new banner
 

Home
Services of Scenic Solutions
Completed landscape projects
Papers and Presentations
Beautiful landscapes
Australian and International studies of landscape quality
The science of scenery
Landscape miscellany
Landscape resources
About Dr Andrew Lothian
Contact us
Site map

FINDINGS OF INTERNATIONAL LANDSCAPE STUDIES

LANDSCAPE PREFERENCES

Introduction

The influence of various aspects of the landscape on human preferences is reviewed under the following headings (click on these):

References cited may be found here.

Water

Water has long been recognised as an important element in landscape preferences. This section summarises some of the studies.

Calvin et al [1972] used the semantic differential technique and factor analysis to analyse responses to photographs of landscapes including several that incorporated water. Figure 1 summarises the attribute scores of each scene for the natural scenic beauty factor [Ibid, 465] and indicates that those with water were among the highest positively scoring scenes, although algae in streams was regarded negatively.

Calvin

Source: Calvin et al, 1972; Note: factor scores signs reversed to correspond with positive & negative perception of scenes. Factor scores shown are for Factor 1 - natural scenic beauty factor
Figure 1 Scores for Landscape Scenes

In their study of the Nigerian city of Warri, Choker & Mene [1992] found that, in natural scenes, the most preferred landscape was “a natural river or water scenery with a surrounding natural and well-preserved tropical rainforest vegetation” [Ibid, 253]. The presence of water and a river was, after trees and flowers, the most important determinant of landscape quality, although dirty water or waterlogged conditions were regarded negatively. The importance of nature for fishing, farming, hunting and other economic needs was the reason given for the appeal of natural landscapes rather than concern for nature.

Gobster & Chenoweth [1989] derived factor loadings on four factors for river landscapes, forest landscapes and agricultural landscapes. The four factors identified were: Factor 1 artistic descriptor: complexity, uniqueness, topography, calmness of water, awe, arousal; Factor 2 affective-informational: land use variety, degree of alteration, unity, balance; Factor 3 spatial structure: distance, river width, land use variety, enclosure, mystery; Factor 4: river sinuosity. The Factor 1 descriptors together accounted for 61% of the variance and all four Factors accounted for 90%.

Gregory and Davis [1993] identified 22 factors that affected the scenic quality of riverscapes, some positively and some negatively. Scenic attractiveness was increased by the proportion of trees in the photograph, the number of tree trunks and the depth of water. Conversely, water colour, channelisation of the bank, percentage riverbank, the sinuosity of the channel, and amount of litter decreased scenic quality. Using regression analysis, Gregory and Davis derived the equation to describe the scenic preferences of riverscapes. It indicated that nearly 90% of the average scenic preference variation could be defined by the water colour, the stability of the channel banks and the average depth of water [Ibid, 181].

A definitive study of water preferences was undertaken by Herzog [1985]. Using factor analysis of preference ratings, he identified four waterscape types: mountain waterscapes; swampy areas; rivers, lakes and ponds; and large bodies of water. Based on S. Kaplan’s theories of information processing (see Theory), the study used as predictor variables: spaciousness, texture, coherence, complexity, mystery, and identifiability with preference used as a criterion variable.

Figure 2 summarises the mean ratings obtained for each predictor variable showing how they varied across each type of waterscape. These indicate that:

  • mountain waterscapes were distinguished by low textures
  • swampy areas were distinguished by low spaciousness
  • rivers, lakes & ponds were distinguished by high identifiability
  • large water bodies were distinguished by spaciousness, texture and coherence but were low in complexity and mystery

Herzog4

Source: Herzog, 1985
Figure 2 Rating of Variables for Various Water Bodies

Herzog found that only spaciousness and coherence were significant predictors of preference [Figure 2]]. Regression analysis of the variables against the criterion variable of preference indicated that “waterscapes high in spaciousness, coherence, and mystery, but low in texture (i.e. featuring coarse or uneven ground surface), were preferred to waterscapes with the opposite characteristics.” [Ibid, 235] The six predictor variables accounted for 71% of preference variance in mountain waterscapes, and for 74% in swampy waterscapes.

In terms of content, “mountain lakes and rushing water are the people’s choice, whereas swampy areas are unlikely ever to attract an enthusiastic following“ [Ibid, 237]. In terms of predictor variables, the most preferred waterscapes were high in spaciousness, coherence and mystery but low in texture. Large water bodies and mountain waterscapes, both high in spaciousness were the most preferred while swampy areas are lowest in this variable and in preference.

In a later study, Herzog and Bosley [1992] included a wider range of scenes to evaluate the role of tranquility on preference. Predictor variables used were mystery, coherence, spaciousness and focus, with tranquility and preference the criterion variables. The preference means for the different landscapes are summarised in Figure 3 and indicates that in terms of both tranquility and preference, water ranks highest among the landscapes evaluated.

Herzog5

Source: Herzog & Bosley, 1992
Figure 3 Comparison of Mean Scores for Tranquility and Preference

Correlations between the descriptor variables and preference for the landscapes evaluated [Figure 4] indicated high correlations for coherence and, to a lesser degree, focus. Mystery and spaciousness were negatively correlated for rushing water. Not surprisingly, the authors found that the turbulence in rushing water decreases the sense of tranquility. While turbulence can focus one’s attention thereby aiding preference, it also conveys a lack of calmness that decreases tranquility  [Ibid, 125].

Herzog6

Source: Herzog and Bosley [1992]
Figure 4 Correlations of Preference and Variables

Using tape recorders and visitor photography, Hull & Stewart [1995] surveyed trail users on the views they encountered. Feeling states were recorded by participants en route and were classified thus: beauty, satisfied, relaxed, and excited. Figure 5 summarises the average rating of these. It indicates that the water bodies contributed most in terms of beauty and were also rated high for satisfaction and relaxation. However, the water bodies ranked lowest for excitement  - which probably reflects the placid types of lake and river encountered.

