Environ. Sci. Technol. 2004, 38, 3971-3976
Aesthetics of Simulated Soiling Patterns on Architecture CARLOTA M. GROSSI* AND PETER BRIMBLECOMBE School of Environmental Sciences, University of East Anglia, Norwich, NR4 7TJ, U.K.
Two desk-top exercises investigated public acceptability of idealized soiling patterns on buildings, using methodologies typical of the psychology of art. The exercises used a range of simulated soiling patterns around a simple architectural element, a pedimented window. In the first experiment respondents were asked to arrange the images from the “most acceptable” pattern to the “least acceptable”. Results hinted at the importance of certain soiling features in driving the ranking. The second exercise explored the characteristics of soiling patterns that most affected their acceptability. In this experiment the images were organized in pairs. People were requested to choose the pattern they found more acceptable in each pair. Uniform patterns and those which created shadowing effects proved more acceptable. Patterns with noninteger fractal dimension that obscured architectural forms were less acceptable. There was usually a preference for images showing less soiling, whereas vertical features and lumpiness were not as acceptable. Results gave an insight into spatial factors that might influence the acceptability of soiling on real buildings. Thus suggesting it is necessary to consider both the level and the distribution of soiling when trying to gauge public reaction.
Introduction Buildings of urban and industrial areas are disfigured from the accumulation of particulate matter, mainly soot, which originates through the incomplete combustion of fossil fuels. This damage, often termed soiling, constitutes a visual nuisance and leads to a loss of architectural value through the darkening of exposed visible surfaces or public-visible areas. Soiling seems to imply a reversible darkening, an aesthetic change, but can also involve an irreversible process. This latter process implies physical damage, such as the formation of black crustssa combination of carbon and gypsumson stone and mortar surfaces (1-3). Despite the potential for real physical decay, building managers frequently see the need for cleaning arising from perception as much as a worry about degradation. In particular, there is a sense of concern about how the public reacts to the appearance of buildings. A good example of this would be the Tower of London, where a major building within the complex is known as the “White Tower” which creates a sense of expectation that the building appear white in a city with a high degree of potential soiling, most particularly from diesel soot. Management decisions usually must be based on an assumed public perception because most managers do not * Corresponding author phone: 00 44 1603 592706; fax: 00 44 1603 507719; e-mail:
[email protected]. 10.1021/es0353762 CCC: $27.50 Published on Web 06/09/2004
2004 American Chemical Society
systematically sample public opinion before cleaning work is undertaken. The decision-making aligns with the notion that there is some point at which this visual nuisance becomes unacceptable and the building must be cleaned. Decisions about cleaning would benefit from generally applicable parameters that frame public perception of soiling. These could allow managers to make consistent and perhaps more justifiable decisions regarding cleaning. Some research on public perception of soiling and cleaning of facades is available and shows a positive public evaluation for cleaned facades, especially in cases of heavy soiling. However, some respondents consider cleaning as detrimental when soiling is light or moderate (4, 5). Earlier work also suggests that public disquiet over the appearance of building facades increases with the amount of soiling (6, 7). Coatings (e.g., paint) may affect the soiling process (8), as some protective coatings can accelerate the discoloration. Soiling amounts have typically been seen as a monotonic exponential loss of reflectance which is readily treated mathematically (1, 9, 10), but perceptions may not be so readily reduced to simple equations. Thresholds for discrimination of soiling and initiation of complaint are not only related to darkening or loss of surface reflectance, but also to the color contrast of surfaces (11, 12). The perception of soiling also depends on the individual and on the general condition of the local environment. It was said that people are much more sensitive to the soiling effect in relatively clean environments than in areas with large particle loads, such as industrial regions (3, 13). Furthermore, it became clear from comments of visitors to notable buildings that soiling patterns themselves can alter the acceptability of a given amount of soiling, e.g., one viewer commented of some soiling streaks that the building could be darker or lighter, but not like that. Thus it appeared likely that soiling perception depends not only on the degree of blackening, but on patterns of soiling exhibited on the building. Soiling patterns have been less frequently investigated than the amount of soiling. Young et al. (14) have assessed the soiling distribution on facades, whereas Davidson et al. (15) have examined some physical controls on soiling patterns and drawn attention to two competing processes: the deposition of pollutants and the removal of soiled material by rain. Water runoff patterns on the fac¸ ades are influenced by the nature of the surface material, architectural features, and microclimate effects (16), and ensure that soiling does not occur in a uniform manner across the entire surface of a building. Thus, there is the potential for a complex interaction between soiling, architecture, and