data or through field observation. Indicators of Disturbance Disturbance refers to the lack of contextual fit and coherence in a landscape. The Information Processing Theory identifies, as presented earlier, coherence as one information factor (Kaplan & Kaplan, 1989). A high degree of disturbance is likely to result in a low level of coherence. The Biophilia hypothesis of Kellert and Wilson (1993) states the human biological need to affiliate with nature, and the consequences of disturbance in the context of human well-being (Kellert, 1996). Indicators of disturbance could be divided into: 1. Presence of Disturbing Elements.. Attributes classified as disturbance (Arriaza et al., 2004; Gulinck et al., 2001) 2. Visual Impact of Disturbance.. Area visually affected by disturbance (Gulinck et al., 2001; Hopkinson, 1971; Iverson, 1985) To apply indicators of disturbance it is necessary to identify which landscape elements are perceived as disturbing (see Table 5). This could be done both for land cover data, landscape photos and field observations, while orthophotos are less able to pick this up. Indicators of Stewardship Stewardship refers to the sense of order and care present in the landscape reflecting active and careful management. Care is a central concept in the aesthetics of care developed by Nassauer (1995, 1997) where visual cues of care are used to explain preference. Indicators of stewardship describe the degree of care and upkeep in the landscape. The literature suggests two groups of indicators for stewardship: 1. Level of Management for Vegetation. This has been described as the level of cultivatedness (van den Berg et al., 1998), and the following indicators are suggested in the literature:. Level of abandonment/stage of succession (Nassauer, 1995).. Presence of weed (Nassauer, 1995; van Mansvelt & Kuiper, 1999).

96 A. Ode et al. Table 4. Coherence suggested indicators and application using different data sources Concept Data source Coherence Landscape photos Orthophotos Land cover data Field observations 1. Spatial arrangement of water. Presence of water % of water cover % of water cover % of water cover Proportion of water cover. Correspondence land form and water location % of area in correspondence % of area in correspondence % of area in correspondence Proportion of area in correspondence 2. Spatial arrangement of vegetation. Correspondence with natural conditions % of area in correspondence % of area in correspondence % of area in correspondence. Fragmentation Fragmentation indices a Fragmentation indices a. Repetition of pattern across the landscape Presence of repeated patterns Proportion of area in correspondence Autocorrelation indices b Autocorrelation indices b Presence of repeated patterns a A range of fragmentation indices are suggested in landscape metric software such as FRAGSTAT (McGarigal et al., 2002) and IAN (DeZonia & Mladenoff, 2004) developed within landscape ecology. b Autocorrelation indices are found within different GIS software packages, such as ArcGIS.

Capturing Landscape Visual Character Using Indicators 97 Table 5. Disturbance suggested indicators and application using different data sources Concept Data source Disturbance 1. Presence of disturbing elements. Landscape elements classified as disturbed Landscape photos Orthophotos Land cover data Field observations Density of disturbing elements in the view % of area classified as visually disturbed % of area classified as visually disturbed Density of disturbing objects 2. Visual impact of disturbing elements. Area visually affected by disturbance % of area visually affected % of area visually affected % of area visually affected. Management type (Sheppard, 2001; van Mansvelt & Kuiper, 1999).. Management frequency (van Mansvelt & Kuiper, 1999; Weinstoerffer & Girardin, 2000).. Management detail (Nassauer, 1995; Sheppard, 2001; van Mansvelt & Kuiper, 1999). 2. Status and Conditions of Man-made Structures in the Landscape. Within the group, two indicators have been distinguished.. Status and maintenance of structures such as farm buildings and fences (Laurie, 1975; Nassauer, 1995; Weinstoerffer & Girardin, 2000). The application of different stewardship indicators depends on the available data (see Table 6). Land cover data depend on the level of detail in the classifications that permits different forms of reclassification based on succession and management type. Orthophotos depend on the identification of characteristics for different levels of stewardship. Field observations and landscape photos could provide information of different levels of management, although accurate measures related to area can be less readily available through these data sources. Indicators of Imageability Imageability reflects the ability of a landscape to create a strong visual image in the observer and thereby making it distinguishable and memorable. Imageability can be a product of the totality of a landscape or its elements. Imageability is related to theories of spirit of place (Bell, 1999), genius loci (Lynch, 1960) and vividness (Litton, 1972) and Topophilia (Tuan, 1974). Two groups of indicators are distinguished in the literature:

