The USGS Land Cover Institute (LCI)
DEVELOPMENT OF THE TEMPORAL TRANSPORTATION DATABASE FOR THE ANALYSIS OF URBAN DEVELOPMENT IN THE BALTIMORE-WASHINGTON REGION [2]
Susan C. Clark, U.S. Geological Survey
John Starr, Timothy Foresman
William Acevedo, U.S. Geological Survey
Carol Solomon, U.S. Geological Survey
ABSTRACT The U.S. Geological Survey is participating in a joint effort with the University of Maryland Baltimore County and other government agencies to construct a dynamic database of urban development for the Baltimore-Washington region from the late-1700's to the 1990's. The multi-theme temporal database includes a principal transportation data layer that documents the primary roads, railroads, and other transportation features that provided the infrastructure for urban development. A geographic information system was used to collect from various maps at various scales. Compilation criteria, such as connectivity, mobility, lineage, and alignment, were developed to accommodate limitations in the source materials. Visualization techniques such as animation and flight simulation are used to portray the changing transportation infrastructure and its effect on urbanization. The database will be used to study the evolution of the Baltimore-Washington metropolitan area. The principal transportation data layer is essential to illustrate and analyze the regional and temporal aspects of human-induced land transformations. Modeling applications will explore the urban growth process and its correlation to key transportation development.
INTRODUCTION
As part of its Global Change Research Program, the U.S. Geological Survey (USGS) conducted a study in the San Francisco-Sacramento region that documents human-induced land transformation (HILT) over a 140-year period (Kirtland and others, 1994). The USGS committed to build upon this effort in the Baltimore-Washington region through a project within the Mapping and Information Sciences Research Program (Acevedo, et al, 1994). The Baltimore-Washington Spatial Dynamics and Human Impacts Study will augment the initial HILT research by creating additional data layers and ensuring Federal Geographic Data Committee (FGDC) metadata compliance (FGDC, 1994). Data layers such as principal transportation, Census population data, and hydrography will be included in this study.
The primary objective of this study is to build a spatially referenced temporal database that reflects urban growth for the Baltimore-Washington region from 1792 to 1992. The USGS has joined the University of Maryland Baltimore County (UMBC) to assemble this database and to develop the techniques necessary to depict urban growth consistently through time. This database will be used to analyze urban growth patterns and rates, as well as to explore the relationships between themes.
Another objective of this study is to promote and demonstrate the use of various USGS spatial data in a geographic information system that contains data collected from the USGS historical map archive. These historical maps are a finite and fragile window into our past. By converting these maps to digital format, we can preserve this resource for further educational and research applications. Ultimately, the database resulting from this study will be used to produce visualizations illustrating how expanding human impact has affected urban, transportation, and hydrographic patterns through time. The depiction of urban development on regional environments and ecosystems is a key goal of this effort.
The Baltimore-Washington region, a 2-degree latitude by 2-degree longitude area centered on Baltimore, Md., and Washington, D.C., was selected as the study site. The area, which includes parts of the Chesapeake Bay, has undergone extensive urban growth over the past 200 years. Rich in both history and natural resources, it has been heavily affected by human presence (fig. 1). The initial focus of this work (Phase I) was performed on a 15-minute by 30-minute cell, centered on the city of Baltimore. Phase I provided the opportunity to test and refine approaches and techniques that would be applied to the broader extent of Phase II. Subsequent phases III and IV are projected to repeat the process of prototype and regional application for other themes. Themes being considered for this activity include wetlands, watersheds, soils, geology, land use, bathemetry, economics, and environmental health parameters.
This paper's focus is on Phase I and the concepts, criteria, and methods necessary to develop a prototype temporal database of principal transportation data. This initial phase provided an opportunity to test and refine approaches and techniques that would be applied to the broader expanse of Phase II.