Hull&Stewart

Source: Hull & Stewart [1995]
Figure 5  Feeling States along Trails

Palmer [1978] reported the results of an extensive landscape research project in Connecticut River valley led by Ervin Zube. The study identified 22 landscape dimensions, including water/land edge density per unit area and percentage water area per unit area. About 50% of the variation in scenic resource value was explained by seven of these dimensions. Scenic value was found to increase with naturalism [regression coeff = 0.59], landform variation [0.58], water/land edges [0.42] and the length of views [0.33]. Findings related to specific land uses included:

  • Farm landscapes - water area density had a major negative influence, suggesting that farm views dominated by large areas of water were not as scenic as those with smaller areas or water accents.
  • Open water landscapes - scenic value increased with water/land edge and decreased as the proportion of water surface area increased. An elevated viewer position, increased difference between elevations within the view, and increased naturalism contributed to scenic quality.
  • Wetlands and streams landscapes - scenic value increased with naturalness. In contrast to most studies, it was found that diversity in land use and contrast in naturalism decreased scenic quality.

Schroeder [1991] analysed the meaning that the Morton Arboretum in Chicago had for its many visitors. The Arboretum includes water features - lake, pond, stream and river. These, together with the forest and colours were the most frequently mentioned features. Serenity was a word used to describe places with water. The “ability of trees, other vegetation, and bodies of water to function as ‘natural tranquilizers’ may be one of the most significant human benefits of preserving nature...” [Ibid, 245].

In his analysis of landscape photographs used in the development of a regression equation, Shafer et al [1969] found through factor analysis that water features had among the highest factor loadings of any of the variables in a 26 X 26 correlation matrix. The area of the water features - stream, waterfall and lake, yielded slightly higher loadings than the perimeter of these features. Shafer’s regression equation contained ten terms and the water area featured in three of these, thereby indicating the importance of water in the landscape.

Urlich’s [1981] study found that while attentiveness declined regardless of the environment viewed, “the drop was significantly less when the scenes contained water” [Ibid, 543]. He considered that water had “greater attention-holding properties” [op cit]. He also found that whereas scenes of urban areas increased feelings of sadness, that water had a stabilising effect on emotions and, in particular, sharply reduced feelings of fear [Ibid, 544].

Yang & Brown [1992] found the most preferred scenes to be those with a dominance of water and a Japanese garden style. Reflections across the water of surrounding trees were a common feature.

In contrast to other researchers who used photographs, Brown and Daniel [1991] used 12-second video clips to capture the dynamic nature of stream flow not apparent in still photographs. Although the study focused on the influence of stream flow volume to scene quality, the researchers took care to ensure that this was not apparent. Paired comparisons were used, one showing a higher stream flow than the other, and the respondent choosing the most attractive. Regression analysis was used to analyse the influence of a range of variables in the landscape estimated from the video scenes. These included the proportion of sky, water, exposed riverbed, stream channel width and vegetation in the scenes. The results indicated that scenic beauty increased with stream flow to a mid point and then diminished [Figure 6].

Brown & Daniel

Source: Brown & Daniel, 1991
Figure 6 Influence of River Flow on Scenic Beauty

In two groups sampled, the scenic beauty was maximised at 1285 cubic feet per second [cfs] in the Fort Collins case and 1092 cfs in the Tucson case. Scenic beauty estimate (SBE) ratings were similar for low flows at 100 cfs as for high flows at 2000 cfs [all p < 0.001]. The findings indicated that flow quantity influences riparian scenic beauty up to a point and then decreases at higher flows. This finding was consistent across a wide range of vegetation, topographic and scene compositions.

Hetherington, Daniel & Brown [1993] replicated the above finding using sound as well as videos of river flow.

Summary - Influence of water on landscape preferences

It is evident from the range of studies that water has a profound effect on landscape preferences. The studies reported that scenic value increased with:

  • Water edge [Anderson et al, 1976; Palmer, 1978; Whitmore et al, 1995]
  • Water area [Anderson et al, 1976; Brush & Shafer, 1975]
  • Channel stability & depth are important factors in river scenic quality [Gregory & Davis, 1993]
  • Moving water [Craik, 1972; Dearinger, 1979; Hammitt et al, 1994; Whitmore et al, 1995]

In the Rockies, Jones et al [1976] found that water bodies were the third most important landscape component in defining preferences after the high mountains and forests. In New Zealand, Mosley [1989] found water ranked fifth in importance after forests, view angle, relative relief and alpine components [e.g. snow and ice]. Significantly he found the river environment to be more important than the river itself in determining preferences. In the less spectacular landscape of the Connecticut River valley, Palmer & Zube [1976] found that after landform, water was the second most important dimension.

Herzog [1985] assessed the preferences for different kinds of water bodies and found in order: mountain waterscapes; large water bodies; rivers, lakes & ponds; with swampy areas last.

Factors which were found to decrease the scenic value of water included pollution and waterlogging [Choker & Mene, 1992], water colour [Gregory & Davis, 1993], and litter, erosion, water quality and structures [Nieman, 1978]. Interestingly Hodgson & Thayer [1980] found that water bodies labelled as artificial rather than natural [e.g. reservoir instead of lake] scored lower than natural labels.

Serenity and tranquility contrasting with awe and arousal were found to be psychological factors deriving from water bodies [Gobster & Chenoweth, 1989; Herzog & Bosley, 1992; Schroeder, 1991]. Water holds one’s attention and has a stabilising effect on emotions [Urlich, 1981].

Overall, water was found to be a major and positive factor by Calvin et al [1972]; Choker & Mene [1992]; Dearinger [1979]; Dunn [1976]; Herzog & Bosley [1992]; Hull & Stewart [1995]; Orland [1988]; Shafer et al [1969]; Urlich, 1981; Vining et al [1984]; and Zube [1973].

Go to top

Mountains

Given the revolutionary change in Western attitudes towards mountains that occurred during the 18th century it could be expected that studies would indicate that mountains affect preferences positively.