98 A. Ode et al. Table 6. Stewardship suggested indicators and application using different data sources Concept Data source Stewardship Landscape photos Orthophotos Land cover data Field observations 1. Level of management for vegetation. Level of abandonment % of vegetation in different stages of abandonment (1 4) a % of vegetation in different stages of abandonment (1 4) a % of vegetation in different stages of abandonment (1 4) a % of vegetation in different stages of abandonment (1 4) a. Presence of weed Density of weed Density of weed Density of weed. Management type % of area under different management regimes. Management frequency Number of highly maintained features % of area under different management regimes % of area under different management regimes % of area under different management regimes Number of highly maintained features 2. Condition of man-made structures. Condition/maintenance Number of structures in Number of structures of structures such as different conditions (1 4) a in different fences, buildings conditions (1 4) a a e.g. 1 ¼ highly maintained/no abandonment; 2 ¼ partly maintained; 3 ¼ poorly maintained; 4 ¼ no maintenance/total abandonment.

Capturing Landscape Visual Character Using Indicators 99 1. Spectacular, Unique and Iconic Elements are the focus for the first type of indicators. The following indicators are suggested:. Spectacular, unique or iconic built features (Coeterier, 2002; Green, 1999). Landmark (Green, 1999). Water (Hammitt et al., 1994; Litton et al., 1974). Historical elements (Jessel, 2006) 2. Viewpoints are the second type of indicators that are connected to this concept. The following indicators have been suggested:. Density of viewpoints (Gobster, 2001) Most of the indicators related to spectacular, unique and iconic elements cannot be estimated using orthophotos or land cover data, and hence need field observation or other data sources, as shown in Table 7. Proportion or percentage of area with water could be estimated using both orthophotos, landscape photos (in view) and land cover data. The number of viewpoints could be calculated through visibility analysis using orthophotos or land cover data together with terrain data. Viewpoints could also be estimated through field observations. Indicators of Visual Scale Visual scale describes landscape rooms/perceptual units in relation to their size, shape and diversity, and the degree of openness in the landscape. According to Appleton s prospect-refuge theory (Appleton, 1975), human beings, with the role of both predator and prey, have through evolution adapted to landscapes offering both prospect (the ability to get overview and hunting opportunity) and refuge (the ability to hide and escape from predators). The prospect-refuge theory is related to the habitat theory which links aesthetic pleasure to fulfilment of biological needs. For assessing visual scale two groups of indicators have been suggested in the literature: 1. Open Area, which focuses on the proportion and the size of open space in the landscape. This could be measured through:. Proportion of open land (Palmer, 2004; Weinstoerffer & Girardin, 2000). Viewshed size (De la Fuente de Val et al., 2006; Germino et al., 2001; Gulinck et al., 2001; Palmer & Lankhorst, 1998; Vining et al., 1984). Depth of view (Germino et al., 2001; Gulinck et al., 2001) 2. Obstruction of the View referring to objects that are seen as blocking the view.. Density of obstructing objects (Palmer & Lankhorst, 1998; Weinstoerffer & Girardin, 2000). Degree of visual penetration of vegetation (Weinstoerffer & Girardin, 2000)

100 A. Ode et al. Table 7. Imageability suggested indicators and application using different data sources Concept Data Source Imageability Landscape photos Orthophotos Land cover data Field observations 1. Spectacular, unique and iconic elements. Density of spectacular, Density in view Density unique or iconic built features. Density of landmark Density in view Density. Proportion of water % of water in view % of water % of water Proportion of water. Density of historical Density in view Density elements 2. Viewpoints. Density of viewpoints Density of viewpoints through visibility analysis Density of viewpoints through visibility analysis Density of viewpoints

