IN
DEGREE PROJECT THE BUILT ENVIRONMENT, SECOND CYCLE, 30 CREDITS
STOCKHOLM SWEDEN 2020,
Patterns in the city - A tool for pattern correlation
Mönster i staden - En metod för mönsterkorrelation
ORESTIS-KOSMAS CHARALAMPIDIS
Abstract (English)
Livable cities are in the frontline of the work of architects and urban designers and livable streets occupy a large and important part of where city life happens. The study of existing streets and the life on them, a part of urban morphology's field of research, could bring light to aspects that will evolve the design process. This thesis suggests a method that aims to contribute to a better understanding of how we experience street environments. It is an effort to analyze and quantify the rhythms that occur through our every day life experience along the streetscapes. The patterns of our built environment's elements contribute to the creation of such urban rhythms. The method is comprised by a mapping process for data collection and a mathematical model which analyzes the data and provides with quantitative results that are used as comparative indexes for the correlation of patterns along the facades of selected paths. The method is tested on a sample of three paths in the city of Stockholm. The results of the test are considered satisfying for the technique to be considered functional. The test, though, limits itself to physical, perceptible objects. Therefore, the method's contribution would be more valuable inside a broader context and in combination with methods and data of a more inclusive study, which will provide a more holistic analysis. Main obstacles for the method's implementation are the lack of information about connection of existing patterns to urban space qualities and the inefficient ways of mapping patterns in a large scale. However, technological advancements and further research might create a fertile ground for development.
Abstract (Swedish)
Beboeliga städer är i frontlinjen i arkitekters och stadsdesigners arbete och beboeliga gator är en stor och viktig del av var stadslivet sker. Studerandet av befintliga gator och livet på dem, en del av urban morfologins forskningsområde, skulle kunna få fram aspekter som kommer att utveckla designprocessen. Den här avhandlingen föreslår en metod som syftar till att bidra till en bättre förståelse av hur vi upplever gatumiljöer. Det är ett försök att analysera och kvantifiera rytmer som uppstår genom vår dagliga livsupplevelse längs gatorna. Mönstren i elementen i vår byggda miljö bidrar till skapandet av sådana urbana rytmer. Metoden består av en kartläggningsprocess för datainsamling och en matematisk modell som analyserar data och ger kvantitativa resultat vilka används som jämförande index för korrelation av mönster längs fasaderna på utvalda gator. Metoden testas på ett urval av tre gator i Stockholms stad. Resultaten av testet anses vara tillfredsställande för att tekniken kan anses vara funktionell. Testet begränsar sig dock till fysiska, synliga element.
Därför skulle metodens bidrag vara mer värdefullt i ett bredare sammanhang samt i kombination med metoder och data av en mer inkluderande studie - vilket kommer att ge en större helhetsanalys. Huvudhinder för metodens implementering är dels bristen på information om samband mellan befintliga mönster och urbana spatiala kvaliteter och även de ineffektiva tillvägagångssätten att kartlägga mönster i stor skala. Tekniska framsteg och ytterligare forskning kan emellertid skapa en god grund för vidare utveckling.
Table of Contents
1. Introduction 3
2. Theoretical framework 6
3. Methodology 10
4. Test 14
5. Results 34
6. Discussion 41
7. Conclusions and recommendations 43
8. References 46
9. Appendix 50
I would like to especially thank my supervisor Todor Stojanovski for his guidance, motivation and excitement. His role was vital in the completion of this thesis.
I would also like to thank my friend Alexios Kotsakis for his contribution to this thesis and our creative collaboration.
Thanks goes also to my examiner Andrew Karvonen for his persistence in urging me to improve this thesis to a great degree.
Finally, I would like to thank Paul Sanders for his feedback on this thesis. His work was the inspiration for the idea of the thesis.
1. Introduction
1.1 Context
Livable cities are in the frontline of the work of architects and urban designers and livable streets occupy a large and important part of where city life happens. Nearly everyone in the world lives on a street. People have always lived on streets (Appleyard, 1981). The role of the streets is vital for the residents' quality of life. City life happens largely on the streets and the form and qualities of this particular city space are of great importance to urban design. Qualities such as walkability, safety, hygiene, accessibility etc.
should be taken into consideration when designing streets, a controversial place in the city that has been approached by different design perspectives, with a character that might adjust to the needs, priorities and realizations of every era. The study of existing streets and the life on them, a part of urban morphology's field of research, could bring light to aspects that will evolve the design process. There are several tools, that become even more sophisticated along technological advancements, which help urbanists to identify and analyze urban space qualities, important parameters for the sustainable development of livable streets, the heart of livable city environments. This thesis suggests a tool that aims to contribute to a better understanding of how we experience street environments.
It is an effort to analyze and quantify the rhythms that occur through our every day life experience along the streetscapes, information that might contribute to the design of more livable streets.
1.2 Problem Formulation
The idea for the thesis was inspired after reviewing several methodologies for urban space analysis. Those that appear to be based more on a mathematical - analytical approach, usually on the larger scale (area, city, country, global scale) and those who appear to be structured in a more descriptive way (including inventories, tabulation of physical elements, visual and conceptual representations), usually on a smaller scale (street, facades, building elements). From analytical and evaluative methods that study urban areas in relation to transportation (Circuity and TOD) (Boeing, 2019, as cited in D’Acci, 2019; Stojanovski, 2017, 2019b, 2020) to computational models for urban pattern analysis such as CIty Induction (Duarte, Beirão, Montenegro & Gil, 2012) which is based on Alexander's pattern language (Alexander, 1977), Hillier's space syntax (Hillier & Hanson, 1989) and Stiny's shape grammars (Stiny & Gips, 1971), significant work landmarks in the study of urban patterns.
