Earth Science Datacasting: Informed Pull and Information Integration.

The intent of Datacasting is to empower consumers of Earth science data with the ability to extract from a stream of data granules (or files) precisely those granules that are required to meet a predefined need, for example, "Acquire from a MODIS L2 data stream only the granules that contain information about a wild fire in Southern California." Our approach to solving this problem has been to take the concept of Really Simple Syndication (RSS) feeds, for delivering regularly changing web content, and extend this to represent a stream of data granules and deliver regularly changing Earth science data content. In essence, this project is doing for Earth science what Podcasting has done for audio and video. Where Podcasting extended RSS to revolutionize how users access audio and video content provided by various media outlets, so Datacasting extends RSS to provide users with the ability to download data granules provided by Earth science data providers as the data are made available. Moreover, we have taken the concept one step further by creating a solution for filtering on the metadata of a feed in order to identify granules of interest based on user-defined criteria. In this paper, we also show how Datacasting feeds can be combined with other RSS-based feeds to identify relationships between information sources and extract new knowledge, as well as aid the development of new geo-based web services not currently envisaged. [ABSTRACT FROM AUTHOR]