aesthetics. Research conducted by Andrew (4) on perception of soiling patterns concluded that light and moderate soiling around architectural details could improve the visual appearance of the building by increasing contrast or adding shadowing effects. Heavy soiling eventually leads to a uniform blackening which reduces the visual information or architectural details and obscures the color, texture, and a sense of the moulding. Questionnaires can investigate public perception of soiling by utilizing observation of real or simulated images. Many survey questionnaires have used photographs of building facades (4, 5, 17) or are run at the building site. Other studies have chosen more abstract images and have evaluated responses for examples of simulated amounts of deposition on plain backgrounds (12). The studies on simulated soiling tend to quantify the level of soiling through the concept of effective area coverage, that is, the percentage of area covered by light absorbing dust. VOL. 38, NO. 14, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3971
FIGURE 1. Images for the first desk-top exercise: (a) pedimented window frame used to simulate soiling patterns, and (b) examples of designed images and the corresponding real facades used for guidance (White Tower, Tower of London). Our own work suggested a need to determine which particular patterns might create the greatest offense. We decided on a desk-top exercise following the abstract approach of Bellan et al. (12). This allowed us to explore a range of soiling patterns while holding everything else constant, i.e., a methodology long familiar in the psychology of art (18). We decided not to use “doctored” photographs, but to present soiling patterns as drawings related to a simple architectural element, in this case a pedimented window.
Experimental Section An initial desk-top exercise to investigate the public response to different soiling patterns was designed around a drawing of a pedimented window (Figure 1a). Soiling was added to the area around this simple architectural feature, which represented the patterns in a purely 2-dimensional form. The first group of images was designed by taking guidance from the types of patterns observed in photographs of real buildings (Figure 1b). There were 10 images in this group (the A group) which had soiling patterns that concentrated in and around the window frame. A further six images (the B group) had soiling more dispersed (i.e., at a lower density) on the wall and around the window frame. The basic window frame was edited in Paint Shop Pro and the sixteen possible soiling patterns were designed to have approximately the same amount of soiling (or overall blackness). The images were printed onto small pieces of stiff paper, slightly larger than playing cards. These were presented to people face-up on a table, under diffuse indoor lighting, and they were asked to arrange them in order of preference. They were to rank the images from the “most acceptable” pattern to the “least acceptable”. We made it clear to participants that this was an aesthetic exercise about the black soiling and we were not asking them to assess which patterns related to physical damage. Although they were not constrained by time they were asked to rely on their first impressions. This was particularly important when offering the exercise to respondents who had some knowledge of conservation, as they would consider underlying decay. One hundred and nineteen respondents were tested, and these included people from a number of European countries. For comparative reasons, respondents were divided into the following two categories: (1) general public (46 cases: different professions, different countries), and (2) experts (72 cases). The group “experts” consisted of specialists or students in areas related to the conservation of heritage. 3972
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 14, 2004
FIGURE 2. Examples of pairs of images for the second desk-top exercise: (a) amount; (b) orientation; (c) separation; (d) symmetry; (e) featheriness/lumpiness. It was clear as we completed this first exercise that certain features seemed to be driving the choices made by respondents. We continued the study with a second desk-top exercise that was designed to explore the particular features of the soiling patterns that caused them to be more acceptable. Specifically, we investigated the relative influence of five aesthetic variables that seemed important from the results of the first exercise. The selected variables related to the following: (a) amount of soiling, although this was not explored in the first exercise (higher vs lower) (Figure 2a); (b) orientation of the soiling pattern (horizontal vs vertical streaking) (Figure 2b); (c) separation of the soiling patterns from the window frame (separated vs attached) (Figure 2c); (d) symmetry reflected about the vertical central axis of the window (symmetric vs asymmetric) (Figure 2d); and (e) feathery (higher fractal dimension) patterns vs lumpy or blocky patterns (Figure 2e). The new images were also designed in Paint Shop Pro using the previous A group images as a base. The exercise consisted of 25 pairs of images, with five pairs testing each of the variables noted above. The 25 pairs were randomized so that subsequent pairs did not test the same variable. Only one variable was tested by each pair in each group while the rest of the variables were kept (roughly) constant, apart from the “fractal dimension”, which was too difficult to control. One hundred and two (102) people were asked to choose the pattern they found more acceptable of each the 25 pairs and once again the exercise sampled people from a variety of European countries (mostly Spain and the U.K.).