Brush [1981] re-tested Shafer’s original photographs with a similar group of campers and found a strong relationship between landform and scenic preference [Table 1]. Brush found the Kendal’s rank order correlation between landform class and scenic preference was -0.37 and was very highly significant. The correlation is negative because Shafer’s method results in low preferences shown by high scores, thus preference scores decrease as relative relief increases [Ibid, 302].

Table 1 Frequency of Scenes by Landform and Scenic Preference Score

  Preference score Flat land Low hill Steep hill Mountain
High preference
60 - 89
1
90 - 119
1
3
120 - 149
1
4
4
150 - 179
1
3
10
2
180 - 209
4
2
2
1
Low preference
210 - 239
1
1
Total
6
6
18
11

Source: Brush [1981]

Buhyoff & Wellman [1980] tested a range of regression functions - linear, exponential, power and loge - against preference data. They found the logarithmic scenic preference functions result in the highest r2 for differing scenes [Table 2]. 

Table 2  Regression Coefficients for Specific Landscape Dimensions

Landscape Dimensions Linear Exponential Power Log e
Rolling Mountains
.12
1.4E-06
.08
.33
Sharp Mountains
.39
.11
.17
.47
Snow
.14
.06
.14
.48
Foreground Vegetation
.007
.01
.04
.15

Source: Buhyoff & Wellman [1980]

With the log function, preferences for snow and sharp mountains rose rapidly as the proportion of their area increased to about 15 - 20% and after the inflection point the functions flatten; in other words, a small amount produces a substantial lift in preferences (The log functions are: snow y = 0.1 log x; sharp mountains y = 0.09 log x; rolling mountains y = -0.08 log x; and beetle damaged vegetation y = - 0.14 log x where x is the proportion of area having these landscape attributes and y is the scenic preference). The converse occurs with rolling mountains and beetle damaged vegetation - the largest drop in preferences occurs as their area increases to about 15 - 20% and then flattens out. Buhyoff & Wellman suggest that scenic management should focus on these ranges when the proportion of area of the given attribute is relatively small because “small changes in landscape dimensions ... will cause substantial changes in preferences” [Ibid, 270].

The implications of the log function are twofold. Firstly, “a small quantity of a dimension that is perceived as positive can substantially improve people’s perceptions of the overall landscape if a great deal of it does not already exist.” [op cit]. Secondly, the curves imply that landscapes can become “aesthetically damaged” rather quickly [op cit].

Buhyoff et al [1983] assessed the preferences for mountain scenes among four nationalities [see section on culture]. Figure 7 summarises the preferences for the Appalachian landscapes and Rockies. The range of values is slightly greater for the Appalachian mountainscapes than for the Rockies, suggesting that the stronger forms and greater scale of the Rockies yields more definite evaluations.

Buhyoff

Source: Buhyoff, et al [1983]
Figure 7 Preference by Nationalities for Mountain Landscapes

Fines [1969] used a small group to rank 20 photographs representing the world’s landscapes [!], although he was vague regarding his methodology. His description of the results and the exponential scale of landscape values he derived [0 – 32] corresponded closely with the height and spectacular appearance of the mountains. His low values were accorded to flat, gently undulating landscapes, marshes, low hills, and coastal cliffs; his higher scores to high hills, lower mountains [e.g. Britain] and great mountains, canyons and waterfalls.

Kane’s [1976, 1981] study of the South Australian landscape showed that mountainous scenes ranked among the highest of the 46 scenes subject to the rating. The top five scenes were of mountains and steep gorges in the Flinders Ranges.

Given the prominence of mountains, it is surprising that in most studies their presence in the landscapes studied appears to be little more than a backdrop to the particular landscape, rather than as a key focal point. Shafer, for example, described mountains in the scenes as the distant non-vegetation zone [i.e. not even designating them with their proper name] in his derivation of a mathematical model. Relatively few studies sought to derive preference ratings for the mountain components of scenes per se – rather, they are treated merely as a part of the scene.

Generally, the studies found mountains provide a positive influence on preferences. While some studies found that mountains were ahead of water in terms of preferences [e.g. Hull & Stewart, 1992] others found water to be the leading preference [e.g. Herzog & Bosley, 1992]. Herzog [1987] also found ‘spacious canyons’ out-performed mountains. However, his example was the well-known Grand Canyon, which may have attracted preferences on the basis of familiarity. He found snowy mountains to be high in spaciousness but low in complexity while smaller mountains were high in both spaciousness and identifiability.

In one of the few detailed analyses of mountain landscapes, Hammitt et al [1994] found that multiple ridges of a range, disappearing into the distance were preferred over a single ridge. A Spanish study using scenes from a variety of locations, however, found mountains to be associated with aridity and roughness [DeLugio & Mugica, 1994].

Overall, it can be assumed that mountains have a positive influence on preferences, but this is not as emphatic as water or vegetation because mountains sometimes create negative reactions.

Go to top

Trees

The ubiquity of trees in studies is striking - out of the 227 surveys, 178 [78%] included trees. Trees are among the most familiar of content elements in landscapes and their contribution to scenic quality is generally positive. This section has examined the findings of surveys relating to the preferences for trees and vegetation.

Because foresters conducted many of the surveys, a majority focussed on how forest management can affect scenic quality. The following summarises the findings of these.

