A prototype spatial data environment for urban design

A Prototype Spatial Data Environment for Urban Design Bharat Dave and Gerhard Schmitt Department of Architecture, Swiss Federal Institute of Technology, CH-8093 Zurich, Switzerland A b s t r a c t . This paper describes an interdisciplinary research project aimed at developing a prototype system for representing spatial data. The project involves researchers and teachers in urban design, computer aided design and photogrammetry. The paper first highlights the spectrum of information used in urban design contexts that guided the approach we adopted in developing the prototype system. We next describe how the acquired data are transformed into useful information by associating with data items (i) textual records stored in a relational database management system and (ii) image, video and audio data. Finally, we present extensions of the prototype as a design decision support tool and close with some open research issues.

1

Introduction

This paper describes a project which is aimed at answering some specific needs for information representation and manipulation in the domain of urban design. In very brief terms, the user needs are to be able to appreciate, assess and operate upon representations of 3D spatial data, supported by computing media. The paper is organized as follows. First, we present the information context, i.e., what designers do, what kinds of information they work with, and examples of representations they use in the process. This discussion identifies a set of desired features that need to be supported by computer systems intended for use in urban design tasks. After briefly explaining the data acquisition process, we then describe the prototype implementation and an overview Of its various functionalities. Finally, we present example applications of the prototype and close with some open research issues.

2

Information Context

In a typical urban design task, various kinds of information are required as well as generated in the design process. For example, a design brief m a y call for the development of a particular site. Some typical questions one m a y ask in this context are: what are the current land-use patterns in the vicinity of this site? W h a t kinds of circulation and development patterns for buildings will fit in the existing fabric? W h a t is the desired spatial character of interior spaces? Even for such a small set of questions, it m a y be observed that the questions being asked

134

are varied, they involve different kinds of information, and they also require different kinds of representations (Figure 1). Additionally, we also note that the role of information being used is not only to represent existing spatial contexts but also to imagine new and as yet unrealized potentials of such spatial contexts.

Fig. 1. Example representations used in a design task

Based upon these observations, we identified the following three as our primary concerns for the prototype development. First, we need to enable the users to get as rich an appreciation of 3D spatial contexts as possible. We need ways to represent and visualize not only 3D environments but also to support information such as textual data and still images, video sequences and audio data. Second, we need ways to support abstractions of data and operations on such abstractions, i.e., when a user needs it, it should be possible to generate views of information with just the right amount of details. Third, we need a computing environment that is open and parsimonious. It should be possible to make use of diverse computing solutions that already exist and can be easily replaced. It should be also possible to give the end-user the control of how particular data items are to be organized, viewed and used. C o l l a b o r a t i n g G r o u p s Three groups of researchers worked on this project: urban designers, photogrammetrists and computer-aided design (CAD) group. The urban design team specified what kinds of information they are typically interested in, e.g., topography, movement network, parcels, builtup volumes, etc. These data requirement~ were handed to the photogrammetry team who conducted aerial photographic surveys, and from these images were extracted 3D models of various features. Our job in the CAD team has been to take the acquired data and transform it in ways that are useful to the urban designers.

135

3 3.1

Prototype Architecture Software Model

At the outset, we had identified some requirements that we considered p r i m a r y for developing the prototype system. The data should be in 3D vector f o r m a t and there should be a means to link 3D vectors with other data such as text, images, video, and audio. For these purposes, we also required a software platform that supports a true p r o g r a m m i n g interface as well as a direct access to the internal data structure, external programs and the operating system. In response to these requirements, we decided to use a 3D modeling p r o g r a m that serves as the graphics engine and then developed the features that it lacked on top of it. Thus, our prototype combines a set of features that are most useful to us from both the CAD and GIS software. The issues concerning CAD and GIS integration are also discussed at length elsewhere, see [3], [4], [8]. 3.2

Data Acquisition

As part of this project, digital data for two different sites were acquired by the p h o t o g r a m m e t r y team. The first site is the historic town of Avenches that dates back to the R o m a n settlement known as Aventicum, and measured approximately 16 km ~. The second site is the valley between Oensingen and Olten, and measures about 200 krn 2. This phase involved conducting aerial flights over the areas and taking high resolution images which were then used on an. analytical plotter to extract the following 3D vector data (Figure 2): topography, water and vegetation coverage, roads, streets and railways, and builtup volumes.

