The application of GIS to mapping real estate values.
In recent years, scientists, planners, and resource managers have increasingly recognized geographic information systems (GIS) as a useful research and planning tool for applications such as resource inventory surveys, change pattern detection, resource routing, and land-use/environmental planning. Because GIS is highly suited for manipulating, analyzing, and presenting spatial information, it can make significant contributions to a wide variety of industries and businesses, particularly in this age of "niche" marketing. The objective of this article is thus to introduce the fundamental concepts of a GIS, and to demonstrate its application in a case study involving the mapping of residential properties sold in Memphis and Shelby County in 1980 and 1990, and the portrayal of its change pattern during the period.
GEOGRAPHIC INFORMATION SYSTEMS
Definition and overview
The concept of a GIS has evolved from the traditional manual procedures for registering and overlaying map layers in the late 1950s. This popular use of maps as an analytical tool provided an entirely new facet in map use - from mere physical description of geographic phenomena to spatial modeling applications. The evolution of GIS technology has since been intimately linked to the advent and development of computer technology. In its present form, GIS technology incorporates database management, spatial modeling, and computer graphics in a hardware/software configuration for managing geographic features [ILLUSTRATION FOR FIGURE 1 OMITTED]. A typical GIS software can provide the mechanisms to, capture, encode, edit, analyze, compose and display digital information organized as map layers in a GIS database.(2)
The key to the application of GIS technology is the database. A GIS database contains map layers representing geographic themes in a digital format.(3) These layers are geocoded to a standard coordinate system such as the Universal Transverse Mercator (UTM) or State Plane systems in use today. The system may be conceptualized as a stack of floating maps tied to a common map base in a way that enables features from individual layers to be correctly referenced to one another from a spatial standpoint [ILLUSTRATION FOR FIGURE 2 OMITTED].
Map layers in a GIS database can be organized in a variety of ways; the two most commonly used formats are the "raster" and "vector" data structures.(4) In each map layer, spatial data are represented using the three geometric primitives: points, lines, and polygons; these primitives are defined differently in the raster and vector formats.(5)
Raster data structure
In a raster format, each map layer is represented by a matrix of grid cells [ILLUSTRATION FOR FIGURE 3 OMITTED]. The origin of the matrix is normally at the top left corner (i.e., row 1, column 1), which corresponds to the northwest corner of the map layer.(6) Each cell is coded to represent a feature or value on the map layer. A point feature for instance, is represented by a cell that indicates its existence within the cell. Similarly, a line feature is represented as a string of contiguous cells, while a polygon feature is represented by a group of contiguous cells. This cellular structure gives features in a raster format a "blocky" or "jagged" appearance, and is therefore not cartographically pleasing. Of course, the jagged appearance can be minimized by increasing the cell resolution (i.e., decreasing the cell size). However, this will exponentially increase the number of cells required to map a given area, which in turn increases the processing time involved. Therefore, when deciding on a cell size, one should consider the tradeoff between cell resolution and data volume (and processing time).
Vector data structure
A vector format provides a more precise approach to feature representation through the use of X, Y coordinate strings [ILLUSTRATION FOR FIGURE 4 OMITTED]. A point feature is represented by a single X, Y coordinate pair indicating its exact location on the map. A line feature is represented by an ordered string of consecutive X, Y coordinates where the beginning and end points (or nodes) are the starting and ending coordinates. A polygon feature is similarly represented, except that the beginning and the end nodes have the same X, Y coordinates to indicate an enclosed entity.(7) Therefore, unlike a raster dataset where the polygon boundary is implied, the polygon boundary in a vector dataset is explicitly defined. Each point, line, and polygon feature is usually topologically organized. In the topological organization, information about neighboring polygons, common boundaries, and nodes is stored. This allows the relationships among entities to be maintained, and reduces redundancies of commonly shared points or linear features.(8)
Another characteristic of the vector data structure is its use of a relational database management system (RDBMS). Each feature (point, line, or polygon) is uniquely identified with its attribute information stored in related database files. For instance, the database file related to a map layer of road network will incorporate information such as road name, road type, physical distance, class type, and address range linked to each road feature. Through the RDBMS, information about each feature can be accessed for editing and updating as each feature is identified.(9)
In developing a GIS database, the choice of the data structure is an important decision. While each has its merits and limitations, the choice is often dictated by the software and the type of analysis required. A trend toward an integration of both formats in a GIS is currently emerging, however, particularly with the increasing use of digital satellite imagery.(10)
GIS and spatial modeling
A GIS is more than just a tool for designing or producing maps. It offers a range of operations for analyzing spatial data by manipulating map layers, individually or in combination, to derive solutions to spatial problems that will assist the user in decision making. As the real estate industry essentially deals with spatial data, GIS technology presents the most natural tool in the business. The mapping capabilities incorporated with a RDBMS provide the ideal tool for managing and displaying property locations. The database can be queried, and a property that meets a set of required conditions can be quickly isolated and mapped together with its attribute information.