Results and Discussion The first exercise revealed that some soiling patterns tend to be preferred over others. Figure 3 summarizes the ranking of the images (1, most acceptable, to 16, least acceptable). There also appear to be differences in the way individuals respond to the A group (soiling close to the window frame) and the B group images (soiling dispersed). This is clearly shown in Figure 4 where the responses are classified using hierarchical clustering. This clustering technique identifies
FIGURE 3. Ranking of the sixteen images (1, most acceptable, to 16, least acceptable) illustrating the two series: soiling close to the window frame (a) and dispersed (b). The image numbers refer to the ten A group and the six B group images.
FIGURE 4. Hierarchical clustering of the images (Centroid method) based on the ranking given by respondents. The clustering shows two differentiate groups of preferences. Some individuals found patterns of A group images more acceptable, whereas others preferred B images. possible homogeneous groups of images based on the ranking given by each of the individual respondents. Ranking was analyzed using Kendall’s coefficient of concordance (W). The value W can range between 0 (no agreement) and 1 (complete agreement). In this case of all sixteen images, W was 0.23, suggesting some agreement between individuals that is significant (p < 0.01). The sequence is even more obvious in the case of A group images, where W is 0.34 (significant at p < 0.01). In the case of diffuse B group images W is lower, at 0.12, showing it is much more difficult to make decisions about preference, yet agreement is still significant at p < 0.01. Table 1 summarizes the ranking and the characteristics of A group images, and the same is summarized for the B group images in Table 2. This was an area where there was a noticeable difference between the answers given by “general public” and “experts on heritage conservation”. In general, “public” preferences were equally divided between A group images and B group images, whereas “experts” tended to prefer B group images, although these differences do not seem to be very relevant in the overall ranking. “General public” responses are more variable when ranking B Group images than the responses of “experts”. General public include respondents from
different backgrounds and education: scientific, arts, economics, law, etc. The sample was too small to establish national, gender, and educational differences in responses. However, interestingly, the respondents often told us it would depend on such variables although we gained no strong sense of this. Hierarchical clustering suggests that the A group images can be divided into two groups: a7, a1, a9, and a8, which are more acceptable, and the rest which are less acceptable. The preferences for the A group images are shown when they are arranged in order of rank in Table 1. Image a7 stands out as preferred by both experts and the public. This is an interesting image, which has a soiling pattern that strongly delineates the window frame. The preference for this feature follows Andrew’s (4) conclusion that soiling patterns could improve the visual appearance of the building by increasing contrast and creating shadowing effects. At the other extreme there is a series of images (a10, a6, a2) that are least preferred. These images are characterized by vertical rain-washed patterns of soiling. The lack of acceptance of these features is preserved over both the “public” and the “expert” rankings. As we examined the responses we began to sense that soiling features that offended the architectural lines, such as vertical streaking, were not liked. This was the natural compliment to Andrew’s position that acceptable patterns were those which somehow delineated the architecture. The notion of delineation seems to imply systems that have an integer fractal dimension (i.e., 1 lines, 2 areas, 3 volumes). Thus systems where the fractal dimension of the soiling pattern was not an integer could confuse architectural lines and be less acceptable. The fractal dimension of the perimeter of the A group images was measured using software designed by Volodymyr Kindratenko to determine the fractal contours of objects (available on the web at