Visualizing Urban History Using GIS

Barbara M. Parmenter
Community and Regional Planning Program
The School of Architecture
The Univesity of Texas at Austin

Abstract

Visualizing places and the way they evolve is a fundamental part of what planners, urban designers, landscape architects, and geographers do. Without being able to visualize a region as a whole, it is often difficult for citizens to plan for the future. This paper describes the design and construction of a GIS database that visualizes the history of landscape changes in Austin, Texas, going back to the early 20th century. The database enables users to see spatial change over time and explore economic, cultural, and political contexts. The paper discusses design and implementation issues.

Acknowledgements: The data layers discussed in this paper were developed by students in a Community and Regional Planning Applied GIS class, spring semester, 2002, at the University of Texas at Austin. They are: Brook Kintz, Jessica Trelogan, Sarah Harris, and Larry Diamond (land cover/urban built-up data development team); Nan Ding, Hyungwook Kim, and Yun Young Na (downtown data development team); Suk Yoon Kang and Sang-Hee Park (commercial-institutional development team); Salila Vanka, Ryan Fennell, David Ebel (Neighborhood Team); David Canright (downtown floodplain development); John Oswald (historic sites); Jim Weiler (multimedia).

Introduction

In 1892, Patrick Geddes set up a “Civic Observatory and Laboratory" known as “the Outlook Tower” in Edinburgh, Scotland. Geddes was of that generation of Victorian scientists, reformers, and visionaries who easily traversed the not-yet-formalized boundaries between biology, geography, town planning, and sociology (among other accomplishments, he created the master plan for Tel Aviv in 1925). Inspired by French geographers and sociologists, he proselytized on behalf of “synoptic” understanding of places based on careful study of history, economy, landscape, and culture.  While others of his generation embraced a crude physical determinism, Geddes knew that relations between people and their environment were much more complex, and yet he insisted that they must be studied together.  In the top part of the tower, a camera obscura projected Edinburgh and its environs on the walls. The next floor down was devoted to the history, present condition, and potential future of the city, including models, drawings, and photographs. The floor below that was devoted to Scotland, the next floor to language, the next to Europe, and the ground floor to the world. "The first contribution of this Tower towards understanding life is purely visual,” Geddes wrote, “for from here everyone can make a start towards seeing completely that portion of the world he can survey. He can also grasp what a natural region actually is and how a great city is linked to such a region" (Geddes 1968, 321-328).  The “Outlook Tower” was one of many devices that Geddes created and utilized in his effort to promote a “synoptic” vision of the complex interconnections, from micro to macro scales, between economy, society, and geography. Later he collaborated with French geographer and anarchist Elisée Reclus to build a huge globe at the 1900 World Exhibition in Paris that would act as a continuous frame of information, to be constantly updated with data from ongoing exploratory surveys.  His concept of an “Index Museum” which would be a sort of built encyclopedia sounds very much like an interactive hypermedia hall (Meller 1993).

The constant theme in Geddes’ work was the necessity of involving citizens in the careful empirical study of economics, culture, and environment together as the basis of sound planning. His promotion of the civic survey influenced decades of town planning methods and in Britain, the school curriculum. Most intriguing, he seemed to be groping for analytical and visualization tools that only came into existence decades later. Like British mathematician Alan Turing, he was able to imagine the things a computer could do long before the computer was invented.

A Return to Place-Based Planning

Over the last decade, the planning profession has been challenged from several directions to re-engage both planners and citizens with the artful physical shaping of communities that was once one of the profession’s strengths. The popularity of new urbanism, whatever its many faults, demonstrates that a focus on the tangible three-dimensional substance of community life at the very least stirs the public’s and the media’s imaginations. The definition of sustainability as being based on three components (environment, economy, and equity) also requires a tangible connection to our living space rather than sole consideration of resource allocation (Campbell 1996). The necessity of creating denser urban and suburban environments to combat sprawl after a century of devoting ourselves to getting away from congestion challenges planners to artfully combine design, development,  and resource conservation.  A 1997 MIT urban planning faculty seminar identified the planning profession’s major challenge as the integration of spatial and social-economic planning, to productively bring back to planning its connection with place and design, while building on what planners have learned in terms of socio-economic development and process mechanics and facilitation. This is typically referred to planning in three dimensions rather than two. The hope is that new information and visualization technologies will aid this integration (Rodwin and Sanyal 2000).

