phillip S. Kane
Applied Geography (1981), 1,77-96
Department of Geography, California State University, Northridge 91330, USA
Abstract
Techniques for assessing landscape attractiveness are becoming increasingly important in environmental planning. They are a manifestation of the growing need to monitor landscape deterioration, to help preserve natural beauty, to learn about our cultural perceptions, and to satisfy an ever-increasing bodv of environmental law.
This report describes two such evaluation techniques developed by the author and used by the National Trust of South Australia to classify and record natural and rural landscapes. One method, a component checklist, relies on a detailed appraisal of individual landscape elements. The other method, using bipolar semantic differentials measures 'whole-landscape' emotional responses. Although very different in design, and in how viewer preferences are ascertained, the two techniques provide vista scores that correlate very highly (Sr = 0.81 for n = 46). For either method, scenes whose scores are more than 6 points apart can be regarded as significantly different in aesthetic qualities. As with all present evaluation methods, there are still problems with both technique design and the transformation of vistapoint data to spatial (mappable) information. Nevertheless, both the checklist and the bipolar methods seem to be reliable and face-valid tools for measuring South Australian landscape preferences.
Techniques for quantifying landscape attractiveness are becoming increasingly important in environmental planning. The purpose of this paper is to describe the results of two such techniques developed by the author for use by the National Trust of South Australia. One --thod centres around a landscape component checklist; the other is a sequence of bipolar ... semantic differentials. Although quite ditterent in design and administration, the two techniques have proved reasonably successful for appraising vistas in a wide variety of landscapes and seascapes.
Why evaluate landscape?
Why evaluate landscapes in terms of their scenic and aesthetic qualities? The answer to this question should be dealt with first, before addressing the more particular matters of technique design and application.
There can be little doubt nowadays that natural and rural landscapes are being degraded by the economic pursuits upon which our modern societies rest. Because of these pursuits ... some of the integrity of the landscape as a continuum is disturbed and some of the varicts offers is lost. With loss of variety we become less able to provide for the varied expectations of users. We are adept at replacing quality with mediocrity (Litton 1972: 262).
Urbanization. highwav development, surface mining. the siting of dams, the layout of utility corridors, the building of power plants-all of these activities go on without much regard for the landscape's potential as a visual resource. This degradation is not only sad, it is also probably dangerous. Scenic landscapes should be regarded as an important natural resource, just as vital as timber, water, soil. mineral ores and fossil fuels. In fact the preservation of diverse natural and rural landscapes may turn out to be a necessity for mentally, stable and physically healthy societies. How else are we humans to judge how far we have come, what changes we have effected, and in which direction we are heading?
Without great expectations or justification we have often acted to preserve, protect, and even create beautiful and unique landscapes. Numerous national parks, green belts, parklands, wilderness areas. and scenic preserves have already been set aside. But this is not enough because such preservation has been, for the most part, largely unplanned and irrational.
For about the last decade, landscape evaluators have been attempting to replace an unplanned and irrational framework with a more structured one. Both the theoretical and the applied aspects of the field have become areas of research for many planners, academics, recreationists, resource managers, architects and environmentalists. To be sure, a theoretical vacuum is still very much with us (Appleton 1975), but the relationship between observer and landscape has already intrigued many (see Lowenthal 1961; 1967; Fines 1968; Litton 1972; Bagley et al. 1973; Tuan 1974; Cunningham 1975; Unwin 1975). The applied part of the field has also attracted numerous researchers; landscape assessment methods, in fact, now number well over 1 50 (Arthur and Boster 1976, Dearden 1977), and many attempts have been made in England, Scotland, Australia and the United States to incorporate some of these methods into various planning and research programs. (Reviews and discussions of some of these applications can be found in Clark 1968; Lovejoy 1973; Zube 1973b; Penning-Rowsell 1975; Turner 1975; Elsner and Smardon 1 979.)