Capturing Landscape Visual Character Using Indicators 101 Several indicators are proposed for assessing openness and are applicable for all four datasets (see Table 8). For the analysis of viewshed a terrain model is necessary for both either orthophotos or land cover data. To assess the obstruction of the view, we identified no indicators using land cover, while density of obstructing objects could be assessed using orthophotos, landscape photos (in view) and field observations. Indicators of Naturalness Naturalness describes the perceived closeness to a preconceived natural state. The Biophilia hypothesis of Kellert and Wilson (1993) states the importance of naturalness as man s biological need to affiliate with nature. This is defined as people s innate tendency to focus on life and lifelike processes. Biophilia is believed to have developed through evolutionary history as a consequence of its functional significance. Environmental psychologists see naturalness as an important aspect of restorative environments, which are environments enhancing recovery of mental energies and effectiveness (Kaplan & Kaplan, 1989; Hartig et al., 2003). Indicators of naturalness could be divided into three types. 1. Naturalness of Vegetation focuses on the quality of the present vegetation in relation to its perceived naturalness. To indicate this, two types of indicators have been suggested in the literature:. Percentage of natural vegetation (Arriaza et al., 2004; Ayad, 2005; Brabyn, 2005; Palmer, 2004; Schu pbach, 2002). Level of vegetation succession (Palmer, 2004, Schu pbach, 2002; van Mansvelt & Kuiper, 1999). Shape of vegetation (e.g. Palmer, 2004; van Mansvelt & Kuiper, 1999) 2. Pattern in the Landscape, as perceived as natural or not. This could be estimated using:. Fractal indices (Hagerhall et al., 2004; Antrop & Van Eetvelde, 2000). Fragmentation indices (Taylor, 2002) 3. Water in the landscape is often used as an indication of naturalness.. Proportion of water in the landscape (Palmer, 2004; van Mansvelt & Kuiper, 1999) The estimation of naturalness of vegetation or stages of succession using orthophotos or land cover data relies on a reclassification of vegetation (see Table 9). Proportions of the landscape with perceived naturalness can then be estimated. Level of vegetation intactness needs field observation or landscape photos for its estimation. Edge shape can be interpreted in terms of naturalness using landscape photos or field observations, while for land cover data or orthophotos a range of different indices are available. For estimating pattern in the landscape in

102 A. Ode et al. Table 8. Visual scale suggested indicators and application using different data sources Concept Data source Visual scale Landscape photos Orthophotos Land cover data Field observations 1. Open area. Proportion of open land % of open land % of open land % of open land Proportion of open land. Viewshed size Size of viewshed Size of viewshed. Viewshed shape Classification of view Shape index of shape (1 3) a viewshed. Depth/Breadth of view Estimation of depth of Length of radius view (1 3) b of view Shape index of viewshed Length of radius of view Classification of view shape (1 3) a Estimation of depth of view (1 3) b 2. Obstruction of the view. Density of obstructing objects. Degree of visual penetration of vegetation Density of obstructing objects Proportion of vegetation with different levels of visual penetration (1 4) c Density of obstructing objects Density of obstructing objects Proportion of vegetation with different levels of visual penetration (1 4) c a e.g. 1 ¼ one large open area; 2 ¼ split open area; 3 ¼ patchy open area. b e.g. 1 ¼ short; 2 ¼ medium; 3 ¼ long. c e.g. 1 ¼ blocked; 2 ¼ dense; 3 ¼ semi-open; 4 ¼ open.