But to understand our future options in a way that builds upon the full dimensionality of a particular place we must regain the “synoptic” vision that Patrick Geddes espoused. This includes exploring how places have evolved and what that evolution presents in terms of diverse opportunities and problems across the landscape. True place-based planning requires the fourth dimension of time. Historical explorations of morphology, economy, ethnography, and ecology are critical for understanding variations between places and within places, and for recognizing specific opportunities for change or continuity.  Yet planning sees itself largely as a present and future-oriented professional science, and considering the kinds of time and budget pressures under which planners work, there is scant opportunity for detailed historical exploration.

A virtual outlook tower?  Building a temporal database for the Austin area

One place where we are afforded this kind of luxury, to get our historical bearings, take time to explore without a rigid agenda, and read broadly, is in our educational system. And here again, Geddes has preceded us to the same conclusion. His constant calls for “civic surveys” in the schools influenced the British school system for decades, and were echoed in the United States by regionalists like Lewis Mumford and Benton MacKaye. Being responsible for teaching the first-semester core course in our Community and Regional Planning Program at the University of Texas at Austin, I have found myself seeking ways to help students visualize both how cities evolve and how planners intervene in that evolution. In the Austin area we are particularly well-endowed with planning-related GIS data, available for free from city, regional, and state agencies. This gives us a wonderful current view of our region. My particular goal was to create an historical context for this current view. In thinking about how to visualize the history of my home region of Central Texas, Geddes and the early regionalists were my first inspiration. Could information and visualization technologies, geographic information systems first among them, be fashioned into a “virtual Outlook Tower”? 

In looking for examples of this kind of work elsewhere, projects in the Baltimore and the Albuquerque regions came closest to what I had in mind at the regional scale. Both were initially affiliated with the USGS’ Urban Dynamics Research program (USGS 2002).  The rationale for this approach is that urbanization is a significant part of land cover change, but we know very little at a detailed level about the interplay between economic, social, and physical processes of urban regions (Acevedo, Foresman, and Buchanan 1996). The Baltimore project has continued and expanded as part of the National Science Foundation’s (NSF) Long Term Ecological Research (LTER) program (BES 2002).  Phoenix also become an LTER urban site and is likewise developing an historical database (CAP LTER 2002). In both the USGS and LTER cases, the historical databases are primarily intended to serve environmental scientists, by linking land use changes to ecological dynamics. But both programs recognize the critical connections between ecological and social systems. In Baltimore, for example, one study examined possible correlations between social stratification (using income, home ownership, and ethnicity indices) and vegetation cover within a Baltimore area watershed (Grove and Birch 1997).

Yet the regional view is only one scale in which planners and citizens are interested, and it is a difficult scale for studying the interplay the physical and socio-economic contexts in detail. Urban cultural geographers have tackled these issues and, in my opinion, offered useful avenues for further exploration (Ford 1994; Arreola and Curtis 1993). In our initial exploratory efforts, I left it up to class participants to propose data development efforts that would examine the city at a finer scale. Below I briefly discuss both the regional data development and the more exploratory efforts.

Regional scale data development

I chose to follow the Urban Dynamics template as a partial basis for our historical database. The USGS is engaged in a project to study the Edwards/Trinity Aquifer (ETA) that includes much of the Austin / San Antonio corridor and the hill country region lying to the west, and area that largely overlaps with my scope of interest. Land cover change and modeling will be an important part of the project, especially regarding the impact of change on aquifer recharge and spring/stream flow on which much of this fast-growing region relies for water.  In both projects, researchers used cellular automaton models to explore past and examine possible future growth scenarios.  In the Albuquerque case study, the Clarke Urban Growth Model (SLEUTH) was selected to model future land use probabilities (Hester 1999). The model requires a set data that became part of our framework layers (see below). In addition several other layers were developed as basic framework data. The rationale for developing these data layers was that 1) it follows existing templates in other regions, and 2) the data, if of appropriate quality, can be utilized by other researchers, including the USGS ETA project (e.g., in urban growth modeling or estimating development impacts on water resources).