'Morphogenetic analysis of architectural elements within the townscape' (Sanders &
Woodward, 2015) was a very influential study for the idea of this thesis. According to Sanders & Woodward, a formula for ensuring that new development relates to its context so as to achieve congruent outcomes is still lacking, despite the established principles and methods of urban morphology that enable the systematic analysis of the built environment. They propose their method as a solution to this problem. Their measurements and tabulation of the amounts of classified architectural elements that exist on the facades along several streetscapes in Brisbane, follow a quantification approach that is common through the literature. The attempt for a different approach gave the idea of creating a technique (to be referred from now on as "PCT" - Pattern Correlation Tool) that will somehow quantify the type of sequence and the recurrence of these elements
along streetscape facades, instead of measuring their amounts. A mathematical model that will be able to use the data of such descriptive inventories and measurements to provide information about the patterns they form.
Whereas a quantitative inventory of elements, such as this by Sanders & Woodward, provides insight through the approach of a more static experience of the facade, by the sense of providing information on amounts of elements (amount being a spatially static concept), PCT provides insight through the approach of experiencing the facade while being mobile; experience of the change of the elements, their rhythm through time. This approach focuses on the mobile state the user is in, inside public space. In this state, the user's perceptive image of the environment is shaped dynamically.
Time is an important factor that has an established position in transportation planning or, on a smaller scale, when designing for residents' accessibility to facilities for example. In that way time becomes a shaping factor for our built environments. But time is not only translated as a notion of distance by also a notion of change. And our built environments are experienced through time, through the change of our perspective towards them. Our movement through and inside them and our interaction with each other in relation to them as well as our interaction with them is in itself an action of change. It is very seldom that we find ourselves standing still, sitting on a bench and stare at a very certain city view, just looking at it and perceive it as an unchanged landscape painting, digesting the long lasting architectural characteristics fixed and motionless as they stand. It is much more common to perceive them relatively to our course or to the exchange we might have with them. A door that we enter, a glass front that makes us stop for a while to look at something we find interesting and then it's so long that makes us stop again after a few meters, someone that just came out from a building and met us, that wall corner that we stood by and waited for our friend, that alley that we turned to and moved on, that café with the beautiful art nouveau ornamentations that we always stop to get a coffee before moving further to reach our workplace. There is a certain rhythm about how we live in relation to our built environment, how we understand it and experience it. It is not just an image or an amount.
The patterns of our built environment's elements contribute to the creation of such urban rhythms. PCT is a method that tries to quantify such urban patterns that might then be translated into rhythms experienced by the users, an area with a lot of prospect for research within the field of environmental psychology as well as urban studies.
Paul Sanders refers to his highly detailed work as a sort of hyper-specificity. The term
‘micromorphology’ has been used to describe such studies of form at the level of elements of individual houses (Larkham, 2006; Whitehand, 2001; Whitehand, Morton & Carr, 1999).
Sanders emphasizes that his aim is to understand the 'now' and how it can be a resource for future design/change. This thesis, influenced by his work, follows along the same lines.
The research questions are:
1. How can we improve the analysis of urban space from a mobile perspective?
2. How can we structure a quantification method that captures the image of the facades along a path being experienced dynamically, rather than statically as a motionless scenery?
3. How can this quantitative tool be a resource, used to inform architects and urban designers and affect future design?
1.3 Purpose And Objectives
The thesis aims to contribute to the existing body of analytic applications and mapping techniques used in urban morphology and in assessment of urban design studies. Ye &
Van Nes (2014) present a very interesting combination of methods to assess socioeconomic performance of urban areas. Talen (2002a) explores the transect strategy.
The work of Ewing & Handy (2009) and Talen (2002b) is more oriented towards more qualitative characteristics such as walkability and pedestrian access. This thesis, however, relates probably more to the work of Mehta because of the focus on the street, the user's experience and the use of indexes as assessment criteria. Mehta (2008) examines the relationship of the physical, land‐use, and social characteristics of the environment at the microscale to people's behavior and perceptions toward walking. Mehta (2009) studies the experience of the street especially from the users' perspective, using extensive behavior mapping. In his 2014 paper 'Evaluating public space', he creates a public space index to assess the quality of public space. The thesis contribution might be small but important in terms of approaching the city from a different perspective and providing a concrete methodology of analysis. PCT aims to lay the groundwork for further development of pattern oriented analysis and to inspire debate for design of livable cities.
This thesis is developing and testing a method (PCT) rather than producing empirical results. PCT has a descriptive character, providing quantitative data that could be used in an explanatory, evaluative or prescriptive manner. The thesis though, following an explorative approach, aims to focus only on the formulation of PCT and the testing of its performance and doesn't intend to elaborate further, i.e., on the use of the acquired data explanatorily or interpreting them into conclusions of qualitative character. The test of PCT is performed on three streetscapes in the city of Stockholm in this thesis. This test though is not an actual evaluation of the three streetscapes but rather an illustration of how PCT works. This decision corresponds to similarly suggested assessment methods, such as the Identifying and Measuring Urban Design Qualities Related to Walkability by Ewing, Handy, Brownson, Clemente & Winston (2006). As explained in that paper, the overall goal of that project was to develop operational definitions . . . that take the form of statistically-derived equations that link objectively measured physical features of the environment to ratings of urban design qualities. The main purpose of the project was not to test the association between these qualities and walkability or walking itself, but rather to develop measurement methods that can be used by researchers to study these relationships.
