Second Generation Spatial Information Warehousing Architectures

KEIGHAN, Edric (Ekeighan@cubewerx.com), CubeWerx, Inc., R-13, 200 Montcalm Hull, Quebec, Canada; and KUCERA, Henry (hkucera@ibm.net), Mercator Systems, Ltd., 2936 Phyllis Street, Victoria,V8V 4L8, B.C., Canada

Key Words: multiresolution, seamless, spatial, warehousing, distributed

The challenge for spatial data users is to access accurate data-trusted information in support of their specific needs. This paper describes the architecture for spatial information access based on a data warehouse driven by an open data access interface. Requested data sets, in vector, raster, matrix, and textual formats are accessible through on-line network gateways, compliant with international, national, industry, and government standards for exchange and interconnection.

Data Warehousing differs from traditional reporting and transactional systems in three significant ways. A warehouse provides a separate decision support database developed by integrating data from one or more operational systems. This supports synthesis and integration of the data. A warehouse also allows decision makers efficient on-line navigation into time-variant corporate data without impacting the operational systems. The warehouse also can be optimized for speed to allow distributed access to information that would be unthinkable in today's GIS or statistical applications. These three principles fundamentally alter the way business and government users interact with corporate data and how data can be leveraged. With data warehousing, users access data directly, when and how they want; instead of submitting requests to their Information Systems (IS) that might take weeks to fulfill. Decision makers can execute queries and build reports on their own workstations or thin clients, freeing the IS department to focus on other tasks such as maintenance and administration. If the spatial data warehouse is properly configured, users can issue one query after another in an ad-hoc manner to explore trends, identify problems, evaluate market opportunities, and/or order data to respond to their application specific requirements. Within a distributed data warehousing architecture, a spatial database populated with a basic set of feature data provides the framework while intensification or vertical integration of subjects are performed through access to federated databases in a seamless design.

The authors believe this type of system is dependent on a database architecture that is extensible (multiple data types), scalable (terabytes plus) and multiresolution (drilldown and rollup on the fly). The system also requires open interfaces to support thin clients and must be supported by a metacontent repository to support user interaction with the information. The first generation of this type of system was delivered through a project in Canada called Mercator I. The next generation architectures are presented and discussed within the context of ongoing research in Canada and the U.S.