| GISdevelopment.net ---> Application --- > Urban Planning --- > Overview |
Object-oriented GIS Data Modelling for Urban Design
Simon Yanuar Putra
Department of Architecture, School of Design and Environment,
National University of Singapore, 4 Architecture Drive, 117566 Singapore.
Tel: +65-6874-4536 Fax: +65-6779-3078
Email: sdep1258@nus.edu.sg
LI Wenjing
MA Research Student
Department of Architecture National
University of Singapore, 4, Architecture Drive, 117566 Singapore
Tel: +65-6874-4536 Fax: +65-6779-3078
E-mail: g0203425@nus.edu.sg
Perry Pei-Ju Yang
Assistant Professor
Department of Architecture
National University of Singapore 4, Architecture Drive, 117566 Singapore
Tel: +65-6874-3465 Fax: +65-6779-3078
E-mail: akiyangp@nus.edu.sg
Simon Yanuar Putra
Department of Architecture, School of Design and Environment,
National University of Singapore, 4 Architecture Drive, 117566 Singapore.
Tel: +65-6874-4536 Fax: +65-6779-3078
Email: sdep1258@nus.edu.sg
LI Wenjing
MA Research Student
Department of Architecture National
University of Singapore, 4, Architecture Drive, 117566 Singapore
Tel: +65-6874-4536 Fax: +65-6779-3078
E-mail: g0203425@nus.edu.sg
Perry Pei-Ju Yang
Assistant Professor
Department of Architecture
National University of Singapore 4, Architecture Drive, 117566 Singapore
Tel: +65-6874-3465 Fax: +65-6779-3078
E-mail: akiyangp@nus.edu.sg
Abstract
The latest advancement of object-oriented (OO) data modelling in Geographic Information Systems (GIS) and Computer Aided Design (CAD) has opened an opportunity in data modelling for urban design. In the CAD realm, the concept of OO data modelling has been materialized in a new concept called “model-based” design, the latest development in CAD-based design methodology. Although OO data modelling has been developed in GIS realm long before its CAD counterpart, it has never been design-oriented since GIS was not initially developed for spatial design purposes. In this research paper, the Database Management Systems (DBMS) is proposed as the foundation for the development of GIS-based urban design support system, which is surprisingly as comprehensive as the GIS application to urban planning. The design of database for GIS is so important and fundamental that it affects the performance of the proposed GIS design support system. We argue that the urban design practice has not fully and intensively utilized the GIS tools because the database has not been designed appropriately for the urban design requirements.
Through data modelling, which have entered a new dimension in OO data modelling, we can design a specific type of database to support the GIS application of urban design. OO data modelling allows us to define our own types of objects through topological, spatial, and general relationships, which can help capture how these objects interact with other objects. The data models were generated programmatically using ArcObjects components and Visual Basic for Application (VBA). The OO modelling will be conducted in Unified Modelling Language (UML) scheme style, with a possibility to use CASE tool. Data modelling runs through conceptual, logical and technical stages, which will be finalized in a relational spatial database. A chosen database for utilization is Geodatabase from ESRI, with other possible implementations such as XML Schema.
The paper also discusses the benefit of designing the data model for urban design. The data model design is regarded as an essential part of the whole urban design process, where Patrick Geddes’ traditional design methodology “survey-analysis-plan making” are redefined as “data model - GIS spatial analysis – design decision making”. With the adoption of object-oriented approach, GIS data modelling will be able to capture the spatial and non-spatial relationships among urban physical objects. The experiment will use latest data modelling techniques, such as semantic ontology and object-oriented approach to test the applicability of data modelling to urban design decision-making.
To what extent the new techniques of data modelling could be applied to urban design process? The paper will discuss the potentials and limitations of the applications of OO data modelling to urban design. Based on the results, a critical analysis of urban design guidelines, principles, and their model-ability will be discussed. The long-term objective of the research is to develop the “data model”-based urban design as a new urban design methodology.
1. Introduction
GIS (Geographic Information System) data modelling is a methodology for designing spatial databases, a type of design activity of its own. However, the way database designer designs a database is distinctively different from the way urban designer designs a city. Why do we need database, the spatial database in particular, for urban design? How could we make use of the spatial database management system (DBMS) technology in the urban design application? To what extent the currently available DBMS technology could be applied to the urban design process? Based on these preliminary questions, the GIS database research will be moved a few steps further through the elaboration of theory and urban design case studies.
Judging from the vast amount of information concerned by urban designers, urban design definitely needs tools to manage this information. GIS has a track record in handling urban information, especially in the field of urban planning (Brail & Klosterman, 2001; Han & Kim, 1990; Zaki, 1998, Laurini, 2001). In the recent development of GIS technology, it begins to show potentials for the design-oriented uses. The way GIS will support design process is through its spatial database, with its competitive strength. The main function of spatial database is to store urban information in digital format for further GIS process, which composes queries, analysis, visualization, and simulation. A new design-supporting GIS function is proposed in the research work, which will use the spatial database as a sketching platform for urban design proposals.
To start with the discussion of data modelling, there is a crucial need to clarify the definitions of model and data model, when the term “model” has different implications and possible meanings in different disciplines. The definition of the model, according to system science, is the “representation of reality”, which selectively expresses certain relevant characteristics of the observed reality, and consists of objects or systems that exist, have existed or may exist. As a representation or an abstraction of the observed reality, a model is always a simplification of the complexity of reality, which makes the reality more understandable and operational (Crowther & Echenique, 1972).
Models may be classified as descriptive, predictive, explorative, or planning according to its purpose. Models can be classified based on their constructed elements, including physical models (solid materials such as architectural wood model), analogue models (drawings and plans), conceptual models (words or mathematical functions) and digital models (computer aided design drawings or data model). A model can be either static or dynamic regarding to its nature towards temporal elements.
A data model is not a physical building model, a city map, urban design drawings or a mathematical model of urban population growth. A data model is an abstraction of the real world that employs a set of data objects that support map display, query, editing and analysis (Zeiler, 1999). The data model can be applied to different aspects of the design process, such as procedure models, predictive models, digital urban models, behavioural models, dynamic agent-based models, etc. A data model, as discussed in this paper, is mainly descriptive, although it can function as explorative through customization. It is made of digital database and may be represented by analogical materials such as digital maps, plans and graphs. We propose that the GIS data model is to be designed as a bridge between urban reality and its spatial representation in GIS. It is the premeditated way to organize and represent the information of urban reality in a spatial database. Since all GIS-based analyses are supported by the spatial database, the data model is regarded as the foundation for urban design analysis and design decision-making.
In the area of computer aided architectural design (CAAD), the data modelling can now generate building system database for building management purpose. The data modelling can be utilized to support model-based building design using object-oriented CAD. For urban planning, the data modelling can be applied to the development of planning support systems (PSS). Unfortunately, the data modelling is not yet available for urban design purpose. A data model is designed for representing the relationships between buildings and their components. It is yet to be developed for an object-relational system (Cote, 2002). Using the test case of Singapore Management University’s (SMU) urban design, the paper will explore how the data modelling could be applied to the management of urban design guidelines.
2. Development of Data Modelling for Urban Design
The recent development of the model-based design in computer-aided building design has become the motivating example for the experiment of using a similar approach for urban design. Autodesk, Bentley and Graphisoft have developed “object-oriented” CAD packages capable of performing model-based design, which is also known as building information model, virtual building, or parametric building modeller. The motive of this paper is to utilize data modelling technology to support computerized urban design process.
Computerized urban design utilizes spatial modelling technologies such as CAD, and most recently 3D GIS. Most computerized urban design activities will need spatial database of some kind, where data are stored in certain format. The structure, format, and technology level of the spatial database will determine the capacity and capability of the software application used for urban design. Through data modelling, spatial database should be designed to accommodate urban design analysis and design decision-making process.
