new banner

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


Given the differences between photographs and field observations, it is not surprising that Carlson states “It goes without saying that photographs are not landscapes and landscapes are not photographs” (1977). Some surveys have sought to answer the question, how adequately do photographs represent landscapes? These are summarised below.

Brown et al (1988) found that scenic ratings taken directly in campgrounds were consistently higher than ratings based on colour photographs of the same areas (Figure 1). T-tests for each of the samples indicated that the direct ratings were all significantly higher than the photo-based ratings (p < 0.001). A second test, undertaken the following year and using ranking of scenes instead of rating, derived similar results.


Source: Brown et al, 1988
Figure 1   Comparison of Direct vs. Photo-based Ratings

Coughlin and Goldstein [1970] found that the ratings of field observers correlated with photograph-based assessments (r = 0.64, p < 0.001); however the field test used only two observers while the photograph-based test used eleven. Based on this limited study they ambitiously claimed that “there is no reason to suppose that ratings of actual environments in the field would be substantially different from those of photographs.” The study also found that photographs taken in one place but in different directions tended to receive similar ratings, suggesting that a single photograph could represent a locality.

Zube et al (1975) reported on a series of studies including the responses from field vs surrogate assessments. Using a range of techniques (semantic scales, rank order and Q-sort) and groups of field and non-field populations, they found high correlations between field and non-field assessments. Comparing the field and non-field evaluations for eight views, the average R2 was 0.92. ANOVA indicated that only one of the eight views had a significant difference (p = 0.05), a view with an R2 of 0.68. The findings were, they stated, “generally impressive, at least in reference to the use of color, wide-angle photography for assessing scenic resource values.”

Daniel and Boster (1976) used their Scenic Beauty Estimation (SBE) method to compare results produced by on-site vs slide judgements of forest landscapes. The SBEs derived from on-site judgements were generally slightly lower([i.e. the scenes were judged to be of higher quality) than those derived from slide judgements. The correlation coefficients were highly significant statistically.

Dunn (1976) included an evaluation of the effectiveness of photographs at representing landscapes as part of an assessment of landscape preferences. He asked visitors to six parks near Birmingham, England to indicate how the site compared with its depiction in the photograph. Many respondents were unable to answer this question due to the difficulty they encountered in comparing a single photograph with the “three-dimensional reality of the interview site.” Table 1 compares the means obtained for the on-site and photographic evaluations.

Table 1   Comparison of On-site and Photograph Ratings




On-site mean


Photo mean


On-site as % of photo mean

Wyre Forest




Highgate Common




Kinver Edge
















Source: Dunn, 1976 Note: 800 participants. Means derived from preferences for the 6 sites

On-site evaluations were invariably lower than by photograph but the difference was generally held to be within acceptable limits [no statistical tests given]. At only one site were the differences substantial due to the photograph not capturing the full attributes of the site and the poor quality of the particular photograph. Dunn concludes that results supported “the proposition that photographs may be used to accurately represent landscapes. "

Kellomaki and Savolainen (1984) used a variation of the semantic differential method to assess the scenic values of selected tree stands in Finland. Three groups of participants evaluated the scenic values:

  • a Basic Group of forestry students assessed the scenic values in the field and laboratory
  • a Comparison Group, also students, assessed the values only in the laboratory
  • 2 groups of City Dwellers only assessed the values in the laboratory

The results indicated very close assessments between the three groups(p < 0.01) (Table 2)

 Table 2   Comparison of Field and Laboratory Assessments


Mean value

Mean deviation

Range of variation

Basic group      - field



44 - 63

    Basic group  - laboratory



42 - 65

Comparison group - lab



43 - 63

City dwellers - lab



40 - 66

City dwellers - lab



40 - 65

Source: Kellomaki and Savolainen, 1984

While only one group rated the scenes in the field, the mean value of their assessments was only marginally higher than the laboratory assessments but the rating variability was slightly less. This result contrasts with Dunn [1976], who found the field assessments to be slightly lower than photographic assessments.

