The GMRT observations listed in Table 4.1 were carried out using an observing bandwidth of 32 MHz subdivided into 256 channels 1. During the first part of
the PhD project we reduced GMRT datasets with AIPS, then, after acquiring the expertise on the CASA data reduction, necessary to handle JVLA datasets, we started using CASA also for the GMRT, mainly to take advantage of the wide field imaging and of the advanced flagging techniques. Specifically, we reduced the observations of RXC J1322.8+3138, PSZ1 G019.12+3123, RXC J2051.1+0216, PSZ1 G139.61+24.20, A1443, RXC J0510.7-0801 and A402 with AIPS and those of A384, Zwcl 1028.8+1419 A1733, MACS J2135-010, A2355, A1437, A2104, Zwcl 2120.1+2256, RXC J0616.3-2156, Zwcl0634.1+4750, A3888 and A1451 with CASA. Regardless of the software we used, the calibration procedure is essentially the same and it is outlined in the following. The flux density scale was set following Scaife
1Only A2104 was observed with the old GMRT setup, namely with the simultaneous observa-
and flagdata (mode=manual) or with the AIPS tasks UVFLG, WIPER and SPFLG. The central channels were averaged to a smaller number of channels each 1 − 2 MHz wide to reduce the size of the dataset without introducing significant band- width smearing within the primary beam. A number of phase-only self-calibration rounds was carried out to reduce residual phase variations. A final amplitude and phase self-calibration was applied. We applied further manual editing to the data, flagging bad self-calibration solutions and visually inspecting the residual uv -data. Wide field imaging was implemented to account for the non-coplanarity of the base- lines. In particular we used the wprojection algorithm in CASA, while in AIPS we subdivided the field of view in tens of facets (the exact number of facets depends on the frequency, the resolution and the presence of bright sources). The facets were imaged separately, with a different phase centre, and then recombined. In CASA, wide band imaging (mode=mfs, nterms=2) was also used to consider the combination of the sources spectral index and primary beam attenuation. To deal with the bright sources in the field of view that typically reduce the dynamic range of the image we adopted the so called “peeling” technique. Specifically, in CASA, we subtracted the components of all the imaged field, except the source that we want to peel. We shifted the phase center of the visibilities on the position of the source, thus creating a dataset containing only that source and centred on its po- sition. This dataset is imaged and self calibrated several times to obtain direction dependent amplitude and phase solutions for that source. Then the peeled source is subtracted from the uv -data and the phase center is shifted back on the pointing coordinates. In AIPS, a similar procedure is implemented in the task PEELR, except that PEELR does not subtract the “peeled” sources.
We used the “Briggs” weighting scheme with robust=0 (both for AIPS and CASA) throughout the self-calibration and we produced final “full-resolution” im- ages whose properties are listed in Table 5.1. We subtracted all the sources outside the central ∼ 300 to reduce the size of the images and facilitate the imaging pro- cess. We then subtracted all the compact sources. First we made high resolution images excluding the baselines sensitive to the emission on scales larger than ∼ 250 kpc (uvrange' 2 − 3 klambda depending on the cluster redshift). We subtracted the clean components of the sources detected in the high resolution images and we used the new dataset to produced low-resolution images by using different weighting
Two of the GMRT datasets (marked with “S” in Table 4.1) have been processed with the Source Peeling and Atmospheric Modelling (SPAM) package. Here we provide a brief description of the SPAM pipeline, we refer to Intema et al. (2009), Intema (2014) and Intema et al. (2017) fore more details.
SPAM is an AIPS-based, semi-automated pipeline that uses the ParselTongue interface (Kettenis et al. 2006) to access AIPS tasks, files and tables from Python. The pipeline consists of two main steps: the pre-processing part that converts the raw data into pre-calibrated visibility datasets for all the observed pointings and the main pipeline which converts pre-calibrated visibilities into self-calibrated continuum images.
The aim of the first step is to obtain the instrumental calibrations for the best available primary calibrator and transfer them to the target field. The flux density scale is set according to Scaife & Heald (2012). After the flagging of the visibilities affected by severe RFI, gain solutions and bandpass solution are determined for the calibrator. The flagging, gain calibration, bandpass calibration loop is repeated several times to improve the results. The calibration solutions are transferred to the target field. In particular, the phase solutions from the calibrator are filtered to separate the instrumental and ionospheric contributions and only the instrumental phase calibration is applied to the target field. Data are averaged in time and frequency and converted to Stokes I to speed up processing. At this level, only simple clipping of excessive RFI is performed on the target visibilities.
The purpose of the main pipeline is the self-calibration. The initial phase cal- ibration of the target field is based on a multi-point source model derived from the NVSS (Condon et al. 1998), WENSS (Rengelink et al. 1997) and VLSS (Co- hen et al. 2007; Lane et al. 2014) radio source catalogues and then several runs of direction independent self-calibration are carried out. To take into account the non-coplanarity of the array, the wide field of the GMRT is divided into facets covering the primary beam and the fields around the bright sources outside the primary beam. The imaging is done using the “Briggs” weighting scheme with the AIPS ROBUST parameter set to -1. This slightly uniform weighting scheme produces a nearly Gaussian PSF, suppressing the broad wings typical of centrally condensed arrays such as the GMRT, at the cost of a reduced sensitivity to the large scale emission. During the self-calibration process, the excision of bad data is carried out
facet. The target field is re-imaged and deconvolved while applying the appropriate correction table per facet. At the end, a number of additional calibration (includ- ing an amplitude self-calibration) and flagging operations are done to solve for the residual direction-independent time and frequency calibration errors.
We re-imaged the datasets produced with SPAM to subtract the compact sources and to make low resolution images with appropriate ROBUST and TAPER parameters to increase the sensitivity to the extended emission.