The Backbone of A City Information Model (CIM) : Implementing A Spatial Data Model For Urban Design
The Backbone of A City Information Model (CIM) : Implementing A Spatial Data Model For Urban Design
The Backbone of A City Information Model (CIM) : Implementing A Spatial Data Model For Urban Design
INTRODUCTION
In light of the challenges of sustainable urban development we have been witnessing an increased interest in a more holistic approach to urban design practice and education. This approach has multiple dimensions and challenges (Gil and Duarte, 2008): To integrate the activities of a wide range of stakeholders; To integrate analysis and evaluation in a performance based design process; To manage the various outputs from the various stages of the design process; To facilitate access to information from a variety of sources and in different data formats; To manage very large quantities of information, facilitating its manipulation and visualization by different users. The field of building design has responded to similar challenges with the development of various Building Information Modeling (BIM) technologies and there have been calls for the creation of something equivalent in urban design, a City Information Model (CIM) (Khemlani, 2007; Gil et al. 2010). The CIM would extend the use of Geographic Information Systems (GIS) in urban planning as decision support tools (Webster, 1993; Batty et al., 1998) through the integration with Computer Aided Design (CAD), to become a design support tool (Dave and Schmitt, 1994; Maguire, 2003). The City Induction research project has been focusing on the development of such a system, based on an urban design method that integrates the stages of formulation, generation and evaluation of urban designs supported by a CAD/GIS software platform (Duarte et al., 2011). In this paper, we present a spatial data model for urban design practice that can serve as the backbone of a CIM. This is a general component of the City Induction project based on previous work carried out on urban design ontologies (Beiro et al. 2009;
City Modelling - eCAADe 29 141
Montenegro and Duarte, 2009) and we demonstrate its integration with the evaluation module. Firstly, we describe the urban design data model and the structure of the spatial database, highlighting the possibilities that it offers by integrating urban environment and urban design process feature classes, and provide an example of its implementation in PostGIS using datasets for the Randstad region in the Netherlands. Secondly, we demonstrate its application to urban design analysis and evaluation through the implementation of a tool for AutoCAD Map 3D. We then discuss the benefits and the potential applications in urban planning and urban design education, and point to further work required to test its integration with the complete City Induction framework.
The information is managed in a single central repository, avoiding data duplication and ensuring that the most up-to-date version is accessed by everyone.
However, to use such an infrastructure in the urban design process, it needs to be configured with an adequate data model that supports the kinds of information, methods and user roles that make up the urban design process.
physical or administrative geo-spatial entities. The second domain contains data that is produced by the planning and design process. This information is not descriptive but rather prescriptive, analytic or interpretative of the urban environment, and is specific to an urban design project. Within each domain we define a series of groups to facilitate the understanding of the content of and the relations between the various feature classes (Figure 1). These are largely derived from previous work on urban ontologies carried out in the City Induction project (Beiro et al. 2009; Montenegro and Duarte, 2009; Duarte et al., 2011). The urban environment domain (ue) has the following groups: 1. Landscape system (lnd) 2. Built system (blt) 3. Mobility networks system (mbn) 4. Boundaries system (bnd) 5. Information system (inf ) Groups 1-4 represent the environment and group 5 contains spatial and pseudo-spatial data. This complements the urban environment domain with essential descriptive information of the environment, its population and activities, in support of site and context analysis for creation of the program and for evaluation of the plan.
Figure 1 Schematic structure of the urban design data model, with two main schemas for the urban environment description and the design process information, and a collection of groups and feature classes belonging to each domain.