Although studies had shown little difference between field and photographic assessments, Trent et al (1987) suggested that this might be due to their use of closed-end questions that did not allow the respondent to offer free responses. They believed that an open format may result in more evocative assessments, more descriptive terms and a greater focusing on ephemera absent in photographs. They tested their hypothesis in an urban environment. Using closed-end questions and a 5-point rating scale they found an identical rating. Using closed-end questions and the semantic differential method they found no statistical difference between site and slide tests (p = 0.05).

Open-ended responses, in which participants averaged 10.4 responses for sites and 6.8 for slides (a statistically significant difference), were evaluated on the basis of whether they were positive, neutral or negative terms. The responses were very similar and no differences were statistically significant although the findings (Table 3) suggest that site assessments gave somewhat more extreme results.

 Table 3   Preferences in open-ended responses 


 Site %  Slide % Total %

















Source: Trent et al, 1987

Table 4   Content in open-ended responses



Site %


Slide %

Total %





















Source: Trent et al, 1987 Note: Site/slide differences significant at p < 0.05. Interpretive refers to design characteristics.

Comparison of the content of the open-ended responses (Table 4) indicated that slide views were significantly higher in physical descriptors and lower in evocative responses than site views. Site views were also marginally more ephemeral. The results support the notion that “site views are richer in stimuli, both because they appeal to all the senses and because they are dynamic.” The site views tended to produce more subjective responses (e.g. evocative, ephemeral) while slides produced more objective results (e.g. design and physical descriptors) - otherwise the “results do not lead one strongly to favor one mode of presentation over the other.”

A definitive study on the use of photographs as a surrogate of field observations was undertaken by Shuttleworth (1980). Being concerned that many of the studies that examined this issue used different populations to assess the sites and the photographs, Shuttleworth used the same group in both situations. The study used landscapes in rural areas and on the urban fringe (East Anglia, England). Colour and black and white prints were used as surrogates. Semantic differential (SD) and bipolar scaling techniques were applied. The sample population of students (n = 93) was divided into two groups, all of whom visited all the field sites; one half viewed the colour and one half viewed the b & w photographs. Various techniques were used to ensure randomness [e.g. changing the sequence of field vs photograph assessments] and to enable within-group and between-group analysis.

Shuttleworth found no significant differences between groups in responses to landscapes in the field and found little difference in responses to the photographs. However, he did detect distinctly more differences between responses to b/w photographs and field views than between colour photographs and field views. He found that, with black and white photographs, participants tended to “make much more definite and differential responses by reinforcing likes and dislikes; responses to them thus tended far more to extremes of opinion than did responses to colour photographs.”

Shuttleworth concluded that the results “indicated that there were very few differences of significance between the reactions to and perceptions of the landscapes either when viewed in the field or as photographs” with any differences being explainable by content. He concluded that photographs can be used, providing they are in colour and are wide-angled to provide a lateral and foreground context.

Stewart et al (1984) were concerned that analyses based on group responses can mask individual differences. A small group of observers evaluated visual air quality in Denver, using both field and photographic assessments, repeating the assessments several times over a period of five months. Stewart et al found that the photographic correlations of visual air quality, clarity and cloud cover matched field judgements, regardless of whether these judgements were by the one observer or were the average of several observers. In one test, correlations averaged 0.76 between field observations and judgements of photographs while inan additional test, correlations averaged 0.73 and 0.78.

Stamps (1990) used meta-analysis to review 11 studies that used photographs as well as in situ preferences and derived a correlation of 0.86. This represented the value from 152 environments evaluated by more than 2,400 participants.

In conclusion, with few exceptions surveys have established that photographs can provide a viable surrogate of landscape, although there are slight differences in responses and certain rules should guide their use. Photographs tend to provide more objective, dispassionate responses, while site assessments can yield a more subjective response influenced by a range of site factors unrelated to landscape quality. Black and white photographs can reinforce likes and dislikes and produce more extreme responses than colour photographs  - they may be useful in discriminating landscapes of close similarity of character and quality. Generally, photographs should be in colour and provide a wide view to provide sufficient context.

Go to top


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