Capturing Landscape Visual Character Using Indicators 103 Table 9. Naturalness suggested indicators and application using different data sources Data source Concept Naturalness Landscape photographs Orthophotos Land cover data Field observations 1. Naturalness of vegetation. Proportion of natural vegetation % of natural vegetation in the view. Level of succession % of vegetation in different stage % of natural vegetation % of natural vegetation Proportion of natural vegetation % of vegetation in different stage (0 3) of succession a % of vegetation in different stage (0 3) of succession a Proportion of vegetation in different stage (0 3) of succession a Shape indices c Shape indices c Interpretation of (0 3) of succession a. Shape of edges Interpretation of edge types b edge types b 2. Pattern in the landscape. Fractality Fractal indices c Fractal indices c. Fragmentation Fragmentation indices c Fragmentation indices c 3. Water. Proportion of water % of water in the view % of water % of water Proportion of water a e.g. 0 ¼ no succession; 1 ¼ primary succession; 2 ¼ intermediate succession; 3 ¼ climax. b e.g. geometrical, intermediate complex shapes; complex shapes. c A range of diversity, evenness, edge density, aggregation, shape and size distribution indices are found within landscape metric software such as FRAGSTAT (McGarigal et al., 2002) and IAN (DeZonia & Mladenoff, 2004) developed within landscape ecology.

104 A. Ode et al. relation to its fractality, McGarigal et al. (2002) present several indices that could be used. Indicators of Historicity Historicity describes the degree of historical continuity and richness present in the landscape. Historical continuity is reflected by the visual presence of different time layers, while historical richness focuses on the amount and diversity of cultural elements. The importance of historic landscapes and landscape heritage has been stressed by several researchers (Lowenthal, 1979, 1985; Fairclough et al., 1999). Historical association is also considered as important for the appreciation of scenery in Tuan s theory of Topophilia (1974). Tuan focuses on the cultural dimension of preference. Indicators of historicity describe both the time depth present in the landscape, the historical richness and their impact in the landscape. Within the literature, three groups of indicators have been identified. 1. Vegetation with Continuity. This could be described as:. Proportion of landscape with long vegetation continuity (Jessel, 2006). Proportion of landscape with traditional land use (Jessel, 2006; Gulinck et al., 2001) 2. Organization of Landscape Attributes as described through:. Field size (Fairclough et al., 2002; Darlington, 2002). Field shape (Fairclough et al., 2002; Darlington, 2002). Spatial arrangement of vegetation (Kuiper, 2000) 3. Landscape Elements, focusing on the presence of historical features in the landscape. This could be described with:. Density of cultural elements (Van Mansvelt & Kuiper, 1999). Shape of line features (Darlington, 2002; Fairclough et al., 2002) In order to apply these indicators we need to establish what a traditional landscape would have contained with regards to land use patterns and cultural elements. This could then be used both to identify areas and elements but also to compare with shape, size and aggregation indices (see Table 10). Field observations and landscape photos will focus on the presence or absence of landscape elements, while for the land cover data and orthophotos a range of indices based on McGarigal et al. (2002) have been suggested in addition to the proportion of the landscape with traditional land use and vegetation continuity. Indicators of Ephemera Ephemera refer to landscape changes related to season or weather. Within restorative environments, there is a fascination factor, where so-called soft

Capturing Landscape Visual Character Using Indicators 105 Table 10. Historicity suggested indicators and application using different data sources Concept Data source Historicity Landscape photos Orthophotos Land cover data Field observations 1. Vegetation with continuity. Proportion of landscape with continuity of land cover. Proportion of landscape with traditional land use % of view with continuity of land cover % of view with traditional land use % of area with traditional land use % of area with continuity of land cover % of area with traditional land use Proportion of area with continuity of land cover Proportion of area with traditional land use 2. Organization of landscape attributes. Field size Presence of small fields Size indices a Size indices a Presence of small fields. Field shape Presence of traditional. Spatial arrangement of vegetation field shapes Presence of traditional spatial arrangement 3. Landscape elements. Density of cultural elements Density of cultural elements Shape indices a Shape indices a Presence of traditional field shapes Aggregation indices a Aggregation indices a Presence of traditional spatial arrangement Density of cultural elements Density of cultural elements. Shape of linear features Shape indices a Presence of traditional shapes a A range of size, shape and aggregation indices are found within landscape metric software such as FRAGSTAT (McGarigal et al., 2002) and IAN (DeZonia & Mladenoff, 2004) developed within landscape ecology.

