ASKAP software engineers solve telescope commissioning measurement mystery


CSIRO software engineers have solved a mystery that astronomers have been puzzling over for more than a year - we now understand why ASKAP has been measuring incorrect positions for the radio sources detected in our images.

This position offset problem was originally discovered in images made using the BETA prototype array, but much to our dismay it did not vanish with the deployment of updated hardware in 2017. As more antennas are brought online, the sensitivity of the telescope improves and we have been able to study this particular problem in more detail. One of the standard tests that we have developed as a measure of image quality is a comparison of detected source positions with respect to existing catalogues made by other observatories. Our science teams reported that this comparison often shows a shift of all the sources in an image by a small amount, indicating that ASKAP did not agree with well-established results. The shift is not always the same, but tends to be similar for each observation of the same area of the sky.

Many different theories were advanced to explain the position offsets, so to narrow the possibilities we went back to basics and started to look closely at the data we use to calibrate the array. Upon looking for subtle variations with time, we could see clear trends during a long observation of a bright source, indicating an error or omission in the system used to align the signals from each antenna in the array. Although clear evidence of a problem, this particular kind of error could be caused by atmospheric or ionospheric effects external to the telescope, or an internal problem with the software or hardware. The systematic nature of the observed trend strongly suggested an internal problem, so our engineers started to dig deeper into the system. Eventually, we found that the software responsible for predicting the precise location of each antenna was not correctly accounting for the subtle motion of the Earth's rotational pole.

The diagrams below show the difference between the position of all the sources detected in one of our ASKAP test images (note how many different sources are visible in a single image!) and the position of the same source as recorded in a catalogue produced by the NRAO VLA telescope. Prior to the fix, the sources cluster around a position offset from the centre. After the fix, the distribution is now neatly in the middle.

Although much improved, there is still some evidence for smaller offsets in recent observations and we are continuing to investigate the situation. Tests like this help to ensure that the telescope will produce high quality, accurate data for our international science teams when it embarks on its major survey campaigns.