Deep wide-field GMRT surveys at 610 MHz

In recent years we have used the GMRT (e.g. Pramesh Rao 2002) to make relatively deep, wide area observations of several fields at 610 MHz, with a resolution of about 5 arcsec. The deepest of these fields is the Spitzer 'Extragalactic First Look Survey' (xFLS) region1 , which is about 4 square degrees in extent, and is in a direction with very little Galactic infra-red foreground emission. This field already has deep VLA observations at 1.4 GHz available with a resolution of about 5 arcsec (see Condon et al. 2003), for which the GMRT provides complementary 610-MHz observations at a similar resolution. The other fields are selected areas from the 'SWIRE' (Spitzer Wide-area InfraRed Extragalactic) survey (Lonsdale et al. 2003). These areas have deep infra-red observations from the Spitzer Space Telescope, and often other deep complementary optical observations, but do not have deep, wide-field radio observations available.

Table 1 gives a summary of the xFLS and SWIRE fields which were observed. Given the primary beam of the GMRT at 610 MHz is ≈ 43 arcmin, each pointing covers ≈ 0.4 degree 2 . For the ELAIS-N1 field, four of the nineteen pointings (which overlap regions where deeper optical observations are available; Warren et al. 2007) were observed more deeply, to provide a noise value of ≈ 40 µJy beam -1 , rather than the ≈ 70 µJy beam -1 of the other pointings in this field. Also, observations of an additional 26 outer pointings in the Lockman Hole field have been obtained -in order to cover the whole ≈ 11 degree 2 observed with Spitzer -and analysis of these is ongoing. All of the observations were made with two 16-MHz sidebands, in both right and left circular polarised emission. Each sideband was divided into 128 spectral channels, so that any narrow band interference could be identified and removed.

The analysis of these observations was performed using classic AIPS. Initial careful editing of the data was necessary in order to remove interference, and other bad data (e.g. a few systematically poor baselines, due to correlator hardware problems). Bright ‘primary’ calibrators (e.g. 3C286 and/or 3C48) were observed at the start/end of each observation run, in order to define the flux density scale, and also derive antenna-based bandpass corrections. The bandpass corrections were applied, and several central channels in each sideband were averaged together, in order to make a ‘pseudo-continuum’ channel (‘channel 0’ in VLA notation). The antenna-based amplitude and phase calibrations of the telescope were derived using the pseudo-continuum channel, from the observations of nearby, compact ‘secondary’ calibrators, which were observed every 30 min or so. The antenna-based amplitude/phase and bandpass corrections were then applied to the observed visibilities, and 10 channels were averaged together, in order to reduce the size of the uv datasets before imaging.

The wide field-of-view of the GMRT means that multiple small ‘facets’ need to imaged, and then combined together. Generally 19 facets were used (i.e. a central facet, 6 in a surrounding ring, and 12 in a larger ring on a hexagonal grid), to cover the full observed field. The quality of the synthesised images were improved by several cycles of self-calibration, at 10, 3 and 1 min for phase only, and finally 10 min for amplitude and phase. This lead to images with dynamic ranges -peak to r.m.s. away from bright sources -of several thousand to one, with r.m.s. noise values away from bright sources that are close to the expected thermal noise. Near bright sources, however, the quality of the images is limited, which is believed to be due to variations in the pointing of the telescope antennas (see below).

The early analyses of the overlapping pointings from the xFLS field revealed two problems: (1) the time stamps of uv-data were slightly wrong (by about 7 s), meaning that the uv coordinates were also slightly wrong, leading to a distortion (nearly a rotation) in the synthesised images; (2) comparison of the flux densities of compact sources in overlap regions between adjacent pointings showed systematic differences, which are thought to be due to antenna pointing offsets during the observations, particularly at low elevations. For the former problem, an AIPS task was written 2 to correct affected data, by adjusting the time stamps and recalculating the uv coordinates (and the real-time GMRT system was corrected in the summer of 2006). The latter problem was correctly empirically, using a slightly shifted effective primary beam position. An The grey-scale ranges between -0.2 and 1 mJy beam -1 , and the resolution is 5.8 × 4.7 arcsec 2 , at a position angle +60 • .

improved pointing model is currently being implemented at the GMRT to address this problem.

Further details of the analysis procedures used are given in Garn et al. (2007b) and Garn (2009). Users of classic AIPS should also note that there were problems with the tasks SPLIT and SPLAT whic

…(Full text truncated)…

This content is AI-processed based on ArXiv data.