1.4 Organization of the thesis
The thesis is organized into the following sections:
Theoretical Framework
A review on literature related to (i) the usability of pattern analysis in the field of Urban Morphology; (ii) how we perceive patterns in our environment; (iii) how design can learn from existing well-tried patterns; and (iv) where and how patterns occur in urban spaces.
Methodology
PCT, which is a comparative technique for measuring and quantifying urban pattern correlations, is presented in this section. Its purpose is described, the context of its use is set (the nature of the patterns it will be applied to is defined), the technique itself (which is comprised by a mapping method for data collection and a mathematical model for their analysis) is explained.
Test
PCT is applied and tested on a sample of patterns along three paths in the city of Stockholm. The whole process is presented and the decision making is explained. The decision making regards structuring of the process and limiting of the test in order for it to be feasible for the scope of this thesis. The limiting includes the extent of the mapping process but in a way that will provide sufficient information which, at first, can describe the patterns adequately, and secondly, is capable of being quantified and used with PCT.
It will be attempted for the operational part of the test (input of collected data to the mathematical model, performance, and retrieval of results) to be automated.
All the inventories of the collected data can be found in the Appendix.
Results
The results of the test are tabulated, put into graphs and presented in this section.
Discussion
Commentary regarding the test outcome. Evaluation of PCT's performance and the validity of the results.
Conclusions and recommendations
Discussion about problems that occurred through the process and about potential applications and evolution of PCT in future research as well as its incorporation with other methods.
2. Theoretical framework
2.1 Urban Morphology
Urban morphology is a field that can contribute to a more insightful design process by providing knowledge on the description of the urban environment (Hall & Sanders, 2011;
Kropf, 2011; Kropf & Malfroy, 2013; Sanders & Baker, 2016). A description can be one of the physical environment but also of the abstractions that relate to it (e.g. function, use, perception, relations). Urban morphologists use a variety of techniques (both qualitative and quantitative) to identify, map and evaluate the elements that the urban environment consists of and can be partitioned to. Elements that own an inherent meaning, given to them just by their design and creation itself, or elements that acquire meaning by the very process of their identification as such. According to Marshall & Çalişkan (2011), three key applications of morphology are: for it to function (i) as an investigative or explanatory technique; (ii) as a diagnostic or evaluative tool, a means of studying successful or unsuccessful kinds of urban fabric; and (iii) as a means of identifying exemplars, types or
According to Marshall & Çalişkan (2011) in terms of scale, urban morphology covers a wide spectrum from the scale of buildings to metropolitan areas. It is usually interpreted as an analytic activity that refers to the study of the urban fabric of buildings, plots, street patterns and is about pattern recognition, contrary (or complementary) to urban design which is about pattern creation.
2.2 Environmental Psychology And Perception
Every aspect of human existence occurs in one environment or another, and the transactions with and within them have important consequences both for people and their natural and built worlds (Gifford, 2014). Through research in the fields of psychology as well as urbanism, it becomes evident, how differently the physical environment is perceived between individuals. As Canter (1977) cites, in the context of psychological research on how common features inside a community are perceived by members of it, in contrast to foreign ones: Whenever material visually presented purports to be representative of some common object, but contains certain features which are unfamiliar in the community to which the material is introduced, these features invariably suffer transformation in the direction of the familiar, finds Bartlett (1932) through the process of showing a drawing to different people and asking them to reproduce it from memory, which in its turn will be reproduced again by another participant in that experiment.
Bartlett's explanation of this is based on Head's (1920) notion of schemata. As described by Head, schemata are organized models of ourselves formed by stored past impressions of the individual in the sensory cortex that can remain outside central consciousness. Such schemata modify the impressions produced by incoming sensory impulses . . . with a relation to something that has gone before. Lee (1954) applies the notion of schemata to the context of the physical environment. Aside from Canter's reference to schemata, Krier (1979) mentions that visual and sensory habits, which vary from one individual to the next, are augmented by a vast number of socio-political and cultural attitudes. Furthermore, Lynch (1960) suggests his notion of imageability i.e. the mental image of the city held by its citizens. In his explanation of the term he describes it as the ease with which the city's parts can be recognized and can be organized into a coherent pattern. Considering the referred research, it can be argued here the value of resulting with a common image of the city between its users, despite the individual schemata and visual and sensory habits that may vary indefinitely among them; the easiness of the city being identified and organized as a distinct pattern by any of its users. Lynch stresses the importance of imageability when it comes to environments on the urban scale of size, time and complexity, suggesting that we must consider not just the city as a thing in itself, but the city being perceived by its inhabitants.
Regular or irregular patterns can be deliberately used during the urban design process or occur through time inside the city, on various scales. Krier (1979) refers to Fischer von Erlach on how he deliberately placed side by side isolated buildings and urban spaces of a very different character, in order to underline the diversity and richness of the urban morphology. He also refers to the 60's clinical separation of urban functions into city zones.