Framework layers - these cover most of Travis County:

These data layers, some of which are still being completed, provide a good regional view. We encountered a number of issues identified in the literature for other urban dynamics mapping projects, particularly the problem of being driven to some extent by available source material. Satellite imagery was used for the land cover mapping; hence we have no land cover prior to the use of satellite sensors. Experiments with land cover digitizing from geo-referenced aerial photos proved that this method would be prohibitive at a regional scale in terms of time investment. Nonetheless it is quite possible to do for smaller neighborhood analyses. The historical topographic maps are from slightly different years for each time period (1950s, 1960s, 1970s, 1980s) and there is not a complete set for the entire region.

For the major roads layer we took a different approach than other projects reported in the literature. We initially tried several approaches, including 1) digitizing fresh from the historical DRGs, and 2) adding temporal attributes to the current road centerline data set based on visual comparisons against the DRGs. Both methods were somewhat inaccurate and time-consuming, but the second approach had the most promise. We soon discovered, however, a tabular database at the Texas Department of Transportation that had dates for all state-maintained road construction activities (by what TxDOT calls “control sections”) going back to 1920. We were able to acquire the control section routes in ArcInfo format, and eventually join the database to the control section features so that we can specify and map both date of initial construction and major modifications for major roads in the Austin region. The result is a very accurate data layer at least for state-maintained roads. For major city or county maintained roads we will still need to do manual digitizing and/or attribute entry.

The graduate students working to develop the land cover and urban built-up layers did an excellent job but were stymied by data acquisition issues. The available data sets were not ideal in terms of time series or season, and there was a great deal of discussion concerning which of the several land cover classification schemes to use. In the end they decided to use a modified version of the USGS’ National Land Cover Data classification scheme (USGS 2001) as follows:

The students also attempted to derive urban built-up zones from the historical DRGs for years prior to satellite imagery availability. The initial results were promising, but interpretation issues arose that still need to be resolved. The USGS 1:24,000 topographical maps have pink tints for built-up areas; outside these areas, individual structures are shown. It is easy enough to extract the pink tint, but more difficult to determine what other areas should be included. Although these earlier layers showing urban built-up land are interesting, the more important period influencing current growth (going back to the 1980s) can be shown using the satellite imagery, and we will concentrate our efforts there.

Urban Built-up Land 1983

Urban built-up land 1983

Urban Built-up Land 2000

Urban built-up land 2000

While these data layers provide excellent visualization and general analysis opportunities, we concluded that a more appropriate time series for satellite imagery in terms of both year and season will be necessary for further research requiring high accuracy. Nonetheless, to my knowledge, the historical land cover and urban built-up data layers are the best data yet showing land cover change through time in our region.

The census tract data layers for 1970 and 1980 were created from aggregating 1990 census tracts back up to their earlier tract boundaries based on paper maps of Travis County census tracts for 1970 and 1980. Population and housing data were then manually entered for each tract.

The digital elevation model and data layer for areas excluded from development are both for a single time period (as required by the Clarke Urban Growth Model). It would be interesting to have terrain models for different time periods but this is beyond the scope of our time investment possibilities. The excluded areas include parks, for which it would be relatively easy to add date of establishment (although many parks were designated as such early on but not actually developed as parks until decades later). Other excluded areas include water bodies, conservation trust lands, and other lands purchased by local agencies for environmental protection purposes. In Travis County the excluded area is fairly extensive due to large conservation land purchases by the city of Austin plus park facilities provided by local governments.

Exploratory data development

While adhering to existing templates for the “framework” data development, I encouraged students to experiment in developing data layers to facilitate more detailed urban historical analysis. We followed no set template; instead we simply wanted to play with different possibilities. Different data layers cover varying parts of the Austin region. The results were fascinating and have generated more extensive ideas for future research.

Exploratory thematic layers:

Historical maps and aerial photos:

These materials combined with current data sets widely available in the region can provide at least the beginnings of virtual view of our city and region and show some of the changes that have taken place. But the larger aim is to provoke questions and ideas that then require more in-depth exploration in the community and its history. The following are just a few examples of what can be done.

Analysis example – downtown floodplain development

One student in the class created a presentation showing the transformation of the Colorado River floodplain in downtown Austin from flood-ravaged gravel pits and brick kilns to our current hike and bike park system looping what is known as Town Lake. The project shows a partially successful result of planning the re-development of a hazardous industrial landscape into a very popular waterfront park system. The effort has been so successful, and the population of Austin has such a high proportion of recent arrivals, that most people have no idea what this park area was once like.  But it also raises questions about environmental equity regarding the establishment of low-income minority residential neighborhoods in the floodplain, the role of a controversial power plant located in the middle of this community, and for which the dam forming Town Lake was actually built, and the construction of the hotels and offices on the south shore, effectively blocking a complete park linkage. The ability of GIS, hypermedia, and photo-editing software to bring these stories to life and raise additional questions is enormous. These kinds of stories abound throughout our region. And they serve to connect the issues that we have identified in our School and our profession as being crucial – morphology, ethnography, economy, and ecology.

