Spatial Representations of Meteorological Data for Intelligent Retrieval

Abstract

Problems in quickly identifying relevant data and managing the massive amount of data involved have long prevented the use of
the historical record of meteorological data. This research proposes a novel method for storing and retrieving weather data that
takes these issues into account. A knowledge base that synthesizes the information of a much bigger archive is built using AI
case-based reasoning techniques. Using abstract definitions of its contents, this database enables intelligent retrieval of historical
data. Analyses and summary views of the data may be provided immediately when relevant data are obtained. Data sets in their
whole may also be obtained from slower medium like CD-ROM or magnetic tape in the event that a more thorough study is
necessary. The methodology is put into action in MetVUW Workbench, an intelligent retrieval and presentation solution for
historical weather data. Meteorologists will be able to use the technology as a “memory amplifier” to quickly find, analyze, and
present relevant past events.