http://www.ncsa.uiuc.edu/ ∼kindr/phd/software.html#ref2). The fractal dimension was determined by the box counting approach, and, as seen in Figure 5, the images with fractal dimensions closer to unity tend to be more acceptable. The second exercise was designed to explore particular features of the soiling patterns that might cause them to be more acceptable. First, and most significantly, it was used to check whether respondents preferred images with lower amounts of soiling. As can be seen in Table 3, this expectation was satisfied and in agreement with earlier work (4, 11). Some VOL. 38, NO. 14, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3973
TABLE 1. Average Ranking of Preference Within the Ten A Group Images group
rank
total
rank
public
rank
expert
characteristics
more acceptable
a7 a1 a9 a8 a4 a5 a3 a10 a6 a2
2.4 3.9 4.0 4.0 5.8 6.5 6.9 7.1 7.2 7.2
a7 a9 a8 a1 a4 a5 a3 a10 a2 a6
2.2 3.9 4.2 4.2 5.5 6.0 7.0 7.1 7.2 7.6
a7 a1 a8 a9 a4 a5 a3 a6 a10 a2
2.5 3.8 3.9 4.0 5.9 6.8 6.8 6.9 7.1 7.2
evenly on the frame even- slightly uneven: horizontal even- slightly uneven: horizontal even- slightly uneven: horizontal uneven: horizontal uneven: horizontal uneven: horizontal uneven: vertical uneven: vertical uneven: vertical
less acceptable
TABLE 2. Average Ranking of Preference Within the Six B Group Images rank
total
rank
public
rank
expert
b5 b6 b2 b1 b4 b3
2.5 3.3 3.4 3.5 4.1 4.2
b6 b5 b2 b4 b1 b3
2.4 3.0 3.2 3.8 4.3 4.5
b5 b1 b2 b6 b3 b4
2.1 3.1 3.5 3.9 4.0 4.4
FIGURE 5. (a) Averaged ranking of A images related to the fractal dimension. (b) Averaged fractal dimension of the most acceptable A group images (a7, a1, a9, and a8, see Table 1) and the less acceptable ones (the other six). The most acceptable images have an average fractal dimension closer to unity than the other six. 84% of the five choices made by 102 respondents were for the less soiled image of the pair. Few respondents (only 12 of 102) chose the more soiled image of a pair more than once out of the five pairs they examined (see Figure 6a). Around 6% of the respondents were unable to decide which image they preferred. This simple test of perception of amount was useful in indicating that the desk-top exercise gave answers that were consistent with expectations from studies of observations of real buildings (19). The orientation of the soiling pattern gave similar clear results, with three times more respondents preferring 3974
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 14, 2004
horizontal marking to vertical rain-washed markings. This was much in line with the indications from the first exercise. As with the amount of soiling, around 5% of the time respondents were unable to decide which they preferred. The separation of the soiling pattern from the window frame also evoked a clear response, with few being unable to decide which they preferred. Most respondents liked patterns that were separated from the window and we presume these were ones that delineated the architectural feature more clearly. However, a smaller number of respondents consistently preferred patterns that were attached to the architectural feature. Thus, although on average patterns separated from the window were preferred, there are some individuals who seem to have a distinct preference for soiling attached to the feature (note the sixth column in Figure 6c). Questions about images that contained different symmetry or lumpiness show a higher numbers of indecisive answers, indicating that the respondents could not choose which they preferred (slightly higher than 10%). Despite this, asymmetrical patterns were preferred overall. A feathery appearance was clearly preferred over a lumpy appearance to soiled areas. This latter result did not agree with the evidence of the first exercise that the higher the order of the fractal the more displeasing the image, because the lumpy images in this case had fractal dimensions closer to unity. However, it is possible that these observations are not inconsistent, rather it is that the respondents simply did not like lumpy soiling patterns, because they also obscure architectural form. It probably remains true that among feathery patterns there is a preference for those with fractal dimensions closer to unity, as established in the first