The design process domain (dp) consists of these main groups: 6. Zones (zn) 7. Axes (ax) 8. Focal points (fp) 9. Site and design regulations (sdr) 10. Site and design analysis (sda) 11. Evaluation goals (evg) 12. Evaluation outcomes (evo) Groups 6-8 contain graphical elements of the spatial framework or master plan, produced in the formulation stages, and additional explanatory or conceptual information about a design proposal, produced in the design stages. Groups 9-12 contain design support information that is produced by the formulation and evaluation stages, providing design constraints, regulations, performance goals and assessment outcomes for the various design proposals. The two knowledge domains (urban environment and design process) are represented in the database by two separate schemas, and each design proposal is stored in its own schema with a structure identical to that of the urban environment schema. The main urban environment schema stores information on the existing environment and is used for site context
143
analysis and visualization. The proposal schemas store the proposed environments within the limits of the site boundary and are directly manipulated by the design generation process. The groups of each domain provide a prefix for the naming convention of data tables, thus grouping together in the database management environment the data tables that are related. Examples of the data structure The urban design spatial data model is defined following principles from spatial data modeling (Yeung and Hall, 2007). Figure 1 represents a
Class Schema Table Entity Land cover urbanenvironment ue_lnd_landcover polygon Attribute id type permeable structure Class Schema Table Entity Land cover urbanenvironment ue_blt_buildings polygon Attribute id lot_id area height floors gf_function uf_function units heritage status year Class Schema Table Entity Land cover designprocess ue_sda_indicators_site polygon Attribute id proposal ind_1 ind_2 ... ind_n
schematic structure of the urban design data model with the main relations between feature classes, the core data tables and some examples of custom tables that can be created to accommodate specific project or local characteristics. This expansion possibility is indicated by the ellipsis () punctuation in the groups that support it. In this section we provide examples of feature classes of the two domains, namely the land cover and building feature classes of the urban environment domain, and the site analysis and evaluation scores of the design process domain (Table 1). The
Type char char char char Type char char double double integer char char integer char char integer Type char char double double Description unique identifier land cover class surface permeability class surface structure Description unique identifier lot/parcel unique identifier area in square meters height in meters total number of floors ground floor function upper floors main function total number of units heritage class current status class year of construction Description unique identifier name of design proposal indicator 1 measurement indicator 2 measurement Table 1 Description of the land cover and buildings feature classes of the urban environment schema, and of the site analysis indicators feature class of the design process schema.
double
indicator 3 measurement
144
complete functional data structure of each table is provided, including the object classes, their geometry, basic attributes and relations. The data structure is typical of spatial relational databases, with the important addition of a geometry field that can store the urban forms geometry in the form of points, lines and polygons.
de Pijp in Amsterdam, Ypenburng in de Hague and Houten near Utrecht. The amount and diversity of information and the need to make it available to different users were sufficient reasons to opt for a managed data solution in the form of a database. The data sets used were obtained via the DANS EASY service [2]: Bestand Bodemgebrauk 2006, for land use and land cover (CBS and Kadaster, 2006) Kadastrale Kaart, for property subdivision and address points (Kadaster, 2008) TOP10 NL, for topographic layers (Kadaster, 2009) Wijk- en buurtkaart 2009, for census data (CBS and Kadaster, 2009) Initially, we started preparing the data sets using a commercial GIS platform and the files provided in Shape format. However, it became clear that certain operations of merging tiles, deleting records, and editing tables either took too long or the operation failed. As a result we decided to load the data directly into the database and perform the data preparation operations using the native SQL functions and the PostGIS functions. The database proved to be far more stable and efficient in performing those
Table gebouw_vlak gebouw_vlak ue_bnd_lots ue_blt_buildings gebouw_vlak Bbg2006 ue_blt_entrances gebouw_vlak gebouw_vlak Attribute geometry identi id from geometry hoogte / hoogteklas BG2006_a parsed from door numbers sequence typegebouw status -
145
operations, reducing processing times in some operations from 40 minutes to 4 minutes. The other two main tasks of data preparation were to make a correspondence between data sets, assigning to the data structure defined in the urban design data model the attributes and geometry of the existing data sources (Table 2), and to reclassify the attribute values themselves, because they were in different languages or were following a very specific classification. In some cases the attribute data is not available in the original data sets and it has to be synthesized from other attributes, complemented by local surveys or simply ignored with consequences to the kinds of analysis that one can perform. The result is a spatial database that follows the specification of the urban design data model, ready to be manipulated by various stages of the urban design process. Integrating the spatial database and CAD for urban form analysis and evaluation In order to demonstrate the use of the database and the integration of CAD and GIS we measure the urban form and evaluate the performance of the different pilot study areas using a tool developed for