The principal transportation classification scheme adopted for this study is a modification of the Anderson Level II Classification System for land use and land cover (Anderson and others, 1976). It was the consensus of the research team that this system be adopted to maintain a consistent and cohesive classification scheme (table 2). This classification scheme is flexible and can be expanded to accommodate evolving modes of transportation. All transportation features for Phase I are linear except railway stations, which are collected as polygons, and aircraft facilities, which are collected as point features.
APPROACH
The compilation of principal transportation features in a temporal database presents unique and challenging issues. The study must define principal transportation in a way that transcends variations in source materials but that also supports the objective. Principal transportation is defined as the transportation network that supports the development of the urban core. Development of the urban core requires both people and materials; therefore, features that facilitated the transport of either of these elements and met compilation criteria were accepted.
Methods were established to build the temporal transportation database using the urban delineations defined in this study (Crawford and others, 1996) as an ancillary reference source. The time-frames selected for this study were distributed over three ranges: 1792-1890, 1900-1966, and 1972-1992 (table 1). These ranges were defined by the unique nature of the historical sources used for each period. A team was established for each time period to collect historical sources and to develop methods based upon these sources and available resources. Participants included cartographers, geographers, a geologist, and an historian. The integration of concepts and insights from each of these disciplines provided a robust forum for the research.
Table 1. Data collection periods for the transportation database Range I Range II Range III 1792 1900 1972 1801 1925 1982 1822 1938 1992 1851 1953 1878 19661890
The availability of historical source maps determined the periods for which the data were collected. Each discrete view was evaluated for consistent data representation and synthesized into a coherent sequence. Data complies with FGDC metadata content standards. Visualization techniques were used to portray the resulting database in both static and dynamic views.
Table 2. Modified Anderson classification system for principal transportation.
Level I 1xxx = Urban or Built-up Land
Level II 14xx = Transportatation
Level III 141x = Roads 142x = Railroads 143x = Hydrographic Transportation 144x = Aircraft Facilities
Level IV 1411 = Principal Route 1419 = Highway Interchange
1421 = Main Railway 1422 = Rail Yard Line 1423 = Rail Station 1424 = Metro/Light Railway 1425 = Government Railway
1432 = Seaport* 1433 = Canal* 1434 = Ferry Lane
1441 = International Airport 1442 = Airfield 1443 = Military Airfield 1444 = Seaplane Base
*Although integral to the study, seaport and canal features were not included in the scope of Phase I. However, they will be included in the regional database of Phase II.
COMPILATION CRITERIA
The characteristics of a principal road change significantly through time. Because properties of transportation features evolve, we adopted a set of both implicit and explicit compilation criteria. Implicit criteria can be inferred by the characteristics and the patterns found in the source. Explicit criteria can be recognized readily on the map source as a cartographic feature.
IMPLICIT CRITERIA
Connectivity If principal transportation routes are defined as the network that supports the urban core, then connectivity between urban centers is the most important criterion specified for principal transportation. However, connectivity is not defined only as routes that link urban centers. These routes may also join other modes of transportation, such as ports, railways, or airports.
As a corollary to connectivity, system continuity was maintained when possible. The Department of Transportation, Federal Highways Administration (FHA) states, "The arterial system should be completely integrated, with stub ends occurring only at the urban boundary (in which case they connect with a rural arterial or a rural collector) or in areas having unusual topographic features, such as sea coasts." (U.S. Dept. of Transportation, 1989)
Lineage Feature lineage is derived from a line of descent. This regression can be observed as the consistent presence of a feature in the historical source maps. Additionally, by determining the predominant function of a route, researchers can presume its significance and the volume of its use. The historical functions of many of the principal transportation features were investigated by the study's historian. This information was particularly useful when selecting principal routes in the early years, when hierarchical road classes were often difficult to discern from our map sources.
The lineage of an enduring transport route, such as the old National Road, which would ultimately become US 40 (now MD 144), and then be replaced in the late 1950's and 1960's with Interstate 70, illustrates the continued significance a route may have over time.