Lynch (1960) also speaks further about patterns in his list of urban qualities. While explaining the quality of Continuity he notes repetition of rhythmic interval (as a street- corner pattern); similarity, analogy, or harmony of surface, form, or use (as in a common building material, repetitive pattern of bay windows, similarity of market activity, use of common signs). These are the qualities that facilitate the perception of a complex physical
reality as one or as interrelated, the qualities which suggest the bestowing of single identity, he argues. There are more of Lynch's listed qualities that a relation to patterns can also be implied for them, such as the Directional Differentiation or Visual Scope. It is however the quality of Time Series where the notion of pattern is mentioned more as a time depended entity. And it is this dependence that makes patterns perceived by the user as such, as one moves through space-time. Patterns that exist along paths can not be experienced statically. They exist by a sequential manner, being perceived through time like events on a timeline, or making time itself being perceived by the user while one experiences them. Time Series are explained by Lynch as series which are sensed over time, including both simple item-to-item linkages, where one element is simply knitted to the two elements before and behind it (as in a casual sequence of detailed landmarks), and also series which are truly structured in time and thus melodic in nature (as if the landmarks would increase in intensity of form until a climax point were reached). The former (simple sequence) is very commonly used, particularly along familiar paths. Its melodic counterpart is more rarely seen, but may be most important to develop in the large, dynamic, modern metropolis. Here what would be imaged would be the developing pattern of elements, rather than the elements themselves.
Furthermore in literature, the effect of the city environment and its patterns on the individual, as well as on the group, is described in Gehl's (1987) observations on public space and the street. As he notes: A stretch of 500 meters viewed as a straight, unprotected, and dull path is experienced as very long and tiring, while the same length can be experienced as a very short distance if the route is perceived in stages. About pedestrian's stops he mentions: Fewest stops were noted in front of banks, offices, showrooms, and dull exhibits of, for example, cash registers, office furniture, porcelain, or hair curlers. Conversely, a great number of stops were noted in front of shops and exhibits that had a direct relationship to other people and to the surrounding social environment, such as newspaper kiosks, photography exhibits, film stills outside movie theaters, clothing stores, and toy stores. . . . If spaces are desolate and empty - without benches, columns, plants, trees, and so forth - and if the facades lack interesting details - niches, holes, gateways, stairs, and so on - it can be very difficult to find places to stop. Therefore Gehl stresses that it is of utmost necessity for the designer to be very careful with every single foot of facade or pedestrian route.
2.3 Learning From History
This concern about careful design is broad among urbanists. According to Lynch (1960), we need an environment which is not simply well organized, but poetic and symbolic as well. It should speak of the individuals and their complex society, of their aspirations and their historical tradition, of the natural setting, and of the complicated functions and movements of the city world. Krier (1979) speaks about the introduction and the success of the orthogonal town-plan in Greece which is attributed to Hippodamus of Milet (5th century B.C.). This plan was subsequently widely imitated, especially in new urban settlements. Patterns of certain acquired qualities have been widely reproduced throughout history. Lessons that history can teach us are vital according to Krier and Gehl.
Krier (1979) argues that each period in art history develops gradually out of the assimilated functional and formal elements which precede it. The more conscious a society is of its history, the more effortlessly and thoroughly it handles historical elements of style. He
to turn out something new which risks causing people suffering. The logical and attractive spatial structures left to us by anonymous architects have been improved upon by countless succeeding generations. Gehl (1987), similarly, mentions that in Europe, well- preserved cities from nearly all periods within the last thousand years still exist. . . . They did not develop based on plans but rather evolved through a process that often took many hundreds of years, because this slow process permitted continual adjustment and adaption of the physical environment to the city functions. The city was not the goal itself, but a tool formed by use. . . . The result of this process, which was based on a multitude of collected experiences, was urban spaces that even today offer extremely good conditions for life between buildings.
The study of existing urban settlements, environments that carry within them such valuable information, may reveal to us guidelines for prosperous design or, on the contrary, examples to make sure we avoid during the design process. As Krier (1983) remarks, in every cultural era there are two camps, the one of the traditionalists and the other of the avant-gardists. . . . The one cautiously weighing tradition, the other boldly questioning tradition. Regardless which camp the designer belongs to, there is always valuable insight provided by urban morphology's research on the existing urban fabric.
2.4 Designing With Patterns
Patterns, as a way for humans to identify and organize the physical environment, a designing tool that improves mobility and orientation, have an integral part in the research of urban morphologists who identify exemplars through history.
Various propositions and instructions are found throughout literature regarding the use of patterns in the city. Lynch (1960) while suggesting several qualities that should exist along city paths (key lines of movement) describes: If one or more of these qualities is employed consistently along the line, then the path may be imaged as a continuous, unified element. It may be a boulevard planting of trees, a singular color or texture of pavement, or the classical continuity of bordering facades. The regularity may be a rhythmic one, a repetition of space openings, monuments, or corner drugstores. He continues: Several check points improve the definition. Or a quality (such as the space of the corridor) may have a modulation of gradient at a changing rate, so that the change itself has a recognizable form. The way of organizing a path, according to Lynch, might be called "melodic", in analogy to music. The events and characteristics along the path - landmarks, space changes, dynamic sensations - might be organized as a melodic line, perceived and imaged as a form which is experienced over a substantial time interval.