Developers can make use of GIS modeling capabilities to identify and analyze sectional or regional change patterns, to determine accessibility to potential development sites, and to predict the market potential related to site selection for development. In the latter application, the greatest advantage of the use of GIS is its ability to pose "what if" scenarios for predictive analysis. For example, GIS can be used for a site suitability study - to identify locations suitable for a proposed subdivision development by analyzing the interactions and spatial coincidence of development objectives as well as environmental, demographic, economic, and political factors.
GIS APPLICATIONS - A CASE STUDY
Integrated with a relational database system, GIS can be incorporated into decision-making processes for presenting spatial data as well as for portraying changes that have occurred in different time periods. In this study, the fundamental data required for the analysis were the 1980 and 1990 residential property sale information summarized for each of the MLS zones in Shelby County. This information was obtained from individual sale values for the two time periods, and then consolidated for each MLS zone. Specifically, the operations included the following tasks:
1. Development of base maps for the GIS database that included a street and highway map of Memphis and Shelby County, the address and selling price of each residential unit sold in 1980, the address and selling price of each residential unit sold in 1990, and an outline map of MLS zones for Memphis and Shelby County.
2. Geocoding of the residential units sold in 1980 and 1990 within each MLS zone.
3. Identification of residential units in 1980 and 1990 within each MLS zone.
4. Development of summary statistics for each MLS zone based on 1980 and 1990 property data and changes that have occurred during the period.
GIS database development
The basic map layers required for developing the GIS database included:
* HOUSE80 - Houses sold in 1980
* HOUSE90 - Houses sold in 1990
* TNSHELS - Streets/roads of Memphis and Shelby counties
* ZONE - Housing zones
The street and highway map (TNSHELS) was the most critical layer in the database. It contained digital line data of streets and highways obtained from the U.S. Bureau of the Census (USBC) Topologically Integrated Geographic Encoding and Referencing/Line (TIGER/Line) file, and Digital Line Graphs (DLGs) obtained from the U.S. Geological Survey (USGS). The map layer was linked to a database file that included street names and address ranges for each block of streets/roads. This information provided the basis for locating each residential unit based on its respective address.
The database of residential properties sold in 1980 and 1990 (HOUSE80 and HOUSE90) was developed using data provided by MARR. For each respective file, reported sales of residential properties in 1980 and 1990 and their addresses (including zip codes) and selling prices were entered. Once the data had been entered, they were checked for errors and inconsistency. Finally, the MLS zone map (ZONE) was digitized from a base map. Because most parts of the boundary of each MLS zone coincide with existing streets, the street and highway map layer (TNSHELS) was used as the back layer when the ZONE layer was digitized. This was to ensure that zone boundaries and streets were precisely registered to each other.
The initial analysis involved geocoding the locations of the properties sold for the two study periods. This was accomplished by address matching each property listed in the HOUSE80 and HOUSE90 datafiles to the street and highway map layer (TNSHELS). Using the address-matching operation, the address of each property was then automatically matched against the highway/street map layer to derive its X, Y (easting and northing coordinates) location. To ensure successful matching, it was imperative that the original street address be as complete as possible, particularly the inclusion of street identifiers such as road, street, avenue, lane, and drive, and pertinent street directions (i.e., north, south, east, and west). Once the operation was completed, the two datafiles (HOUSE80 and HOUSE90) were transformed into map layers, each consisting of points representing the locations of property sold in the respective year, and registered to the street and highway layer.
Next, the MLS zone on which each property is located was identified. This was obtained by overlaying and intersecting HOUSE80 and HOUSE90 with ZONE; the zoning information was then coded into an item in the HOUSE80 and HOUSE90 datafiles. This makes it possible to identify the zone on which it is located when each property is accessed. Similarly, by querying the database for each zone, properties located within it can be quickly selected to compute summary statistics for each MLS zone. For each of the 1980 and 1990 databases, two map layers representing the total number of units sold and the average sale price for each zone were derived. Figures 5 and 6 show the average sale price for 1980 and 1990, respectively.