  • Anderson [1978] aimed to develop of reliable approach for assessing visual forest resources and found foresters more amenable to scenes of clear cutting, poorly stocked areas and new growth of cutover stands than were residents or students.
  • Arthur [1977] related landscape quality with various forest management treatments and found that large trees, high contrasts and heavy canopies enhanced scenic quality while the amount of slash [i.e. piles of unmarketable wood] affected it adversely.
  • Brown [1987] combined assessments of scenic beauty of pine plantations with management costs to identify efficient combinations for producing scenic beauty and the tradeoffs with timber, forage & water benefits.
  • Buhyoff and his colleagues undertook a series of studies of the influence of southern pine beetle on the scenic preferences of forest landscapes. These showed that preferences varied inversely with the proportion of visible forest damaged by beetles [Buhyoff & Leuschner, 1978], and that knowledge about beetle damage adversely affected preference ratings [Buhyoff, Leuschner & Wellman, 1979; Buhyoff & Riesenman, 1979].  A linearised logarithmic model was derived for preferences of informed subjects [Buhyoff, Leuschner & Arndt, 1980]. A separate model was derived for pines over 9 years old which found that stand age, the diameter of trees and the stocking density of trees were all positively related to scenic quality [Buhyoff et al, 1986] The study also found that scenic quality was optimal for trees of around 1100 - 1200/acre after which scenic quality decreased and also that thin-stemmed trees were regarded negatively.
  • Cook [1972] evaluated walker’s preferences for hardwood forest trees and the extent by which these accorded with timber quality and found generally a good correlation. Favoured characteristics included balanced form, straight trunk and thick crown, however crooked trunks, leaning trees and even lopsidedness were also favoured.
  • Daniel & Boster [1976] developed their Scenic Beauty Estimation method in the ponderosa forests of Arizona.  Daniel & Schroeder [1979] applied it to derive a model of scenic quality in a forest landscape, while Daniel et al [1978] used the SBE method to map the spatial scenic beauty of forest landscapes.
  • Following early efforts to assess temporal change in the scenic beauty of forests by Hull, Buhyoff & Cordell [1983], Hull & Buhyoff [1986] developed their Scenic Beauty Temporal Distribution method, based on the SBE method, to assess the effects of forest management over time. By including the stand age in their regression equation for scenic beauty, as well as tree density and size, they were able to predict the changes to scenic beauty with time. Decreasing stand density, less productive sites, and increasing stand age increases scenic beauty.
  • Schroeder & Daniel [1981] extended Arthur’s [1977] study to develop a valid and useful model for predicting scenic beauty of forest landscapes by including a range of forest mensurations of overstorey, understorey, ground cover and downed wood. The relationship between SBE values and physical forest features provided the basis for the scenic beauty prediction model. The model, derived in Arizona, was applied to another forest in Colorado and performed reasonably well.
  • Schroeder & Brown [1983] tested a range of mathematical forms of scenic beauty regression models (The forms were linear, quadratic [i.e. linear + squares], cross-product [i.e. linear + squares + products], logarithmic of main effects, and square roots of main effects) and found the nonlinear forms [i.e. log & square root] performed only slightly better than the linear forms.
  • Vining et al, [1985] evaluated landowner perceptions of hardwood forest management. They found that the amount of dead and downed wood had a strong negative influence on preferences. Clear cut areas and heavily thinned areas were the lowest in scenic preferences while the lightly thinned stands were comparable with the natural stands.

Other studies of trees that derived interesting findings included the following.

Abello et al [1986] found from their analysis a preference for images that exhibit “simultaneously greater fertility, some pattern or rhythm, and a certain structural legibility” [Ibid, 168]. The authors believed the findings supported a socio-ecological interpretation of landscape aesthetics as the dominant characteristics have survival promoting meaning. 

Kaplan, R. and Herbert [1987] assessed the preferences of students in Western Australia and Michigan for WA jarrah forests. Figure 8 summarises the findings for these students [5-point scale] and indicates a close agreement.

Kaplan

Source: Kaplan, R. & Herbert [1987]; p <.0001 except for open smooth texture & forest vista which were not significant.
Figure 8 Preferences for Australians & Americans for Jarrah Forests

Lyons [1983] examined how preferences for five biomes change with age and found that among adults, the most preferred were coniferous forest closely followed by deciduous forest [Figure 9]. Preferences for rain forest were next and the savanna and desert attracted the lowest preferences. The high ranking attributed to conifers is understandable in the northern hemisphere where they are ubiquitous, but would be unlikely in Australia and New Zealand where conifers are generally regarded as inferior to native hardwood forests [Brown, S., 1985; Kaplan R. & Herbert, 1987].

Lyons

Source: Lyons, 1983
Figure 9 Preferences for Biomes by Age

Schroeder [1991] assessed the preferences of people familiar with the Morton Arboretum near Chicago and using cluster analysis identified four groups or clusters of raters [Figure 9] Groups 1 and 2 were the largest [11 & 17 respectively] and both preferred the woods the most but group 1 had a higher preference for meadows and a lower preference for the formal and tree/lawn scenes than group 2. Groups 3 and 4 were small [3 & 1 respectively]. Group 3 liked the formal and tree/lawn scenes while group 1 particularly liked the tree/lawn scene. Thus overall there was a strong preference for natural scenes but a distinctive sector of the community who preferred the more formal scenes.

Schroeder

Source: Schroeder, 1991
Figure 10 Preferences of Groups for Arboretum Scenes

Shafer et al [1969] developed their model for predicting scenic preferences based on 100 photographs of landscapes taken throughout the US. All were photographed when the trees were in foliage. The regression equation derived [Table 3] included three factors relating to the vegetation:

  • Perimeter of immediate vegetation
  • Perimeter of distant vegetation
  • Area of intermediate vegetation

Table 3 Shafer’s Predictive Model of Landscape Preferences

Y = 184.8 – 0.54X1 – 0.09X2 + 0.02(X1 . X3) + 0.00055 (X1 .X4) – 0.0026(X3 . X5) + 0.0016(X2 . X6) – 0.008(X4 . X6) – 0.0004(X4 . X5) + 0.00067X12 + 0.00013X52

Where X1 = perimeter of near vegetation
X2 = perimeter of middle distant vegetation
X3 = perimeter of distant vegetation
X4 = area of near vegetation
X5 = area of any kind of water
X6 = area of distant nonvegetation

Note: Negative items contribute positively, while positive items contribute negatively; i.e. the lower the score the higher the landscape quality.

Shafer found that factors having a positive influence on the landscape’s aesthetic appeal were the:

    • Perimeters of near and middle distant vegetation
    • Perimeter of distant vegetation multiplied by the area of water
    • Area of middle distance vegetation multiplied by the area of distant non-vegetation
    • The area of middle distant vegetation multiplied by the area of water.

