Hostname: page-component-7b7567c8c-2n9c9 Total loading time: 0 Render date: 2026-11-17T09:36:42.010Z Has data issue: false hasContentIssue false

Enhanced astrometry of the rapid ASKAP continuum survey: Mid and high-frequency epochs

Published online by Cambridge University Press:  24 September 2026

Akhil Jaini*
Affiliation:
Centre for Astrophysics and Supercomputing, Swinburne University of Technology, Hawthorn, VIC, Australia
Adam T. Deller
Affiliation:
Centre for Astrophysics and Supercomputing, Swinburne University of Technology, Hawthorn, VIC, AustraliaARC Centre of Excellence for Gravitational Wave Discovery (OzGrav), Hawthorn, VIC, Australia
Yuanming Wang
Affiliation:
Centre for Astrophysics and Supercomputing, Swinburne University of Technology, Hawthorn, VIC, AustraliaARC Centre of Excellence for Gravitational Wave Discovery (OzGrav), Hawthorn, VIC, Australia
Emil Lenc
Affiliation:
CSIRO Space and Astronomy, Epping, NSW, Australia
Stefan W. Duchesne
Affiliation:
CSIRO Space and Astronomy, Bentley, WA, Australia
Marcin Glowacki
Affiliation:
Institute of Astronomy, The University of Edinburgh, Royal Observatory, Edinburgh, UKInter-University Institute for Data Intensive Astronomy, Department of Astronomy, University of Cape Town, Cape Town, South Africa
*
Corresponding author: Akhil Jaini; Email: ajaini@swin.edu.au
Rights & Permissions [Opens in a new window]

Abstract

Accurate radio astrometry is essential for reliable cross-identification of sources across wavelengths, precision localisation of transient events, and the construction of stable all-sky reference catalogues. In this work, we extend our astrometric correction framework for the Rapid Australian Square Kilometre Array Pathfinder (ASKAP) Continuum Survey (RACS) to its mid- and high-frequency epochs (RACS-Mid1 and RACS-High1), building on our previous corrections to the low-frequency surveys. Using a hierarchical cross-matching strategy with high-precision external catalogues, we remove large-scale systematic positional errors that were present in the uncorrected data and significantly reduce the residual scatter across the sky. After correction, the median positional offsets are effectively eliminated, and the 68% confidence interval of the mean residuals of RACS source positions averaged over $\sim$1 sq. deg. regions is reduced from $\gtrsim$$0.4''$ to $\lesssim$$0.18''$ over most of the survey area for both epochs. Independent validation against multiple external radio astrometric references confirms that individual corrected RACS source positions are accurate to a 1-$\sigma$ confidence level of $\sim$$0.25''$ across the bulk of the sky, with slightly degraded performance within the Galactic plane. While motivated primarily by the need for improved localisation of ASKAP fast radio bursts, these corrections also benefit a wide range of science applications, including transient identification, multiwavelength host association, and studies of Galactic and extragalactic radio populations. Together with our previous work, this establishes RACS as the highest precision arcsecond-resolution all-sky astrometric reference in the southern hemisphere at decimetre wavelengths.

Information

Type
Research Article
Creative Commons
Creative Common License - CCCreative Common License - BY
This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution and reproduction, provided the original article is properly cited.
Copyright
© The Author(s), 2026. Published by Cambridge University Press on behalf of Astronomical Society of Australia
Figure 0

Table 1.The residual offsets for RACS-Mid1 vs WISE for scans below Dec. +30∘$+30^\circ$.Table 1 long description.

Figure 1

Figure 1.The modelled and residual offsets of RACS-Mid1 vs WISE after two stages of crossmatching are shown here. The top row shows the modelled offsets and the bottom row shows the residual offsets in RA (left) and Dec. (right) for the entire sky coverage.

Figure 2

Figure 2.The modelled (top row) and residual (bottom row) offsets for RACS-Mid1 vs WISE for scans below Dec. +30∘$+30^\circ$. The median RA offset (left) improves from 0.21″$0.21''$ to 0.00″$0.00''$ and the median Dec. offset (right) decreases from −0.07″$-0.07''$ to 0.00″$0.00''$. The 68% confidence intervals also narrow down significantly.

