The third International Celestial Reference Frame (ICRF3) is realized by very long baseline interferometry (VLBI) positions of thousands of compact extragalactic radio sources, each catalog position being a weighted mean over up to four decades of geodetic observations. Cigan et al. (2024) built single-session position time series for over 5550 sources and showed that the residual distributions are explicitly non-Gaussian — heavy-tailed Gamma/Fréchet (extreme-value) families whose negative location parameters produce a 'zone of avoidance' around mean positions — and that several variability metrics trend with declination, implicating atmospheric path errors and northern-hemisphere-dominated network geometry. The problem, formulated from that paper's results and its future-work discussion, is the linked research agenda these diagnostics motivate: (1) a quantitative, predictive model explaining the observed non-Gaussian residual-distribution behavior — heavy tails and the deficit of measurements near mean positions — from observing geometry, frequency-dependent source structure (core shift, jets), and time-variable atmosphere; (2) circularity-free incorporation of per-source empirical corrections into global VLBI solutions; (3) catalog-wide quantification of astrometric-variability timescales; (4) empirically validated selection, weighting, and scheduling rules for the next frame (ICRF4), including photometric-variability predictors; and (5) an observing-network design assessment for southern-sky improvement. Follow-up work through 2026 has partially established jet-structure alignment of position excursions and validated optical photometric variability as a stability predictor, but the predictive statistical model, solver incorporation, catalog-wide timescales, and network design remain open.

Background

Very long baseline interferometry (VLBI) is a radio technique in which widely separated telescopes simultaneously record the signals of compact extragalactic radio sources — nearly all of them active galactic nuclei (AGN), the luminous cores of galaxies powered by accreting supermassive black holes with relativistic jets. Cross-correlating the recordings yields the group delay, the frequency-averaged difference in signal arrival time between stations, from which global least-squares solutions determine the source's angular position on the sky together with station positions and Earth-orientation parameters. The International Celestial Reference Frame (ICRF) — the catalog of VLBI positions of several thousand such sources that materializes the celestial coordinate system for astronomy, geodesy, and deep-space navigation — is currently in its third realization, ICRF3 (Charlot et al. 2020), with 4536 sources measured at 2.3/8.4 GHz (the S/X bands). Catalog positions are weighted means over observing histories of up to about 40 years, an estimation that implicitly assumes single-epoch position residuals are Gaussian with correctly known formal covariances.

The sources that realize the frame are not fixed points. Their radio emission comes from parsec-scale jets whose apparent centroid position depends on observing frequency — the 'core shift': the synchrotron-self-absorbed core appears farther from the central engine at lower frequency, displacing apparent positions at the 0.1–1 milliarcsecond (mas) level — and moving jet features or opacity changes shift the apparent position with time, with apparent motions of tens of microarcseconds (µas) expected over monitoring timescales (Titov et al. 2023 catalogued such apparent motions). Optical–radio position offsets are known to correlate with jet direction. On the instrumental side, unmodeled ionospheric and tropospheric propagation delays, combined with the northern-hemisphere-dominated station networks of geodetic VLBI, imprint declination-dependent systematic errors on the measured positions.

Cigan et al. (2024) assembled per-source time series of single-session positions for over 5550 radio sources (including all of ICRF3-S/X) from 6581 geodetic VLBI sessions spanning more than four decades, and quantified astrometric variability with complementary metrics: the covariance-weighted mean offset from the catalog position ('jitter'), the excess variance (unexplained variance beyond the formal errors), robust dispersion measures such as the 68th percentile of normalized offsets (Q68) and the weighted rms (wrms), and the range of a physically motivated smoothed time series (STR). Two findings define the problem. First, the residual distributions are explicitly non-Gaussian: for the 86 best-observed sources (more than 1000 single-epoch measurements each), the distributions of normalized offsets are best fit mostly by Gamma or Fréchet extreme-value families (occasionally Lognormal or Cauchy), and the fitted Fréchet laws have negative location parameters — a 'zone of avoidance', i.e. a local deficit of measurements near the mean position — in direct tension with the Gaussian assumption underlying weighted-mean catalog positions. Second, several metrics (formal errors, jitter, wrms) trend with declination with a cusp near −40°, the approximate southern observing limit of the northern stations, implicating atmospheric path and network-geometry systematics rather than astrophysics.

The tools needed to proceed exist only in fragments. The Allan variance — a two-sample variance from frequency-stability analysis that characterizes noise regimes (white, flicker (1/f), random-walk) as a function of averaging timescale, and whose astrometric and geodetic applications are reviewed by Malkin (2016) — was applied to celestial-reference-frame source positions by Gattano, Lambert & Le Bail (2018). Robust 1-norm periodograms, designed for heavy-tailed, irregularly sampled time series, were applied by Makarov et al. (2024) to 259 well-observed ICRF3 quasars, yielding 49 quasi-periodic position-change candidates above 3σ. On the predictor side, Secrest (2022) and Lambert & Secrest (2024) established that optical–radio position offsets and VLBI position dispersion are inversely correlated with optical photometric variability, suggesting that frame sources could be selected and weighted by their photometric behavior. What does not yet exist is the connective tissue: a physical explanation of the observed residual-distribution families, and a validated route from these diagnostics to the construction of the next-generation frame.