    In an early study of public preferences, Yarrow [1966] assessed the British public’s attitudes about afforestation practices of the Forestry Commission. Interestingly, he found “large majorities” in favour of afforestation of areas such as Snowdonia and the Lake District [Ibid, 62], findings that would probably be very different today. While conifers were supported for upland areas, deciduous trees were preferred for the agricultural areas. Mixtures of conifers and deciduous trees were also favoured. Deciduous trees of mixed heights and conifers of even height were favoured, possibly favouring conditioning by the Forestry Commission. Hard edges of woods were not favoured, and most preferred borders of deciduous and/or mixed trees rather than a continuous border of trees.

    Table 4 summarises the key positive and negative aspects of vegetation as derived from the studies reviewed in this section. The results suggest preferences for substantial trees with height, thickness of trunk, and breadth of canopy, trees that have a significant impact on the landscape.

    Table 4   Summary of Positive and Negative Aspects of Trees and Forest Management

    Positive Negative
    Trees
    • large trees
    • large, heavy canopies
    • thick trunks
    • straight trunks - but also crooked trunks
    • balanced form
    • older trees
    • mixture of trees
    • trees along rivers
    • foreground vegetation
    • native trees
    • deciduous trees
    Trees
    • dead trees
    • thin stemmed trees
    • small trees
    • conifers [Australia]
    • pruned shrubs
    Forests
    • stocking density - optimal 440-480/hectare
    • extent of ground cover
    • visual penetration through trees
    • native trees
    • deciduous trees
    • conifers [US]
    Trees
    • extensive slash
    • clear cuts
    • poorly stocked areas
    • new growth of cutover stands
    • beetle damaged forests
    • downed trees
    • heavily thinned areas

    On the one hand, trees are preferred with order, balance, symmetry and a tidiness about them while on the other, possessing diversity and interest provided by mixed species, crooked trunks and age. Similarly, the range of species preferred - native, deciduous, and in the US, conifers - suggest that any trees are preferred to none. Disliked are trees that lack boldness - scrawny, small, thin trees or those that have been changed artificially from a natural form by pruning.

    Forests are preferred with moderate density, not too dense but also with a spaciousness of openings and the ground cover being visible. People are more definite about what they dislike in forests - images of slash, downed trees, thinning, and especially clear cuts, destroy the illusion of a natural forest and remind the observer that the forest they are viewing is managed for economic ends.

    Go to top

    Naturalness

    Naturalness, the natural qualities of the landscape, is the most prevalent element examined by studies. It is the element underlying the specific attributes of water, trees and mountains examined separately.

    Civco [1979] assessed the natural, rural and urban landscapes of Connecticut by asking respondents to rate [7-point scale] the 32 landscape features contained in landscape photographs, features such as lakes, various types of trees, hills, shore-lines, wetlands, roads, fences, houses. Following this, they were asked to rank another set of photographs in terms of scenic quality. Figure 11 indicates the ratings of the landscape features. This indicates that the natural landscape elements were rated amongst the highest while the artificial elements were rated low.

    Civco

    Source: Civco, 1979
    Figure 11 Ratings of Landscape Features

    Dearden [1980] measured 30 landscape elements per 1 km grid square on Vancouver Island and used respondents to rate the scenic quality of these grid squares. Using regression analysis, the weights of each landscape element were derived. The ‘R’ scores are shown in Figure 12 and their size indicates the correlation between visual quality and the landscape element [Ibid, 63], with the figure indicating whether it is a positive or negative relationship. The ‘R’ value is shown here instead of the R2 to retain the positive or negative relationship.

    Dearden2

    Source: Dearden, 1980
    Figure 12 Regression “R” Values - Landscape Elements Vancouver Island

    Many of the positive features are natural elements of the landscape, the top 8 being: undeveloped coast, oak woods, rocky coastline, sandy coast, pine woods, relative relief, scattered trees and mudflats. Interestingly natural lakes and rivers were ranked lower than some artificial elements and the reasons for this are unknown. 

    Herzog [1984] examined preferences for waterscapes while Herzog [1987] assessed preferences for mountains, canyons and deserts. His findings for these were reviewed earlier and are not repeated here. Suffice to say that both of these natural landscapes produced high preference ratings.

    Hodgson and Thayer [1980] examined the effect that varying the labels on photo-graphs had on preferences. They found that labels indicating natural sites [e.g. lake] were invariably preferred to ones labelled artificial [e.g. reservoir] [Table 5].

    Table 5 Features Viewed from Road in Rockies

    Positive rated scenes Negatively rated scenes
    High mountains 80.5%
    Cliffs, capes, rocks 49.9%
    Canyons 46.3%
    Beaches 44.0%
    Waterfalls/rapids 72.9%
    Ocean 66.4%
    Swift rivers 52.5%
    Snow & glaciers 51.4%
    Evergreen forest 86.9%
    Parks & recreation 53.1%
    Harbors/waterfront 46.1%
    Deserts 44.3%
    Swamps & marshes 49%
    Scrubland 46.9%
    Billboards 78.3%
    Commercial bldgs 62.6%
    Industry & railroads 51.8%
    Suburban houses 43.2%

    Source: Jones, et al, 1976

    Jones et al [1976], surveyed the community on their enjoyment of views from a road through the Rockies in the State of Washington. Prominent among the positive features were natural scenes, while negative scenes included artificial features but also natural features such as deserts, wetlands and scrubland. This suggests that it is not simply naturalness per se which influences preferences, but also the content of the scene.

    It is notable, however, that the ratings given to these negative natural scenes were generally less than for the artificial scenes, suggesting that the feelings against them are not as strong as against, for example, billboards, industry and the like.

    In an early study, R. and S. Kaplan and J. Wendt [1972] examined the preferences for scenes of nature and of urban areas and found a distinct preference for the former [Figure 13]. They found that complexity could not account for the difference in preference values “even though higher complexity values are related to higher preference values within each group.” [Ibid, 355]. Correlations between complexity and preference for nature scenes and urban scenes were significantly correlated: r = 0.69 and r = 0.78 respectively.