The idea of applying data modelling to physical design is not entirely new in the realm of design since the systematic design thinking in the 60s and 70s. Forty years ago, Alexander proposed a “set-subset” theory, where he suggested designers to construct analytical design process through problem classification in hierarchical logic. He initiated the use of constructive or schematic diagram of design variables as design tool, which implied a preliminary idea of design programming and data modelling. He argued that performance standards, which resembles urban design guidelines in their roles, can be set up for every misfit variable that exhibits continuous variation along a well defined scale (Alexander, 1964). In his theory, he experimented on redistribution or reclassification of the design variables, which is comparable to the object-oriented (OO) principles, where variables related with the same object are clustered together. Alexander’s problem solving principle of classification is actually the preliminary stage of current data modelling principle. It can be followed by the transformation of “misfit variables” by semantic and logical deduction into objects, subtypes, attributes, validation rules, connectivity rules and relationship rules, which are the basic components of OO data model.
Further researches following Alexander’s earlier work also supported the concept of data modelling for design process. Geoffrey Broadbent, Anthony Ward, Amos Rapoport, Luckman, Best, etc, have discussed the role of operation research (OR) methodology in design process, which has much compatibility with data modelling techniques.
The developments of computerized urban applications in the prior days were emphasized more on the analytical and predictive aspects of the applications. The development of Planning Support System (PSS) was a significant evidence of this trend (Klosterman 2001). A PSS usually contains several levels of predictive urban modelling or simulation, which requires a database modelling as the foundation. Several DSSs and PSSs were built on ArcView version 3.x GIS platform by customizing ArcView system and adding on its original functions. Relational spatial database was not existed in this version; therefore developers could not design spatial database through data modelling, and have to customize the functionality of relational spatial database through its customization language such as Avenue. In the later version of ArcView, ArcGIS 8.x has the capability to build spatial database designed through the data model of UML and XML and the customization platform ArcObjects with embedded VBA. Planners have benefited from the object-oriented ArcGIS Parcel Data Model recently (ESRI 2001), which was actually designed for cadastral purpose, with more emphasis on land ownership and distribution. Although containing urban objects, ArcGIS Parcel Data Model may not be suitable for accommodating urban design activities, since design process needs sufficient built-up physical data.
Spatial database design is very important in the development of planning support systems, since a GIS system must be built on standard components. The components start from spatial database as the foundation, then editing, analysis, simulation, and finally visualization. However, the design methodology of spatial database hasn’t been developed adequately in urban research compared with the predictive modelling and simulation. This argument is strengthened by the minimum findings of the similar PhD theses currently researching data modelling for urban design. One of them is currently pursuing on “Object-Oriented Data Modelling and Warehousing to Support Urban Design” (Koshak & Flemming, 2002). Another research is pursuing on “Restructuring UrbanSim Model into ArcGIS (Geodatabase) Data Model using UML” (Kyuwon Park, 2002), which emphasized on land-use dynamics simulation modelling of the well-known UrbanSim planning model. Judging from the minimum research efforts in this area, we argue that data modelling is still a new area in urban design research.
In the past few years, data models for editing and managing urban information have largely been in the province of specialized computer aided design (CAD) tools used by engineers and architects. This combination of data modelling in GIS, Spatial Data Engine (SDE), through UML or XML format as bidirectional medium, provides a potential for managing virtually unlimited amounts of details representing many aspects of the built environment of the entire city, along with different future or past scenarios. Cote argued that data models for representing the relationships between buildings and their components have yet to be developed for an object-relational system (Cote, 2002). He perceived the urgent need of an appropriate data model when he was confronted by several problems while developing an agenda “towards modelling broad scale urban scenes in an object-relational database”. The first problem is how to associate primitive shape objects to represent built-environment, i.e. buildings. A representation of a building is usually composed of different features, such as points, lines and polygons. Relationships in the spatial database, such as “the windows, associated with hotel rooms, which have a view of the natural gas tanker”, are examples of information that may be desired from a good urban data model. The second problem is the need for data models, which are interoperable between GIS and CAD. These are the problems that can only be solved through data modelling.
In the next section, the paper will introduce data modelling of urban design guidelines. Most guidelines consist of clear regulatory and advisory statements of design rules and relational rules between design objects. The paper will explain how urban design guidelines can be modelled into data models, which in turn will create spatial database capable of enforcing the guidelines in the design process.
3. Urban Design Guidelines
To implement data modelling in the development process of spatial database for urban design, urban design guidelines were chosen as modelling case study. It is because they contain clear statements of desired relationships and rules between urban objects, available to public. They act as a guide for developers and designers, and possibly other agents, in planning and designing development, prepared by local authorities with the participation of landowners, developers, partnerships, business, and community organizations (Cowan, 2002). They are either the result of negotiated agreements between stakeholders or a direct guidance from local planning authority. They can be used to identify and develop performance criteria to measure effective plan execution through performance-based standards (CONCERN, Inc. et al., 2002).
There are several types of urban design guidelines. The first are guidelines of specific places. There are three main types of these: urban design frameworks, development briefs, and master plans, mentioned in hierarchical order. Urban design framework mainly consists of two-dimensional vision of future infrastructure requirements. Development briefs usually contain a more indicative vision of future development form. Master plans are detailed plans usually prepared by the organization that owns the site. The second type of guidelines relates to specific topics. These, usually called design guides, cover topics such as shop fronts, house extensions, lighting and cycling. Singapore Management University (SMU) New Campus Design Competition Guidelines are considered one of these. The third type of guidelines relates to specific policies, such as policies on conservation areas, transport corridors, waterfronts, promenades and green belts. An established example is the user-centred guidelines, which was derived from user specifications and preferences (Katoshevski & Timmermans, 2001). The fourth type of guidelines relates to a whole local authority area. These may give general urban design guidance for the whole city or county.
There are several roles and benefits of urban design guideline in design process, for example, to promote a visually harmonious community. The urban design guideline is also considered as the most innovative and flexible approach among all development regulations. The development of guidelines will support planning policy, encourage collaborations among stakeholders, create vision for the new environment and guide the development of local design standards. Guidelines were involved in almost all urban design projects, such as new development, rehabilitation and urban public space projects (Cowan, 2002; Moughtin, et al. 1999; Friedman, et. al., 2002; Citizens Party of Hong Kong Central Government, 1999). Most design guidelines contain a comprehensive urban design policy and the full range of design considerations that are important in a locality (Punter, et al. 1994). The guidelines can be classified based on the scales of area affected, including region, metropolis, city, and town, neighbourhood, district and corridor and finally block, street and building (Congress for the New Urbanism, 1996).
The development of database for urban design guidelines require resources such as common urban features, i.e. land coverage, lot size, parking, and their guiding regulations, i.e. FAR, setback, land uses, growth space, additions, sections of lots, and building heights (Friedman, et. al., 2002). The guidelines may provide classification of urban features. For example, the lot types can be classified as corner lots, typical lots and lots with alleys. The development of urban design guidelines database requires an integrated and multi-disciplinary team that understands the importance of quality of urban design.
4. The Role of Data Modelling in Urban Design
Urban design process is essentially a modelling process, where designers use models to represent the urban reality and the design intervention within. Urban design process is where evolution of urban models occurs. The mental models are ideas, theories, and regulations, which are processed in the designer’s mind. The urban design process may not be systematically linear. The designers often use analogue models, such as hand-drawing sketches, or even building massive model, to express their ideas. With the advancement of CAD, design process can utilize digital models, such as computer-based drawings, either in sketches, technical drawings, or 3D models. We argue that ideas and the ruling knowledge behind the urban design process can be systematically represented in the data model as well as the analysis model and design decision model.