106 A. Ode et al. fascination (Kaplan & Kaplan, 1989) has been illustrated by many examples of changes in weather or season. These features, according to Kaplan and Kaplan (1989), enhance the being away aspect of landscape experience. Indicators of ephemera describe seasonal and weather changes per area, frequency of changes and the magnitude of change. Indicators identified in the literature can be divided into three groups. 1. Season-bound Activities, which focuses on events taking place in the landscape in relation to season. These are:. Farming activities with seasonal patterns such as harvest (Brassley, 1998; Jessel, 2006; van Mansvelt & Kuiper, 1999). Presence of animals (Jessel, 2006; Litton, 1968, 1972) 2. Landscape Attributes with Seasonal Change which refers both to natural vegetation and agricultural land. Suggested indicators are:. Seasonal variation in natural vegetation (Ahas et al., 2005; Hendriks et al., 2000; Brassley, 1998; van Mansvelt & Kuiper, 1999). Seasonal variation in crops and fields (Brassley, 1998; Jessel, 2006; van Mansvelt & Kuiper, 1999). Water with seasonal change (Morgan, 1999) 3. Landscape Attributes with Weather Characteristics, focuses on elements that are prone to changes in relation to meteorological changes. Indicators found in the literature are:. Water (e.g. Morgan, 1999; Litton, 1968, 1972) A large number of the indicators rely (regardless of dataset used) on a classification of natural vegetation and farming activities into those with seasonal changes or not, as seen in Table 11. Land cover data are often not detailed enough to make this distinction in relation to agricultural land and farming activities, and hence orthophotos, landscape photos or field observations are needed. Discussion Choosing Indicators Usefulness of a Limited Set The experience of the countryside is holistic and the overall impression of a view is what people observe as visual character. This calls for great care in the application of indicators. In order to capture landscape visual character the selection of indicators requires careful consideration. A wide range of visual indicators are available for assessing landscape change and its visual consequences. In most landscape assessment and monitoring projects it will be unnecessary if not impossible to apply all these indicators. We suggest an approach to indicator selection based on filters.

Capturing Landscape Visual Character Using Indicators 107 Table 11. Ephemera suggested indicators and application using different data sources Concept Data source Ephemera Landscape photos Orthophotos Land cover data Field observations 1. Season-bound activities. Presence of animals Seasonal presence of animals Seasonal presence of animals. Presence of cyclical farming activities % of land cover with cyclical farming activities in view % of land cover with cyclical farming activities Proportion of land cover with cyclical farming activities 2. Landscape attributes with seasonal change. Seasonal variation in % of area with seasonal natural vegetation changing vegetation in view. Seasonal variation on agricultural land % of agricultural land with seasonal variation in view % of area with seasonal changing vegetation % of agricultural land with seasonal variation % of area with seasonal changing land cover Proportion of area with seasonal changing vegetation Proportion of agricultural land with seasonal variation. Water with seasonal change % of water in view % of water % of water Proportion of water 3. Landscape attributes with weather characteristics. Presence of water % of water in view % of water % of water Proportion of water