Figure 3

Figure 3.The corrected offsets in RA and Dec. for Dec. < +30∘$+30^\circ$ for RACS-Mid1 vs reference catalogues, with FIRST comparisons in the top row and RFC in the bottom row.

Figure 4

Table 2.The residual offsets for RACS-High1 vs WISE for scans below Dec. +30∘$+30^\circ$.Table 2 long description.

Figure 5

Figure 4.The modelled and residual offsets of RACS-High1 vs WISE after two stages of crossmatching are shown here. The top row shows the modelled offsets and the bottom row shows the residual offsets in RA (left) and Dec. (right) for the entire sky coverage.

Figure 6

Table 3.Corrected RACS-Mid1 comparison with FIRST and RFC catalogues for regions below Dec. +30∘$+30^\circ$.Table 3 long description.

Figure 7

Figure 5.The modelled (top row) and residual (bottom row) offsets for RACS-High1 vs WISE for scans below Dec. +30∘$+30^\circ$. The median RA offset (left) improves from 0.16″$0.16''$ to 0.00″$0.00''$ and the median Dec. offset (right) decreases from −0.04″$-0.04''$ to 0.00″$0.00''$. The 68% confidence intervals also narrow down significantly.

Figure 8

Table 4.Corrected RACS-High1 comparison with FIRST and RFC catalogues for regions below Dec. +30∘$+30^\circ$.Table 4 long description.

Figure 9

Figure 6.The corrected offsets in RA and Dec. for Dec. < +30∘$+30^\circ$ for RACS-High1 vs reference catalogues, with FIRST comparisons in the top row and RFC in the bottom row.

Figure 10

Figure 7.The corrected RACS-High1 source lists compared to the corrected RACS-Mid1 source lists in RA and Dec. The results show that, post corrections, both the source lists are very comparable to each other.

Figure 11

Figure 8.RACS-Mid1 astrometric offsets in Dec., for ∼$\sim$6000$6\,000$ individual RFC sources, plotted as a function of Dec. The top panel shows the uncorrected offsets having a mean of 0.09″$0.09''$ and 68% confidence intervals between [+0.79″,−0.59″]$[+0.79'', -0.59'']$, exhibiting systematic declination-dependent errors. The panel also shows a 2nd-order polynomial (in green) fitting the scatter, as described in Duchesne et al. (2024). The middle panel shows the offsets after subtracting the polynomial to have a mean of 0.06″$0.06''$ and 68% confidence intervals of [+0.61″,−0.50″]$[+0.61'', -0.50'']$, significantly reducing the largest outliers, but still exhibiting considerable scatter. The bottom panel shows the offsets after applying our corrections to have a mean of 0.04″$0.04''$ and 68% confidence intervals ranging between [+0.25″,−0.16″]$[+0.25'', -0.16'']$, illustrating the both the removal of the systematic trend and a much greater reduction in scatter. The reader should note that these values are computed over the full declination range of RACS-Mid1, whereas the results presented in Table 3 are restricted to declinations below +30∘$+30^\circ$. As a result, the quoted values differ slightly. The right-hand panels show the collapsed histograms of the offsets.

Figure 12

Figure 9.RACS-High1 astrometric offsets in Dec., for ∼$\sim$6000$6\,000$ individual RFC sources, plotted as a function of Dec., following the same structure as Figure 8. The uncorrected data (top) shows significant scatter having a mean of 0.13″$0.13''$ and 68% confidence intervals ranging between [+0.67″,−0.49″]$+0.67'', -0.49'']$, along with a 4th-order polynomial (in green) fitting the scatter, as described in Duchesne et al. (2025). The polynomial corrected plot (middle) shows slightly reduced scatter with a mean of 0.01″$0.01''$ and 68% confidence intervals of [+0.53″,−0.53″]$[+0.53'', -0.53'']$. The offsets are markedly reduced in the data corrected using our methodology (bottom) to a mean of 0.03″$0.03''$ and 68% confidence intervals between [+0.20″,−0.16″]$[+0.20'', -0.16'']$. The reader should note that these values are computed over the full declination range of RACS-High1, whereas the results presented in Table 4 are restricted to declinations below +30∘$+30^\circ$. As a result, the quoted values differ slightly. The right-hand panels show the collapsed histograms of the offsets.

