In December, 2008 a specialist workshop titled Spatial Concepts in GIS and Design was hosted by the University of California, Santa Barbara Center for Spatial Studies (also known as spatial@ucsb). One of the core questions discussed was ‘to what extent are the fundamental spatial concepts that lie behind GIS relevant in design?’ One approach to answering that is to identify the sets of fundamental spatial concepts that lie behind each field—GIS and ‘design’—and find their intersection. We undertook the first step of that procedure last year, and the preliminary results appear in the Concepts section of the teachspatial.org web site. The field of design in this inquiry is limited to urban design and architecture, with concepts drawn from Kevin Lynch’s classic of design theory, Good City Form and Christopher Alexander’s The Nature of Order.
In preparation for the GeoDesign Summit held at ESRI in Redlands in January, 2010 I’ve undertaken the second step, finding the intersection. First, I took a closer look at Good City Form and added 24 spatial terms to the 26 already gathered from that work and from Alexander. This meant taking a more liberal view of what constitutes a ‘fundamental’ spatial concept (versus simply an important one) than I had previously. The result was 50 terms for spatial concepts underlying urban design and architecture.
Sets of concepts
For the spatial concepts underlying GIS, I distilled 81 distinct terms from the 160 found in our earlier investigation of 12 geographic sources. Finding the intersection required some subjective interpretation in mapping of terms with similar meaning. Of 50 design terms, 23 exactly match terms in the GIS list. 17 more could be reasonably, if imperfectly, mapped to a term in the GIS list. Of those, all but two were to GIS terms in the exact match group. The resulting intersection is then a set of 25 terms for important spatial concepts common to the fields of GIS and urban design, alphabetically: adjacency, boundary, center, cluster, connectivity, containment, density, direction, distribution, gradient, grid, hierarchy, identity, neighborhood, network, path, pattern, place, region, scale, shape, space, spatial context, spatial interaction and surface.
The remaining 10 design terms either missing from or not mappable to the geographers’ 81 include: flow, intersection, landmark, symmetry, navigation, part, positive space, structure and viewshed. Of the 17 ‘reasonable’ mappings mentioned earlier, 8 were ‘imperfect’ enough to add to this group of the missing: access, contrast, form, grain, order, orientation, roughness and simplicity.
For the sake of argument, one could say the first set of 25 terms represent concepts supported in GIS, and that the second set are 18 spatial concepts important to design, but absent or insufficiently considered in GIS. At least it is a simplistic starting point for considering how GIS might be made more useful for urban designers. We can walk through each of the 18 and consider whether it corresponds to missing functionality. Before doing that, it’s worth considering why there would be differences between GIS and design lists.
Many of 25 ‘intersection’ terms and all 8 imperfect mappings are understandably used in different senses by the two fields. A large proportion of GIS applications are descriptive and analytical, aimed at learning how things are (or were) and why. Design can also involve description and analysis of existing conditions, but principally to support normative reasoning, which concerns how things should be in order to satisfy human requirements and purpose. The ‘missing 18’ reflect this difference, and the frameworks developed by Lynch in Good City Form are illuminating.
Good City Form
Lynch suggested there are three historical classes of normative theories or models of city form: magical, machine and organism. Each emphasizes particular human purposes, and spatiality is fundamental throughout them all. The magical theory, which addresses needs for spiritual meaning and cosmic connection, is seen particularly in ancient Chinese and Indian design. Orientation to the cardinal directions is meaningful. Order, achieved with grids, spatial hierarchy and symmetry, is important. Gateways and approaches are imbued with magical function. The magical approach is also concerned with protection in enclosures, from the forces of chaos and more concrete enemies.
The machine model concerns parts, wholes and function. It is "still alive…in the powerful concepts of systems analysis, which models the world as a set of distinct parts linked by well-defined dynamic connections, like a giant aeroplane" (p 86). It embodies "explicit rationality," in achieving human purposes like "equity of allocation, good access, broad choice, smooth technical function, productive efficiency, material well-being, physical health and the autonomy of parts…" (Ibid).
The city as organism is a more recent model, which Lynch associates with the "rise of biology" beginning in the 18th century. This organism has a "sharp external boundary," with differentiated but indistinctly bounded parts. It is purposeful and dynamic, its form self-adjusting, self-repairing and self-regulating.
Lynch found many faults in these approaches to city form but reaffirms an underlying principle central to all three: that the spatial form of the city is the single most important determinant in meeting human purposes. He proposes a set of five "performance dimensions…characteristics which refer primarily to the spatial form of the city" (Ibid, p 111-118) and which correspond to five broad classes of human purpose: vitality, sense, fit, access and control. The use of the term dimension is not accidental—the degree to which purpose is afforded by a design is measurable. The desirable forms, methods and measures are spelled out at some length, and not surprisingly these are the spatial concepts found in our list.