Fig. 2. Sample Data

136

3.3

Representation of Associations

Since we use the CAD software as the graphics engine, we need not worry about maintaining a graphics entities database. W h a t we do need, however, is to maintain pointers from the graphics elements into the external database to associate external data types such as text, image and other data. This scheme allows us to traverse and retrieve the necessary data and graphics entities in either direction (Figure 3). The software implementation of this scheme is facilitated by the CAD software which has a number of p r o g r a m m i n g layers, including the one which opens up the graphics data structure, and another one t h a t lets us access and communicate with external database m a n a g e m e n t systems. This is accomplished via standard SQL data types and SQL compliant statements that are passed to a database driver software. The prototype built on top of these communication modules allows the users to create and associate various kinds of information, including 3D vector d a t a and other data items stored in the graphics and external databases in the forms of records of ordered fields in tables.

Table Key columns

"

Tables ~ _ _

~

[

j

A

Table Key columns

Table C Key columns

I

~

A row

Fig. 3. Maintaining data associations

T a b l e C r e a t i o n The definition of a new table is done directly from inside the CAD environment (Figure 4), although it is also possible to use tables that are created separately, e.g., by a DBMS program. There are three data types that are supported: character, integer and float. Although the system can interface with four different DBMS, we have selected the one DBMS that seems to fulfill most of our needs and for which we have also built in some integrity constraint checking while a user defines a new table. For example, floats and integers cannot have width more than 8 digits. It is also possible to substitute these constraints with a different set of constraints that are compliant with another DBMS.

137

Fig. 4. Table creation

R e c o r d A d d i t i o n It is relatively straight forward to add textual records of information. At every stage, the user gets sufficient feedback about the data type and all related information such as the width and decimal digits that are acceptable (Figure 5). In order to associate the graphics elements and textual records of information, the user selects the graphical entities and an appropriate record; subsequently, the system maintains such associations. While making the link, the user also has a choice to link the selected entities with any image, video or audio data files.

Fig. 5. Record addition

138

V i e w i n g L i n k s The most direct ways to retrieve the data are to use the menu to view the links of the most recently edited data items or to first select the graphic entities of interest for which their related links, if any, are displayed on the screen. The textual records are displayed in their entirety whereas for image, video and audio data, appropriate buttons are highlighted. By clicking on one of the enabled buttons, external viewers or players are spawned via calls to the operating system (Figure 6). It is possible to use different external viewers if needed on various hardware platforms by adding them in a configuration file.

Fig. 6. Viewing linked data

Q u e r y S p e c i f i c a t i o n An alternate way of data retrieval involves composing a query using t h e point-and-click interface or by directly keying in appropriate SQL statement. All the tables are displayed in one window and their field names are displayed in another window (Figure 7). An SQL statement may be further qualified (i) by picking a graphic region of interest directly on screen or by keying in coordinates, and (ii) by using four logical compositions made with the buttons at the b o t t o m right. The first one allows the user to retrieve and view only the records that are linked with any graphic entity in the selected region of interest. The second one creates a set of graphic entities in the region as well as records linked with graphic elements that may even lie outside the region of interest. The last two subtract either the graphic elements in the region of interest from the elements that are linked to the retrieved records or vice versa. S Q L S t a t e m e n t E x e c u t o r A similar point-and-click interface for composing SQL statements is also available (Figure 8). The pull-down options at the top enable a user to see templates of SQL statements using keyword selections. The middle pull-down menus enable selection of SQL keywords and modifiers that

139

Fig. 7. Query specifications

are directly inserted in the statement with each selection, whereas the tables and their fields may be selected in the two scrolling windows. Using this interface, any standard SQL-compliant statement may be specified, which is then directly executed by the installed database management system. In this way, the prototype system provides a number of tools that work together in a transparent fashion without overburdening the user.