Analysis example – grocery store development

Another team of students wanted to examine the spread of commercial and institutional development. To focus their efforts on something feasible for a semester’s work, they chose to map grocery stores, schools and medical facilities for a variety of time periods. Working mostly from city directories, they were able to use scanning and optical character recognition software to create tabular data sets that included facility name and address for the different periods. They then address-matched these using our current street centerline files and were relatively successful, achieving 80% or more matching rates. They then used ArcGIS Spatial Analyst to create density maps. This data can be used to examine economic changes through time, and also equity questions in terms of access by different communities with varying income and ethnic characteristics. We can also combine this data with photographic surveys to look at the morphological evolution of commercial strips and nodes.

Grocery store density 1890

Grocery store density 1890

Grocery store density 1930

Grocery store density 1930

Grocery store density 1980

Grocery store density 1980

Analysis example – downtown core development

A third team focused their work on digitizing building footprints for the downtown Austin core, a grid originally laid out in 1839 to be the capital of the Republic of Texas. The basic source for these data layers were Sanborn Insurance Maps, which are now available in black and white digital format online from University Microfilms (UMI) by subscription (UMI 2002).  The Sanborn Maps, originally drawn for insurance purposes, include a wealth of information about type of business, fire and flood hazards associated with a particular parcel or structure, and number of stories. The downtown team converted the digital maps from .pdf to .tif format, geo-referenced them, and digitized building footprints, adding number of stories as an attribute. By using an estimate for typical story height (12 feet) for extrusion purposes, the students were able to produce 3-D images of downtown Austin in 1900, 1935, and 1961. Comparisons can also be made with current (1997) footprints.  The maps of these periods show gradual infill processes, plus some extensive re-development associated with floodplain destruction and reconstruction, and with urban renewal. Again, various economic and equity issues can be examined, plus design problems.

Downtown 1900

Downtown 1900

Downtown 1961

Downtown 1961

Analysis example – history of planning activities

Looking for historical data sources, I acquired several previous comprehensive plans for Austin. We scanned and geo-referenced the major maps included in these plans.  Since most comprehensive plans contain maps of both existing and proposed land use, zoning, transportation, and facilities, these turned out to be very useful for understanding both what existed at the time of the plan, and how city leaders and stakeholders envisioned the city of the future. The 1928 Plan, for example, included an existing “land use” map. The land use classification scheme employed consisted of white residential, miscellaneous residential and business property. That alone reveals much about the concerns and mental frameworks within which planners of the time were working. In the text, the planners conclude that racial segregation is a problem in Austin, the problem being that there is not enough of it, with African Americans in particular being dispersed across the city. The authors note that it is no longer constitutional to zone by race, but that it is legal to establish race-based services (e.g., schools, medical facilities). They recommend therefore that services for African Americans be placed only on the east side of town where there is already a growing African American community centered around the industrial zone of railroad tracks and warehouses (Koch and Fowler 1957). The results of these actions are still very apparent in the physical segregation of Austin neighborhoods.  On a more positive note, the 1928 plan envisioned a parks system that at least in part came to fruition.

1928 Parks Plan overlaid with current city parks layer

1928 Parks Plan overlaid with current city parks

Conclusion

With the possible exception of the land cover change data sets (required for calibrating models of future land use), the kind of data we are developing in this project are not critical to anyone’s daily professional duties, including planners. Nonetheless, they can be powerful tools, and perhaps most powerful for the education not just of planners but of citizens in general. For that reason, university planning, geography, and urban studies programs are an appropriate context for this kind of project. For students (and community residents) who are always questioning why things were done this way or that, why some plans fail and others succeed, these data sets and accompanying texts are potentially very useful. Combined with more detailed looks at specific areas using older and current aerial photos, we can begin to discuss to how and to what extent formal planning interventions have affected the city and its surroundings. Rather than simply bemoaning urban sprawl, we can challenge students to look at an urban fringe area in 1951, consider population and employment growth at the time, and think through how it might have been done differently and what impacts that would have had. Why was it done the way it was, and could it have been done differently? What does that tell us about our planning efforts today? We can examine street patterns, lot size, and structure types in subdivisions from different periods, and through field investigations, devise qualitative and quantitative indices about how well these different places function from economic, ecological, equity, and aesthetic perspectives.

