Digital libraries extend traditional library tools, collections, and metaphors in new and sometimes radical directions. Geoscience information, traditionally found in libraries in the form of maps, reports, books, and journal articles, can potentially gain enormous benefits from digital interfaces that allow geospatial information to be effectively searched and displayed in context. However, to many users of such a collection, the concepts within geoscience represent a barrier to use and understanding. The Georgia Basin Digital Library (GBDL) Project examines community-based decision support for sustainable development, using digital geoscience information as a fundamental component. The context and relevance of geoscientific and other information in the GBDL is explained and represented through the use of semistructured representations based on architectural pattern languages, where a web of connected concepts are each contextualized using storytelling with supporting graphics. The GBDL allows multiple coexisting pattern networks, so that multiple perspectives on an idea, a situation, or a region are possible and in fact preferable. The GBDL interface makes use of geographic information systems (GIS) components to show maps, such that the point of engagement for a user might alternatively be a story, a map, a photo, or a text heading. Linking the mapping component directly to the pattern framework allows for community-level contextualization of ideas and stories. For geoscientists, the ideas in the GBDL are relevant both as novel approaches to sharing geoscience information with the public and as approaches that can be applied to specialized scientific discourse in the future. As the Web evolves, geoscience information systems and communication must evolve along with it; the GBDL shows one such evolutionary pathway.

The Semantic Web for Earth and Environmental Terminology (SWEET) is a NASA-funded prototype to explore the technological challenges in making the semantic Web a reality for earth system science data and information. SWEET includes an extensive collection of ontologies written in the Ontology Web Language (OWL). These ontologies encompass both faceted (decomposed) concepts and integrative (unifying) concepts. Taken together, these enable a representation of the two science processes of reductionism and synthesis. SWEET stores ontology elements optionally in a DBMS (database management system), with a two-way translator between the XML and DBMS representations. SWEET also includes wrappers for large external online databases (gazetteers, earthquakes, etc.) to enable their contents be virtual components of the ontologies ( http://sweet.jpl.nasa.gov ).

Geospatial Network (GeoNet) is a distributed information system accessible through the Internet that will permit users from the geoscience community to share and contribute knowledge from the transition zone between the Colorado Plateau and the Basin and Range province. Specifically, the goals of GeoNet are to facilitate the access, analysis, and visualization of geospatial data for use by scientists, policymakers, educators, and the general public. The types of geospatial data provided through GeoNet include satellite imagery, gravity, heat flow, digital elevation models, remote sensing, seismicity, active faults, and basic geographic data. This paper, written as a primer, describes how state-of-the-art technology is being used to realize GeoNet's goals. It also discusses GeoNet's layered architecture and shows how this architecture supports software applications deployed as service-oriented components. Ontologies based on the emerging OWL Web ontology language standard are used to enhance knowledge discovery. Use-case scenarios describe the features of GeoNet and the expected sequences of interaction between GeoNet and different classes of users. GeoNet's design will permit its functionality and data sets to be integrated with the GEON (Geosciences Network) grid and, thus, will contribute to the larger geoscience community.

A detailed survey of collections and publications for the Florissant fossil beds (Colorado, USA) forms the basis for developing a new relational database and Web site that documents information that had become widely scattered following 130 years of scientific study at Florissant. More than 1700 species that remain valid, mostly of plants, insects, and spiders, had been described in more than 300 publications, and these published specimens had been dispersed among ∼15 museums. Some of these specimens were not well documented in original publications and many of the type specimens had never been illustrated. Catalog data were compiled on-site at museums, specimens were photographed, and all of the publications referring to Florissant specimens were located. Taxonomic classification of the fossils was updated to modern concepts. A relational database incorporates the data into five core tables for specimens, bibliography, references to specimens in publication, taxonomy, and images. The database allows for complex searches to interrelate these categories, enabling new research and facilitating collections management. Examples show that the largest number of scientific publications and new species descriptions appeared from 1890 to 1920 and that most of the originally described insect species, but only about half of the plant species, still remain valid and unrevised. Digital images of the fossils and digital files for pre-1923 publications form an archive that is linked to the data records. A Web site makes the database publicly accessible for technical use, and also provides a less complex application for the layperson as well as a new college-level curriculum.

Data mining is presented in this paper as an example of geoinformatics, to illustrate how a collaborative effort between data-mining experts and geoscientists is essential for analyzing large volumes of data to extract knowledge. There is a wide variety of mining algorithms and tools available for both predictive and descriptive data mining that have been applied on a variety of geoscience problems. This paper provides a high-level overview of the data-mining process, several mining algorithms and tools, and their application to the geosciences. It also describes the geoscience data characteristics that make mining these data a challenge, and details some of the common data preparation operations that are required to make the geoscience data ready for mining. Three data-mining applications are presented in detail in this paper to demonstrate the usefulness of data mining to solve specific geoscience problems. Issues relevant in determining a successful data-mining application in geoscience are presented as lessons learned. Finally, the paper looks at challenges ahead by focusing on two relevant research areas for data mining and geosciences: mining streaming data in real time and exploiting the emerging cyberinfrastructure for mining.