Since the image would be of a total melody rather than a series of separate points, that image could presumably be more inclusive, and yet less demanding. Besides these references to the pedestrian's experience, it is also suggested (Appleyard, Lynch & Myer, 1964), as one of the principal objectives in shaping the highway visual experience, to present the viewer with a rich, coherent sequential form, a form that has continuity, rhythm, and development, and that provides contrasts, well-joined transitions, and a moving balance. Furthermore, in regards to designing along a path, Gehl (1987) advices that a good rule of thumb . . . is that suitable places to sit should be located at regular intervals, for example, every 100 meters. He also remarks that if activities are to be assembled rather than dispersed in city streets, only the entrances to large buildings, businesses, banks, and offices naturally belong on the facade fronting the public area.
Street life is drastically reduced when small, active units are superseded by large units. . . .
In contrast, examples exist of careful planning in which holes and voids are not accepted, where large units are situated behind or above the small units along the facade. It is these patterns on the scale of the path and the facades fronting it, that the thesis intends to focus on.
Contributing to the body of work in the field of urban morphology, PCT aims to capture the dynamic experience of existing patterns in our built environment that create urban rhythms. Complementing works that suggest mapping techniques through a more static approach, through imaging and inventories, this thesis suggests a quantification method that could inform architects and urban designers of how we perceive facades from a mobile perspective and examines the possibility of using such inventories constructively for this purpose. This method could provide insight into our existing environments and how we could learn from them but also into how human perception functions in relation to them.
3. Methodology
3.1 Definitions and specifications
PCT (Pattern Correlation Tool) is defined as a method comprised by a mapping process for data collection and a mathematical model which analyzes the data and provides with quantitative results that are used as comparative indexes for the correlation of an x amount of patterns. A pattern is defined as a geometric motif (in the physical space) that can be described by a sequence (an arithmetic progression).
The geometric motif of physical objects along a path (along the path's facades), that exist in the three-dimensional physical space, will be treated as a linear mathematical sequence. This will be achieved by projecting the physical objects and their path, onto a two-dimensional plane. This projection transforms them into nodes distributed along a line. These distributions will be used with PCT. Paths, as defined by Lynch (1960), are the channels along which the observer customarily, occasionally, or potentially moves. . . . People observe the city while moving through it, and along these paths the other environmental elements are arranged and related.
The application process of PCT is influenced by its purpose in the context of each research. The correlation of patterns along one or several paths may carry limitless possibilities. A meaningful delimitation is to examine the correlation (proximity or deviation) of patterns identified along selected paths/path, to a reference pattern that carries certain distinctive qualities of interest. In this way, the evaluation of one path or a comparison of several paths with each other, in regard to certain distinctive qualities, is achieved. To do so, PCT performs a comparison of the identified patterns, that certain elements form along selected paths, to a theoretical reference pattern or to an existing reference pattern in the real physical space, and the comparisons to this reference pattern are correlated with each other. This process can be adjusted to alternative research purposes (and according to the scale of the research).
Patterns of any kind of physical element (parked cars along the sidewalk, benches, trees,
by the Sanders & Woodward method though, which studies facades, and considering the facade streetscape as a very important factor in shaping the city experience, this thesis suggests PCT as a technique to study patterns along facades. As Krier (1983) notes, the facade is the most essential architectural element capable of communicating the function and significance of a building.
3.2 Steps of the process
Step 1. Select the study's components.
Select the desirable path/paths to study (define their length by setting a start and an end point).
Identify and select the kind of elements along the selected path/paths whose pattern will be studied.
Select the reference pattern, theoretical or existing, that carries certain distinctive qualities of interest.
Step 2. Mapping.
Measure the position of every element's instance (e.g. windows) along its path, and by the process of projection onto a two-dimensional plane, transform the collected data into node distributions along a line (with the line's length being the length of the selected path). This mapping process is done for every kind of element separately. Inventories with the node positions are created. Node positions are expressed in the working unit of measurement for the study (e.g. meters, feet etc.).
The working unit of measurement isn't of any importance to the results since it doesn't affect them. It could be important though in facilitating the measuring process and in determining the means of mapping.
The transformation of the mapped patterns into distributions is done through a simplification process.
The path's height level variations (a path may ascend or descend) or its deviations along a straight horizontal axis (street turns, curvatures) are not taken into consideration since both the observer and the patterns themselves follow evenly these variations. Hence, the path is unwrapped and projected onto a two-dimensional plane, taking the form of a straight line.
The elements' dimension that is collinear to the path is the one taken into consideration (e.g. the window's length along the facade wall and not its height or its frame's width), since the patterns are situated along the path. If this dimension is not parallel to the path (e.g. the window is placed on the facade with a divergent angle in relation to the path) then it is the projection of this dimension to the path that is taken into consideration.
The start and the end point (edges) of this measured dimension are mapped for every element's instance. According to whether there is value to compare the patterns by the instances' edges or not, either the distributions of the edges are used or the distributions
of their midpoints (the midpoints are calculated and simplistically represent every instance as a single node).
Example of elements' projection as nodes along a line.
(The elevation drawing is just an example, found in midsteeplequarter.org)
The distributions occurring from the mapped patterns will be referred to as "actual distributions".
Step 3. Calculate the reference distribution.
After the completion of the actual distributions follows the calculation of the distribution that derives from the reference pattern.
This distribution will be referred to as "reference distribution".
The reference distribution is calculated for every element separately and is depended on the amount of every element's instances along the selected path and on the path's length.
Through this process every element's instance is acquiring two node positions (if it is represented by its midpoint): one along the actual distribution and a corresponding one along the reference distribution. Four node positions are acquired in the case of using a distribution of edges.
