A trial run

To give researchers an idea of how they might eventually use the NVO, Plante and his group unveiled a prototype at the American Astronomical Society meeting in January 2003. The prototype focuses on a particular research problem: identifying the relationship between the morphologies, or shapes, of galaxies within a given cluster and the dynamic state of that cluster as a whole.

When a cluster is analyzed, the metadata is catalogued internally using a flexible XML-table format called VOTable, designed especially for astronomical data. VOTable allows metadata and data to be stored separately but linked so that researchers who would find the metadata useful are not also required to download enormous chunks of data. The format is able to handle different kinds of data, including XML, FITS, and binary data. VOTable is also scalable and capable of managing varying sizes and quantities of data records.


In addition to the table, image cutouts are created containing single galaxies to help eliminate all but the most relevant information about those galaxies. Using Chimera, which Plante describes as “an Alliance-produced technology for creating and executing computational recipes for data,” three parameters for each galaxy are calculated in real time. These parameters include the average surface brightness, the light concentration index, and the asymmetry index (which describes whether a galaxy is spiral or elliptical in shape). Chimera uses Globus technology, also Alliance-produced and integral to the development of grid infrastructure, to manage the data and submit the processing requests to Condor pools (here, computing clusters located at NCSA, the University of Wisconsin at Madison, the University of Southern California, and Fermilab). The resulting analysis produces two files of FITS data and a catalog in VOTable format containing the calculations that users can download and view using a visualization tool such as Aladin or OASIS modified to read VOTable files.

By itself, the information imparted through the prototype analysis is already well known to researchers who study galaxy clusters. But, says Plante, “what we can do is not only make this measurement for one cluster but for many clusters. We can analyze not just hundreds of galaxies but tens of thousands of galaxies. We can build up a lot of statistics that show how the distribution of symmetric galaxies changes with the age of the galaxy, the distance, or many other variables.”

Access to enormous quantities of data, using the diverse and powerful computing capabilities of the Grid, could very well catapult researchers in astronomy and other fields into an entirely new way of framing scientific questions, says Brunner. “We're talking here about questions that we never would have thought about asking before because we didn't have access to the resources and technology that we have now.”

 

This research is supported by the National Science Foundation and NCSA.