Mobility Routes provide mobility. The FHA cites the issue of access versus mobility as integral in understanding the highway network. "Allied to the idea of traffic channelization is the dual role the highway network plays in providing (1) access to property and (2) travel mobility. Access is a fixed requirement, necessary at both ends of any trip. Mobility along the path of such trips can be provided at varying levels, usually referred to as 'level of service.' It can incorporate a wide range of elements...but the most basic is operating speed or trip travel time." (U.S. Dept. of Transportation, 1989) It is this "level of service" that is reflected in road quality and design and that defines mobility.
Arterial routes intersect, sometimes diagonally, auxiliary road networks to provide express mobility. Maintaining as high a speed as possible for the traveler, arterial routes may be designated "limited access," permitting access only at interchanges in order to avoid the impediment of traffic lights. Auxiliary roads, referred to as "collectors" and "locals", work together to provide access to the traveler's final destination.
In earlier times, one characteristic that defined mobility was road quality. Some roads were not passable in automobiles because of large rocks, tree stumps, or even streams. A well-maintained road is crucial to efficient travel. Cartographically, the presence of inns and stations along a route indicates a significant level of mobility. These points provided the traveler an opportunity to restock supplies and prepare for further travel. The appearance of rudimentary urban areas, or seed points, promoted better mobility, which in turn resulted in increased urban expansion. The symbiotic relationship between urban growth and principal transportation is strongly reflected in this criterion.
Alignment Principal transportation arteries are direct and therefore, maintain rectilinear characteristics. This phenomenon can be observed in the data. Key routes enter the city of Baltimore like the spokes of a wheel. Only when these routes begin to converge do they start to lose this quality, typical of inner-urban road patterns. Exceptions to this rule occur when topographic obstructions cause route alterations. Particularly in earlier time periods, principal routes circumvented obstacles such as mountains and water features. As our ability to span these features and modify the landscape evolved, these routes were made more direct. A meandering route may meet criteria such as lineage and connectivity, but unless it is direct, it is doubtful that it is a principal feature. In these cases, there is usually another route that meets the same criteria, yet maintains a more rectilinear nature. The absence of route alignment prompts a closer evaluation of the feature as principal transportation.
Exceptions also occur in the case of circular routes, such as beltways. However, these beltways channel traffic back into aligned routes, which in turn provide mobility to other urban areas. These circular routes generally occurred in the later time periods, when urban areas had become saturated.
EXPLICIT CRITERIA Embedded in our implicit compilation criteria are elements that are explicit in nature. Explicit compilation criteria are shown on maps as cartographic features. The depiction of features that further the effectiveness of a transportation route reveals the significance of that route. For example, in order for an investment to be made in the construction of a bridge, the road over the bridge is probably an important one. The presence of ancillary transportation features, such as bridges, tunnels, and ferries, is one of the factors that support the application of implicit compilation criteria.
The presence and nature of the names applied to these routes also helps determine principal transportation courses. Toponymically, the labeling of a road with its proper name may provide valuable information. For instance, names such as the "Annapolis and Baltimore Road" indicate the significance of this route as a major artery providing mobility between two major centers. The descriptive title of these routes often provides insight into their use. In fact, distinct eras favored the use of certain descriptors for their principal routes: "turnpikes," "highways," and "beltways" are signatures for their respective generations.
The advent of modern cartography and standardized approaches through feature classification and symbolization schemes has alleviated much of the dependence upon implicit compilation criteria. Digital line graphs (DLG), which are attributed vector representations of standard USGS topographic maps, were used to select most of the principal transportation network for the later time periods. For instance, when compiling the 1992 transportation view was being compiled, it was determined that roads designated as "interstates" by the Federal Highway Administration or as "Primary highway, hard surface" on the USGS 1:100,000-scale topographic maps met capture conditions by meeting the implicit compilation criteria.
SOURCE MATERIALS
Locating and retrieving the historical map sources present a continuing challenge. This challenge is a portent of further difficulties in this process, because Phase II requires much more coverage in less documented territory. A comprehensive inventory of all source maps used in this study appears in the metadata library (Baltimore-Washington Regional Collaboratory, 1996.) Although it is our intent to use the USGS's rich collection of historical topographic series maps whenever possible, these sources were not being produced until the late 1800's.
