Pith. sign in

REVIEW 14 references

Mitigating Source Structure in Geodetic VLBI on the Visibility Level

T0 review · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read Source structure in geodetic VLBI can be corrected at the visibility level, immediately after correlation, by subtracting an analytic structure phase, as demonstrated on real VGOS observations.

desk verdict A genuinely new visibility-level structure correction for VGOS, demonstrated on real data, but the validation is partly circular and the quantitative gains are unproven. read the letter →

arxiv 2501.04787 v1 pith:DDHXBKR3 submitted 2025-01-08 astro-ph.IM

classification astro-ph.IM
keywords verylongbaselineinterferometrysourcestructurecorrectionvisibilitydatageodeticVLBIVGOSclosuredelayGaussianbrightnessmodelreferenceframe
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Geodetic and astrometric VLBI still treats most radio-loud active galactic nuclei as point sources, even though interferometers resolve their jets, and source structure is among the dominating error sources at the VGOS accuracy goals of 1 mm station positions. This paper claims that source structure can be corrected at the visibility level, immediately after correlation, by subtracting an analytic structure phase computed from a Gaussian-component brightness model. Tests on real VGOS observations of 3C418 show that the corrected visibilities behave like point-source data: the image collapses to a single component, closure delays become more concentrated, and post-fit geodetic residuals narrow slightly. Because the correction enters before fringe fitting, it cleans not only group delays but also delay rates and differential total electron content estimates, and it works with any downstream processing chain.

What carries the argument

The load-bearing object is the structure phase $$\varphi_s = \frac{2\pi}{\$\lambda$}\mathbf{B}\cdot\mathbf{OP}_0 + \$tan^{{-1}}$\left(-\frac{Z_s}{Z_c}\right),$$ where $Z_s$ and $Z_c$ are the two-dimensional sine and cosine transforms of the source brightness distribution, $\mathbf{B}$ is the baseline, $P_0$ the reference point within the source, and $\lambda$ the wavelength. For a brightness distribution made of Gaussian components these transforms have closed analytic forms, so the phase can be evaluated for every frequency channel and baseline. The correction tool reads the correlator output, subtracts $\varphi_s$ from each visibility phase, leaves the amplitudes unchanged, and writes corrected visibility files that any standard fringe-fitter and geodetic analyser can consume. The mechanism transfers the extended brightness distribution information into the phases and thereby removes the structure-induced delay from the measurement.

What would settle it

Take a well-observed structured source, build its structure model from an image of a different epoch or from an independent imaging database, apply the visibility correction to a geodetic session that was not used to make the model, and check whether closure delays and geodetic post-fit residuals improve. If the residual scatter and closure misclosures are not reduced, or if estimated source positions shift away from the reference-frame positions, the claim that the model removes real structure would be refuted.

Watch

Extended reading notes

Core claim

The central claim is that the impact of resolved source structure on geodetic VLBI is removable at the visibility level, right after correlation. The authors implement the analytic structure-phase formalism for Gaussian brightness components in a software tool that subtracts, for every spectral channel and polarization product, the phase contribution of a model brightness distribution; amplitudes are left unchanged and corrected visibilities are written back in the standard correlator format. Four tests on real VGOS data support the claim: images from corrected visibilities are compact with a frequency-independent core position, deliberate reference-point shifts are recovered one-to-one in estimated source positions, closure delays peak more sharply, and a full geodetic solution gives slightly improved post-fit residuals for the corrected source. The corrected data therefore behave as if they came from point sources, so the whole downstream chain from fringe fitting to geodetic parameter estimation is freed from this systematic effect.

Load-bearing premise

The entire demonstration rests on a Gaussian-component model of 3C418 derived from the very same VGOS session that is later used to judge whether the correction worked; if that model does not faithfully represent the true brightness distribution and its frequency dependence, the measured improvement could be an artifact of the model rather than a real removal of source structure.