The change pattern and its distribution were obtained by overlaying respective map layers from the two different dates. Figure 7 presents the pattern of changes in units sold between 1980 and 1990. The appreciation of home values within the period is reflected in the map, showing changes in average sale prices between 1980 and 1990 [ILLUSTRATION FOR FIGURE 8 OMITTED]. With respect to the number of units sold, the attribute indicates an overall decline in housing sales during the 1980s, most likely caused by high interest rates. The map also indicates a disparity in the change pattern, however, with the greatest increase occurring outside, particularly to the east, of the city. Similarly, the greatest appreciation of house values was recorded in the same sector. Together, the changes depict a migratory trend of the more affluent to the eastern part of the county, and the less affluent to the inner city.
The most significant attribute of real estate analysis is its reference to location. In this respect, GIS technology is inherently highly suited for such applications. Its mapping capabilities serve as a convenient tool for spatially locating and displaying the data. When incorporated with a relational database management system, GIS provides access to attribute information of the mapped features, and its overlay capabilities allow changes over time to be rapidly identified.
The project demonstrates the effective use of GIS technology for consolidating information with spatial data for real estate analysis. It uses the address-matching capabilities of GIS to geocode 1980 and 1990 housing sale information to derive X, Y coordinates. The overlay operations are applied to categorize the housing sale data according to MLS zones. The query and summary operations provide access to sale information for each zone through the use of the RDBMS. Changes in sale values and units sold between 1980 and 1990 obtained by overlaying maps from different dates can be used to support decision making in property investments.
1. Memphis Area Association of Realtors, 1993 Atlas of Memphis Area MLS Maps (Memphis, Tennessee: MARR, 1993).
2. S. Aronoff, Geographic Information Systems: A Management Perspective (Ottawa, Canada: WDL Publications, 1989); and J. Star and J. Estes, Geographic Information Systems: An Introduction (Englewood Cliffs, New Jersey: Prentice-Hall, 1990).
3. D. F. Marble and D.J. Peuquet, eds. "Geographic Information Systems and Remote Sensing," Manual of Remote Sensing, 2d ed. (Falls Church, Virginia: American Society of Photogrammetry, 1983), 923-957.
4. P. A. Burrough, Principles of Geographic Information Systems for Land Resources Assessment (New York: Oxford University Press, 1989).
5. Star and Estes, 34-36.
7. Aronoff, 164-166, and Star and Estes, 48-50.
8. Aronoff, 172-177.
9. J. C. Antenucci, K. Brown, P. L. Croswell, M. J. Kevany, and H. Archer, Geographic Information Systems: A Guide to the Technology (New York: Van Nostrand Reinhold, 1991); and T. E. Avery and G. L. Berlin, Fundamentals of Remote Sensing and Airphoto Interpretation (New York: Macmillan Publishing Company, 1992).
10. Avery and Berlin; and G. Schultink, "Integrated Remote Sensing, Spatial Information Systems, and Applied Models in Resource Assessment, Economic Development, and Policy Analysis," Photogrammetric Engineering and Remote Sensing 58, no. 8 (1992): 1229-1237.
Berry, J. "Computer-Assisted Map Analysis: Potential and Pitfalls." Photogrammetric Engineering and Remote Sensing 52, no. 10. (1987): 1405-1410.
Fung, D.S. "Managing by the Map." Business Perspectives 8, no. 2. (1995): 12-18.
Lo, C. P. and R. L. Shipman. "A GIS Approach to Landuse Change Dynamics Detection." Photogrammetric Engineering and Remote Sensing 56, no. 11. (1990): 1483-1491.
Parker, H. D. "What Is Geographic information System?" Proceedings, Sixth Annual GIS Conference. (San Francisco, California, October 26-30): 72-80.
Schultink, G. "Integrated Remote Sensing, Spatial Information Systems, and Applied Models in Resource Assessment, Economic Development, and Policy Analysis." Photogrammetric Engineering and Remote Sensing 58, no. 8. (1992): 1229-1237.
Tomlin, C. D. Geographic Information Systems and Cartographic Modeling. Englewood Cliffs, New Jersey: Prentice-Hall, 1990.
Devlin S. Fung, PhD, is an assistant professor in the Department of Geography and Planning at the University of Memphis. He received a PhD in geography from the University of Georgia. Specializing in photogrammetry, remote sensing, and GIS and their applications to resource management and land use planning, Dr. Fung has designed GIS workshops for business and environmental planning.
Hsiang-te Kung, PhD, is a professor and chair of the Department of Geography and Planning at the University of Memphis. He received a PhD in geography from the University of Tennessee. His research interests include urban physical environment and regional studies of the Far East.
Melvin C. Barber, PhD, is associate professor at the University of Memphis. He received a PhD in geography from Southern Illinois University and specializes in economic geography with an emphasis on marketing, urban development, and retail and industrial location analysis.