The far-field beam pattern of a single-dish modulates the observed field of view. This pattern is the Fourier Transform of the single-dish aperture illumination. Hence for a rectangular aperture (a 2-d circular aperture is generally approximated by a rectangular aperture in one dimension), the far-field voltage pattern is a sinc function. While designing antennas, engineers aim at reducing the sidelobe levels to the minimum possible which is generally achieved by introducing a taper in the illumination of the dish. Thus the final design is generally a compromise between reduced illumination and reduced sidelobe levels. The beam efficiency factor quantifies the power contained in the main lobe to the total power and higher the factor, better is the efficiency.
The `primary beam' which is what the main lobe of the far-field pattern is generally referred to as in radio astronomy modulates the field of view. All the sources in the field, depending on their location within the beam are multiplied by the gain pattern of the primary beam. Thus a source at the beam center has unit gain whereas the strength of a source located at the half-power point of the primary beam has a gain of 0.5. This means that the strength of the source at the half-power point is reduced by 3 dB because of the primary beam gain. Since this is an instrumental effect, it needs to be removed. This is generally achieved by modelling the primary beam of the antenna by a high-order polynomial or a gaussian. The observed field is then divided by the gain pattern to regain the actual strength of the sources placed at different locations within the beam.
GMRT data is analysed using the NRAO AIPS package. PBCOR is the task in AIPS which is used to apply the primary beam gain correction to the data. For VLA data, a 10th order polynomial is fitted to the primary beam at different frequencies and applied to the data. For GMRT, we find that an 8th order polynomial gives a sufficiently good fit and there is hardly any change between a 8th and 10th order fit. Hence we have used an 8th order fit for the GMRT beams at different frequencies. These fits were found from 1-d grids in the elevation and azimuth directions. The half-power beamwidths found from this data are also listed. Subsequently we aim at obtaining a 2-d rastar scans to model the primary beam.
The polynomial fitted to the data is:
1 + (a/10^3)x^2 + (b/10^7)x^4 + (c/10^10)x^6 + (d /10^13)x^8
where x is in terms of (separation from pointing position in arcmin * frequency in GHz).
a,b,c,d are the coefficients which are required to be specified in PBCOR
when using GMRT data. In AIPS, the adverbs which are to be specified are
PBPARM(3) (=a), PBPARM(4) (=b) and so on to PBPARM(7). PBPARM(7)=0 for GMRT.
Below we give the values for different frequency bands for GMRT.
We recommend you use PBPARM(1)=0.1 and PBPARM(2)=1. However, you can try
a value of PBPARM(1)=0.05.
This work was started with A. Pramesh Rao.
Revised coefficients from data obtained in October 2004 have replaced
the older ones for 325 MHz and 610 MHz. There were errors in the half power beamwidths quoted
earlier which have been corrected here for the corresponding polynomial fit.
Last revised: 5 May 2005
ngk@ncra.tifr.res.in