Daily Image
23-01-2009
Simulated Cosmology
| Submitter: | Tony Willis |
| Description: | Although we hope that the large advance in collecting area of telescopes such as the SKA will reward us by making spectacular discoveries of previously unknown natural phemomena, scientists must still go through the exercise of attempting to guarantee to funding agencies that the huge monetary investment in such an instrument will produce interesting results no matter what! We do this by extrapolating the properties of nearby normal galaxies and radio galaxies, to predict their properties in the distant (and early) universe. The mid-frequency SKA may be composed of ~10 metre dishes with a field of view of about one degree. The left image shows a simulated sky with a diameter of one degree (twice the diameter of the full Moon) where the faintest sources have a flux density of 0.1 microJanskys. The cosmological simulation software (developed by Andrew Hopkins, now at the Anglo-Australian Observatory) predicts that there should be approximately 280,000 sources with an average separation of about 0.1 arcmin in the field of view. In order to properly observe the sky to such sensitivity we must therefore construct a radio telescope whose synthesized beam width would be about one-tenth of the average source separation, or about 0.01 arcmin (0.6 arcsec). (The actual SKA is expected to reach even fainter flux density limits and required resolution limits of 0.1 arcsec have been suggested.) If we have a resolution (PSF) of 0.6 arcsec, a pixel separation of about 0.2 arcsec would be needed in order to properly sample it. This would imply that we must create an image of about 36000 x 36000 pixels on a side before deconvolution (quadrupling the size of the final image). We did not have the computing resources to do this, and our main goal was to test the behaviour of the telescope imaging software. So we 'observed' the simulated sky @1415 MHz with an array having 30 dishes and a maximum baseline of about 6 km. This gives a spatial resolution of about 5 arcsec. At the lowest signal levels we would not be able to distinguish the individual sources. The right image shows the resulting 'observation'. Note that the ring-like structure around the brighter sources is the Point Spread Function (PSF) caused by the array configuration. This simulation has become possible because of a new MeqTrees component called the UVBrick, which calculates the observed visibilities for such a complex field rather efficiently. But the real impact of the UVBrick will come when it is also able to apply "image-plane effects" like station beamshapes and ionosphere. Because of the way MeqTrees is organized, this tricky problem can be tackled in friendly collaboration/competition by a group of widely separated specialists in London, Penticton, Socorro, Cambridge and Dwingeloo. |
| Copyright: | Tony Willis |
| Tweet |  |