The performance dimensions are defined briefly by Lynch as follows: vitality ("the degree to which the form…supports the vital functions, the biological requirements and capabilities of human beings…"); sense ("the degree to which the settlement can be clearly perceived…and structured in time and space by its residents…); fit ("the degree to which the form and capacity of spaces, channels and equipment…match the pattern and quantity of actions that people customarily engage in"); access ("the ability to reach other persons, activities, resources, services, information or places"); and control ("the degree to which the use and access to spaces…are controlled by those who use, work or reside in them."
The Missing 18
Do the concept terms on the ‘missing 18’ list refer to forms, methods and measures not supported well by state-of-the-art GIS software? Again, they are: access, contrast, flow, form, grain, intersection, landmark, symmetry, navigation, order, orientation, part, positive space, roughness, simplicity, structure, territory, and viewshed. A complete answer is beyond the scope of this posting, but I will summarize the meanings found in Lynch and Alexander as a start. Note that Lynch is keen to measure and Alexander seldom mentions objective metrics.
- Access for Lynch refers to the relative cost (in time, money) of reaching locations or acquiring resources from locations. Accessibility in this sense is modeled and analyzed routinely in GIS, using both raster and vector data models.
- Contrast for Alexander "is the thing which creates differentiation, and allows differentiation.
- Flow in the design sense is a very general term referring to the movement of people and both material and non-material resources. It is a key concept which could be handled better–by which I mean more generically–in GIS. For our geographic sources, it is implicitly associated with the terms network and connectivity, but surprisingly absent from spatial term lists. There exist complex data models which handle flow for specialized cases like utility lines and transportation, but many kinds of flows have intrinsic temporal qualities that are not managed well. An ST-Path datatype such as suggested by Miller (2005) has not been implemented commercially that I’m aware of and would be useful.
- Form and structure are closely related. Form is the most general term for spatial (physical) arrangement used in this urban design literature. It subsumes shape and structure, which for Alexander is, "at the scale of a small place…the sense of how its parts fit together." Structure can be specified (in design) or discovered (in analysis). Whether GIS supports well enough the identification, typing and analysis of ad hoc spatial structures (clusters, functional networks, etc.) seems to me an open question.
- Grain is defined by Lynch this way: "…measurements can be made of the influence of surrounding points on any given location. One such measurement is …grain. If (some characteristic) is discontinuous…then the measure is grain, or the fineness of the mix of diverse characters" (p 355). It would seem that several geostatistical methods address spatial influences, some of which are making their way into commercial GIS toolboxes.
- Intersection, landmark and navigation all refer to human movement within a settlement. GIS supports sophisticated transportation network models. Although navigation by means of survey knowledge and landmarks is an active cognitive geography research area, it’s not clear we should expect to model human factors like navigability in GIS.
- Order in the design sense used by Lynch is an overarching concept one might call ‘intentional pattern’ that takes many forms, notably grids and spatial hierarchical arrangements.
- Orientation is used in two senses–as a synonym for cardinal direction (obviously supported in GIS), and in the cognitive sense of someone understanding where they are relative to landmarks and cardinal directions.
- Parts are an important aspect of all the theoretical models touched on earlier. Whether viewed as magical spaces, mechanisms, organisms or dynamic ecological systems, urban settlements are composed of connecting and otherwise related parts.
- Positive space is a term used by Alexander to suggest settlement features have, visually and spatially, strong positive identities. It seems to relate to the perception of figure/ground.
- Roughness as defined by Alexander corresponds to intentional spatial heterogeneity. To quote: "(Roughness) is an essential feature of living things, and has deep structural causes," and "[it] does not seek to superimpose an arbitrary order over a design, but instead lets the larger order be relaxed, modified according to the demands and constraints which happen locally in different parts of the design."
- Simplicity is meant in a geometric sense. For Alexander, "everything unnecessary is removed."
- Symmetry is a powerful design principle; for Alexander it is best used selectively, for example to create a sense of centrality at smaller scales. Measures of symmetry using GIS, while not unknown, are unusual.
- Territory is for Lynch, a region asserted to be under human control. I can imagine theories of control expressed in measures of connectivity, adjacency or containment, but lack the expertise to judge whether GIS functionality is currently adequate for design applications.
- Viewshed calculations are a basic analytical function in commercial-grade GIS.
I expect to continue this inquiry following the GeoDesign Summit, where I expect to learn more about the practice of design at geographic scales, and welcome comments your about this topic!
Alexander, C. (2004). The Phenomenon of Life: The Nature of Order, Book One. New York: Oxford University Press
Lynch, K. (1984). Good City Form. Cambridge, MA: MIT Press
Miller, H. (2005). A Measurement Theory for Time Geography. Geographic Analysis, 37(17-45).