A data model takes role just as digital and analogue models, to represent the mental model. Data modelling is the foundation for the digital modelling in the way that the former becomes the dominating system of the latter. Unlike analogue and digital models, data model is able to represent not only geometric and functional ideas, but also regulative knowledge, which may be derived from design guidelines. By running the data model for “design” part and for “regulation” part comparatively and simultaneously, the information system can acknowledge the design “compatibility” with the regulation.
As the methodology of databases, data modelling is fundamentally a design process, a creative process, having similar concurrent components with other design process, such as architecture and urban design. Data modelling in the enterprise world have been utilized as a “tool for expressing and communicating enterprise rules” consisting of rules and regulations in the businesses and organizations (Simsion, 1994). The similar principles can be referred to the making of urban design guidelines, which are the documents stating the urban design’s “enterprise rules”. Through data modelling, guidelines can be implemented and embedded in a GIS-based spatial database particularly, where all subsequent digital models will abide to their parent data model. This way, data modelling will extend guidelines’ capability as a support system of design process. As a simplified form of urban reality, to what extent the data model is able to represent the components of urban design guidelines? The question is to be verified through of test case of Singapore Management University (SMU) campus design competition guidelines.
5. SMU New Campus Design Competition Guidelines
In the late 1990s, Singapore government reserved 7 plots in this highly valued central district, for the development of a new campus for Singapore Management University. The plots are located at the centre of current Bras Basah Park, which is a prominent side and currently functioning well as an urban ‘oasis’. The park itself maintains openness and visual connection between Singapore Museum & Art Museum. As an effort to get the best design for its future campus, SMU has coordinated a design competition on 2000-2001, which has received a good response of more than 150 participants from 36 countries. The urban design guideline for SMU New Campus Competition was written to guide incoming proposals to suite the requirements of SMU management. The proposals must express the fulfilment of SMU’s policies and needs, regarding its existence in the CBD’s Museum District.
Figure 2 Urban design guidelines for SMU New Campus Design Competition (SMU, 2000)
6. Data Modelling Methodology
All data modelling techniques share the same three stages: conceptual stage, logical stage, and technical stage. The conceptual and logical stage can be combined into a single stage, usually occurs in modelling simple models. Modelling techniques considered under the conceptual stage are Entity Relationship (ER) techniques, such as Chen’s Original ER developed in 1976 and Barker’s Oracle ER developed in 1990. Object Role Model (ORM) was developed by Terry Halpin in 1995 (www.orm.net), and was acclaimed for more user-friendly conventions and could be easily understood by non-technical users. The rules of ORM data models can capture more ‘enterprise’ rules, equivalent to urban design guidelines, and are easier to validate and evolve than data models in other approaches. The Object-Oriented (OO) Model provides examples of the logical modelling stage, where Unified Modelling Language (UML) is designated for off-line database design and eXtended Markup Language (XML) for tentative Internet database design. Both can be classified in the implementation or technical stage of data modelling. The modelling of urban design guidelines can be done through transforming guidelines’ textual-format rules and regulations into data model’s relationship schema. The process is basically a language transformation.
To model guidelines’ statements of relationships and their rules among urban objects, as described in urban design guidelines, we need to transform them into a database model. We can use the methodology developed below for designing ORM, which is called Conceptual Schema Design Procedure (CSDP).
After following Conceptual Schema Design Procedure (CSDP) closely, we can conclude that:
The outcome of Conceptual Schema Design Procedure (CSDP) is the ORM conceptual data model of urban design guidelines for Singapore Management University (SMU) New Campus Competition. This data model must be transformed to a spatial database in order to make it useful for urban design process. This transformation was conducted through technical data modelling, where we’ve found issues to be tackled:
The data modelling experiment explained in this paper is primarily conducted in Object Role Model (ORM) format. Export to other formats, such as UML and XML, were not included in this experiment yet. Future works will investigate these data model formats for urban design guidelines. A direct interpretation of the ORM model has produced a spatial database containing the urban design guidelines for Singapore Management University (SMU) New Campus Competition 2000-2001. The spatial database “SMU_guide_model” contains feature datasets of existing Singapore urban data, existing SMU site, SMU guidelines, topographical data, tables containing Singapore urban data, and SMU campus design proposals from competition finalists. The feature classes within datasets are presented in Appendix.
The experiment has transformed text-based guideline statements to database components. Different statements are transformed into different components, such as feature classes, topological rules, relationship classes, and domains.
7.4 Preliminary review of urban design guidelines model-ability
In the modelling experiment, we managed to transform 85% of the 39 guidelines statements intended to be modelled. The rest of the statements were not modelled because they were not required in the design process, or they could not be modelled properly, or they needed to use network features, or the information was not available. From all 33 modelled statements, 13 indicated the existence of new spatial objects transformable to feature classes, which must be incorporated in the design proposals. Another 3 statements indicated the needs to acknowledge existing objects in the design consideration. Four statements prescribed specific width requirement of movement path were transformable to domains. Another four statements indicated relationships between two spatial objects, which were transformable to relationship classes. The last four statements described the topological relationships between spatial objects, which were transformable to topological rules.
Although a large part of the design guidelines was transformed to guidelines model, they were not sufficient. We had to add 11 additional topological rules to make the first four rules effective. The additional rules were extracted from logical deduction of the statements, since the guidelines often contain “unwritten” logical statements. The four original statements related to proposal’s design integrity actually required eight topological rules to be implemented in the database, while four “unwritten” topological rules are added to maintain spatial integrity between existing spatial objects. Four additional relationship rules were also required. In conclusion, spatial integrity can be validated in GIS database by adding topological and relationship rules based on logical deduction of spatial organization. The additional rules may take up to 40 – 50% of the total rules modelled.
8. Conclusion
The data modelling experiment of urban design guidelines for Singapore Management University (SMU) New Campus Competition 2000-2001 in GIS has been successfully conducted. After implementing the urban design guideline model on the target GIS system, a spatial database can be created, using the Geodatabase format. The guideline rules and regulations have been converted into object-oriented (OO) rules, which consists of topological rules, validating rules (domain), relationship rules, and connectivity rules. The rules were implemented on urban design -dedicated database elements such as feature classes, domains, relationship classes, and topological rules. The implementation of urban design guidelines itself comes to play when a designer sketches his urban design proposals using GIS system, and the modelled spatial database. The GIS system, after minor customisations, will inform the designer about his design “fitness” according to the modelled urban design guidelines.
The materialization of data modelling for urban design process has maximized the use of spatial database for urban design. We may conclude that data modelling gives more benefits if implemented for complex and large-scale urban design, where more issues, problems, and stakeholders are involved. In design process, data modelling works best on large-scale design environment, which involves many designers and drafters handling the same project. Data modelling increases efficiency in teamwork by maintaining design project’s integrity and consistency. The capacity of an informative database may lead design process to a further step of knowledge-based design. With the rapid development of network technology, data models can be shared among designers as the media for sharing design information and knowledge, which will encourage collaborative design.
However, the introduction of data modelling in existing design community may not be a smooth integration, since it is an alien, new design methodology. The designers’ adaptation to data modelling may be slow, since current data modelling techniques are far from perfect, and the immediate financial benefits are not apparent. Even for small-scale simple design, the method can become too complicated, cumbersome, and irrelevant. It will require some time before data modelling can be accepted as an option in design process. Finally, only urban designers who understand the whole “universe of discourse” of urban design can practice data modelling for urban design.