     Kaplan2

    Source: Kaplan, Kaplan & Wendt, 1972 Note: Diamonds = nature scenes, squares = urban scenes
    Figure 13 Nature & Urban Scenes - Complexity vs Preference  

    In addition to the influence of mountains on landscape rating, Kane [1976, 1981] showed naturalness to be a key factor. Table 6 summarises Kane’s descriptions of the topmost 10 scenes and the bottommost 5 scenes and illustrates the strong influence of naturalness.

    Table 6 Influence of Naturalness on Rating of South Australian Landscapes

    Rank Description Checklist Score Bipolar Score
    1 Warren Gorge near Quorn, Flinders Ranges (FR)
    82
    80
    2 Eastern Wilpena Pound rim from Pound floor, FR
    81
    79
    3 Rawnsley Bluff from highway to Hawker, FR
    81
    82
    4 Aroona Valley from Aroona ruins, FR
    79
    71
    5 Parachilna Creek & Gorge, FR
    79
    77
    6 River Murray & cliffs, Memdelbuik Reserve, near Berri
    79
    77
    7 Cliffs along Murray River, Murtho Park
    79
    79
    8 Murray River & floodplain, Headings Cliff
    78
    84
    9 Seascape from above Sellicks Beach
    77
    74
    10 Mallee scrublands on dunes, Overland Corner
    77
    81
       
    41 Martins Bend Picnic Reserve, Murray River, Berri
    55
    71
    42 Oraparinna Barytes mine workings, FR
    52
    41
    43 Main street of small town of Carrieton, FR
    50
    50
    44 Small railroad station, Upper Sturt, Adelaide
    47
    52
    45 Mt Barker Road highway interchange, Crafers
    41
    42
    46 Rubbish heap near Victor Harbour
    33
    29

    Source: Kane, 1981

    Knopp, Ballman and Merriam [1979] assessed the preferences for 39 environmental elements among users in a river environment. Figure 14 indicates the top twelve rankings. Natural landscapes were the equal topmost variable and all of the other elements are aspects of the natural environment.

    Knopp

    Source: Knopp, Ballman & Merrian, 1979
    Figure 14 Preferences for Environmental Variables - River Environment

    Lamb and Purcell [1990] examined the perception of naturalness associated with differing vegetation formations found in New South Wales. Perception of naturalness increased with the height of vegetation and density of foliage cover [Figure 15]. Vegetation with dominant trees of 10 - 30 m in height was judged more natural with dense foliage than medium cover.

    Lamb1

    Source: Lamb & Purcell, 1990. Note: 3 groups were labelled Sparse [10 - 30%], Mid-dense [30 - 70%], and Dense [70 - 100%]. Tall forest/woodland > 30 m.
    Figure 15 Naturalness vs Foliage Cover & Height

    The interaction of height and density is important, not their separate contributions. Low vegetation of 2 - 5 m restrict the extent of view and offer little ‘prospect’ in Appleton’s terms and also limit legibility and mystery of the landscape in Kaplan’s terms: “restriction of mid-ground view causes reduced preference” [Ibid, 347]. Changes to the vegetation structure are detectable by respondents and are perceived to reduce its naturalness. The ability to discriminate changes increases with the vegetation density but decreases with its height.

    The reasons for structural change included grazing, fire, weeds and dereliction due to failed agriculture. Fire was not regarded negatively, which indicates the influence of familiarity with Australian biomes where fire is considered to be part of the ecosystem.

    Grazing and dereliction produced the greatest negative effects on perceived naturalness. The resultant landscapes were “relatively open, park-like, and ordered yet the perceived naturalness is low” [Ibid, 350]. Based on this, the authors suggested that preference and naturalness were not always equivalent. Lamb and Purcell considered that the study indicated the need for an adequately defined naturalness dimension:

    “the inadequate distinction between naturalness and preference may have led to a confusion between the contribution of the simple physical attributes of scenes and more complex psychological dimensions contributing to the experience of landscape.” [Ibid, 350].

    The study also showed that ecological naturalness and perceived naturalness are related but not equivalent.

    In New Zealand, Mosley [1989] found scenic attractiveness to be related to the percentage of the scene in native forest [Scenic attractiveness = 4.6 + 3.56*(% native forest)] with an r2 of 0.41.

    Urlich’s studies of the affective implications of natural landscapes reinforce the preferences for such scenes.

    Stephen and Rachel Kaplan’s The Experience of Nature: a psychological perspective [1989] focused on the preference for natural settings and reviewed a range of relevant studies. Concluding the book the Kaplans wrote: “Viewed as an essential bond between humans and other living things, the natural environment has no substitutes.” [Ibid, 202].

    The studies described in this section support the view that nature exerts a powerful influence on human landscape preferences. In most of its manifestations, whether as coast and sea, rivers and lakes, mountains and hills, trees and forest, the natural element nearly always produces positive preferences. There are, however, some natural scenes that tend to be regarded negatively and these include deserts, marshlands and scrubland. Erosion or vegetation debris in streams, though elements of the natural environment, also tend to depress preferences.

    Where the presence of humans is apparent, such as through clear felling, grazing, pollution of water, dams and structures, the preferences are affected adversely. Some studies, however, found that the natural element is in fact illusory, having been created artificially, such as in an arboretum or park. The highly preferred savanna-like scene with trees amidst grass is often artificial, either through the grazing of introduced stock or by design.

    This suggests that human preference for natural scenes is superficial, that it is concerned with the appearance, not the substance of the scene. Environments that are essentially artificial can be made to appear natural. Such an approach is, of course, well known among theme park designers; the Disneyland creations of contrived nature. Nevertheless, studies have also indicated that human observers are able to discriminate quite finely between natural scenes and those that have human influence.