Editorial extensions

If this is right

  • Fringe fitting of corrected data produces group delays free of structure phase, and the correction also propagates into delay rates and, for VGOS, differential total electron content estimates.
  • A structure model derived from one image can be applied as a standard correction in the correlator-to-geodesy pipeline, since the corrected files keep the usual format.
  • Reference point shifts in the model produce exactly corresponding shifts in estimated source positions, so the phase model carries positional information reliably.
  • Closure delays of a structured source become more concentrated after correction, which removes baseline-specific systematic scatter that station-specific errors cannot explain.
  • The image of 3C418 after correction is reduced to one component with a frequency-independent core, the signature of a point-like effective source.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because the model for 3C418 is derived from the same session used to validate the correction, the test is not fully independent; an overfitted or frequency-inaccurate model could remove real signal without being exposed. A decisive extension would repeat the correction with a model from a different epoch or from an external image database.
  • Since the method adjusts only phases, amplitude-based source effects such as flux-density evolution of individual jet components remain in the data and would still need separate treatment.
  • If the same correction is applied before fringe fitting in routine VGOS processing, structure models become reusable products: once a source is modelled, every future observation of it can be cleaned without re-imaging.
  • The visibility-level placement means the correction is compatible with any downstream geodetic estimator, and the same phase-subtraction principle should transfer to other visibility formats once the tool is extended beyond correlator-native files.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Circularity Check

2 steps flagged · score 6.0 of 10

Same-session validation of the structure correction reduces to subtracting the fitted Gaussian model from the data that produced it; the only cross-session check (vo3012 residuals) lacks significance testing.

  1. self definitional [Section 3.1, Fig. 2 (model derivation and corrected image); Section 2 (VieSOFT pipeline)]
    "We derived a model of frequency-dependent Gaussian components from this image. The image shown in the right panel is derived from the corrected visibilities. In this image the structure is greatly removed and reduced to only one component, and also the core position is the same for all frequencies."

    The Gaussian component model is fitted to the image of 3C418 from session vo2187, and VieSOFT then subtracts exactly those model structure phases from the visibilities of the same session (Section 2: 'subtracts phi_s from the original structure phases'). The right-hand panel is therefore the residual of the fitted model, not an independent reconstruction: a model that reproduces the input image will, by construction, leave a more compact residual image. No independent epoch or different session is used to validate the model, so the compactness of Fig. 2(b) is a consistency check of the subtraction rather than a test of the correction's physical fidelity.

  2. fitted input called prediction [Section 3.3, Fig. 5]
    "We investigated the closure delays resulting from observations of the source 3C418 in the VGOS session vo2187. ... Comparing the two panels in Fig. 5, the distribution of the corrected data has a higher and narrower peak, and many of the misclosures are removed."

    These closure delays are from vo2187, the same session from which the Gaussian model was derived in Sect. 3.1. Source structure contributes to baseline-dependent group delays and hence to closure delays; subtracting the fitted model's structure phases from the same data removes the modeled contribution by construction in the in-sample sense. The observed narrowing is therefore a consistency check that the model phases were applied, not an out-of-sample prediction of the model's correctness. The paper reports no uncertainties, chi-squared values, or p-values for the improvement and explicitly acknowledges that not all misclosures are removed.

full rationale

The paper's phase formula from Charlot (1990), implemented in VieSOFT (Section 2), is not circular. The circularity lies in the validation path: in Section 3.1 the Gaussian brightness model is fitted to the 3C418 image from session vo2187, and the same model is then subtracted from the vo2187 visibilities; the resulting 'corrected' image is therefore the residual image of the fitted model, so its compactness (Fig. 2b) cannot serve as an independent test. Likewise, Section 3.3 applies the same correction to the same session, so the narrowing of the closure-delay distribution is an in-sample consistency check rather than a prediction. The reference-point test (Section 3.2) is a self-consistency check of the phase formula and is not circular. The geodetic post-fit residuals (Section 3.4) do use a different session (vo3012) and thus provide some independent evidence, but only 3C418 residuals are shown, with fitted Gaussians quoted without uncertainties, chi-squared values, or p-values, so the improvement is not statistically quantified. The paper's own wording is cautious ('post-fit residuals seem slightly improved'), and the code is not yet public, which limits independent reproduction but is not itself circularity. The citation to Pérez-Díez et al. (2024) supplies the original image and imaging pipeline with overlapping authors, but the model is taken from the image data rather than from the authority of the citation, so it is not a load-bearing self-citation chain. Overall, one core validation path (image compactness and closure-delay narrowing) reduces by construction, while the cross-session geodetic check retains independent content, giving partial circularity.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The central claim rests on the accuracy of a Gaussian source model fitted to the same data that the correction is applied to, and on the published structure phase formalism. No new physical entities are introduced.