Along the way, many theoretical questions have been raised on evaluating data modelling in urban design process. Is it really useful, or relevant? What can be covered in data models? Will it affect the urban design process? Will the effect be positive or negative? Will it affect the issue of creativity? What is the best implementation in urban design process? Questions about the “model-ability”, the compatibility of urban design guidelines if modelled in data modelling, may be raised as well. Several methods have been developed to hypothetically measure guidelines’ “model-ability”, such as its logical integrity and information system friendliness. We can also compare the guidelines’ model-ability and data model’s compatibility to measure the modelling success rate. Even when the modelling process is successful, the contribution to urban design process still requires explanation. The format of model-based design for urban design also requires further definition. These questions will be the scope of future researches to maximize data modelling and spatial database use in urban design.
References
Database components of Singapore Management University (SMU) New Campus Design Competition Guidelines (2000)
© GISdevelopment.net. All rights reserved.
The latest advancement of object-oriented (OO) data modelling in Geographic Information Systems (GIS) and Computer Aided Design (CAD) has opened an opportunity in data modelling for urban design. In the CAD realm, the concept of OO data modelling has been materialized in a new concept called “model-based” design, the latest development in CAD-based design methodology. Although OO data modelling has been developed in GIS realm long before its CAD counterpart, it has never been design-oriented since GIS was not initially developed for spatial design purposes. In this research paper, the Database Management Systems (DBMS) is proposed as the foundation for the development of GIS-based urban design support system, which is surprisingly as comprehensive as the GIS application to urban planning. The design of database for GIS is so important and fundamental that it affects the performance of the proposed GIS design support system. We argue that the urban design practice has not fully and intensively utilized the GIS tools because the database has not been designed appropriately for the urban design requirements.
Through data modelling, which have entered a new dimension in OO data modelling, we can design a specific type of database to support the GIS application of urban design. OO data modelling allows us to define our own types of objects through topological, spatial, and general relationships, which can help capture how these objects interact with other objects. The data models were generated programmatically using ArcObjects components and Visual Basic for Application (VBA). The OO modelling will be conducted in Unified Modelling Language (UML) scheme style, with a possibility to use CASE tool. Data modelling runs through conceptual, logical and technical stages, which will be finalized in a relational spatial database. A chosen database for utilization is Geodatabase from ESRI, with other possible implementations such as XML Schema.
The paper also discusses the benefit of designing the data model for urban design. The data model design is regarded as an essential part of the whole urban design process, where Patrick Geddes’ traditional design methodology “survey-analysis-plan making” are redefined as “data model - GIS spatial analysis – design decision making”. With the adoption of object-oriented approach, GIS data modelling will be able to capture the spatial and non-spatial relationships among urban physical objects. The experiment will use latest data modelling techniques, such as semantic ontology and object-oriented approach to test the applicability of data modelling to urban design decision-making.
To what extent the new techniques of data modelling could be applied to urban design process? The paper will discuss the potentials and limitations of the applications of OO data modelling to urban design. Based on the results, a critical analysis of urban design guidelines, principles, and their model-ability will be discussed. The long-term objective of the research is to develop the “data model”-based urban design as a new urban design methodology.
1. Introduction
GIS (Geographic Information System) data modelling is a methodology for designing spatial databases, a type of design activity of its own. However, the way database designer designs a database is distinctively different from the way urban designer designs a city. Why do we need database, the spatial database in particular, for urban design? How could we make use of the spatial database management system (DBMS) technology in the urban design application? To what extent the currently available DBMS technology could be applied to the urban design process? Based on these preliminary questions, the GIS database research will be moved a few steps further through the elaboration of theory and urban design case studies.
Judging from the vast amount of information concerned by urban designers, urban design definitely needs tools to manage this information. GIS has a track record in handling urban information, especially in the field of urban planning (Brail & Klosterman, 2001; Han & Kim, 1990; Zaki, 1998, Laurini, 2001). In the recent development of GIS technology, it begins to show potentials for the design-oriented uses. The way GIS will support design process is through its spatial database, with its competitive strength. The main function of spatial database is to store urban information in digital format for further GIS process, which composes queries, analysis, visualization, and simulation. A new design-supporting GIS function is proposed in the research work, which will use the spatial database as a sketching platform for urban design proposals.
To start with the discussion of data modelling, there is a crucial need to clarify the definitions of model and data model, when the term “model” has different implications and possible meanings in different disciplines. The definition of the model, according to system science, is the “representation of reality”, which selectively expresses certain relevant characteristics of the observed reality, and consists of objects or systems that exist, have existed or may exist. As a representation or an abstraction of the observed reality, a model is always a simplification of the complexity of reality, which makes the reality more understandable and operational (Crowther & Echenique, 1972).
Models may be classified as descriptive, predictive, explorative, or planning according to its purpose. Models can be classified based on their constructed elements, including physical models (solid materials such as architectural wood model), analogue models (drawings and plans), conceptual models (words or mathematical functions) and digital models (computer aided design drawings or data model). A model can be either static or dynamic regarding to its nature towards temporal elements.
A data model is not a physical building model, a city map, urban design drawings or a mathematical model of urban population growth. A data model is an abstraction of the real world that employs a set of data objects that support map display, query, editing and analysis (Zeiler, 1999). The data model can be applied to different aspects of the design process, such as procedure models, predictive models, digital urban models, behavioural models, dynamic agent-based models, etc. A data model, as discussed in this paper, is mainly descriptive, although it can function as explorative through customization. It is made of digital database and may be represented by analogical materials such as digital maps, plans and graphs. We propose that the GIS data model is to be designed as a bridge between urban reality and its spatial representation in GIS. It is the premeditated way to organize and represent the information of urban reality in a spatial database. Since all GIS-based analyses are supported by the spatial database, the data model is regarded as the foundation for urban design analysis and design decision-making.
In the area of computer aided architectural design (CAAD), the data modelling can now generate building system database for building management purpose. The data modelling can be utilized to support model-based building design using object-oriented CAD. For urban planning, the data modelling can be applied to the development of planning support systems (PSS). Unfortunately, the data modelling is not yet available for urban design purpose. A data model is designed for representing the relationships between buildings and their components. It is yet to be developed for an object-relational system (Cote, 2002). Using the test case of Singapore Management University’s (SMU) urban design, the paper will explore how the data modelling could be applied to the management of urban design guidelines.
2. Development of Data Modelling for Urban Design
The recent development of the model-based design in computer-aided building design has become the motivating example for the experiment of using a similar approach for urban design. Autodesk, Bentley and Graphisoft have developed “object-oriented” CAD packages capable of performing model-based design, which is also known as building information model, virtual building, or parametric building modeller. The motive of this paper is to utilize data modelling technology to support computerized urban design process.
Computerized urban design utilizes spatial modelling technologies such as CAD, and most recently 3D GIS. Most computerized urban design activities will need spatial database of some kind, where data are stored in certain format. The structure, format, and technology level of the spatial database will determine the capacity and capability of the software application used for urban design. Through data modelling, spatial database should be designed to accommodate urban design analysis and design decision-making process.
The idea of applying data modelling to physical design is not entirely new in the realm of design since the systematic design thinking in the 60s and 70s. Forty years ago, Alexander proposed a “set-subset” theory, where he suggested designers to construct analytical design process through problem classification in hierarchical logic. He initiated the use of constructive or schematic diagram of design variables as design tool, which implied a preliminary idea of design programming and data modelling. He argued that performance standards, which resembles urban design guidelines in their roles, can be set up for every misfit variable that exhibits continuous variation along a well defined scale (Alexander, 1964). In his theory, he experimented on redistribution or reclassification of the design variables, which is comparable to the object-oriented (OO) principles, where variables related with the same object are clustered together. Alexander’s problem solving principle of classification is actually the preliminary stage of current data modelling principle. It can be followed by the transformation of “misfit variables” by semantic and logical deduction into objects, subtypes, attributes, validation rules, connectivity rules and relationship rules, which are the basic components of OO data model.