exercise. This second exercise then complimented the first and indicated a preference for images showing less soiling, horizontal orientation, “feathery” patterns, and soiling patterns that were separated from the window. Although less clear, asymmetric patterns seemed to be preferred. All these distinctions were significant at the 0.01 level. In terms of building management it would be of value to use these observations from a desktop experiment when trying to determine acceptability of soiling on real buildings. However, this has to be done with care as our reactions to soiling in drawings cannot be neatly translated to what we see on a building. Nevertheless, the results give an insight into the kinds of questions we need to ask about soiling on real buildings. The most general conclusion from this is the obvious one. Although the amount of soiling on a building is important, its distribution also affects the way in which we perceive soiling. The strong negative reaction to vertical rain-washed streaks was clear in both sets of images used, which suggests particular attention should be given to these forms of soiling when considering cleaning. More general, our work confirms that those soiling features which obscure or conflict with architectural form induce the most adverse reaction. This suggests that cleaning might first focus on areas which accumulate streaks or other disturbing features. Although
TABLE 3. Results of the Second Desk-Top Exercise: Paired Images amount preference (%)
low high inda
overall preference chi-square significant at p < 0.01 a
ind ) indecisive answer.
b
orientation 84 10 6
hor. vert. ind
73 22 5
separation no yes ind
32 64 4
symmetry no yes ind
featheriness 51 37 13
feath. lumpy ind
64 23 13
lower
horizontal
separated
asymmetric
feathery
424 301b yes
293 134b yes
192 56b yes
93 11b yes
205 102b yes
Excluding indecisive responses.
FIGURE 6. Number of respondents choosing the less selected option of each variable in the second desk-top exercise: (a) amount of soiling; (b) orientation; (c) separation; (d) symmetry; (e) featheriness/lumpiness. deliberate cleaning of such areas could just make the building look patchy, one can imagine approaches that would mitigate the development of patterns or clean the area of the fac¸ ade thus lowering the intensity of the pattern. Perception of soiling is complex so assessing public reaction to soiled surfaces remains difficult, yet the cleaning is often seen to be undertaken as part of a public need. Public perceptions are not easy to integrate with considerations such as available finance, concerns over physical damage due to crust build-up, or the importance of having a clean building for specific events or celebrations. Public opinion is generally established through questionnaires or exercises,
which are time-consuming. Additionally, the current research suggests that the perception of soiling needs to consider the pattern of soiling in addition to the amount of soiling. Expert views of soiling seem to parallel those of the general public, which means that their views do not necessarily impose a bias. Ultimately, it would be convenient to be able to assess the soiling of a building and degree of public concern in a more abstract way that did not require a questionnaire. We have had some success in earlier work establishing perceptions of facades in terms of a gray scale or a color chart (19). The current work suggests that it may be possible to establish VOL. 38, NO. 14, 2004 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3975
a hierarchy for spatial features that would give guidance as to which patterns are likely to be least acceptable. Protocols for assessing damage features on stonework are available (eg., ref 14). The development of a protocol to estimate aesthetic damage in terms of likely public reaction to the soiling and its distribution would be an additional guide to management decisions about cleaning.
(9)
(10)
Acknowledgments
(11)
This paper has benefited from work funded within the CARAMEL project (ENV4-CT-2000-0002). We thank Dr. Cristina Sabbioni and Prof. Alessandra Bonazza of the Istituto di Scienze dell’Atmosfera e` del Clima (Bologne, Italy) and Mr. Harding of the University of East Anglia for assisting with the desk-top exercises.