The nature of the earliest source maps is quite diverse. An example of this is G.M. Hopkins' Atlas of Fifteen Miles Around Baltimore Including Howard County, published in 1878. This atlas, which was acquired from a private collection, contains maps varying in scale from 1:31,680 to 1:158,400. Other source maps for the time period to 1890 were acquired through the Maryland State Archives of Cartography of Maryland, the Library of Congress, the Johns Hopkins University's Peabody Library Collection, and the USGS National Mapping Division historical map collection.
For the period from 1900 to 1966, we relied heavily upon USGS historical topographic maps. Map scales of these sources ranged from 1:24,000 to 1:2,000,000. The majority of the historical source maps scanned were 1:62,500 scale. Maps not scanned for data collection were used as ancillary sources. DLG data (1984) were used to derive the content of the final datasets for the periods 1900 to 1992. The most recent view was compiled by using 1992 Landsat Thematic Mapper data and 1992 commercial road maps.
Interpreting principal transportation features in the early time range was particularly difficult because of the need to discern actual features from proposed features and a lack of consistent symbology. Additionally, many of these maps had been commissioned by private citizens with special interests and purposes. Often, the collection criteria of the historical source were determined by the nature of this commission.
The oldest historical map sources were examples of both the art and science of early cartography. Because there was little standardization of map content or design, symbologies such as line weight and pattern were unreliable criteria. Railroad stations were often depicted in the same manner as city blocks, with only the feature name to distinguish them. Many of the early source maps were actually plans. No cartographic distinction was made between features that actually existed at the time and proposed features. These issues were resolved through the application of one of the implicit compilation criteria: lineage. This criterion proved valuable in our oldest time views and diminished in importance as we moved closer to contemporary views.
Although both implicit and explicit compilation criteria were relied upon, the implicit criterion of connectivity was used as the main factor. The presence of the other criteria further supported compilation decisions. Topographic symbology was also useful in confirming these decisions. In the more recent views, DLG attributes provided the information necessary to extract principal transportation routes. All implicit compilation criteria were met using this method, with little manual modification required.
DATA COLLECTION METHODS
The procedures established are outlined below in chronological order. Urban delineations generated in this study were used as a supplemental reference source.
Using the 1890 USGS topographic map as a source, study participants annotated transportation features that met collection criteria onto a mylar overlay. A separate ARC/INFO coverage was created for roads and rail. Designated features were then digitized into these coverages. Earlier views were derived from these coverages by deleting features that either did not meet collection criteria, or did not exist, in the time period. Any features that could not be derived from the 1890 coverage (roads that existed before 1890, but that no longer existed) were digitized into the final coverages. The road coverages for all views were subsequently integrated into one road coverage. Correspondingly, the rail coverages were also integrated into a parallel coverage.
For other historical maps to be used as sources, transportation features that met collection criteria were annotated onto a mylar overlay for use as a compilation reference. The maps were scanned to be used as the primary source of data. The resulting raster image was
registered, rectified, and used as a reference coverage in a geographic information system (GIS). Separate coverages that contained 1984 1:100,000-scale data were created for both road and rail data. The DLG data were draped over the scanned map. Items to be compiled for each time period were appended to the attribute file. The appropriate item for each arc was selected and valued to indicate feature type.
The annotated mylar and the scanned map served as references to indicate the arcs to be selected. These references were valuable, considering the magnitude of the arc data existing in the DLG data (33,672 arcs for roads and 2,516 arcs for rail.) In the few cases that additional arcs were required, they were manually digitized from the on-screen display. The resulting data were reviewed for consistent application of collection criteria (fig. 2).
For the 1972 to 1992 period, a coverage containing the 1:100,000-scale 1984 DLG data was created for roads, and another for rail. These coverages were used with topographic maps of various scales and commercial road maps as ancillary sources. Thus, the transportation data were primarily derived from the 1984 1:100,000-scale DLG data. When 1972 data was being collected, it was determined that using road features designated as "Heavy Duty, greater than or equal to four lanes" from 1:24,000-scale maps met the defined capture conditions for our features. Compilation of these routes was achieved by using the DLG attribute tables to select specific interstate, U.S., and State routes.