Problem Statement

The problem, formulated from Cigan et al. (2024), is one linked research objective: to convert the diagnostics of astrometric position variability of ICRF3 sources into a physically explained and operationally usable treatment for constructing the next-generation celestial reference frame (ICRF4). For the ~5550-source VLBI position time series assembled in that work (6581 geodetic VLBI sessions spanning more than four decades, S/X band, i.e. 2.3 and 8.4 GHz; the per-epoch position series and their covariances are drawn from the source's quarterly global solution products, publicly downloadable from the USNO quarterly VLBI solution archive at https://crf.usno.navy.mil/quarterly-vlbi-solution, and the catalog positions from Charlot et al. 2020), this requires:

  1. Physical and instrumental modeling. Construct a quantitative, predictive model that explains the empirically observed non-Gaussian behavior of the position residuals — the heavy-tailed distributions and the local deficit of measurements near the mean position (the 'avoidance zone' captured by the negative location parameters of the fitted Fréchet laws) — in terms of observing geometry (northern-hemisphere-dominated station networks, baseline coverage), frequency-dependent source structure (core shift, jet orientation and evolution), and time-variable ionospheric/tropospheric conditions, together with the declination dependence of the variability metrics, including the cusp near −40°.

  2. Global-solution incorporation. Develop and validate procedures that incorporate per-source empirical corrections — excess variances, secular-trend models from smoothed time series, and structure-based position corrections — into global VLBI solution software in a way that avoids circularity (not using time-series diagnostics derived from the same data as a priori constraints) and preserves solvability (enough degrees of freedom to solve for all necessary parameters).

  3. Catalog-wide timescales. Quantify the characteristic timescales and periodicities of astrometric variability across the catalog.

  4. Frame-construction strategy. Determine and empirically validate the optimal strategy for translating these time-series diagnostics into concrete selection, weighting, and scheduling rules for future reference frames such as ICRF4, including how the inverse correlation already established for optical photometric variability, and radio structural information, translate into measurable gains in frame stability.

  5. Network design. Assess how the observing-network design — enhanced southern-station coverage and longer north–south baselines — would improve southern-sky performance, the uniformity of position errors across the sky, and the determination of the declination component.

These items are complementary facets of the same objective: item 1 explains the phenomenon the diagnostics measure; item 2 turns the explanation into usable corrections inside the estimation software; item 3 characterizes the temporal structure that any correction and scheduling rule must respect; item 4 converts all of it into frame-construction rules; and item 5 supplies the observing-network complement required by the same objective. Resolution requires a quantitative, predictive model that reproduces the observed non-Gaussian residual-distribution behavior and the declination dependence and states testable predictions (item 1), together with at least one circularity-free route from the diagnostics to frame construction validated against the time series (item 2 or item 4). Items 3 and 5 are complementary investigations, each assessed independently on whether it delivers the quantitative product it names; a descriptive restatement of the diagnostics resolves no item.