    The ecological-evolutionary viewpoint argues that naturalness preferences are survival-enhancing and that humans are able to discriminate between scenes that contribute to survival and those that may be adverse.

    Go to top

    Wildlife

    One study focussed on the influence of wildlife on landscape preferences, while several other studies have examined this incidentally to their main focus. Hull & McCarthy [1988] examined the effect of wildlife as part of a study of changeable landscape features. The landscape settings were park-like forests, rivers and grassy fields near Melbourne, Australia. Scenes were photographed with and without wildlife, the wildlife comprising kangaroos, wallabies, deer and swans. The study found that these wildlife had a positive, statistically significant but moderate effect,  accounting for less than 10% of the total scenic score [using SBE method]. Wildlife tended to have a greater influence on the less attractive scenes. An expectation of seeing wildlife [e.g. by “wildlife feeding area” signs] can significantly enhance the scenic scores.

    Nassauer & Benner [1984], found that the presence of birds, porpoises and animals were positive features in the Louisiana oil and gas development area, though of relatively small importance overall [e.g. factor loadings 0.145 - 0.190 in one of the factors identified compared with 0.400 for a small bay and 0.643 for a sand bar or oyster bar].

    Schroeder [1991], in his study of the Chicago arboretum, found the presence of birds was one of the frequently mentioned features of the garden.

    These studies, though limited, indicate that wildlife has a positive, if minor, influence on landscape preferences.

    Sky

    The sky generally comprises a substantial area of a photographed scene and it is unrealistic to assume it has no influence on preferences. Lighting or cloud features can generate interest in an otherwise mediocre scene and can enhance colours and the clarity of the view. Mist, on the other hand can engender a sense of mystery that the Kaplans assert is an important factor in preferences. In flat terrain, the sky’s features tend to be more noticeable than in hilly or mountainous landscapes.

    Anderson, Zube and MacConnell [1976] described aspects of a major landscape study in Connecticut that included a cloud cover index [the proportion of sky covered by clouds] and an atmospheric clarity index [the amount of haze, smog etc in the scenes]. Scenic resource values [SRV] of 1.92 and 1.53, respectively, were obtained for the two indexes, compared with 1.09 for land use diversity, 5.63 for a naturalness index and 2.05 for a height contrast. The two indexes were correlated with other features giving a correlation of - 0.55 for the cloud index with the SRV while the atmospheric clarity correlation with SRV was 0.30. Both were low and their elimination had little effect on the regression strength.

    Arthur [1977] examined the scenic beauty of forest environments and included measurements of clouds. She found that most observers responded favourably to clouds  [ß coeff = .78].

    Hammitt et al [1994] assessed visual preferences along the Blue Ridge Parkway in the Appalachian Mountains, landscapes with thick vegetation and rolling ridges. They found the area of sky in a scene was the most important predictor, though negative, in the equation. Their interpretation was that the area of sky is a surrogate for other features, specifically the absence of attractive features such as ridges, rolling plateau and water. Although they suggest that the lack of visual complexity and involvement [in the Kaplans’ terms] of the sky support their hypothesis, it does not appear to provide sufficient explanation. The absence of other features is not something that observers would generally notice.

    Hammitt et al note in passing that the area has considerable haze in summer when the photographs were taken and by reducing the comprehension of the landscape, its value is diminished. This is supported by Herzog & Bosley [1992] who suggested that:

     “... mistiness in a mountain setting reduces both tranquility and preference but the reduction in preference is greater... Mist reduces the ability to comprehend an environment, and in the Kaplans’ scheme any such reduction would be especially damaging to preferences.” [Ibid, 125].

    Mist and haze have a similar effect of reducing the clarity of the scene and its understandability.

    In their study of the landscape encountered while hiking, Hull and Stewart [1995] found that clouds comprised only about 0.9% of the view, a surprisingly low figure [compared with 23.8% for ground, 14% for vegetation and 20% for mountains and valley].

    Shafer’s predictive model of landscape preferences included a measurement of the sky zone area and perimeter and, although their factor loadings were significant, these were not included in the final equation. The factor loadings were equal or greater than for many other features measured:

    Perimeter sky
    -0.67
    Area sky
    -0.83
    Tonal variation sky
    -0.78
    Total count sky
    -0.85
    Perimeter immediate vegetation
    -0.94
    Perimeter distant vegetation
    +0.89
    Perimeter stream 
    -0.83
    Area immediate vegetation
    -0.92
    Tonal count land
    +0.56
    Tonal count water 
    -0.56

    Note: Negative quantities are more attractive

    No explanation of the omission of the sky factors was given.

    Overall, the sky plays literally a backdrop role in landscapes and appears to be generally disassociated from the landscape itself. There has been insufficient study of the effect of clouds, lighting and other atmospheric features to be definitive about their significance, though from the studies cited they appear to be less significant than might be imagined.

    Go to top

    Sound

    In a series of experiments, Anderson et al [1983] evaluated the effect that different sounds had on preferences for outdoor settings. The sounds used were songbirds, crickets, wind in trees, dog barking, children yelling and laughing, sounds of farm animals, traffic, aircraft and power lawnmower. Each was used for about 20 seconds in a field test and 10 seconds in a slide and sound test. The field sites included a hardwood forest and a downtown street.

    Figure 16 illustrates the results of one of the experiments that indicate the effect of different sounds on a botanical site with the hardwood forest.

    Anderson

    Source: Anderson et al, 1983
    Figure 16 Effect of Sounds on Botanical Garden Setting

     The results of this and other experiments led the researchers to conclude that:

    “There is an interaction between acoustic and other features of a setting that modifies the effect of different sounds in determining the quality of the setting. Sounds that, in the abstract might be regarded as enhancing did improve wooded, natural, and heavily vegetated urban settings, but not the downtown and other mostly built sites. Detracting sounds affected the tree-covered sites, but not the highly developed locations.” [Ibid, 561].

    They also found that vegetation creates a higher expectation of environmental quality so that sounds from human sources, including jets, dogs, etc can have a greater detracting effect.