Although several critical problems in methodology remain, the evaluation techniques keep coming. The need for them is great. Just what are the applications, and why, in a practical sense, evaluate landscape at all? The following reasons and needs are suggested:
1. to help establish priority lists of sites and regions that should be preserved as part of our natural heritage;
2. to provide a means of aesthetically comparing sites and regions so that, if desired, human impact can be used to advantage or guided into the least attractive areas;
3. to help monitor deterioration of landscape qualitv for specifip places, by means of periodic evaluations;
4. to provide a means of carrying out 'before and after' studies in order to gauge the impact of particular kinds of human activities and alterations;
5. to define and isolate the perceptual factors and phvsical-iandscape components that are important in environmental perception and. if desirable or necessary, to be able to itemize why a, a particular landscape is or is not aesthetically pleasing.,
6. to collect data on landscape preferences from different cultures and from diverse subpopulations (e.g. male/female, young/old, travelled/untravelled) so as to better understand technique theory, the working of our senses, the differences between various societal groups, and the biases of our cultures;
7. to satisfy a growing body of environmental law in many countries-in the United States, for example:
... all agencies of the Federal Government shall... identify and develop methods and procedures which will insure that presently unquantified environmental amenities may be given appropriate considerations (NEPA 1969: See. 102b).
The needs, then, for methods of assessing landscape amenities are many; and it is more likely than not that these needs will become even more pressing as time goes on. Environmental planning at various levels of responsibility is only just beginning to develop systematic methods of including aesthetic considerations in the planning process.
The techniques
In the mid-1970s it was the decision of the Australian Council of National Trusts that State Branches should pursue the matter of evaluating and classifying landscape, in a manner similar to that in which they evaluate and classify historic buildings. The ultimate aim of assessing the landscapes would be to help protect those which represented the best of the national heritage. In South Australia this invitation was taken up by the Trust's Nature Preservation Committee, with the writer as the consultant in charge of technique development.
Four techniques were developed by the present writer and finally used by the Nature Preservation Committee. Two of them (the most successful ones) form the basis of this paper; they include (a) a component checklist, which records viewer appraisals of scene components, and (b) a bipolar semantic differential list which records viewer 'emotional responses' to entire scenes. Because both methods are based on sampled South Australian preferences, they cannot be expected to provide reliable scores if used for another cultural population or in localities other than South Australia. The techniques are certainly adaptable to any group and area, but the forms, as finalized here with their particular landscape components and adjective-pairs, are not.
The checklist and bipolar evaluation methods are designed to rest, in part, on three tenets:
1. Public preferences-not professional evaluations-can and should serve to determine what constitutes aesthetically attractive landscape (Shafer et al. 1969; Penning-Rowsell 1973, 1975; Zube 1973a; Appleton 1975; Kane 1976).
2. Views and sites should be evaluated by a group of people and by more than one technique so as to keep a check on score reliabilities (Kane 1976);
3. Some techniques can be designed to use either on-site evaluations or evaluations from photographs (Coughlin and Goldstein 1970; Shafer and Richards 1974; Zube et al. 1974; Kane 1976). Several other important concepts and problems were also considered, including technique score-reliabihty, test-retest reliability, and the role of professional esthetic standards (see Osgood 1957; Fines 1968; Linton 1968; Wright 1974; Turner 1975).
The component checklist
The checklist is designed to use the mean response of a sample of at least 10 people as a measure of a view's attractiveness. Because of the nature of some of the selected components, responses must be elicited on site. The view score obtained ranges between 0 Points (least attractive) and 100 points (most attractive). Two views that differ in scoring Y more than 6 points can be regarded as statistically significantly different at the 0.05 level. (This is based on Student's t tests with groups of at least 10 people evaluating 26 views on-site. The mean standard deviation for a particular view was 8.5 points for the checklist and 7.5 points for the bipolar list.) Two types of forms are used to derive a view's checklist score. The first form (Fig.1) is used by a respondent to record his or her evaluations of all of the perceived components of the scene. The respondent does this by first reading through the checklist and indicating all components that he or she thinks are present, and secondly, by rating the I quality contribution' of each one on a scale from -2 to +2 points. Components are grouped into six categories: landforms (coastal or non-coastal), water, vegetation cover, human impact, and compositional factors.
The second form (Fig. 2) is used to determine individual scores and a mean score for the view. As can be seen from the formula on this form (see step 4), respondents' scores are based on the number of items checked, the value of the i'tesms checked, and the perceived diversity of the scene.