(12)
(13)
(14)
Literature Cited (1) Haynie, F. J. Theoretical model of soiling of surfaces by airborne particles. In Aerosols; Lewis: Chelsea, MI, 1986; pp 951-959. (2) Yocom, J. E.; Kawecki, J. M. Overview of soiling and materials damage from aerosols. In Aerosols; Lewis: Chelsea, MI, 1986; pp 913-922. (3) Watt, J.; Hamilton, R. The Soiling of Buildings by Air Pollution. In The Effects of Air Pollution on the Built Environment; Air Pollution Reviews Vol. 2; Brimblecombe, P., Ed.; Imperial College Press: London, 2003; pp 289-334. (4) Andrew, C. Towards an aesthetic theory of building soiling. In Stone Cleaning and the Nature, Soiling and Decay Mechanisms of Stone; Donhead: London, 1992; pp 63-81. (5) Ball, J.; Laing, R.; Young, M. Stone cleaning: comparing perception with physical and financial implications. J. Architectural Conserv. 2000, 6, 47-62. (6) Carey, W. F. Atmospheric deposits in Britain - a study of dinginess. Int. J. Air Poll. 1959, 2, 1-26. (7) Hancock, R. P.; Esmen, N. A.; Furber, C. P. Visual response to dustiness. J. Air Poll. Control Assoc. 1976, 26, 54-57. (8) Meiners, C.; Cabrera, I.; Fichtner, T.; Kuropka, R.; Grottenmuller, R.; Zingerle, H.; Ibarraran, C.; Anquetil, J. Y. Binders for exterior
3976
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 38, NO. 14, 2004
(15)
(16) (17)
(18) (19)
coatings exhibiting low soiling tendency. Macromol. Symp. 2002, 187, 207-214. Pesava, P.; Asku, R.; Toprak, S. Horvath, H.; Seidl, S. Dry deposition of particles to building surfaces and soiling. Sci. Total Environ. 1999, 1-3, 25-35. Brimblecombe, P.; Grossi, C. M. The rate of darkening of material surfaces. Air Pollution and Cultural Heritage; A. A. Balkema: Lisse, The Netherlands, 2003. Lanting, R. W. Black smoke and soiling. In Aerosols; Lewis: Chelsea, MI, 1986; pp 923-932. Bellan, L. M.; Salmon, L.; Cass, R. A study on the human ability to detect soot deposition onto works of art. Environ. Sci. Technol. 2000, 34, 1946-1952. Newby, P. T.; Mansfield, T. A.; Hamilton, R. S. Sources and economic implications of building soiling in urban areas. Sci. Total Environ. 1991, 100, 347-365. Young, M.; Ball, J.; Laing, R.; Cordiner, P, Hulls, J. The Consequences of Past Stonecleaning Intervention on Future Policy and Resources; Historic Scotland: Edinburgh, 2003. Davidson, C. I.; Tang, W.; Finger, S.; Etyemezian, V.; Striegel, M. F.; Sherwood, S. I. Soiling patterns on a tall limestone building: changes over 60 years. Environ. Sci. Technol. 2000, 34, 560565. Andrew, C.; Young, M.; Tonge, K. Stone Cleaning. A Guide for Practitioners; Historic Scotland: Edinburgh, 1994. Pollicino, M.; Maddison, D. Valuing the impacts of air pollution on Lincoln Cathedral. Working Paper GEC 99-03, CSERGE: Norwich, UK, 1999. Pickford, R. W. Psychology and Visual Aesthetics; Hutchinson Educational LTD: London, 1972. Grossi, C. M.; Brimblecombe, P. Aesthetics and perception of soiling. Air Pollution and Cultural Heritage; A. A. Balkema: Lisse, Netherlands, 2004.
Received for review December 10, 2003. Revised manuscript received April 28, 2004. Accepted April 30, 2004. ES0353762