Fig. 8. Executing SQL statements

140

4 4.1

Further

Applications

P l a n n i n g Codes

With the database facilities described above, we have also experimented with other kinds of operations and decision support tools for design. One obvious application was to enhance the prototype so that it works with the existing planning regulations. Used in this way, it is very easy to provide feedback concerning the planning regulations during the early design stages to the designers (which they typically find out after the fact). A more experimental part of this application is shown in Figure 9, in which planning regulations are used to attach and maintain constraints on graphical objects. For example, a designer may choose a building topology which can be parametrically changed and placed on a parcel; the placed instance of the topology (which represents a design proposal) remains visible only until its volumetric and other attributes conform with the planning regulations.

Fig. 9. Using planning codes for interactive design aids

4.2

Dynamic Visualizations

With the software modules that we have available and developed, we still have some needs that cannot be met directly. For example, one of the most often visual appraisal techniques used in the design tasks is to build physical scale models. By walking around a model and viewing it from various directions, we are better able to judge a particular spatial configuration. A digital analog of such a capability can be provided by real-time animation. For this purpose, we

141

use PolyTrim [10] into which our CAD models can be imported in a transparent fashion. Subsequently, it is possible to undertake realtime 3D dynamic visualization studies (Figure 10) including simple flat shading, gourad shading or texture mapping, e.g., by draping scanned green area coverage or m a p s over contour meshes.

Fig. 10. Animation segment

5

Open

Issues

Based on our experiences in this project, we see the following as the m a j o r themes that constitute future research agenda:

1. Abstraction of information using domain specific knowledge bases: Once a context specific information has been retrieved and assessed, we also need to find ways to encapsulate and abstract information [2]. Computational solutions for generating such abstractions need to be developed which are in some sense similar to m a p generalizations techniques [6]. 2. Catalog of primary 3D volumetric operations: The tools should let the user select and compose operations out of the p r i m a r y ones in a language that is more suited to his or her task instead of assuming that there is an ideal user whose needs can be categorized a priori. Much broader perspectives on such issues are available elsewhere [1], [5]. 3. Open Systems: Related to the above is the notion of building open system s in a modular fashion which can be combined as and when needed by the users [7]. More modular the programs and more open their p r o g r a m m i n g interfaces, more acceptable they will become for the users.

~42

4. Networked repositories of data: In future, we expect more and more data to become available over the networks. W h a t the users will need, retrieve and generate will be the views appropriate for their own tasks. At least, two things that need to be addressed in this regard are fast algorithms, e.g., select-compress-transmit-decompress type of operations, and effective user interface designs. 5. Collaborative Work: In the case of large design projects, a number of professionals collaborate during design development and appraisal phases. There is an emerging body of work- computer supported collaborative work, that needs to be absorbed and further developed (for example, see [9]) for spatial data systems. Acknowledgements Foundation.

This project was supported by the Swiss National Science

References 1. Bracken, I., Webster, C.: Towards a typology of geographical information systems. International Journal of Geographical Information Systems 3 (1989) 137-152 2. Buttenfield, B., Ganter, J.: Visualization and G!S: What Should We See? What Might We Miss? Proceedings of the 4th International Symposium on Spatial Data Handling, Volume 1 (1990) 307-316 3. Dave, B., Schmitt, G.: Information Systems for Urban Analysis and Design Development. Environment and Plafining B: Planning and Design 21 (1994) 83-96 4. Geertman, S.: The GIS-CAD Debate within a Physical Planning Context: The Desirability of Integration. Proceedings of EGIS '92, Volume 1 (1992) 493-502 5. Goodchild, M.: Spatial Information Science. Proceedings of the 4th International Symposium on Spatial Data Handling, Volume 1 (1992) 3-12 6. Gruenreich, D., Powitz, B., Schmidt, C.: Research and Development in ComputerAssisted Generalization of Topographic Information at the Institute of Cartography, Hannover University. Proceedings of EGIS '92, Volume 1 (1992) 533-541 7. McAbee, J.: The Importance of "Openness" to Data for GIS Applications. Proceedings of EGIS '92, Volume 1 (1992) 163-172 8. Sadler, G., Hamid, A.: Integrated Analytical Tools to Monitor the Urban Environment. Proceedings of EGIS '92 Conference, Volume 1 (1992) 773-781 9. Schiffer, M.: A Hypermedia Implementation of a Collaborative Planning System. Proceedings of the URISA 1991 Conference, Volume I I I (1991) 121-135 10. Toronto: PolyTrim Reference Manual. Centre for Landscape Research, University of Toronto (1993).