References

  1. Cigan, P., Makarov, V. V., Secrest, N. J., Gordon, D., Johnson, M. C., & Lambert, S. 2024, 'Metrics of Astrometric Variability in the International Celestial Reference Frame. I. Statistical Analysis and Selection of the Most Variable Sources', The Astrophysical Journal Supplement Series 274, 28. DOI 10.3847/1538-4365/ad6772 (arXiv:2408.01373). https://doi.org/10.3847/1538-4365/ad6772
  2. Charlot, P., Jacobs, C. S., Gordon, D., Lambert, S., de Witt, A., Böhm, J., Fey, A. L., Heinkelmann, R., Skurikhina, E., Titov, O., Arias, E. F., Bolotin, S., Bourda, G., Ma, C., Malkin, Z., Nothnagel, A., Mayer, D., MacMillan, D. S., Nilsson, T., & Gaume, R. 2020, 'The third realization of the International Celestial Reference Frame by very long baseline interferometry', Astronomy & Astrophysics 644, A159. DOI 10.1051/0004-6361/202038368. https://doi.org/10.1051/0004-6361/202038368
  3. Makarov, V. V., Cigan, P., Gordon, D., Johnson, M. C., DiLullo, C., & Lambert, S. 2024, 'Partial Alignment of Astrometric Position Excursions of International Celestial Reference Frame Quasars with Radio Jet Structures', The Astrophysical Journal Letters 977, L14. DOI 10.3847/2041-8213/ad94f1. https://doi.org/10.3847/2041-8213/ad94f1
  4. Makarov, V. V., Lambert, S., Cigan, P., DiLullo, C., & Gordon, D. 2024, 'Robust 1-norm Periodograms for Analysis of Noisy Non-Gaussian Time Series with Irregular Cadences: Application to VLBI Astrometry of Quasars', Publications of the Astronomical Society of the Pacific 136, 054503. DOI 10.1088/1538-3873/ad4b9f. https://doi.org/10.1088/1538-3873/ad4b9f
  5. Makarov, V. V. 2024, 'Optical Variability of Gaia CRF3 Sources with Robust Statistics and the 5000 Most Variable Quasars', The Astrophysical Journal Supplement Series 274, 27. DOI 10.3847/1538-4365/ad697b. https://doi.org/10.3847/1538-4365/ad697b
  6. Lambert, S., & Secrest, N. J. 2024, 'VLBI position variability of AGNs is inversely correlated with their photometric variability', Astronomy & Astrophysics 684, A93. DOI 10.1051/0004-6361/202348842. https://doi.org/10.1051/0004-6361/202348842
  7. Secrest, N. J. 2022, 'Optical–Radio Position Offsets Are Inversely Correlated with AGN Photometric Variability', The Astrophysical Journal Letters 939, L32. DOI 10.3847/2041-8213/ac8d5d. https://doi.org/10.3847/2041-8213/ac8d5d
  8. Secrest, N. J., & Lambert, S. 2026, 'Blazars define a stable celestial reference frame', Astronomy & Astrophysics 706, A342. DOI 10.1051/0004-6361/202556690. https://doi.org/10.1051/0004-6361/202556690
  9. Gattano, C., Lambert, S. B., & Le Bail, K. 2018, 'Extragalactic radio source stability and VLBI celestial reference frame: insights from the Allan standard deviation', Astronomy & Astrophysics 618, A80. DOI 10.1051/0004-6361/201833430. https://doi.org/10.1051/0004-6361/201833430
  10. Xu, M. H., & Charlot, P. 2025, 'Variations of Absolute Source Positions Determined from Quad-band VLBI Observations', The Astronomical Journal 169, 173. DOI 10.3847/1538-3881/adb133. https://doi.org/10.3847/1538-3881/adb133
  11. Gordon, D., de Witt, A., Jacobs, C. S., Krásná, H., Bertarini, A., McCallum, J., Quick, J., García-Miró, C., Jung, T., Hodgson, J. A., Cheong, W. Y., Lee, S.-S., & Byun, D.-Y. 2026, 'The Celestial Reference Frame at K Band: The CRF-K-2025 Catalog', The Astronomical Journal 172, 38. DOI 10.3847/1538-3881/ae6efe. https://doi.org/10.3847/1538-3881/ae6efe
  12. Osetrova, A. A., Titov, O. A., & Melnikov, A. E. 2024, 'Extragalactic Radio Sources with Large Changes in Their Apparent Coordinates', Astronomy Letters 50, 657–670. DOI 10.1134/S1063773725700045. https://doi.org/10.1134/S1063773725700045
  13. Petrov, L. Y., & Kovalev, Y. Y. 2025, 'The Radio Fundamental Catalog. I. Astrometry', The Astrophysical Journal Supplement Series 276, 38. DOI 10.3847/1538-4365/ad8c36. https://doi.org/10.3847/1538-4365/ad8c36
  14. Hodgson, J. A., Krásná, H., de Witt, A., van Zyl, P., & Valverde, J. 2026, 'The Astro2Geo Project I. Radio astrometric offsets correlated with Gamma-ray brightness', arXiv:2605.19507. https://arxiv.org/abs/2605.19507
  15. Titov, O., Frey, S., Melnikov, A., Shu, F., Xia, B., González, J., Tercero, B., Gurvits, L., de Witt, A., McCallum, J., Kharinov, M., Zimovsky, V., & Krezinger, M. 2023, 'Astrometric Apparent Motion of High-redshift Radio Sources', The Astronomical Journal 165, 69. DOI 10.3847/1538-3881/aca964. https://doi.org/10.3847/1538-3881/aca964
  16. Malkin, Z. 2016, 'Application of the Allan Variance to Time Series Analysis in Astrometry and Geodesy: A Review', arXiv:1607.04712. https://arxiv.org/abs/1607.04712
  17. Makarov, V. V., Cigan, P., & Corcoran, K. A. 2026, 'The Peculiar Velocity of Messier 87 from Microarcsecond Geodetic Very Long Baseline Interferometry Astrometry', The Astrophysical Journal 1003, 209. DOI 10.3847/1538-4357/ae62f4. https://doi.org/10.3847/1538-4357/ae62f4
  18. McCarthy, T., McCallum, L., & Schartner, M. 2025, 'Investigating a continental Australian VGOS VLBI intensive for UT1-UTC determination', Earth, Planets and Space 77, 58. DOI 10.1186/s40623-025-02182-0. https://doi.org/10.1186/s40623-025-02182-0
  19. Laha, A., Schartner, M., Böhm, S., Krásná, H., Soja, B., Böhm, J., Balasubramanian, N., & Dikshit, O. 2025, 'Optimal Placement of VGOS Telescope in India: Simulation Insights for 24-Hour and 1-Hour VLBI Sessions', EGU General Assembly 2025 abstract. DOI 10.5194/egusphere-egu25-285. https://doi.org/10.5194/egusphere-egu25-285