Dr. Anna Scaife from the University of Southampton visited NAOC on September 15, 2014 and gave a lecture about challenges associated with the large volume of data that will be produced by the Square Kilometre Array (SKA). Dr. Scaife started by giving an overview of the SKA, which will have radio receivers in South Africa and Australia during the first phase of operation, and more telescopes may be added in other countries later. The SKA will incorporate forefront technologies to achieve both wide bandwidth and a large field of view. Two projects currently in operation, MeerKat in South Africa and ASKAP in Australia, are precursors for the SKA which will aid in further development.

Dr. Scaife continued by describing how huge radio telescopes need huge data processing abilities. Conservative estimates show that during its first phase, the SKA will produce petabits of data per hour. Observations from the receivers must first be correlated in order to derive their interferometric properties. The data must then be ingested into a data processing pipeline that can apply Fourier-transform based techniques to deconvolve the data and produce data that are interpretable for scientific targets. According to Dr. Scaife, the current algorithms that are used in data processing applied to radio observations are too slow because they are inherently serial operations that have not incorporated distributed computing. In addition, they have not utilized new technologies like GPU processing which can dramatically speed up calculations. She said that early tests from her research group show a speed up of three orders of magnitude for part of the pipeline.

Dr. Scaife concluded the talk by saying this project represents a true “Big Data” challenge that requires extensive development of existing algorithms into highly distributable forms. As part of an international consortium, her work represents a valuable contribution to the SKA effort.