DOOR’S MIDPOINT
GLASS FRONT’S EDGE START POINT GLASS FRONT’S EDGE END POINT
FACADE’S GROUND FLOOR ZONE
Step 4. Compare the distributions.
The comparative method is comprised by the calculation of two statistical indexes, the Mean Absolut Error and the Standard Deviation. For every node, calculate the difference of its position on the actual distribution from its corresponding position on the reference distribution. The difference might be of negative value due to the event of the actual positions occurring before the corresponding reference positions, while mapping towards the direction of the path. Therefore, the absolute value of every difference is taken into consideration, since its sign is not relevant, due to the context of the study belonging to the physical world.
This absolute value is represented by the term "error" in the Mean Absolut Error index.
The term "error" though won't be used in this methodology because it's not an accurate description of what this difference represents. The deviation of the actual distribution from the reference one is not inherently negative to be referred to as an error. Hence, this absolute value will be referred to as "DV" (Deviation Value).
Mean Absolute Error index:
For every element (for n instances), calculate the Mean Absolute Error (to be referred to as "MAE").
n DV
i=1
n
MAE is an index that measures the actual distribution's amount of deviation from the reference distribution. The index has no intrinsic significance. Its significance is relative to other MAEs since it acquires it through the comparative process.
For example, the MAE of the window pattern along a certain path being 27 doesn't provide any actual information unless it is compared with the MAE of the window pattern along another path which might be 42, meaning that the first path's window pattern is leaning closer the reference one.
The length of the selected paths is irrelevant (e.g. all 1km or not) since it doesn't affect quantitatively the correlation of the actual distributions to the reference distribution. The reference pattern can potentially be customized to any path length as well as to any amount of element instances. If it can not, it might be due to its inapplicability to the study thus its use should be questioned. The length of the paths can be of significance however, in the beginning of the process, while selecting the paths. At that stage it is reasonable to set a length that encompasses a meaningful amount of element instances or the whole of an already identified pattern or one that is sufficient for comparison to the reference pattern.
Standard Deviation index:
Additionally, for every element (for 𝑛 instances), calculate the standard deviation of its DVs (to be referred to as "SD").
(DV - MAE)2
ni=1
n
This index measures the dispersion of the actual distribution's correlation to the reference distribution. How inconsistently the actual distribution deviates from the reference distribution. MAE can not provide this information on its own. SD provides further information on what MAE represents, because even in the case of having similar (or even same, although very improbable due to the complexity of the physical environment) MAEs between several paths, let alone them being very different, their actual distributions might differ substantially regarding how they are structured (e.g. sporadically, scatteredly, orderly, densely etc.).
The results of PCT, MAE and SD combined, provide quantitative information that describes the amount of deviation of a pattern from another and how inconsistent this deviation is. These results are comparable between paths similarly to other methods, such as the street network analysis in respect to Circuity, a term which is defined as the ratio of network distances to straight-line distances and is an important measure of urban street network structure and transportation efficiency (Huang & Levinson, 2015). The Circuity method measures the amount of deviation a distance between two points in the city has when measured via the actual city street network, from their theoretical straight-line distance. This method, applied on large scales, provides useful information about the efficiency of walking and driving route networks of whole cities.
4. Test
4.1 Steps of the process
Step 1. Select the study's components.
Select the paths:
PCT will be tested on a sample of three paths in the city of Stockholm. The chosen paths are three streets, two of them situated in the city core and the third being suburban. The amount of the paths was decided empirically during the mapping process. It was an amount that seemed sufficient for the test demands and at the same time easy to handle inside the time constraints of the thesis. Regarding the kind of the paths, the two city streets were selected because of their large variety of physical elements and potential patterns due to their plurality of uses and construction periods. The third street was
interesting parameter for the comparative nature of the study. The parts of these streets to become the three paths and their length, were decided based on an effort to bundle their user's experience. It was presumed that more users are to complete more often a full walk between two connective nodes of public transport, or a public transport station and an area of high activity. So, this segment of the street is more often experienced as a whole and independently from the rest of the street. According to this principle the start and end point of the paths were chosen to be, for example, a metro station or a major crossroad. In other words, where the path meets an edge or a node, according to Lynch's categorization.
The paths can of course comprise several streets with a variety in direction, connecting two points inside the city's street network, that will be unwrapped and subsequently projected onto a straight line. For the convenience of the thesis scope however, the three selected paths comprise three autonomous streets without major direction changes. The selected paths are:
1) Along Birger Jarlsgatan.
Start: The fork junction with Karlavägen.
End: The Östermalmstorg metro station.
Moving towards the path's direction, the left side facade was mapped.
2) Along Hornsgatan.
Start: The junction with Götgatan.
End: The Zinkensdamm metro station.
Moving towards the path's direction, the right side facade was mapped.
3) Along Tranebergsvägen.
Start: The junction with Bengt Tranas väg.
End: Tranebersplan.
Moving towards the path's direction, the right side facade was mapped.
From now on, the three paths will be also referred to with an abbreviation of the street names they belong to, i.e. BJ, HG and TV.
