Statistics and Social Sciences Statistics and Social Sciences[Ed: Links to web pages and/or e-mail addresses which have become inactive since the publication of this article have been enclosed in curly brackets { }. Replacement links have been provided where possible.]
A Geographic Information System (GIS) can be defined as a computerized database system for the capture, storage, retrieval, analysis and display of tabular and spatial data. GIS is used by many disciplines, including geography, urban planning, engineering, landscape architecture, environmental sciences and sociology. It provides individuals from varied disciplines with a set of tools to improve the efficiency and effectiveness of working with map information of spatial and non-graphic tabular attributes.The origins of GIS date back to the 1960s at Harvard University, with the successful development of SYMAP. This product was raster-based and had the capability to shade slope maps. In the 1970s, Harvard's computer graphics lab produced Odyssey, a primitive GIS with polygon overlay functions. These two products were the first to be identified as having GIS functionality. It was the combination of computer-aided design (CAD) technology supporting a variety of design and drafting tasks, and database management system technology, for the manipulation of digital tabular data, that led to the development of GIS.
The Internet is another valuable tool for accessing information. The Internet was originally developed about 30 years ago by the United States Department of Defense as an experimental network for military research. Universities and government agencies soon demanded access to this technology as well. In the 1980s, the National Science Foundation developed a network called NSFNET that connected universities via telephone lines for scholarly research. In 1987, with growing network traffic and maintenance costs, a management and upgrade contract was awarded to Merit Network Inc. in partnership with IBM and MCI. The old network was replaced with faster computers and telephone lines. Today, Internet access is available to anyone with a modem and a personal computer. In fact, it has become a ubiquitous resource.
With increased use of both GIS and the Internet, the marriage of these two technologies was inevitable. The incredible potential for the use of geographic data on the Web has led GIS vendors to develop new software that brings GIS functionality to the desktop of Internet and intranet users. The information superhighway is becoming a valuable medium for sharing spatial data. Government agencies including the U.S. Environmental Protection Agency, the U.S. Geological Survey and the Census Bureau are in the process of providing direct access to their geographic databases via the Web. With the use of new web-based GIS technologies, geographic information is being deployed and accessed via the Internet and organizational intranets by a variety of users for diverse applications.
The raster data model, on the other hand, divides the earth's surface into a grid consisting of individual cells with an associated value. A raster layer can be an aerial photograph or a satellite image. Both are commonly used as base layers upon which the geometry and geographic coordinates of other layers are built.
Raster layers can also represent individual attributes. In this case, the value of the individual cell indicates the value of the attribute it represents. Furthermore, each raster layer only represents one theme. Topography and soils for an area would be stored in separate layers where each cell has a value for elevation or soil type. Operations on multiple raster layers involve the retrieval and processing of the data in corresponding cells positioned in different layers. In order to find all cells with an elevation greater than 1,000 feet and having a sandy soil type, for example, each cell in the elevation layer and each corresponding cell in the soils layer would be identified and output to a new combined layer (Figure 1).
Calculating the least-cost path is an example of a possible application of the raster data structure. In this procedure, each cell in a layer has an associated cost to traverse it. An example of layers in the raster model could be geology, vegetation, slope, aspect, soils and land use. These layers would be mathematically combined to create a final cost surface. A cost surface consists of a grid of cells containing the summation of the cost of each corresponding cell from all the layers in the model.
The next step in calculating the least-cost path requires defining the cells of origin and destination. Once this is done, the software runs an algorithm which derives the unique least-cost path (Figure 1). This is particularly useful when planning a road or a pipeline. Finding the least-cost path is a common function available in commercial GIS packages supporting the raster data structure. Two such packages available at ACF are ARC/INFO and GRASS. ARC/INFO is produced by Environmental Systems Research Institute (ESRI) and is the most commonly used high-ended GIS package available. GRASS is public domain software developed by the U.S. Army Corp of Engineers.
Dr. Smotritsky's solution, EARL, (Environmentally Acceptable Route Location), has been developed with funds from a National Science Foundation grant. It has been moved to the Web with the help of the Statistics and GIS group at ACF.
Once the calculations are complete, the solutions are displayed on a graphic report window that also displays the numeric representation of the total costs. From these numbers, one can determine which solution yields the lowest cost. In addition, the cost associated with specific cells can be changed independently to respond to specific needs, and a new path will then be calculated almost immediately. This is useful if factors other than lowest cost should be considered when planning a route. For example, if a planned road traverses an Indian reservation that had not previously been noted, the planner can set the cost of the cells in the reservation artificially high, to force the program to find an alternate route around the area.
The second method is more involved and allows the system to create the map dynamically in response to the user's request. Objects on the map are linked to a database that can be queried by the user. Mapping engines are served by data stored by spatial engines that directly plug into relational databases. These mapping and spatial engines play a role in developing a GIS website.
The software provides an application programming interface (API), based on the C programming language, for building spatial queries. It supports Oracle, Informix, Sybase, IBM DB2 and Microsoft SQL. SpatialWare, another example of a spatial engine, was developed by Blue Bell and Unisys, and was purchased by MapInfo in October 1997. SpatialWare is based on object-oriented technology, and stores data as a spatial abstract data type in an Oracle server. Unlike SDE, SpatialWare stores the genuine geometry instead of a series of vertices for a geographic object.
Despite these recent advances in spatial engines, challenges still remain. For example, spatial engines can not effectively handle raster-based geographic data.
Various operations, such as address matching, map composition, spatial queries and database links, can be performed with MapObjects because it is a programming environment. In other words, it is possible to create a system in which complex spatial operations are performed at the user's request, with the results sent out as a map. MapInfo's equivalent to MapObjects is called MapX. Windows NT has proven to be an optimal platform for map servers. Windows NT custom controls can be easily developed and deployed inside industry standards tools such as Visual Basic.
![[ C ]](../icons/CSmSum98.gif){kind=icon}
Posted May 18, 1998. Revised May 24, 2004.
|
|
|
|