Geospatial knowledge discovery using semantic web services

Abstract

Modern-day satellites and other data acquisition systems have collected an overwhelming volume of Earth and space science data. The data are processed and managed by a variety of geographically distributed data providers. NASA’s Earth Observing System (EOS), for instance, has been generating on the average almost 100 gigabytes of imagery per hour for the past decade. It releases over 900 Earth science data products at more than a dozen data centers. It is extremely valuable for innovative scientific researches and decision-making processes to extract useful information and knowledge from these distributed massive volumes of data. A geospatial model in which prior domain expertise is encoded formally as computable algorithms can facilitate knowledge discovery by detecting and interpreting patterns and regularities, discovering classification rules, and inferring causation. With complex spatial and/or temporal dynamics, geospatial knowledge discovery commonly requires a capability beyond that of an individual geospatial model. Specifically, it involves a complex workflow that requires the integration of various geospatial models and distributed multi-disciplinary, multi-source, and multi-scale science data. For example, to predict fire behavior and estimate possible damage, decision makers and firefighters must effectively combine satellite observations, weather data, geographic data, census data, and simulation models from various sources. The best model and the most appropriate data must be selected in order to predict the fire spread in near-real time or real time. Therefore, the interoperability of geospatial models and data is becoming a critical issue for geospatial knowledge discovery.