108 A. Ode et al. The filters are criteria that indicator application should meet. The filtering will identify a suitable set of visual indicators for application within a specific project or landscape context. The filter approach is useful for its ability to make the process of indicator selection transparent. The approach allows for changes in the criteria (filters) for the selection of indicators should the context or aim of assessment change. If the aim is to assess the totality of the visual character of a landscape, this will require an extensive set of indicators covering all visual aspects. Sometimes, a landscape is monitored with regards to one or a few visual aspects, for example, visual scale or disturbance. In such projects an indicator set restricted to just analysing these aspects might suffice. The filters will help to ensure the appropriate selection of indicators to be applied in specific landscape assessment and monitoring projects. We suggest the following six initial filters for visual indicator selection (see Figure 1). The first filter, clear theoretical base, represents a criterion that we believe should apply for all indicators and projects. This implies that indicators should be theory driven rather than data driven, and is an expression of the need to know what we want indicators to indicate. Second, we suggest that visual indicators should be transferable between landscapes, meaning that they should not be landscape type specific. This criterion makes possible comparisons between landscapes. Third, visual indicators should be quantifiable, so that they can be measured and compared. Fourth, visual indicators should be mappable, meaning that they should be possible to locate spatially and express through maps. Filters 2 4 are suggested to make the use of indicators transparent and repeatable. All indicators presented in our framework fulfil the criteria of these first four filters. Finally, we suggest two filters dependent on project features. First, relevance, this is a filter for indicator selection that is project and context sensitive. Indicators will be selected as a consequence of the particular interests involved, points of focus, stakeholder interests, etc. The relevance filter will be particularly important in relation to public participation and landscape planning or impact assessment. The relevance filter will determine how narrow the set of indicators can be in terms of encompassing a few or many visual concepts. Finally, a filter in the selection of indicators will always be data availability. This will depend on project resources, and on the visual aspects to be assessed or monitored in the project. Some concepts have a wider range of indicators to choose from than others, for example, complexity has a vast range of possible indicators while coherence has only a few. Some indicators are easy to apply because their data are readily available through land cover data bases. Others, such as stewardship indicators, require additional effort, for example, field surveys. Scarcity of data creates the danger of making the process of visual analysis data driven rather than theory driven. Lack of data could limit the possibility of getting a valid expression of landscape character. It is crucial to keep the focus on what we want indicators to indicate, and identifying the data required to make this possible. To obtain a valid expression of landscape character through the use of indicators we suggest using several data sources, for example, land cover data, orthophotos, landscape photos and field observations, thus widening the set of potentially applicable indicators.

Capturing Landscape Visual Character Using Indicators 109 Figure 1. Selection of indicators through suggested filters. Concept Interrelationships Although the visual concepts and their visual indicators are presented independently, they are interrelated, and landscape changes altering indicator values related to one concept may cause an increase or decrease in indicator values of another concept. This needs careful consideration in application and interpretation. Figure 2 shows a map of the visual concepts, and how some are closely related while others could be seen as opposites. An example of closely linked and sometimes overlapping concepts is historicity and imageability, as elements creating strong imageability are often, but not always, cultural elements. Another example are the concepts of complexity and naturalness, where complexity is used as a description of naturalness (e.g. Hagerhall et al., 2004; Purcell & Lamb, 1998). On the other hand, some concepts are opposites, such as naturalness and stewardship, when stewardship decreases naturalness increases and vice versa. This relationship between naturalness and stewardship has been identified by, for example, Nassauer (1995), and signs of abandonment (increasing naturalness) are used as indicators of decreasing stewardship. Other opposing concepts are coherence and disturbance, where the absence of disturbance has been used as an indicator of disturbance (van Mansvelt & Kuiper,

110 A. Ode et al. Figure 2. Map of concepts where dotted lines represent dependencies between the concepts, e.g. perceived disturbance is dependent on the visibility of the disturbing element, which is determined by the visual scale of the landscapes.