The major Steps involved in the development of the checklist began with the collection and examination of previously Published evaluation lists. These included Weddle (1969), the West Midlands Regional Study (see Lovejoy 1973), and especially the Countryside Commission of Scotland (1971). This was followed by a period of on-site group testing in which several 'forced rank-ordered (A forced rank order is the rank order obtained when. respondents are asked point-blank (that is, forced) to rank the elements in a series without havulg to explain their choices. For landscape scenes, such a forced rank ordering can be accepted as valid for a population because it is, in fact, I most truthful' ordering of the series there can be. Evaluation techniques which can closely duplicate this forced ordering are said to be 'face valid' ) Scenes were compared with scores frorn experimental checklists. Further testing and refinements led to the arrangement shown here, which appeared to be able to not only duplicate the face-valid forced ran k ordering of views, but also to realistically score them. In final test-runs involvin 6-10 scenes, the Kendall coefficient of concordance between checklist ranks and for g ced ranks averaged 0.74.
Component checklists in general suffer from some important problems: components whose correlation with landscape aesthetics is actually unproved, the equal weighting of the components, and the assumption that their values are additive. Nevertheless, the checklist method described here has provided one way to measure South Australian landscape preferences.
The bipolar list
Like the checklist, the bipolar sequence is designed to use the mean response of at least 10 people as a measuer of that Views attractiveness. Responses can beeliceed either on-site or from colour slides (the use 's attractiveness. A view's score can range up to 100 points, and two scenes hat ediffer in score by more than 5 points are, significantly different at the 0.05 level of probability (see footnote to p. 79).
Two types of forms are employed to calculate a bipolar score. The first one (Fig. 3) is used by each respondent to record his or her emotional responses to the scene being evaluated. The form consists of 21 adjective-pairs, 14 of which seem significant to South Australians (see below). The other seven bipolars are decoys placed in the form to help prevent the viewer from detecting any kind of pattern in his or her selections. The 6 arrangement of all 21 pairs-some with their 'attractive end' at the left, others with their attractive end' at the right-is also designed to help hinder patterned responses.
Mean responses for each of the 14 significant bipolars are recorded on the second form (Fig. 4), and these in turn are used to calculate the vista score, Weighting factors take into account both the relative importance of each adjective-pair and which end of each bipolar correlates with 'attractive landscape'.
Bipolar scales have been used previously to explore and deal with psychological concepts (Osgood 1957) and, in landscape work, to compare groups of people (e.g. Zube et al. 1974) or to compare responses to different media (such as on-site versus colour slide responses, as in Shafer and Richards 1974). Their use as a tool in ac ual landscape appraisals has been very restricted so far (Kane 1976; Echelberger 1979), atreflection, one an guess, of the fact that the 'real meaning' of the adjective-pairs and' the factors they enerate is not clear. Even without knowing the deeper meanings of the bipolar semantic differentials, however, it is still possible to incorporate them into an evaluation tool. One does not necessarily need to know what a meter is in order to measure with it.
The main research problem in developing a list for evaluation purposes was identifying significant adjectives from a starting-pool of about 40 pairs. This was accomplished in the following steps: (a) Kolmogorov-Smirnov tests to eliminate non-significant pairs; (* A significant pair was considered to be one whose mean scaler value for a particular view differed at the 0.01 level more than 12 per cent of the time when compared both to the next most attractive and the next least attractive views. The views used consisted of the 10 scenes described in Fig. 5; their forced rank ordering had already been determined and was considered to be face valid. The 14 significant pairs accounted for 45 per cent of the variance in the responses).
(b) determination of correlation coefficients between significant pairs and the pairs 'beautiful/ugly' and 'like/dislike'-this yielded which word in a pair corresponded with the quality of attractiveness; (c) factor analysis (principal components analysis), in order to assign weight to the significant pairs-weights, when rounded off, ranged from 1-5.
