Back to Volume
Paper: Demonstration of a Data Distribution System for ALMA Data Cubes
Volume: 485, Astronomical Data Analysis Software and Systems XXIII
Page: 7
Authors: Eguchi, S.; Kawasaki, W.; Shirasaki, Y.; Komiya, Y.; Kosugi, G.; Ohishi, M.; Mizumoto, Y.; Kobayashi, T.
Abstract: The Atacama Large Millimeter / submillimeter Array (ALMA) is the world's largest radio telescope in Chile. As a part of Japanese Virtual Observatory (JVO) system, we have been constructing a prototype of data service to distribute ALMA data, which are three or four dimensional cubes and expected to exceed 2 TB in total size, corresponding to 75 days at world-averaged Internet bandwidth of 2.6 Mbps, in the next three years. To utilize the limited bandwidth, our system adopts a higher dimensional version of so-called "deep zoom": the system generates and stores lower resolution FITS data cubes with various binning parameters in directions of both space and frequency. Users of our portal site can easily visualize and cut out those data cubes by using ALMAWebQL, which is a web application built on customized GWT. Once the FITS files are downloaded via ALMAWebQL, one can visualize them in more detail using Vissage, a Java-based FITS cube browser. We exhibited our web and desktop viewer “fresh from the oven” at the last ADASS conference (Shirasaki et al. 2013). Improvement of their performance and functionality after that made the system nearly to a practical level. The performance problem of ALMAWebQL reported last year (Eguchi et al. 2013) was overcome by optimizing the network topology and applying the just-in-time endian conversion algorithm; the latest ALMAWebQL can follow up any user actions almost in real time for files smaller than 5 GB. It also enables users to define either a sub-region or sub-frequency range and move it freely on the graphical user interface, providing more detailed information of the FITS file. In addition, the latest Vissage now supports data from other telescopes including HST, Subaru, Chandra, etc. and overlaying two images. In this paper, we introduce the latest version of our VO system.
Back to Volume