Further researches following Alexander’s earlier work also supported the concept of data modelling for design process. Geoffrey Broadbent, Anthony Ward, Amos Rapoport, Luckman, Best, etc, have discussed the role of operation research (OR) methodology in design process, which has much compatibility with data modelling techniques.
The developments of computerized urban applications in the prior days were emphasized more on the analytical and predictive aspects of the applications. The development of Planning Support System (PSS) was a significant evidence of this trend (Klosterman 2001). A PSS usually contains several levels of predictive urban modelling or simulation, which requires a database modelling as the foundation. Several DSSs and PSSs were built on ArcView version 3.x GIS platform by customizing ArcView system and adding on its original functions. Relational spatial database was not existed in this version; therefore developers could not design spatial database through data modelling, and have to customize the functionality of relational spatial database through its customization language such as Avenue. In the later version of ArcView, ArcGIS 8.x has the capability to build spatial database designed through the data model of UML and XML and the customization platform ArcObjects with embedded VBA. Planners have benefited from the object-oriented ArcGIS Parcel Data Model recently (ESRI 2001), which was actually designed for cadastral purpose, with more emphasis on land ownership and distribution. Although containing urban objects, ArcGIS Parcel Data Model may not be suitable for accommodating urban design activities, since design process needs sufficient built-up physical data.
Spatial database design is very important in the development of planning support systems, since a GIS system must be built on standard components. The components start from spatial database as the foundation, then editing, analysis, simulation, and finally visualization. However, the design methodology of spatial database hasn’t been developed adequately in urban research compared with the predictive modelling and simulation. This argument is strengthened by the minimum findings of the similar PhD theses currently researching data modelling for urban design. One of them is currently pursuing on “Object-Oriented Data Modelling and Warehousing to Support Urban Design” (Koshak & Flemming, 2002). Another research is pursuing on “Restructuring UrbanSim Model into ArcGIS (Geodatabase) Data Model using UML” (Kyuwon Park, 2002), which emphasized on land-use dynamics simulation modelling of the well-known UrbanSim planning model. Judging from the minimum research efforts in this area, we argue that data modelling is still a new area in urban design research.
In the past few years, data models for editing and managing urban information have largely been in the province of specialized computer aided design (CAD) tools used by engineers and architects. This combination of data modelling in GIS, Spatial Data Engine (SDE), through UML or XML format as bidirectional medium, provides a potential for managing virtually unlimited amounts of details representing many aspects of the built environment of the entire city, along with different future or past scenarios. Cote argued that data models for representing the relationships between buildings and their components have yet to be developed for an object-relational system (Cote, 2002). He perceived the urgent need of an appropriate data model when he was confronted by several problems while developing an agenda “towards modelling broad scale urban scenes in an object-relational database”. The first problem is how to associate primitive shape objects to represent built-environment, i.e. buildings. A representation of a building is usually composed of different features, such as points, lines and polygons. Relationships in the spatial database, such as “the windows, associated with hotel rooms, which have a view of the natural gas tanker”, are examples of information that may be desired from a good urban data model. The second problem is the need for data models, which are interoperable between GIS and CAD. These are the problems that can only be solved through data modelling.
In the next section, the paper will introduce data modelling of urban design guidelines. Most guidelines consist of clear regulatory and advisory statements of design rules and relational rules between design objects. The paper will explain how urban design guidelines can be modelled into data models, which in turn will create spatial database capable of enforcing the guidelines in the design process.
3. Urban Design Guidelines
To implement data modelling in the development process of spatial database for urban design, urban design guidelines were chosen as modelling case study. It is because they contain clear statements of desired relationships and rules between urban objects, available to public. They act as a guide for developers and designers, and possibly other agents, in planning and designing development, prepared by local authorities with the participation of landowners, developers, partnerships, business, and community organizations (Cowan, 2002). They are either the result of negotiated agreements between stakeholders or a direct guidance from local planning authority. They can be used to identify and develop performance criteria to measure effective plan execution through performance-based standards (CONCERN, Inc. et al., 2002).
There are several types of urban design guidelines. The first are guidelines of specific places. There are three main types of these: urban design frameworks, development briefs, and master plans, mentioned in hierarchical order. Urban design framework mainly consists of two-dimensional vision of future infrastructure requirements. Development briefs usually contain a more indicative vision of future development form. Master plans are detailed plans usually prepared by the organization that owns the site. The second type of guidelines relates to specific topics. These, usually called design guides, cover topics such as shop fronts, house extensions, lighting and cycling. Singapore Management University (SMU) New Campus Design Competition Guidelines are considered one of these. The third type of guidelines relates to specific policies, such as policies on conservation areas, transport corridors, waterfronts, promenades and green belts. An established example is the user-centred guidelines, which was derived from user specifications and preferences (Katoshevski & Timmermans, 2001). The fourth type of guidelines relates to a whole local authority area. These may give general urban design guidance for the whole city or county.
There are several roles and benefits of urban design guideline in design process, for example, to promote a visually harmonious community. The urban design guideline is also considered as the most innovative and flexible approach among all development regulations. The development of guidelines will support planning policy, encourage collaborations among stakeholders, create vision for the new environment and guide the development of local design standards. Guidelines were involved in almost all urban design projects, such as new development, rehabilitation and urban public space projects (Cowan, 2002; Moughtin, et al. 1999; Friedman, et. al., 2002; Citizens Party of Hong Kong Central Government, 1999). Most design guidelines contain a comprehensive urban design policy and the full range of design considerations that are important in a locality (Punter, et al. 1994). The guidelines can be classified based on the scales of area affected, including region, metropolis, city, and town, neighbourhood, district and corridor and finally block, street and building (Congress for the New Urbanism, 1996).
The development of database for urban design guidelines require resources such as common urban features, i.e. land coverage, lot size, parking, and their guiding regulations, i.e. FAR, setback, land uses, growth space, additions, sections of lots, and building heights (Friedman, et. al., 2002). The guidelines may provide classification of urban features. For example, the lot types can be classified as corner lots, typical lots and lots with alleys. The development of urban design guidelines database requires an integrated and multi-disciplinary team that understands the importance of quality of urban design.
4. The Role of Data Modelling in Urban Design
Urban design process is essentially a modelling process, where designers use models to represent the urban reality and the design intervention within. Urban design process is where evolution of urban models occurs. The mental models are ideas, theories, and regulations, which are processed in the designer’s mind. The urban design process may not be systematically linear. The designers often use analogue models, such as hand-drawing sketches, or even building massive model, to express their ideas. With the advancement of CAD, design process can utilize digital models, such as computer-based drawings, either in sketches, technical drawings, or 3D models. We argue that ideas and the ruling knowledge behind the urban design process can be systematically represented in the data model as well as the analysis model and design decision model.
A data model takes role just as digital and analogue models, to represent the mental model. Data modelling is the foundation for the digital modelling in the way that the former becomes the dominating system of the latter. Unlike analogue and digital models, data model is able to represent not only geometric and functional ideas, but also regulative knowledge, which may be derived from design guidelines. By running the data model for “design” part and for “regulation” part comparatively and simultaneously, the information system can acknowledge the design “compatibility” with the regulation.
As the methodology of databases, data modelling is fundamentally a design process, a creative process, having similar concurrent components with other design process, such as architecture and urban design. Data modelling in the enterprise world have been utilized as a “tool for expressing and communicating enterprise rules” consisting of rules and regulations in the businesses and organizations (Simsion, 1994). The similar principles can be referred to the making of urban design guidelines, which are the documents stating the urban design’s “enterprise rules”. Through data modelling, guidelines can be implemented and embedded in a GIS-based spatial database particularly, where all subsequent digital models will abide to their parent data model. This way, data modelling will extend guidelines’ capability as a support system of design process. As a simplified form of urban reality, to what extent the data model is able to represent the components of urban design guidelines? The question is to be verified through of test case of Singapore Management University (SMU) campus design competition guidelines.
