Besides traditional responsibilities for education and research, the Center for Advanced Spatial Technologies (CAST) at the University of Arkansas, Fayetteville, is also a National Center for Resource Innovation (NCRI). One of six Congressionally funded NCRI sites, CAST is involved in helping schools and state and local governments decide which GIS and related spatial technologies they need and how to buy them. Particular effort is directed toward rural governments. CAST's director, Fred Limp, pointed out that rural county governments often have critical need for spatial technologies, but have a low tax base or a low technology base.
"For the last 10 years, much of what we have been dealing with involves interacting with this group and trying to find ways to develop the technologies appropriate to their needs," Limp said.
Working on the premise that resource sharing is the key issue, CAST Technical Director Jim Farley and technologists are designing the Seamless Warehouse of Arkansas Geodata (SWAG). When operational, the system will be accessible via the Internet or intranet, using any geoprocessing software supporting OpenGIS interfaces, a standard being developed and supported by major vendors of GIS technology. SWAG is designed around the spatial interoperability specification of the OpenGIS Consortium (OGC). The system runs on a Sun Enterprise 5000 Server, with UNIX; an Oracle Enterprise Server 8.i database engine using an object-relational data model; and Oracle's Spatial Cartridge, a database extension designed specifically for spatial data. When fully populated, SWAG will hold more than a terabyte of spatial data.
SWAG combines spatial data from various systems in a single database environment that allows users to locate and access them, along with associated metadata, from specific categories, or domains, such as roads, elevations, infrastructure, etc. Users can carve out a subset of data and insert it seamlessly in their own application. They will also have the ability to seamlessly overlay vector images on small or large-scale raster images.
Not Just Data Storage
Farley stressed that SWAG is not a data store. "If you throw an aggregation of data on a disk in a database, you have a data store. What we're building here is a warehouse," he said. "The difference between the two is driven by the amount of intelligence and domain expertise the interface presents to the user. In a data store, which is simply a large database, the onus is on users to have enough familiarity with the information ... to efficiently find what they need. In a data warehouse, an intelligent interface assists users in asking questions and getting information from a selected domain."
An example illustrates Farley's point and underscores the versatility SWAG achieves by melding an intelligent interface, OpenGIS implementations and a database using an object-relational data model: Users can go to the domain of cities, select one, draw a five-mile buffer around the downtown center, then ask for state highways within that area. If the original data contains different geographic scales and multiple attributes, or data layers, users can display a subset of the highways with the desired data layer and scale, and insert this in their application.
Initially, CAST will load SWAG with neutral framework data -- elevations, roads, hydrography, political boundaries, school districts, demographics, and, where available, local parcel information -- the type of data, Farley said, that many geospatial applications are likely to need. A project team has already loaded SWAG's Sun Enterprise 5000 Server with the complete TIGER (Topologically Integrated Geographic Endcoding and Referencing) Line files of Arkansas. Eventually, the SWAG database's wide range of imagery will include digital line graphs and orthophoto quads.
"But framework data is SWAG's first mission," Farley said. "We want the data to be neutral and useable. Arkansas supports the Land Information Board, an organized forum whose mission includes a mandate to develop a policy concerning digital spatial