Indicator Green areas per inhabitant Number of light railway/tram/trolley stops Bike paths lengths per inhabitant/dwelling Dwelling per hectare Connected community Street network length Ground space per inhabitant/dwelling Impermeable areas Average distance between building entrances Distance to nearest kindergarten Number of kindergarten in close neighbourhood Distance to nearest light railway/tram/trolley Number of access points to public transportation
this purpose in AutoCAD Map 3D. This tool is the proof of concept implementation of the City Induction evaluation module. The purpose of the module is to measure a series of sustainable urban form indicators and offer a variety of means to evaluate the results to support decision-making during the design process. The proof of concept tool implements a sample of indicators, listed in Table 3, identified in a review of existing sustainable urban development assessment tools (Gil and Duarte, 2010). This sample aims to test indicators of different dimensions of sustainability (environment, economy, and society) that are calculated at different levels of spatial resolution. The tool offers at this stage a basic urban design workflow following the City Induction structure. In the Formulation section it includes the functionality to set up a new urban design project and load data in the spatial database, to configure project settings, such as the acceptable walking distance, to define the project site and analysis boundaries, and to setup the design requirements in terms of the weight and performance levels of each indicator (Figure 2). In the Generation section the tool offers the possibility of starting a new design proposal by creating a new schema in the database and loading the design tools
Spatial unit site site site site site site site site street segment entrance entrance entrance entrance Result unit m2 n m n n m m2 m2 m m n m n Table 3 List of the sample urban form indicators implemented for analysis of the neighbourhood data stored in the urban design spatial database.
146
toolbars. Finally, it has an Evaluation section with a collection of tools relating to the analysis and evaluation of these proposals. This includes tools to run the analysis on the data sets, display those results visually through thematic maps of each indicator, compare the results between several different design proposals showing a radar plot (Figure 2), compare the proposals against a reference case (not necessarily from the same project), and score the results based on the previously defined performance levels, aggregating those scores towards an overall proposal performance score. The overall score is a weighted mean at various levels of aggregation and these are summarized in a multilevel pie chart for an individual proposal.
The user interface is implemented using AutoLisp and DCL. The data visualization functionality uses AutoCAD Map3D workflows because the Map API is not accessible from AutoLisp, and finally the various indicator calculations have been implemented as plSQL functions in the PostGIS database itself. We have decided to keep the data analysis and calculations in the database because of the performance gains observed in our previous experience and to keep the data processing functionality independent from any specific user interface. The AutoLisp front-end creates PostgreSQL instructions that are sent directly to the database via the command line using DOSLib 8.6 thus performing any number of
Figure 2 Screenshot of the proof of concept tool implemented in AutoCAD Map3D. It is displaying an urban neighbourhood orthophoto, overlayed with the results of the Number of access points to public transport indicator in a thematic map. The proposal comparison radar chart shows this designs performance against other options. The information that is mapped in this view is being read from the PostGIS database.
147
operations on the data. The charting functionality to visualize the evaluation outcomes has been implemented in Python using the Matplotlib library and connecting directly to the database using the Psycopg library.
CONCLUSIONS
In this paper we have presented an urban design data model to enable the development of spatial databases to support the urban design process. The proposed urban design data model has unique characteristics because it combines the more common urban environment description with a schema to support information relating to the design process. Testing the proposed urban design data model by building a spatial database has shown that these databases offer a suitable platform for an integrated approach to urban design, in line with the idea of a City Information Model (CIM), becoming the backbone of such a system. We demonstrate their use in the analysis and evaluation of designs integrated in a CAD environment, but we still need to test and demonstrate their use in a complete set-up that includes the formulation and the generation of urban design proposals.
ACKNOWLEDGEMENTS
This work is part of the City Induction project funded by Fundao para a Cincia e Tecnologia (FCT), Portugal, and hosted by ICIST at the Technical University of Lisbon (PTDC/AUR/64384/2006). The project is coordinated by Jos Pinto Duarte. Jlio Almeida is assistant researcher and Jorge Gil is main researcher in the project, funded by FCT with grant SFRH/BD/46709/2008.