5. SMU New Campus Design Competition Guidelines
In the late 1990s, Singapore government reserved 7 plots in this highly valued central district, for the development of a new campus for Singapore Management University. The plots are located at the centre of current Bras Basah Park, which is a prominent side and currently functioning well as an urban ‘oasis’. The park itself maintains openness and visual connection between Singapore Museum & Art Museum. As an effort to get the best design for its future campus, SMU has coordinated a design competition on 2000-2001, which has received a good response of more than 150 participants from 36 countries. The urban design guideline for SMU New Campus Competition was written to guide incoming proposals to suite the requirements of SMU management. The proposals must express the fulfilment of SMU’s policies and needs, regarding its existence in the CBD’s Museum District.
- The guidelines described clearly the existing site situation made up by parcels, building types, canals, roads, and tunnels. The designated parcels for campus development are referenced as Parcels A North, A South, B, C1 West, C1 East, C2, C3, D1, D2, and E. The guidelines also described the location of Mass Rapid Transit (MRT) stations and network, and the location of site boundary lines.
- The proposals should consider the future infrastructure works in the vicinity. The location of designated 3.6 meter-wide colonnaded-covered walkway, connected to a future integrated bus stop. The open spaces were regulated to have a 7.5 meter-wide setback for pedestrian and specific area to be landscaped.
- The open space over Bras Basah Park was regulated to maintain direct visual pedestrian connection to Fort Canning Park on top of the hill. For maintaining visual character between National Museum and Singapore Art Museum, the guidelines enforced proposed open space on Bras Basah Park bounded by sight lines. The open space becomes the indicative location of direct visual pedestrian connection from Bras Basah Park to Fort Canning Park.
- The proposals should include possible inter-parcel underground connections below street level on designated zones. Possible subterranean development was also designated under several public spaces. Possible overhead connection between land parcels D1 & D2 was designated.
- There are also proposed roads to be constructed by SMU. For each development parcels, service access points and vehicular ingress/regress points were designated along zones. The guidelines also recognized government’s regulations such as Land Transport Authority’s (LTA) road requirement.
Figure 2 Urban design guidelines for SMU New Campus Design Competition (SMU, 2000)
6. Data Modelling Methodology
All data modelling techniques share the same three stages: conceptual stage, logical stage, and technical stage. The conceptual and logical stage can be combined into a single stage, usually occurs in modelling simple models. Modelling techniques considered under the conceptual stage are Entity Relationship (ER) techniques, such as Chen’s Original ER developed in 1976 and Barker’s Oracle ER developed in 1990. Object Role Model (ORM) was developed by Terry Halpin in 1995 (www.orm.net), and was acclaimed for more user-friendly conventions and could be easily understood by non-technical users. The rules of ORM data models can capture more ‘enterprise’ rules, equivalent to urban design guidelines, and are easier to validate and evolve than data models in other approaches. The Object-Oriented (OO) Model provides examples of the logical modelling stage, where Unified Modelling Language (UML) is designated for off-line database design and eXtended Markup Language (XML) for tentative Internet database design. Both can be classified in the implementation or technical stage of data modelling. The modelling of urban design guidelines can be done through transforming guidelines’ textual-format rules and regulations into data model’s relationship schema. The process is basically a language transformation.
- Textual-format statements of guidelines are basically information expressed in a written natural language. Using a technique called Natural language Analysis Model (NlAM), the textual-based information can be transformed into schema-based information, in the format of a data modelling language called Object Role Model (ORM).
- ORM is the conceptual stage of data modelling, which can be furthered by transforming into more technical data models such as Unified Modelling Language (UML). Although ERM is the most common data model convention in the market, ORM provides better transformation from natural languages (such as English).
- The key task in building conceptual and logical data model is to precisely define the set of objects of interest and the relationships among them. They will then be validated against the user’s requirements for entering, updating, and accessing data and by testing it against urban design practices and procedures. Basic data modelling principles must be applied, such as no duplication, simplicity, clarity, and different views of data could be accommodated for different user groups.
- The GIS system provides three options for implementing the technical data model:
- Option 1: From ORM conceptual model through UML modelling and “ArcInfo repository” software component, with the help of Computer-Aided Software Engineering (CASE) supported by MS-VISIO 2000.
- Option 2: From ORM conceptual model through programming with ArcObjects and Visual Basic for Applications (VBA), without using UML.
- Option 3: From ORM conceptual model through eXtended Markup Language (XML) format to contain the data model, with the help of Geodatabase Designer developed by The Applications Prototype Lab, ESRI(R) Redlands.
- After conducting the first option, the GIS system will create a spatial database, containing objects and relationships, based on the rules and regulations in guidelines modelled in the data model. The spatial database, namely geodatabase, is an object-oriented spatial database. It contains database objects and their relationships, which can be modelled to have the characteristics of urban objects and relationships between urban objects. Rules and regulations in urban design guidelines are transformed into spatial database rules, which consist of topological rules, validity rules, relationship rules, and connectivity rules. (The connectivity rules are not included in this experiment.) The spatial database will be mounted on existing Relational Database Management System (RDBMS) such as MS-Access, Oracle, etc. ArcSDE (Spatial Data Engine) will provide database handlers for storing ArcGIS data in existing RDBMS.
- The implementation of urban design guidelines itself comes to play when a designer sketches his urban design proposals using GIS system and the modelled spatial database. The GIS system, after minor customisations, will inform the designer about his design “fitness” according to the modelled urban design guidelines. The enforcement of urban design guidelines in GIS system can be set to minor or major enforcement by weighting factor, which determines the degree of freedom in design process.
To model guidelines’ statements of relationships and their rules among urban objects, as described in urban design guidelines, we need to transform them into a database model. We can use the methodology developed below for designing ORM, which is called Conceptual Schema Design Procedure (CSDP).
- Transform familiar information examples into elementary facts (objects), and apply quality checks. Each urban object stated in the guidelines, such as building, street, and parcels can be transformed into spatial database objects, which are represented by (1) a spatial entity such as polygon, polyline, or point, and (2) an attribute entity: a row in a table.
- Draw the fact (object) types, and apply population check. The spatial data represents the spatial characteristic of the urban object, such as polygons for buildings, or lines for pipelines. The attribute data represents non-physical characteristics of urban objects, such as land use classification or different building purposes (commercial, residential, or industrial), or a measurement (pipeline diameter or building’s number of stories).
- In determining urban object’s characteristics, we have to refer to (a) documents containing the written definition of urban objects, such as: what is the definition of a commercial building. Or we can refer to (b) existing condition, if documents cannot be found. In most cases, the local authority will reserve documents containing regulatory definition of urban objects, most commonly in urban design guidelines.
- Check for entity types that should be combined, and note any arithmetic derivations
- Identify roles (relationships) and object classification. Spatial relationships can be modelled as topological relationships. For non-spatial relationships, they can also be modelled as relationship classes. Both spatial relationship and non-spatial relationship can be modelled together in a single data model. Relationships, or roles in ORM, are modelled by connecting two or more objects with a relationship line, which must have an arity value that can be unary, binary, or ternary, depending on the number of objects connected by the relationship. Arity is the number of arguments a function or operator takes (Howe 1993).
- Sketch the ORM diagram using VISIO ORM template. By sketching the ORM diagram down, we can re-evaluate logical roles between objects more easily.
- Add topological relationships between objects, and re-evaluate them. The topological relationships, or topological rules in ArcGIS, are classified according to the feature class geometry type they supported, which are polygon rules, line rules, and point rules. There are 9 polygon rules, 12 line rules, and 4 “point” rules, in total 25 rules. The topology class also provides ranking system of feature classes, to control which features may be moved to other features during validation.