For the 1982 view, roads designated as "Primary highway, hard surface" on USGS 1:100,000-scale topographic maps met the defined capture conditions. Compilation of principal features was achieved by using DLG attribute tables to select all primary routes.
For the 1992 view, roads designated as "interstates" by the Federal Highway Department or as "Primary highway, hard surface" on the USGS 1:100,000-scale topographic maps met the defined capture conditions. The DLG data were updated using 1992 Landsat Thematic Mapper data and 1992 commercial road maps as a supplemental source. The same DLG attributes used for the 1982 view were used to select the appropriate arcs. Any recently built interstates not found on the DLG data but visible on the Thematic Mapper imagery and confirmed from commercial road maps were manually digitized using the imagery as a reference coverage. Rail features were also selected by DLG attributes. Arcs attributed as "sidings" in DLG's were deleted.
The process outlined above provided templates for each view. These templates were then edited manually to remove minor omission or commission errors resulting from the batch selection of arcs by DLG attribute rather than by review of each feature.
DATA SYNTHESIS
Because this study is a collaborative endeavor between the USGS and UMBC, data collection was conducted at various work sites. The on-site team completed the initial review and evaluation of this work. Any inconsistencies detected in the data were resolved. The resulting data files were made available through Internet file transfer protocol sites. Because the work was preliminary, these sites were made accessible only to the research teams.
By accessing the data from these locations, each team was able to compare its data to the other teams' results. As the various datasets neared completion, teleconferences were held while the teams viewed all of the datasets in sequence to confirm that a smooth and natural evolution of features was occurring in the data representation. This was done in an iterative manner, further refining the data with each round of evaluations and subsequent modifications.
Inconsistencies encountered in the data resulted primarily from lapses in connectivity. Because the transportation data rely heavily upon the defined urban extent, the principal routes may vary slightly to accommodate new urban data. Additionally, transportation routes believed to link urban centers contained in the outlying Phase II region had to be corroborated with corresponding historical urban delineations. A case of disconnectivity occurred over a few of the early time views, as routes entering Baltimore from the west appeared to merge before they actually entered the urban center. As road density increases near the city, it becomes increasingly difficult to discern a clear through route, especially without the benefit of road class symbology.
METADATA COMPLIANCE
In order to comply with the FGDC Content Standards for Digital Geospatial Metadata, the Baltimore-Washington Regional Study has decided to document this aspect of our work fully. Metadata is a vital part of this study. Although it requiring a significant investment of time and resources, metadata are particularly important in a study of this complexity. Because of the multi-variate nature inherent in a temporal database developed in a collaborative effort, the need to define the study's parameters, assumptions, and sources is essential. The intent of our study will be outlined in the metadata. Principal transportation has been defined only for the scope of this project, and should not be misconstrued for other purposes. In this way, metadata are integral to the successful application of this body of work to subsequent research and development.
RESULTS Results shown in figure 3 are urban, transportation, and hydrographic data. These views were created in ARCPLOT. The temporal aspect of the shoreline is shown only where it is adjacent to an urban area. Natural changes in the shoreline, such as erosion, were not within the current scope of this study, but may be considered for subsequent phases.
In the years 1792 and 1822, four routes emanate from Baltimore to the west. As discussed in the data synthesis section, two of these routes meet before the urban center--one of the anomalies addressed in the synthesis process.
In 1851, a substantial railroad network existed in the city. This network demonstrates ongoing expansion through 1938. In the late 1800's, suburban areas began to appear along the transportation corridors. The electric railways, or interurbans, contributed to much of this development. In 1966, the urban data for this frame have virtually saturated the Phase I area.
In 1966, the city's beltway appears. In the three subsequent frames, 1972 through 1992, the data depicted exceed the Phase I limit. This reflects the broader extent encompassed by Landsat Thematic Mapper data.