free parameters (1)
  • Gaussian component model for 3C418 = not specified (derived from image of session vo2187)
    The brightness distribution model parameters (positions, amplitudes, sizes, frequency dependence) are fitted to the image of 3C418 from the same session and used to compute the structure phases that are subtracted. These parameters are not reported in the paper.
assumptions (3)
  • standard math Charlot (1990) structure phase formalism (Eq. 1) correctly describes the phase contribution of a resolved source.
    The paper builds entirely on this published formalism, citing Charlot (1990); it does not re-derive or test it.
  • domain assumption The source brightness distribution is adequately represented by a sum of frequency-dependent Gaussian components.
    The correction quality depends on the model being a good representation of the true source structure; the paper derives the model from the same image, so the assumption is not independently tested.
  • domain assumption Subtracting the structure phase while leaving amplitudes unchanged is sufficient to correct group delays.
    The paper states amplitudes are left unchanged; it is assumed that amplitude structure does not affect the group delay observables used in geodesy.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Mitigating Source Structure in Geodetic VLBI on the Visibility Level." pith.science (2026). https://pith.science/paper/DDHXBKR3

@misc{pith2026250104787,
  author       = {Pith},
  title        = {Pith review of: Mitigating Source Structure in Geodetic VLBI on the Visibility Level},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DDHXBKR3}},
  note         = {Machine review of arXiv:2501.04787}
}
read the original abstract

Geodetic and astrometric VLBI has entered a new era with the implementation of the VLBI Global Observing System (VGOS). These broadband and dual linear polarization observations aim at an accuracy of station coordinates of 1 mm and a reference frame stability of 0.1 mm/year. Although the extended brightness distribution of many of the radio-loud active galactic nuclei observed during geodetic VLBI sessions is resolved by the interferometer, the established processing chain still treats these objects as point sources. We investigate the impact of source structure on the visibility level and develop tools to remove the structure from the visibility data, right after correlation. Here we present our approach and show results obtained from observational VGOS data.

Figures

Figures reproduced from arXiv: 2501.04787 by the authors.

Figure 2
Figure 2. (a) Original image of 3C418, obtained from an￾alyzing VGOS session vo2187. This figure was first pub￾lished in P´erez-D´ıez et al. (2024). (b) Image of the same source after removing the structure phases from the ob￾servational data of the same session. 3.1. Imaging [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 5
Figure 5. Histogram of the triangular closure delays from observations of 3C418 in the VGOS session vo2187. Left: Original data without any correction. Right: Structure phases removed. where all of the group delays τ entering this sum have to refer to the same wavefront. All station-specific error sources cancel out in this equation, so in the absence of any baseline-specific systematics, the closure delays should be distribu… view at source ↗
Figure 3
Figure 3. VLBI image of 1156+295 at X-band. -6 -4 -2 0 2 4 6 -6 -4 -2 0 2 4 6 y [mas] x [mas] Configured Estimated y [mas] x [mas] -15 -10 -5 0 5 10 15 -15 -10 -5 0 5 10 15 [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Configured reference points (purple open circles) lead to estimated source positions (green filled circles) for 1156+295, as indicated by the arrows. The estimated source positions are relative to the ICRF3 S/X position of 1156+295. In the left panel we changed the ref…
Figure 6
Figure 6. Figure 6: In the left-hand panel of each of the sub-figures (a) [PITH_FULL_IMAGE:figures/full_fig_p003_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