If we take up Patrick Geddes’ challenge, we also need to extend access to these data layers and visualization projects to younger students and to the community at large. Current educational theory calls for place-based education, where students use their own community for interdisciplinary investigations (Smith 2002). Current planning theory and practice requires widespread community participation in all phases of plan-making. Again, neither K-12 schools nor public agencies have the budget nor time to produce such data sets. But we do have the resources in our universities, and the development of these kinds of historical GIS data sets is educational in and of itself, forcing in-depth investigations of community history, data sources and data development techniques. By undertaking these kinds of projects and sharing them with our communities, we may achieve at least in part the “synoptic” vision that Geddes hoped us to have.

References

Acevedo, W., Foresman, T.W., and Buchanan, J.T. 1996. “Origins and Philosophy of Building a Temporal Database to Examine Human Transformation Processes.” Proceedings, ASPRS/ACSM Annual Convention and Exhibition, Baltimore, MD, April 22-24, 1996, Vol. I, p. 148-161. Available online: http://landcover.usgs.gov/urban/umap/pubs/asprs_wma.html (Accessed June 15, 2002).

Arreoloa, Daniel D., and Curtis, James R. 1993. The Mexican Border Cities: Landscape Anatomy and Place Personality. Tucson: The University of Arizona Press.

BES. 2002. Baltimore Ecosystem Study Web Server. Available online: http://www.ecostudies.org/bes/ (Accessed June 15, 2002).

Campbell, Scott. 1996. “Green Cities, Growing Cities, Just Cities? Urban Planning and the Contradictions of Sustainable Development.” Journal of the American Planning Association, vol 62(3): 296-313.

CAP-LTER. 2002. Central Arizona-Phoenix Long-Term Ecological Research Web Server. Available Online: http://caplter.asu.edu/ (Accessed June 15, 2002).

Ford, Larry. 1994. Cities and Buildings: Skyscrapers, Skid Rows, and Suburbs. Baltimore: John Hopkins University Press.

Geddes, Patrick. 1968. Cities in Evolution: An Introduction to the Town Planning Movement. New York: Howard Fertig.

Grove, J.M. and Burch, W.R. Jr. 1997. "A Social Ecology Approach to Urban Ecosystem and Landscape Analyses." Journal of Urban Ecosystems 1(4): 259-275.

Hester, David. 1999. “Modeling Albuquerque’s Urban Growth (Case Study: Isleta, New Mexico, 1:24,000-Scale Quadrangle).”  US Geological Survey Middle Rio Grande Basin Study—Proceedings of the Thrid Annual Workshop, Albuquerque, New Mexico, February 24-25, 1999. Open File Report99-203: 8-12. Available online: http://rockyweb.cr.usgs.gov/public/mrgb/forecast.html (Accessed June 15, 2002).

Koch and Fowler. 1957.  A City Plan for Austin, Texas. Koch and Fowler, Consulting Engineers. Reprint of the 1928 edition. Austin: Department of Planning,

Meller, Helen. 1993. Patrick Geddes : social evolutionist and city planner. London: Routledge.

Moody, Walter D. 1916. Wacker’s Manual of the Plan of Chicago: Municipal Economy. Chicago: Calumet Publishing Company.

Rodwin, Lloyd and Sanyal, Bishwapriya, eds. 2000. The Profession of City Planning, Changes, Images, and Challenges: 1950-2000. New Brunswick: Center for Urban Policy Research.

Smith, Gregory A. 2002.Place-Based Education: Learning to Be Where We Are.” Phi Delta Kappan. Vol 83(8): 584-594.

UMI. 2002. Digital Sanborn Maps. Available online: http://sanborn.umi.com/ (Accessed June 15, 2002).

USGS. 2002. Urban Dynamics Research Program: Analysis of Land Use Change in Urban Environments. Available online: http://landcover.usgs.gov/urban/intro.html (accessed June 15, 2002).

USGS. 2001. NLCD Land Cover Class Definitions. Available online: http://landcover.usgs.gov/classes.html (Accessed June 15, 2002).