By means of these steps a list was finally produced which contained 14 adjective-pairs judged to be significant to South Australians when viewing landscapes. Test runs involving 6-10 scenes yielded Kendall coefficients of concordance between bipolar rank orders and forced rank orders of around 0.93. During these on-site testings, colour slides were taken for the purpose of determining the possibility of using such pictures for actual appraisals. These slides were evaluated by other groups at a later date (using the same bipolar form), and it was established that the correlation between on-site and picture scores was very high (Pearson's r=0.96). This produced the regression line shown in Fig. 5. Notice that although the correlation is very good, slide scores are less extreme than scores obtained directly in the field.
The two techniques compared
Since their research, design and implementation in 1976, the component checklist and the bipolar sequence have been used by the Nature Preservation Committee to help classify several landscape vistas. The results of four trips made by the Committee for the purpose of evaluations show a well-estabhshed correlation between scores obtained by the two different methods for the same scenes. Types of landscape evaluated by both techniques on these trips include mountains (the Flinders Ranges), semi-arid fluvial landscapes (the Riverland), coastal scenery (the Eyre Peninsula), and rural and urban features (in the local Adelaide area and in the Flinders foothills). Figure 6 shows a plotting of checklist and bipolar scores from 46 scenes, selected from these different types of landscapes. A listing of scores, ranks, and scene descriptions for these localities is given in Table 1. Referring to the regression line in Fig. 6, Spearman's rank correlation coefficient is very highly significant (S, = 0.81). and it is not unreasonable to conclude that these two evalutation methods will reliably rank order and quite satisfactorily score the landscape preferences of South Australians. Considering the rather difterent 'tasks' the respondents are asked to do for each method, this high association is striking. (Examples of checklist and bipolar scores for selected scenes are given in Plates 1-8, see pp. 91-94).
In application, evaluators have found it practical to average together the checklist and bipolar values to derive a single score for each scene. Based upon the available results, the National Trust of South Australia's Nature Preservation Committee has suggested the following breakdown of these averaged scores for their classification programme (McBriar 1977):
1. Scores of 70 and higher: classified scenes (i.e. those parts of the physical environment, both natural and man-made, that in the Trust's opinion are essential to the heritage of Australia, and must be preserved).
2. Scores between 69 and 60: recorded scenes (i.e. those parts of the physical environment, both natural and man-made, that contribute to the heritage of Australia, that should be recorded, and whose preservation is encouraged).
3. Scores less than 60: too low to record.
The problem of data transformation
Field-based evaluation techniques are designed to provide vista scores-not data directly mappable in any traditional way. The problem of data transformation whereby isopleth maps of landscape attractiveness are produced from 'point data' still remains unsolved. Of course, some techniques already available are purportedly designed to produce regional evaluations or even detailed maps, but both the selection of field-sampling sites as well as the mapping procedures have not been altogether explained or thought out. Problems in the selection of sampling sites, for example, have engaged several researchers (see Fines et al. 1969; Shafer et al 1969; Shafer and Rutherford 1969; Leopold 1969; Coughlin and Goldstein 1970; Litton 1972; Wright 1974; Daniel and Boster 1976). Examples of attractiveness maps' or attention to mapping problems can be found in Fines (1968), Linton (1968), Brancher (1969), Coventry-Solihull-Warwickshire project (1971), Crofts and Cooke (1974), Wright (1974), Unwin (1975), Daniel and Boster (1976), and in several contributors to Eisner and Srnardon (1979).
The component checklist and the bipolar sequence described in these pages were intended to be used to produce reliable view scores. To this end they appear to be very satisfactory. They may perhaps be adaptable to regional (map) evaluations by treating small tracts or physiographic entities as mapping units. But this possible utilization nee much more investigation before becoming applied research.
Conclusion
The component checklist method is the more analytical and 'objective' of the two techniques described here. It provides not only attractiveness scores but also details the 'landscape reasons' why a certain score is what it is. The bipolar method provides less meaningful information (at least so far), but because it can be used with colour slides and because the correlation between the two techniques is so high, the bipolar method can be employed in place of the checklist if taking evaluation groups into the field for on-site work is not practical.