Capturing Landscape Visual Character Using Indicators 111 1999). Kaplan and Kaplan (1989) discussed the trade-offs between coherence and complexity, stating that an overly complex landscape can be considered messy or lacking coherence. According to Kaplan and Kaplan, the relationship between the two concepts is however not straightforwardly opposed, as a scene can also be high in complexity and coherence at the same time, it being rich but organized. The map in Figure 2 should be seen as a suggestion of what the interrelationships of the concepts of visual character might look like. The visual indicators do further show a relationship between each other and to other concepts. Visual scale is a concept that influences several indicators including both disturbance ( area perceived as disturbed ) and imageability ( density of viewpoints ). When using visual indicators for analysing character, few studies have focused on the nature of the relationship. Studies related to preference research have found that the relationship between indicators and preference is not necessarily a linear relationship (e.g. Kaplan & Kaplan, 1989). This is probably also the case for the relationship between character and indicators, where the indicator values are not reflected in changes of character or remain unnoticed until a threshold value is passed. Identifying the nature of the relationship between character and indicator values is an important area for further studies. Topography and Water The concept indicator framework developed in this study suggests that certain landscape elements and features are indicators of landscape character as expressed through several of the visual concepts. One such element is water, which is believed to contribute to naturalness, coherence, imageability and ephemera. Topography is another feature of importance within several of the concepts, for example, complexity (landform diversity), visual scale (depth/breadth of view), imageability (view points) and disturbance (visibility of disturbing elements). The special importance of both topography and water are also evident in the LCA as developed for England and Scotland (Swanwick, 2002). In the LCA land form and water are often used as the key landscape elements for distinguishing different character areas. The special importance of water and topography for the experience of landscape has also been evident in preference research (e.g. Arriaza et al., 2004; Brush, 1981; Nasar & Li, 2004; Wherrett, 2000). Landscape Type Though all the presented concepts are important for the formation of landscape character, there is reason to believe that the relative importance or weighting of the different concepts is dependent on landscape type (Purcell et al., 2001). When assessing different landscape types in the Netherlands, de Groot and van den Born (2003) found examples of landscape type dependency in relation to perceived naturalness. Gulinck et al. (2001) report a need for local tuning of indicators of disturbance. What is seen as intrusions in a rural landscape will depend on local context. This implies that even if indicators should be transferable between landscapes, one needs to take the local context into account in the interpretation of indicator values.

112 A. Ode et al. Importance of Scale Visual indicators are sensitive to the choice of landscape scale in visual assessment (e.g. mapping scale, photo frame) and this relates to all types of media used. The perceived grain size of a small part of a landscape might easily differ greatly from the impression given by the same landscape at a larger scale. This requires care in situations where photographic samples are used to reflect the content of whole landscapes. When using land cover data or orthophotographs it has been shown that calculation of landscape metrics (e.g. to capture complexity) is very sensitive to scale and resolution of the data (Li & Wu, 2004; Lausch & Herzog, 2002). When applying indicators we recommend the use of a consistent scale within a project in order to allow for comparison across areas and over time. Data Sources Which data source will give the most useful information about a landscape or landscape feature is dependent on the purpose of the study. For preference studies assessing general preferences for a given landscape feature, photographs are valid, practical and frequently used representations of landscapes (Trent et al., 1987; Wherrett, 1998). In such studies, it is not the sites per se, but the character they represent that is being rated. Indicator values in the landscapes represented can be manipulated in the photographs to allow assessment of the selected landscape feature and gain control of the photo content. For scenario assessment photos can also be useful representations of, for example, planned developments. For monitoring and assessing the visual character of particular landscapes, photographs have limitations as a data source. It is difficult to capture the totality of a real landscape using photographs as discussed by Palmer and Hoffman (2001). GIS-based indicators are very common in visual landscape assessment (e.g. Germino et al., 2001; Gulinck et al., 1999; Lynch & Gimblett, 1992). Using land cover data it is possible to measure changes in visual indicators taking whole landscapes into account. However, land cover data cannot directly represent what people see, and there is a danger of misinterpreting the effect of changes in land cover on visible landscape character. The value of land cover data for assessing different visual aspects depends on the level of detail in the classifications. Assessments of stewardship, for example, require detailed information on succession and management type. Orthophotos can be valuable supplements to land cover maps, and often add detail of smaller visual features, particularly linear and point features. Field observations and landscape photos can often provide greater detail in the status or presence of particular features, but can sometimes be difficult to link the whole landscape area in an accurate measure. In many studies a combined approach using several data sources will be the most appropriate (see e.g. Palmer & Lankhorst, 1998). Links between Indicators and Theory As can be seen from the display of different indicators in the previous sections we have identified a wide range of visual indicators. The nature of these indicators