- Add uniqueness constraints, and check arity of fact types. In ORM models, the arity is translated back into uniqueness constraint.
- Add mandatory role constraints, and check logical derivations. A mandatory constraint governs the “must” or “may” aspects of relationships.
- Add value, set subtyping constraints. These conditional values can be preset, or pre-programmed in the data model.
- Add other constraints and perform final checks. Finalizing Urban Design Guidelines Model.
After following Conceptual Schema Design Procedure (CSDP) closely, we can conclude that:
- In the transformation of facts to objects, we must be careful in determining objects and attributes separately. Attributes can be modelled as values, which become the attribute of the related feature class. The roles related to the values are translated into “non-object” relational rules. Model integrity based on the guidelines’ original statement should be well preserved.
- Several non-spatial relationships, typically “(A) has (B)”, may not be required since it’s redundant with the spatial topological relationship. This is also debatable, since we may need to implement the “has” relationship to indicate ownership. Replacing the non-spatial relationship “has” with topological relationship, such as “must be covered by feature class”, which is a better definition, can reduce the redundancy.
- Technical data modelling in UML may have problem in representing a “unary” role from ORM. A “unary” or “one-ary” role doesn’t create any relationship class between objects, but instead may be translated into an attribute, most probably a Boolean attribute. An example is from “Building is maintained” unary role can be transformed into “Maintained (Yes/No)” attribute field.
Figure 3 Conceptual Data Model – Guidelines (left) translated to ORM Diagram (right)
- The ORM drawing convention uses “boxes” to represent roles, which can be found “space-consuming”. After several modelling experiments, we realized that some roles, typically “has” (ownership), is the most common roles modelled in spatial database. For simpler and better graphical-representations of ORM models, the role “has” can be represented by a simpler graphic, such as directional arrow. In this way, ORM models can be more “readable”.
- Objects that can’t be derived from the guideline statements can be generated if they are required.
- Topological relationships are usually not stated in the guidelines statement; therefore they may need to be recognized by deducting from modelled-relationships, namely logical deduction. Topological relationships are very important in maintaining design integrity based on spatial implementation of urban design guidelines.
- In the modelling process, a “sub-model” may be identified from the guidelines statements. For example, Land Transport Authority (LTA) requirements are found, which another guideline transformable as a sub-set guidelines model is. The sub-model requires a separate modelling process of the sub-set guidelines.
- Validity rules from the guidelines can be implemented through object subtypes. The subtypes will enforce a specific value or a domain of limited values in any attribute fields subjected to validating rules.
- Several objects that have similar traits can be transformed to subtypes of a “larger” parent object. Example, the previous objects “underground pedestrian link” and “surface pedestrian link” can be transformed into the subtypes of the object “Pedestrian Link”.
The outcome of Conceptual Schema Design Procedure (CSDP) is the ORM conceptual data model of urban design guidelines for Singapore Management University (SMU) New Campus Competition. This data model must be transformed to a spatial database in order to make it useful for urban design process. This transformation was conducted through technical data modelling, where we’ve found issues to be tackled:
- Objects and relationships modelled will contain data of existing urban condition, rules of urban design guidelines, and sketches and data of design proposals. Originally, the object classes should be separated into three feature datasets: existing, guidelines, and proposed design datasets. For the guidelines to take effect on the proposed design dataset, the classes can’t be separated to different datasets because the topology class where the topological rules reside can only affect classes in the same dataset. Therefore, the class organization should not use separate datasets between guidelines and proposed design objects.
- Ternary (3rd level of arity) roles, such as “… between … and …”, usually can’t be modelled directly to relational classes. Ternary rules often indicated network relationships between objects, which should be modelled in network classes.
- Several topological rules were not modelled because the change of hierarchy of objects previously intended in the conceptual stage. If those rules were modelled, they will create redundancy with rules of their superseding objects or other topological rules related.
- Several topological rules were wrongly modelled in the conceptual stage, such as “must intersect”, “boundary must overlap with”, etc. It is mainly because the topological relations are limited to certain object geometry types; e.g. “must intersect” rule is not available between polygon and line objects.
- Combining several subtypes to a super-type object will also combine separate topological rules related to the subtypes into a single topological rule.
The data modelling experiment explained in this paper is primarily conducted in Object Role Model (ORM) format. Export to other formats, such as UML and XML, were not included in this experiment yet. Future works will investigate these data model formats for urban design guidelines. A direct interpretation of the ORM model has produced a spatial database containing the urban design guidelines for Singapore Management University (SMU) New Campus Competition 2000-2001. The spatial database “SMU_guide_model” contains feature datasets of existing Singapore urban data, existing SMU site, SMU guidelines, topographical data, tables containing Singapore urban data, and SMU campus design proposals from competition finalists. The feature classes within datasets are presented in Appendix.
The experiment has transformed text-based guideline statements to database components. Different statements are transformed into different components, such as feature classes, topological rules, relationship classes, and domains.
- Statements indicating the existence of a type of physical reality are transformed to feature classes.
- Statements indicating the spatial arrangement between urban objects are transformed to topological rules.
- Statements indicating non-spatial relationship between two objects are transformed to relationship classes.
- Statements indicating measured limitations and range of choices are transformed to domains.
7.4 Preliminary review of urban design guidelines model-ability
In the modelling experiment, we managed to transform 85% of the 39 guidelines statements intended to be modelled. The rest of the statements were not modelled because they were not required in the design process, or they could not be modelled properly, or they needed to use network features, or the information was not available. From all 33 modelled statements, 13 indicated the existence of new spatial objects transformable to feature classes, which must be incorporated in the design proposals. Another 3 statements indicated the needs to acknowledge existing objects in the design consideration. Four statements prescribed specific width requirement of movement path were transformable to domains. Another four statements indicated relationships between two spatial objects, which were transformable to relationship classes. The last four statements described the topological relationships between spatial objects, which were transformable to topological rules.
Although a large part of the design guidelines was transformed to guidelines model, they were not sufficient. We had to add 11 additional topological rules to make the first four rules effective. The additional rules were extracted from logical deduction of the statements, since the guidelines often contain “unwritten” logical statements. The four original statements related to proposal’s design integrity actually required eight topological rules to be implemented in the database, while four “unwritten” topological rules are added to maintain spatial integrity between existing spatial objects. Four additional relationship rules were also required. In conclusion, spatial integrity can be validated in GIS database by adding topological and relationship rules based on logical deduction of spatial organization. The additional rules may take up to 40 – 50% of the total rules modelled.
8. Conclusion
The data modelling experiment of urban design guidelines for Singapore Management University (SMU) New Campus Competition 2000-2001 in GIS has been successfully conducted. After implementing the urban design guideline model on the target GIS system, a spatial database can be created, using the Geodatabase format. The guideline rules and regulations have been converted into object-oriented (OO) rules, which consists of topological rules, validating rules (domain), relationship rules, and connectivity rules. The rules were implemented on urban design -dedicated database elements such as feature classes, domains, relationship classes, and topological rules. The implementation of urban design guidelines itself comes to play when a designer sketches his urban design proposals using GIS system, and the modelled spatial database. The GIS system, after minor customisations, will inform the designer about his design “fitness” according to the modelled urban design guidelines.
The materialization of data modelling for urban design process has maximized the use of spatial database for urban design. We may conclude that data modelling gives more benefits if implemented for complex and large-scale urban design, where more issues, problems, and stakeholders are involved. In design process, data modelling works best on large-scale design environment, which involves many designers and drafters handling the same project. Data modelling increases efficiency in teamwork by maintaining design project’s integrity and consistency. The capacity of an informative database may lead design process to a further step of knowledge-based design. With the rapid development of network technology, data models can be shared among designers as the media for sharing design information and knowledge, which will encourage collaborative design.