Because of the purpose of the Trust's work (to identify classified scenes), neither the classified nor the unrecorded scores were subdivided into smaller categories. Depending upon one's objectives, though, many planning situations can be envisaged in which it would be useful and valuable to have, for example, an 'especially attractive' category (scenes scoring 85 or higher), as well as various degrees of unattractiveness.
Although very different in design and in how viewer preferences are ascertained, the two techniques described in this report provide vista scores that correlate very highly. To be sure, some problems remain in application and interpretation, but this significant correlation suggests that the two methods are reliable and face-valid tools for measuring the landscape preferences of a particular population.
Plates tables and figures
Figure 1 Part 1 Part 2
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Plate 1 Rawnsley Bluff from the highway to Hawker, (checklist) and 82 (bipolar). Rank order of 46 scenes: 2J (checklist) and 2 (bipolar).
Plate 2 Parachilna Gorge and Creek, Flinders Ranges, South Australia. Score: 79 (checklist) and 77 (bipolar). Rank order of 46 scenes: 5.5 (checklist) and I 1.5(bipolar).
Plate 3 Hillside with abandoned homestead, Upper Sturt, Adelaide, South Australia. Score: 74 (checklist) and 73 (bipolar). Rank order of 46 scenes: 16 (checklist) and 19.5 (bipolar).
Plate 4 View down a country road towards the Mt. Remarkable area, from the Melrose-Port Augusta road, South Australia. Score: 68 (checklist) and 71 (bipolar). Rank order of 46 scenes: 28 (checklist) and 25 (bipolar).
Plate 5 View towards downtown Adelaide from the suburb of Springfield, South Australia. Score: 64 (checklist) and 69 (bipolar). Rank order of 46 scenes: 33.5 (checklist) and 31 (bipolar).
Plate 6 The town of Clare, Flinders Ranges, South Australia. Score: 63 (checklist) and 64 (bipolar). Rank order of 46 scenes: 355 (checklist) and 37.5 (bipolar).
Plate 7 The Mt. Barker Road interchange at Crafers, Adelaide, South Australia. Score: 41 (checklist) and 42 (bipolar). Rank order of 46 scenes: 45 (checklist) and 44 (bipolar).
Plate 8 A rubbish heap near Victor Harbour, South Australia. Score: 33 (checklist) and 29 (bipolar). Rank order of 46 scenes: 46 (checklist) and 46 (bipolar).
References
Appleton, Jay ( 1975) Landscape evaluation-the theoretical vacuum. Transactions of the Institute of British Geographers 66, November, 120-123.
Arthur, L. M. and Boster, R. S. (1976) Measuring scenic beauty-a selected annotated
bibliograp&v. Fort Collins, Co.: US Forest Service, General Technical Report RM 25.
Bagley. M. D., Kroll, C. A. and Clark, Kristin (1973) Aesthetics in ent,ironitienial planning.
Washington, D.C.: US Environmental Protection Agency, Report EPA-600/5-73-009.
Brancher, D. M. (1969) Critique of K. D. Fines: landscape evaluation-a resource project in East Essex. Regional Studies 3, 91-92.
Clark, S. B. K. (1968) Landscape survey and analysis on a national basis. Planning Outlook 4,
15-29.
coughlin, R. E. and Goldstein, K. (1970) The extent of agreement among obsert.ers on ent,ironnzetital attractit,eness. Philadelphia: Regional Sciences Research Institute, Discussion Paper 37.
Countryside Commission for Scotland (1971) A planning classification of Scottish landscape
resources. Countryside Commission for Scotland, Occasional Paper. Coventry-Solihull-Warwickshire Sub-regional Study Group (1 97 1) A strategy for the sub-region.
Coventry, Solihull and Warwickshire councils (England).
Crofts, R. S. and Cooke, R. U. (1974) Landscape et.aluation-a comparison of techniques. London:
University of London, Department of Geography, Occasional Paper 25.
Cunningham, F. F. (1 975) The human eye and the landscape. Landscape 20(1), 14-19.
Daniel. T. C. and Boster, R. S. (1976) Measuring landscape esthetics-the scenic beautj, estitnation
method. Fort Collins, Co.: US Forest Service, Research Paper RM-67.
Dearden, Philip (1977) Landscape aesthetics, an annotated bibliography,. Victoria: University of
Victoria.