    This limited research indicates that sounds can influence aesthetic preferences, either enhancing or diminishing them. 

    Hetherington et al [1993] extended the work of Brown & Daniel [1991] in using videos to examine preferences for wild rivers to the use of sound together with videos and slides. The study found [Figure 17] that the function curve for the video with sound condition was similar [r = 0.98] to that found by Brown & Daniel (see Figure 6) while that for the no-sound condition appear “more linear than polynomial functions” [Ibid, 289].

    Hetherington

    Source: Hetherington, 1993
    Figure 17 Influence of Sound on Scenic Ratings

    The results indicated that both sound and motion affect judgements of scenic beauty while the results for motion without sound are similar to a static image.

    Go to top

    Landscape Component Preferences – Conclusions

    This section examined the preferences for key components of the landscape.

    Water generally has a significant positive influence on preferences, particularly when in the form of moving water, lakes, edge of water; pollution and colouration of water can diminish its attraction. While open areas of water yield positive responses, swamps and wetlands often provoke negative responses.

    Mountains generally have a positive influence on preferences and while some studies found mountains ahead of water in terms of preference, this was not as consistent as for water or vegetation. In some circumstances mountain scenes provoke negative reactions [e.g. being associated with aridity or roughness]. Most studies tend to treat mountains as backdrops to the landscape, rather than the most dominant object, and few studies derived preferences for the mountain components per se.

    Trees, whether singly or in forests, generally have a positive influence on preferences. Preferred are substantial trees with height, thickness of trunk, and breadth of canopy - trees that have a significant impact on the landscape. Trees are preferred with order, balance, symmetry and tidiness about them but trees possessing the diversity provided by mixed species, crooked trunks and age also enhance preferences. Similarly the range of species preferred - native, deciduous, and in the US, conifers - suggest that any trees are preferred to none. Disliked are trees that lack boldness - scrawny, small, thin trees or those that have been changed artificially from a natural form by pruning.

    Forests are preferred with moderate density, not too dense but with a spaciousness of openings and the ground cover being visible. People are more definite about what they dislike in forests - images of slash, downed trees, thinning, and especially clear-cuts destroy the illusion of a natural forest and remind the observer that the forest being viewed is managed for economic ends.

    Naturalness, the natural qualities of the landscape, underlies the specific attributes of water, trees and mountains. There is no doubt that naturalness exerts a powerful influence on human landscape preferences. Whether present as coast and sea, rivers and lakes, mountains and hills, trees and forest, the natural element nearly always produces positive preferences. Some natural scenes are regarded negatively including deserts, marshlands, scrubland and erosion or debris in streams. Human influences in otherwise natural scenes also affect preferences adversely. Paradoxically some artificially created landscapes, such as parklike scenes, can be highly preferred. The naturalness preference may therefore only be superficial, focussing solely on the appearance, not the substance of the scene.

    Wildlife appears to have a positive though minor influence on landscape preferences.

    Skies provide a backdrop to the landscape and are generally disassociated from the landscape itself. Although there have been only a few studies of the influence of clouds, lighting and other atmospheric conditions, they appear to have less influence on preferences than might be imagined.

    Sound and motion appear to have a positive influence on preferences in scenes involving moving water.

    Overall the key component affecting landscape preferences is naturalness, as expressed by the specific content of water, trees and mountains. Water, in most of its forms [lakes, rivers, sea] contributes a positive influence on preferences, the main exceptions being swamps and wetlands, and polluted or coloured water. Mountains also generally have a positive influence on preferences though, surprisingly, not as consistently as for water or vegetation. Mountains tend to be treated as a backdrop to a scene rather than as the dominant landscape element.

    Trees that make a significant impression on the landscape by virtue of their height, thickness of trunk, and breadth of canopy are preferred, but so too are trees that have an order, balance, symmetry and a tidiness about them. Trees possessing diversity provided by mixed species, crooked trunks and age also enhance preferences. In the US, native, deciduous and conifers enhance preferences, however in Australia where conifers are non-native, they detract from preferences. Trees that are scrawny, small and thin lower preferences, as do trees that have been pruned into an artificial shape. Forests with openings and ground cover are preferred over dense dark forests, and signs of human forest management are disliked.

    These features - water, trees and mountains are attributes of naturalness, the most powerful influence on landscape preferences. Apart from deserts, marshlands and scrub, natural scenes almost invariably enhance landscape preferences. Scenes which have been artificially created but which appear natural, such as park-like landscapes, are also preferred, suggesting that the naturalism preference is only superficial, focussing on appearance not substance.

    Arising from this review of landscape studies of the 20th century, several further conclusions can be drawn:

    Most of the studies have focussed on analysing the contribution of specific attributes to landscape preferences and have isolated these from other factors in order to assess their contribution. This is an essential stage in research - to understand the influence of each factor in isolation through controlling the effect of other factors. However, this means that the synergistic effects of factors tend to be overlooked or minimised. It means the approach is essentially reductionist, not holistic, although a number of studies used multiple regression with high levels of explanatory capability for the given landscape.

    Relatively few studies have examined the assessment of landscapes in regions. Being focussed on the contribution of given factors, their emphasis tends not to have been on the assessment of regional landscapes. While equations were derived for some, few go to the next stage: mapping regional landscape quality based on this information.

    Relatively few studies have been undertaken in Australia. Lamb & Purcell’s studies, Williamson & Chalmers’ assessment of NE Victoria, Kane’s studies of South Australian landscape, Hull & McCarthy’s work on the influence of wildlife, and the Kaplan & Herbert study on Western Australian forests are the sum total of studies in Australia.

    The studies covered indicate that a considerable amount of headway has been achieved over the past 20 to 30 years in developing an understanding of what it is in landscapes that people appreciate. The next step is to apply this knowledge in the appraisal of landscapes.

    References cited may be found here.

    Go to top

 

 

 

 

 
 

© 2009-14 Scenic Solutions   Privacy   Terms of Service   Site Map   Contact us