Two-dimensional homography-based correction of positional errors in widefield MRT images

TL;DR

This paper presents a homography-based method to correct systematic positional errors in widefield MRT images, improving accuracy without reprocessing raw data, and investigates array geometry errors affecting source positions.

Contribution

The study introduces a two-dimensional homography correction technique for MRT images and re-estimates array geometry errors using astrometry principles, enhancing positional accuracy.

Findings

Positional errors reduced to within 10% of beamwidth after correction.

Detected a 1 in 1000 stretch in declination, indicating array geometry errors.

Estimated array geometry errors of ~1 mm/m and a 40 arcsec inclination.

Abstract

A steradian of the southern sky has been imaged at 151.5 MHz using the Mauritius Radio Telescope (MRT). These images show systematics in positional errors of sources when compared to source positions in the Molonglo Reference Catalogue (MRC). We have applied two-dimensional homography to correct for systematic positional errors in the image domain and thereby avoid re-processing the visibility data. Positions of bright (above 15-{\sigma}) point sources, common to MRT catalogue and MRC, are used to set up an over-determined system to solve for the homography matrix. After correction the errors are found to be within 10% of the beamwidth for these bright sources and the systematics are eliminated from the images. This technique will be of relevance to the new generation radio telescopes where, owing to huge data rates, only images after a certain integration would be recorded as opposed to raw visibilities. It is also interesting to note how our investigations cued to possible errors in the array geometry. The analysis of positional errors of sources showed that MRT images are stretched in declination by ~1 part in 1000. This translates to a compression of the baseline scale in the visibility domain. The array geometry was re-estimated using the astrometry principle. The estimates show an error of ~1 mm/m, which results in an error of about half a wavelength at 150 MHz for a 1 km north-south baseline. The estimates also indicate that the east-west arm is inclined by an angle of ~40 arcsec to the true east-west direction.