However, the introduction of data modelling in existing design community may not be a smooth integration, since it is an alien, new design methodology. The designers’ adaptation to data modelling may be slow, since current data modelling techniques are far from perfect, and the immediate financial benefits are not apparent. Even for small-scale simple design, the method can become too complicated, cumbersome, and irrelevant. It will require some time before data modelling can be accepted as an option in design process. Finally, only urban designers who understand the whole “universe of discourse” of urban design can practice data modelling for urban design.
Along the way, many theoretical questions have been raised on evaluating data modelling in urban design process. Is it really useful, or relevant? What can be covered in data models? Will it affect the urban design process? Will the effect be positive or negative? Will it affect the issue of creativity? What is the best implementation in urban design process? Questions about the “model-ability”, the compatibility of urban design guidelines if modelled in data modelling, may be raised as well. Several methods have been developed to hypothetically measure guidelines’ “model-ability”, such as its logical integrity and information system friendliness. We can also compare the guidelines’ model-ability and data model’s compatibility to measure the modelling success rate. Even when the modelling process is successful, the contribution to urban design process still requires explanation. The format of model-based design for urban design also requires further definition. These questions will be the scope of future researches to maximize data modelling and spatial database use in urban design.
References
- Alexander (1966) The coordination of the urban rule system, Center for Planning and Development Research, U. of California, Berkeley.
- Alexander (1987) A New theory of urban design, Oxford University Press, New York.
- Alexander, C. (1964). Notes on the Synthesis of Form. Cambridge: Harvard University Press.
- Alexander, C. (1965). A City is not a Tree. In Architectural Forum. New York: Billboard Publications.
- Alexander, Ishikawa and Silverstein (1977) A pattern language towns, buildings, construction, Oxford University Press, New York.
- Alexander, Ishikawa, Silverstein and C.f. (1968) A pattern language which generates multi-service centers, Center for Environmental Structure, Berkeley, Calif.
- Alexander, Poyner and G.B. (1967) Atoms of environmental structure ; Twenty-six entrance relations for a subarban house ; Relations for an office entrance, Ministry of Public Building and Works, Directorate of Development, London.
- Best, G. (1969). Method and intention in architectural design. In Broadbent & Ward, Design Methods in Architecture. London: Lund Humphries.
- Brail, R.K. & Klosterman, R.E. (2001). Planning Support Systems. ESRI Press.
- Broadbent, G. & Ward, A. (1969). Design Methods in Architecture, Architectural Association Paper Number 4. London: Lund Humphries.
- Broadbent, G. (1969). Notes on design method. In Design Methods in Architecture. Architectural Association Paper Number 4. London: Lund Humphries.
- Chen, W. (1999). 3-D city: prototyping techniques for urban design modeling (abstract). Macaulay Land Use Research Institute, Aberdeen. http://www.geovista.psu.edu/sites/geocomp99/Gc99/002/abs99-002.htm
- Citizens Party of Hong Kong Central Government, (1999). A walk on the wild side: Better planning for pedestrians in Hong Kong. Planning document.
- CONCERN, Inc., & Environmental Simulation Center, US. Dept. of Energy, US. Dept. of Housing and Urban Development, (2002). Placemaking: tools for community action.
- Congress for the New Urbanism, (1999). Charter of the New Urbanism. New York: McGraw-Hill.
- Cote, P. & Ervin, S. (2002). Real Infrastructure for Virtual Cities. Harvard Graduate School of Design. http://www.gsd.harvard.edu/~pbcote/gsd6329/pbcote/arcscene/
- Cowan, R. (2002). Urban Design Guidance: urban design frameworks, development briefs and master plans. Heron Quay: Thomas Telford Publishing.
- Crowther, D. & Echenique, M. (1972). Development of a model of urban spatial structure. In Urban Space and Structure. L. Martin & L. March. Cambridge University Press.
- ESRI (2001). http://support.esri.com/index.cfm?fa=downloads.dataModels.byLabel&dmid=11
- Friedman, A. et al. (2002). The development process of urban and architectural guidelines for the rehabilitation of an inner-city neighbourhood: Le Village, Cornwall, Ontario, Canada. Journal of Urban Design, Vol. 7, No. 1 (p.5-34). Abingdon, Oxfordshire : Carfax Publishing, Taylor & Francis
- Greed, C. & Roberts, M. (eds). (1998). Introducing Urban Design: Intervention and Responses. Addison Wesley Longman Ltd.
- Han S. Y. and Kim T. J. (1990). Intelligent Urban Information Systems: Review and Prospects. In Kim T.J., et al. (eds). Expert Systems: Applications to Urban Planning. New York: Springer-Verlag,
- Harris, B. and Batty, M. (1993). Locational Models, Geographic Information and Planning Support Systems. Journal of Planning Education and Research 12 (pp. 184-198). Association of Collegiate Schools of Planning.
- Howe, D. (1993). Free Online Dictionary of Computing, http://foldoc.doc.ic.ac.uk/foldoc/index.html
- Institute of Highway Incorporated Engineers, (2002). Home Zone: Desogm Guidelines. Institute of Highway Incorporated Engineers.
- Katoshevski, R. & Timmermans, H. (2001). Using Conjoint Analysis to Formulate User-centered Guidelines for Urban Design: The Example of New Residential Development in Israel. In Journal of Urban Design, vol. 6, No. 1 (pp. 37-53). Abingdon, Oxfordshire : Carfax Publishing, Taylor & Francis
- Klosterman R.E., (2001), Planning Support Systems: A New Perspective on Computer-aided Planning. In Planning Support Systems, ed. Brail R.K. and Klosterman R.E. ESRI Press
- Laurini, R. (2001). Information System for Urban Planning: A hypermedia co-operative approach. Taylor & Francis.
- Luckman, J. (1969). An approach to the management of design. In Design Methods in Architecture. London: Lund Humphries.
- Mohd Zaki, Z. (1998). GIS and Land Use Planning in Singapore. In Planning Singapore: From Plan to Implementation. Singapore Institute of Planners. Ed. by Yuen, Belinda.
- Moughtin, C. et al. (1999). Urban Design Method and Techniques. Architectural Press.
- Punter, J. et al. (1994). Design policies in development plans, Urban Design Quarterly, No. 51, (pp.1-15).
- Putra, S.Y. & Yang, P. P. J. (2002). Data Modeling for Urban Design Guidelines. Unpublished research paper
- Putra, S.Y. (2001). Building A Geodatabase For Keko Mwanga Informal Settlement In ArcGIS 8. ITC Publications, Master Thesis, August 2001
- Rapoport, A. (1969). Facts and models. In Design Methods in Architecture. London: Lund Humphries.
- Simsion, G. (1994). Data Modeling Essentials: Analysis, Design, and Innovation. International Thomson Computer Press.
- Stone, A. (1997). The New Essex Design Guide, Urban Design Quarterly, No. 62, (p.31-35).
- Urban Design Associates, (1996). The Urban Design Handbook: Techniques and Working Methods. Urban Design Associates
- Urban Redevelopment Authority, (2000). Sparking off architectural creativity. Singapore: Skyline, May/June. (p.10-11).
- Yang P P J., Putra S.Y. (2002). GIS Analysis for Urban Design and Redevelopment, Two Cases in Singapore, Proceedings of the 8th Conference on Computers in Urban Planning and Urban Management (CUPUM) in Sendai, March 2002
- Zeiler, M. (1999). Modeling Our World: The ESRI Guide to Geodatabase Design. ESRI Publications.
Database components of Singapore Management University (SMU) New Campus Design Competition Guidelines (2000)