14 extracted references · 6 canonical work pages

  1. [1]

    doi:10.1007/s00190-023-01738-w

    Altamimi Z., Rebischung P., Collilieux X., M \'e tivier L., Chanard K., 2023, JGeod, 97, 47. doi:10.1007/s00190-023-01738-w

  2. [2]

    M., Xu M

    Anderson J. M., Xu M. H., 2018, JGRB, 123, 10, 162, 190. doi:10.1029/2018JB015550

  3. [3]

    o hm J., B \

    B \"o hm J., B \"o hm S., Boisits J., Girdiuk A., Gruber J., Hellerschmied A., Kr \'a sn \'a H., et al., 2018, PASP, 130, 044503. doi:10.1088/1538-3873/aaa22b

  4. [4]

    doi:10.1086/115419

    Charlot P., 1990, AJ, 99, 1309. doi:10.1086/115419

  5. [5]

    S., Gordon D., Lambert S., de Witt A., B \"o hm J., Fey A

    Charlot P., Jacobs C. S., Gordon D., Lambert S., de Witt A., B \"o hm J., Fey A. L., et al., 2020, A&A, 644, A159. doi:10.1051/0004-6361/202038368

  6. [6]

    T., Brisken W

    Deller A. T., Brisken W. F., Phillips C. J., Morgan J., Alef W., Cappallo R., Middelberg E., et al., 2011, PASP, 123, 275. doi:10.1086/658907

  7. [7]

    doi:10.1088/1538-3873/abeca4

    Gruber J., Nothnagel A., B \"o hm J., 2021, PASP, 133, 044503. doi:10.1088/1538-3873/abeca4

  8. [8]

    K., Mart \' -Vidal I., et al., 2021, Proceedings of the 25th European VLBI Group for Geodesy and Astrometry Working Meeting, 14-18 March 2021 Cyberspace, Gothenburg, Sweden, Ed

    Jaron F., Bernhart S., B \"o hm J., Gonz \'a lez Garc \' a J., Gruber J., Choi Y. K., Mart \' -Vidal I., et al., 2021, Proceedings of the 25th European VLBI Group for Geodesy and Astrometry Working Meeting, 14-18 March 2021 Cyberspace, Gothenburg, Sweden, Ed. R. Haas, ISBN: 978-91-88041-41-8, pp. 19-23

Show all 14 references
  1. [9]

    doi:10.1029/2018RS006617

    Niell A., Barrett J., Burns A., Cappallo R., Corey B., Derome M., Eckert C., et al., 2018, RaSc, 53, 1269. doi:10.1029/2018RS006617

  2. [10]

    doi:10.1007/s00190-016-0950-5

    Nothnagel A., Artz T., Behrend D., Malkin Z., 2017, JGeod, 91, 711. doi:10.1007/s00190-016-0950-5

  3. [11]

    H., et al., 2024, A&A, 688, A151

    P \'e rez-D \' ez V., Mart \' -Vidal I., Albentosa-Ruiz E., Gonz \'a lez-Garc \' a J., Jaron F., Savolainen T., Xu M. H., et al., 2024, A&A, 688, A151. doi:10.1051/0004-6361/202348633

  4. [12]

    Petrachenko B., Niell A., Behrend D., Corey B., Boehm J., Charlot P., Collioud A., et al., 2009, NASA/TM-2009-214180

  5. [13]

    H., Soja B., 2023, JGeod, 97, 17

    Schartner M., Collioud A., Charlot P., Xu M. H., Soja B., 2023, JGeod, 97, 17. doi:10.1007/s00190-023-01706-4

  6. [14]

    J., Fanselow J

    Sovers O. J., Fanselow J. L., Jacobs C. S., 1998, RvMP, 70, 1393. doi:10.1103/RevModPhys.70.1393

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.