BUILDING THE AUSCOPE AUSTRALIAN EARTH SCIENCE GRID

In 2006 the Australian Government announced a new funding initiative, the National Collaborative Research Infrastructure Strategy (NCRIS). NCRIS aims to provide Australian researchers with access to major research facilities, supporting infrastructure and networks necessary for world-class research. As an element of this strategy, $42.8 million was allocated to the Australian Earth Science Research community to build an integrated national Geoscience infrastructure system called AuScope. The NCRIS AuScope funding has several parts. One part of the funding is to develop an advanced national infrastructure for acquiring and analysing geophysical and geochemical data. Components in this infrastructure include a geotransect facility for imaging large-scale cross sections of the earth's crust, an Ion Probe for advanced analysis of earth samples and the development of a virtual library of drill core samples from across Australia. Another part of the AuScope infrastructure is the development of a High Precision Positioning System, which will build an enhanced national geospatial reference system.

To draw together information from this new national infrastructure and from other existing sources, the AuScope NCRIS funding will also be used to develop a world-leading Geoscience geoinformatics network. This network will be called the AuScope Grid and will comprise a Data Grid and a Compute Grid. Combined, they will provide a distributed infrastructure that will enable the dynamic construction of an open-access, four-dimensional model of the Australian continent.

The Earth Science Data Grid is a national Geoscience data network which will enable online access to information from new NCRIS Geoscience infrastructure and from other sources in academia, industry and government. The Data Grid will use open geospatial standards to allow real time access to data, information and knowledge stored in distributed repositories. A key objective for the Data Grid is that it will be built on ‘end-to-end' Science principles (aka open access principles) whereby there will be access to the highly processed information and knowledge as well as the original raw data and the processing programs used to generate the results.

The goal of the Compute Grid is to facilitate quantitative Geoscience analysis by providing an infrastructure and tools for advanced data mining and online computational modeling and simulation. Computationally demanding Geoscience programs will be made available as web services, and distributed across computing and storage resources in a manner that requires limited knowledge of the physical infrastructure.

The key to linking components and resources on the AuScope Compute Grid with the Data Grid will be service based access to the Geoscience information holdings using a common service interface and information models, including GeoSciML. Further development, maturing and formalization of GeoSciML are essential to the success of the AuScope Grid. GeoSciML is being developed through the Interoperability Working Group of the Commission for the Management and Application of Geoscience Information (CGI), a commission of the International Union of Geological Sciences (IUGS). Similar information models are also required for geophysical and geochemical data. As with GeoSciML, it is desirable that these additional information models be developed as Open Standards and under the auspices of the relevant International Scientific Organizations.

There are no obvious technological barriers to what has been proposed in building the AuScope Grid. Nearly all the required technical elements have been trialed in recent years in a series of interoperability test bed projects. Potential limitations are now seen as ‘social engineering' issues. One social limitation to developing the AuScope Grid to its full potential is the requirement to bring the relevant communities together at an international level; to work collaboratively to develop open standards for information exchange of all of the required scientific content. Developing and ratifying these standards at an international level is only the first step: they also have to be widely accepted and adopted by the community.

More fundamentally, the proposed Grid requires a transition to fully distributed systems whereby all components (tools, applications, compute resources and data) are available online as globally distributed resources. This paradigm shift contrasts the current culture of ‘monolithic silos' whereby all the required components for a data mining exercise or for a modeling and simulation experiment are hosted by a single organization. In some areas, this social transition is proving the most difficult challenge of all as distributed systems are sometimes perceived as reducing control of assets and resources. Distributed systems also rely heavily on mutual trust and open collaboration, and require that once resources are exposed as web services that these are maintained and made available 24*7.