Select the elements:
The selection of the elements begun with a filtering process, narrowing down the focus area of the streetscape. Through this process it was decided that the test will focus on the ground floor level of the facades. According to Krier (1983) the base of a building, its ground floor zone, is without doubt the most important urban element of a facade. . . . The ground floor has a particular importance in urban life. Because this area is most directly perceived by people, it often serves for the accommodation of shops and other commercial enterprises. Given the nature of business, such a ground floor zone is also subjected to frequent change. But, not only it seemed reasonable to delimit the test to the ground floor zone because of its interactive interest but also because of its impact from a sensory perspective. Gehl (1987) explains: The senses are essentially frontally oriented, and one of the best developed and most useful senses, the sense of sight, is distinctly horizontal. . . . The downward field of vision is much narrower than the horizontal, and the upward field of vision is narrower still. The field of upward vision is reduced further because the axis of vision when walking is directed approximately ten degrees downward, in order to see where one is walking. A person walking down the street sees practically nothing but the ground floor of buildings, the pavement, and what is going on in the street space itself. Furthermore, Lynch (1962) mentions that is unpleasant if the line of sight just at eye level is an ambiguous one. The ambiguity may be caused by a narrow visual barrier at that level, or by a vertical surface terminating there. Preferably, vision is either kept clear at this sensitive elevation, or is decisively blocked. Therefore, for this test, any reference to the facade will imply only its ground floor level, which, according to Karssenberg (2016), clearly impacts public life. As he elaborates, in front of active facades, pedestrians move slower, more people stop, and more activities take place on the friendlier, more populated street segments. When we add it all up, we see that the number of stops and other activities is seven times greater in front of active rather than passive facades. One main conclusion, then, is that lifeless, closed facades pacify while open and interesting facades activate urban users. It is important to note that the activity level of a street is quantitative in principle—a measure of how many people and how much life and activity available. However, a higher activity level does not necessarily indicate a better urban quality.
The selection of the physical elements was not done according to a descriptive architectural classification, but instead, it was based on the user's interpretation from a sensory perspective. Other studies, such as the one mentioned before by Sanders &
Woodward, approach the physical elements from a more architectural or structural point of view. Building parts, materials or elements of design (etc. windows, roofs, beams, stairs, pillars, pediments, tiles, stones, murals) are inventoried. Colors or shapes (etc. convexity, concavity, niche, extrusion) are registered. On the contrary, in this test, only elements that affect significantly the user's interaction with the streetscape were selected. The elements that, as Sanders & Woodward (2015) describe, can be distilled into the elementary components of the . . . street facade system, that are measurable and can be tabulated into a syntax for evaluation and/or application. Lynch (1962) mentions that spaces vary in effect by the way in which they are entered, passed through, and left behind, and by the related spaces that precede and follow them. Moreover, window openings, according to Krier (1983), which repeat themselves again and again in succession with the wall elements, create the contrasts of open-closed, dark-light, smooth and rough surfaces. It is the syntax
of all these fundamental aspects of the physical space that affect the senses of touch and vision, the main senses that shape the user's experience of the physical space in relation to their mobile state. The elements were interpreted and classified according to such aspects.
The ones decided for the test are the following pairs of opposites: Accessibility - Blockage, Transparency - Opaqueness, Interactivity - Dissociation. They describe mostly the physical character and visual depth that vary between the elements. Consequently, the initially selected elements are:
Doors
Commercial Residential Public Private Windows
Shop windows (to be referred to as glass fronts) Roads
Parks Alleys
Recessed fronts Landmarks Balconies Outdoor tables
Spaces may be enclosed by opaque barriers, or by walls which are semitransparent or broken by gaps or windows mentions Lynch (1962). Additionally, according to Krier (1983), every opening, whether door or window, means the violation of the wall. These violations . . . give direction and appropriate meaning. Doors play a decisive role in this context because they prepare the visitor for the spatial event to come. . . . A crucial pre-condition for our reflection is to recognize the door as being an important symbol. Thereafter, if we consider the walls as a barrier, an element identified with absolute blockage and opaqueness, it is the rest of the elements, carrying any other gradation of these parameters, that create any rhythmical alteration along the paths. This gradation may vary not only between different kinds of elements but also among instances of the same element. For instance, and in reference to doors, Krier (1983) points out that beyond the many versions of door formats available, the meaning of the door can vary according to its purpose. He explains that it is not always the scale of the human body which determines the size of the door. . . . Quite often, for everyday purposes, a door within a larger door was conceived, which could be used easily by people just wanting to go through. But when major events occurred, the entire over-dimensioned door was opened.
Hence, commercial doors are identified with high accessibility and interactivity, whereas private doors with respectively low. Residential doors with high accessibility but low interactivity. Windows with average transparency (due to restricted visual depth by curtains, shutters or lack of light) when glass fronts with significantly higher. Roads are interpreted with the highest transparency as well as with very high accessibility.
Here, it should also be mentioned that although outdoor tables are not a part of official urban design processes but rather a more direct, bottom-up urban intervention, it is an element worth of including in the study since it affects significantly the experience of the
The life on the sidewalk in front of the café is the prime attraction. Almost without exception café chairs throughout the world are oriented toward the most active area nearby. Sidewalks are, not unexpectedly, the very reason for creating sidewalk cafés.
Select the reference pattern:
The selected reference pattern is a theoretical one. It is the pattern of the equidistant distribution of element instances along the path. A geometric motif that is described by the sequence x2 = x1 + c (xi = position of the element's instance, c = a constant value), which means that all the instances of this arithmetic progression are equidistant.
For this test, the reference distribution will be referred to as "theoretical distribution".