Echelberger, H. E. (1979) The semantic differential in landscape research. Proceeding of our national landscape-a conference on applied techniques for analysis and management of the visual resource, (G. H. Elsner and R. C. Smardon, eds), pp. 52 4-531. Berkeley, Calif.: US Forest Service, General Technical Report PSW-35.
Elsner. G. H. and Smardon, R. C. (1979) Proceeding of our national landscapes conference on applied techniques for anal ' ;,sis and management of the visual resource. Berkeley, Calif.:
US Forest Service, General Technical Report PSW-35.
Fines, K. D. (1968) Landscape evaluation-a resource project in East Essex. Regional Studies 2,
41-55.
Kane. Phillip S. (1976) Evaluating landscape attractiveness - a review of problems and methods, and a technique developed for the National Trust of South Australia. Adelaide: National Trust of South Australia.
Lcopold, L. B. (1969) Quantitatit,e comparison of some aesthetic factors among rivers. Washington, D.C.: US Geological Survey, Circular 620.
Linton, D. L. (1968) The assessment of scenery as a natural resource. Scottish Geographical Magazine 84, 219-238.
Litton, R. B., Jr (1972) Aesthetic dimensions of landscape. In Natural environments (J. V. Krutilia, ed.), pp. 262-29 1. Washington, D.C.: Resources for the Future.
Lovejoy, Derek (ed.) (1973) Latiduse and landuse planning. London: Leonard Hill Books /International Textbook Company.
Lowenthal, David (I961) Geography, experience, and imagination-towards a geographical
epistemology.AssocialionofAmericanGeographersAnnals51,241-260.
Lowenthal, David (1967) Environmentalperception and behat,ior. Chicago: University of Chicago,
Department of Geography, Resource Paper, No. 109.
McBriar, E. M. (ed.) (1 977) Landscape studil. Adelaide: National Trust of South Australia.
Osgood, C. E. (1957) The measurement of meaning. Urbana, Ill.: University of Illinois Press.
Penning-Rowsell, E. C. (1973) Alternative approaches to landscape appraisal and evaluation.
Enfield: Planning Research Group, Middlesex Polytechnic at Enfield, Report, No. I 1.
Penning-Rowsell, E. C. (1975) Constraints on the application of landscape evaluations. Transactions of the Institute of British Geographers 66, November, 149-155.
Shafer, E. L., Hamilton, J. F., Jr and Schmidt, E. A. (1969) Natural landscape preferences-a predictive model. Journal of Leisure Research 1, 1-19.
Shafer, E. L. and Richards, T. A. (1974) A comparison of viewer reactions to outdoor scenes and photographs of those scenes. In Psychology and the built environment (D. Canter and T. Lee, eds), pp. 71-79. London: Architectural Press.
Shafer, E. L. and Rutherford, W. (1969) Selection cuts increased natural beauty in two Adirondack forest stands. Journal of Forestry 67, 415-419.
Tuan, Yi-Fu (1974) Topophilia. Englewood Cliffs. N.J.: Prentice-Hall.
Turner, J. R. (1975) Applicdtions of landscape evaluation-a planner's view. Transactions of the
Institute of British Geographers 66, November, 15 6-16 1.
Unwin, K. K. (1975) The relationship of observer and landscape in landscape evaluation.
Transactions of the Institute of British Geographers 66, November, 130-134.
Weddle, A. E. (1969) Techniques in landscape planning. Journal of the Town Planning Institute 55,
387-398.
Wright, G. (1974) Appraisal of visual landscape qualities in a region selected for accelerated growth. Landscape Planning 1, 307-327.
Zube, E. H. (1973a) Rating everyday rural landscapes of the northeastern United States.
Landscape Architecture 63, 370-375.
Zube, E. H. (1973b) Scenery as a natural resource. Landscape Architecture 63, 126-132.
Zube, E. H., Pitt, D. G. and Anderson, T. W. (1974) Perception and measurement of scenic resources in the southern Connecticut River Valley. Amherst: University of Massachusetts, Institute fly Man and Aids
Click to return to GG209 practical
(