Figure 13

Figure 10.The ensemble mean localisation offsets for 1 000 simulated FRBs randomly distributed across the ASKAP sky (−80∘$-80^\circ \lt \mathrm{Dec.} \lt +25^\circ$). The offsets obtained using the corrected RACS-Mid1 (top) and RACS-High1 (bottom) source lists are shown relative to those derived using the corrected RACS-Low1 catalogue. The left column shows the differences in RA, while the right column shows the corresponding differences in Dec.

Figure 14

Figure 11.An example of determining the positional offsets of the mid-band FRB20220918A while using the corrected RACS-Low1 catalogues (left) and the corrected RACS-Mid1 source lists (middle) with CELEBI. After updating the RACS epoch, the estimated offset correction based on the field sources changed from 0.05″±0.23″$0.05'' \pm 0.23''$, −0.17″±0.16″$-0.17'' \pm 0.16''$ to −0.19″±0.15″$-0.19'' \pm 0.15''$, −0.45″±0.11″$-0.45'' \pm 0.11''$ in RA and Dec. respectively. After including an assumed 1-σ$\sigma$ systematic uncertainty of 0.3″$0.3''$ (as the source is outside the Galactic plane) and the statistical FRB position uncertainty, the final position of the FRB is updated from 1h10m22s.10±0.35″$1^{\rm h}10^{\rm m}22^{\rm s}.10 \pm 0.35''$ to 1h10m22s.06±0.35″$1^{\rm h}10^{\rm m}22^{\rm s}.06 \pm 0.35''$ in RA and from −70∘48′40″.94±0.37″$-70^\circ48'40''.94 \pm 0.37''$ to −70∘48′41″.21±0.37″$-70^\circ48'41''.21 \pm 0.37''$ in Dec. The effect of this change on the FRB’s localisation within the host galaxy is shown in the VLT R-band image of the host (right), where the black error ellipse indicates the localisation with the corrected RACS-Low1 catalogues (Shannon et al. 2025), and the green ellipse shows the localisation achieved with the corrected RACS-Mid1 source lists.

Figure 15

Figure 12.An example of determining the positional offsets of high-band FRB20211212A while using the corrected RACS-Low1 catalogues (left) and the corrected RACS-High1 source lists (middle) with CELEBI. After updating the RACS epoch, the estimated offset correction based on the field sources changed from −0.06″±0.20″$-0.06'' \pm 0.20''$, 0.44″±0.11″$0.44'' \pm 0.11''$ to 0.63″±0.15″$0.63'' \pm 0.15''$, −0.28″±0.22″$-0.28'' \pm 0.22''$ in RA and Dec. respectively. A large part of the shift in the offsets can be contributed to the additional source that was crossmatched while using the RACS-High1 source lists. This could be because our heuristic is not as refined with the mid- and high-band source lists when compared to the low-band heuristic, due to the lack of mosaicked source catalogues in addition to the Gaussian components represented by the source lists. After including an assumed 1-σ$\sigma$ systematic uncertainty of 0.3″$0.3''$ (as the source is well outside the Galactic plane) and the statistical FRB position uncertainty, the final position of the FRB is updated from 10h29m24s.19±0.40″$10^{\rm h}29^{\rm m}24^{\rm s}.19 \pm 0.40''$ to 10h29m24s.24±0.42″$10^{\rm h}29^{\rm m}24^{\rm s}.24 \pm 0.42''$ in RA and from +01∘21′37″.55±0.40″$+01^\circ21'37''.55 \pm 0.40''$ to +01∘21′37″.71±0.42″$+01^\circ21'37''.71 \pm 0.42''$ in Dec. The effect of this change on the FRB’s localisation within the host galaxy is shown in the VLT g-band image of the host (right), where the black error ellipse indicates the localisation with the corrected RACS-Low1 catalogues (Shannon et al. 2025), and the green ellipse shows the localisation achieved with the corrected RACS-High1 source lists.