The choice of the reference pattern, to be a theoretical one and not an existing, is due to the lack of information of an existing pattern that carries certain distinctive qualities of interest. The equidistant distribution carry the inherent properties of absolute symmetry between its instances and of absolute constancy of its motif. Properties that are mostly of mathematical interest and probably not that interesting from an urban morphologist's perspective since there is no argument that the equidistant placement of elements in our environment is in fact desirable. The equidistant placement of building elements however, has been a common practice in classical and neoclassical architecture, for example, but this practice is mostly limited to the extent of a single building. The test's context extents on the scale of kilometers along the streetscape, so no assumption that would connect architectural qualities of symmetry to this larger scale is made. Nevertheless, since this thesis doesn't have an evaluative character regarding the results themselves, this reference is selected for the test. Additionally, the distinctness of its character, being a very simple distribution, may provide very clear data about the nature the actual distributions.
Furthermore, it facilitates the acceleration of the process because it simplifies the calculations.
Step 2. Mapping.
The process of mapping was physically done by myself, walking along each one of the three paths and taking measurements for every instance of all the selected elements.
Consequently, the unit of measurement was steps. It is reasonable to lack precision but the alternative of taking precise measurements with an instrument, possibly in meters or even centimeters, seemed infeasible for this thesis and it is even futile for the purpose of PCT.
Steps is a convenient unit that determined the means of mapping to be: walk and count.
Example of the mapping process along BJ.
Measured lengths of the three paths:
Birger Jarlsgatan: 1480 steps Hornsgatan: 1263 steps Tranebergsvägen: 925 steps
The mapping was placed on separate days for every path, starting around midday. It was done on a slow walking pace, trying carefully to keep a steady step range that would be used as the measurement unit. Every time I reached a midpoint or an edge I stopped and recorded the measurement. The first path was BJ and the data collection was done on 12 May 2020 from 11:00-16:00. For HG it was done on 15 May 2020 from 12:00-15:00 and for TV on 20 May 2020 from 16:00-18:00. The process was faster after the first path because it was done more confidently and of course the other two paths were shorter. For recording the measurements, a linear outline of the path was used, sketched on a notebook. Every instance of every element was marked on it with a pen, together with the number of the step at which it was measured.
Example of the linear outline used for recording the measurements along BJ.
MIDPOINT MARKER EDGE START MARKER EDGE END MARKER COMMERCIAL DOOR RESIDENTIAL DOOR
PRIVATE DOOR
GLASS FRONT
OUTDOOR TABLES number of step
At this point, the way of how the elements will be used with PCT had been already thought and decided. Some of these decisions helped to optimize the mapping process and avoid ending up with incomplete inventories (it was more efficient to know what to measure, how to measure it and what not to measure instead of just make a general data collection and clear it up afterwards). For example, mapping some elements by their midpoint already during this stage, than calculating it afterwards from their edges, made the process more time efficient.
An important parameter, was whether an element was common for all the three paths or not. Therefore, the elements of Balconies and Landmarks were left out of the study since they only appeared on one or two paths and without a significant frequency. A comparative study concerning the common elements among the paths seemed more comprehensive. Outdoor tables were kept despite their appearance only on two paths.
Another important decision concerned the elements' classification. Roads, Parks and Alleys, would not be compared separately but instead they would be merged into the group of Building Block Breaks (together with any other open space already identified as a block break). Comparing elements grouped according to the parameter of Accessibility - Blockage (meaning open reachable space - building or wall as a barrier) seemed more insightful about the accessible options the user has along the path, than just comparing the patterns of the Alleys independently for example. Similarly, Roads was also merged with Garage Doors, a group named into Vehicle Access. The pattern of the accessible points for vehicles is of interest here, concerning the likelihood for the path's user to come across a vehicle.
It was also decided which elements will be compared by their edges or by their midpoint. Given their relatively small size, Doors define a rather confined space where their functional role is completed (entering - exiting). Taking into consideration their very similar size and how it is determined by the scale of the human body, it was decided for them to be marked by their midpoint already during the mapping process, which was accelerated considerably given the large amount of the Door instances. The same principle applied also to Windows. On the contrary, Glass Fronts, depending on their length and content, may affect the user's experience extensively. Therefore, Glass Fronts were marked by their edges.
Similarly, during the transformation into distributions, Roads were represented by their midpoints when used in the Vehicle Access group. They function here as points which hold a potential interaction with a vehicle. Their length is not important. On the contrary, when used in the Building Block Breaks group, their length was taken into consideration, as part of the cumulative length of the grouped elements' instances. This group was projected by its edges.
Doors leading to spaces of public gathering, considered as an access point to public space, were grouped with entry points to parks and pedestrian roads into the group Public Doors, represented by their midpoints.
After the filtering process the elements that are going to be used are:
By their midpoint:
Commercial Doors Residential Doors Public Doors Private Doors All Doors Windows Vehicle Access
By their edges:
Building Block Breaks Glass Fronts
Recessed fronts Outdoor tables
According to the previous decisions, the inventoried elements were transformed into distributions of correspondingly 1480, 1263 and 925 steps in length. The actual distributions were completed, ready to be used with PCT.
Amount of instances
Actual distribution
midpoints Path start 0 steps 1st instance 665 steps 2nd instance 669 steps 3rd instance 673 steps 4th instance 744 steps 5th instance 747 steps 6th instance 754 steps 7th instance 758 steps Path end 1480 steps
Example of tabulated actual distribution of Windows along BJ.