REFERENCES
Batty, M, Dodge, M, Jiang, B & Smith, A 1998, GIS and urban design, CASA UCL, London. Beiro, JN, Montenegro, N, Gil, J, Duarte, JP and Stouffs, R 2009, The city as a street system: A street description for a city ontology, in Proceedings of the 13th Congress of the Iberoamerican Society of Digital Graphics, Sao Paulo, Brazil. Centraal Bureau voor de Statistiek (CBS) and Kadaster 2006, Bestand bodemgebruik 2006 BBG06, Persistent identifier: nbn:nl:ui:13-0an-1ei.
Centraal Bureau voor de Statistiek (CBS) and Kadaster 2009, Wijk- en buurtkaart 2009, Accessible from: http://www.cbs.nl/nl-NL/menu/themas/dossiers/nederland-regionaal/publicaties/ geografische-data/archief/2010/2010-wijk-enbuurtkaart-2009.htm. Dave, B and Schmitt, G 1994, Information systems for urban analysis and design development, Environment and Planning B: Planning and Design, 21 (1), p.p.83 96. Duarte, JP, Montenegro, N, Beiro, JN and Gil, J 2011, City Induction: a model for formulating, generating, and evaluating urban designs, in J. Halatsch et al. (eds.) Digital urban modelling and simulation, Communications in Computer and Information Science Series, Springer-Verlag, Berlin. (forthcoming) Gil, J, Beiro, JN, Montenegro, N and Duarte, JP 2010, Assessing computational tools for urban design: Towards a City Information Model, Future Cities, Proceedings of the eCAADe conference, ETH Zurich, Switzerland, pp.361-369. Gil, J and Duarte, JP 2008, Towards an Urban Design Evaluation Framework, Architecture in Computro, Proceedings of the eCAADe conference, Antwerpen, Belgium, pp.257-264. Gil, J and Duarte, JP 2010, A review of urban design sustainability evaluation tools, in HJP Timmermans and B de Vries (eds) 10th International Conference on Design & Decision Support Systems in Architecture and Urban Planning, Eindhoven University of Technology. Groger, G, Kolbe, TH, Czerwinski, A & Nagel, C 2008, OpenGIS City Geography Markup Language (CityGML) Encoding Standard, Open Geospatial Consortium Inc. Hamilton, A, Wang, H, Tanyer, AM, Arayici, Y, Zhang, X and Song, Y 2005, Urban information model for city planning, Journal of Information Technology in Construction, 10 (Special Issue From 3D to nD modelling), pp.55-67. Kadaster 2004, Kadastrale kaart, Persistent identifier: urn:nbn:nl:ui:13-oxg-wgy
Kadaster 2009, TOP 10NL - digitaal topografisch bestand, Persistent identifier: urn:nbn:nl:ui:13-fft-twq. Khemlani, L 2007, Autodesk University 2007, AECbytes Newsletter, (32). Kolbe, TH 2009, Representing and Exchanging 3D City Models with CityGML in J Lee and S Zlatanova (eds), 3D Geo-Information Sciences, Springer-Verlag, Berlin, Heidelberg, pp.15-31. Koshak, N and Flemming, U 2002, Object-Oriented Data Modeling and Warehousing to Support Urban Design, in Proceedings of DDSS 2002, Ellecom, The Netherlands, p.18. Maguire, D 2003, Improving CAD-GIS Interoperability, ArcNews, Winter, 2003 (24), pp.4. Montenegro, N and Duarte, JP 2009, Computational Ontology of Urban Design: Towards a City Information Model, Computation: The New Realm of Architectural Design, Istanbul Technical University, Turkey, pp.253-260. Pandit, P 2009, Using PostGIS/PostgreSQL for Managing CAD and GIS Data, Autodesk. Stadler, A, Nagel, C, Konig, G and Kolbe, TH 2009, Making Interoperability Persistent: A 3D Geo Database Based on CityGML, in J Lee and S Zlatanova (eds) 3D Geo-Information Sciences, Springer-Verlag, Berlin Heidelberg, pp.175-192. Webster, CJ 1993, GIS and the scientific inputs to urban planning. Part 1: description, Environment and Planning B: Planning and Design, 20 (6), pp.709 728. Yeung, A and Hall, GB 2007, Spatial Database Systems: Design, Implementation and Project Management, Springer-Verlag, Berlin. [1] http://resources.arcgis.com/content/data-models [2] http://easy.dans.knaw.nl/
149
150