In order to provide an intuitive view of the data, NASA and the UMBC animation lab have developed a perspective fly-through of the study cell. For this animation, Thematic Mapper images are used in conjunction with digital elevation models (USGS raster elevation data) to provide a background for the study's GIS data (Masuoka and others, 1996). The visualization resulting from this regional study is not only a dynamic window into the Baltimore-Washington area's past, but also a view into land surface characterization's future.
CONCLUSION
One of the primary objectives of this project is to illustrate the impact that urban development has had upon the study region's environments and ecosystems. As urban growth accelerates globally, the scientific community must respond to the fundamental need for dynamic geographic information. The documentation of historical land transformation is a foundation for analytical and predictive modeling. Socioeconomic, environmental, and land-use/land-cover issues are a cross-section of the themes that could be explored temporally. The potential applications of this work would broaden in scope as the data diversify. This database, and the products generated from it, will provide the user community with the elements necessary to begin to explore the impact urban growth has had on the Chesapeake Bay region over the past 200 years.
The Baltimore-Washington Spatial Dynamics and Human Impacts Study has successfully demonstrated a collaborative approach to building a multithematic temporal database in Phase I. An integral component of success for this project is the development of a productive and mutually beneficial partnership. The combined technical and scientific expertise of the study partners provided a solid foundation for this research. Furthermore, these associations may be used for future activities. As Phase II ends, a host of potential partners are expressing interest in participating in this study. In times of reduced resources, this approach can yield a much more significant product than an individual agency's effort to address land surface characterization issues of this scope.
REFERENCES
Acevedo, W., Foresman, T.W., 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, (in press).
Acevedo, W., Bell, C., De Cola, L., Dynamic Mapping of Urban Ecosystems: Proposal to National Mapping Division Office of Research, 1994.
Anderson, J.R., Hardy, E.E., Roach, J.T., and Witmer, R.E., 1976, A land use and land cover classification system for use with remote sensor data: U.S. Geological Survey Professional Paper 964, 28 p.
Baltimore-Washington Regional Collaboratory, 1996, The Baltimore-Washington Regional Spatial Dynamics and Human-Impacts Study Metadata Library: Baltimore (in press).
Crawford, J.S., Acevedo, W., Foresman, T.W., and Prince, W., 1996, Developing a Temporal Database of Urban Development for the Baltimore-Washington Region, Proceedings, ASPRS/ACSM Annual Convention and Exhibition, Baltimore, MD, (in press).
Federal Geographic Data Committee, 1994, Content Standards for Digital Geospatial Metadata (June 8): Washington, Federal Geographic Data Committee.
Kirtland, D., Gaydos, L., Clarke, K.C., De Cola, L., Acevedo, W., and Bell, C., 1994, An Analysis of Human-Induced Land Transformations in the San Francisco Bay/Sacramento Area: World Resource Review, Vol. 6, No.2, pp. 206-217.
Masuoka, P., Acevedo, W., Fifer, S., Foresman, T.W., and Tuttle, M.J., 1996, Techniques for Visualizing Baltimore Regional GIS Data, Proceedings, ASPRS/ACSM Annual Convention and Exhibition, Baltimore, MD, (in press).
U.S. Department of Transportation, Highway Functional Classification: Concepts, Criteria and Procedures, Federal Highway Administration, March 1989.
ACKNOWLEDGMENTS
The authors express appreciation to the following people who contributed to the Baltimore-Washington Spatial Dynamics and Human Impacts Study: Janis T. Buchanan, Johnson Controls World Services; Walter Prince, and Dana Porter, and Helen Wiggins, University of Maryland Baltimore County; Janet Crawford, Steve Kambly, and Mimi Willan, USGS National Mapping Division, Mapping Applications Center. |
U.S. Department of the Interior |
U.S. Geological Survey
URL: http://landcover.usgs.gov
Page Contact Information: LCI Administrator
Page Last Modified: May 2010
