Pith. sign in

REVIEW 4 major objections 6 minor 16 references

The International Lunar Reference System

T0 review · 4 major / 6 minor · reviewed 2026-08-01 · deepseek-v4-flash

Pith's one-line read The paper delivers ILuRF2026, the first realization of a standardized lunar reference frame, built by a weighted combination of three leading lunar ephemerides.

desk verdict ILuRF2026 is a genuinely useful first realization of a lunar reference frame; the internal-consistency caveat is real but already half-admitted, and the paper deserves peer review as a standards document. read the letter →

arxiv 2607.27762 v1 pith:BCRPOL6N submitted 2026-07-30 astro-ph.EP astro-ph.IM

classification astro-ph.EPastro-ph.IM
keywords lunarreferenceframeephemeridesILuRF2026variancecomponentestimationlaserrangingplanetaryephemeriscombinationtimescaleTCLpositionnavigationtiming
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

The paper, issued by a joint working group on lunar reference frames, argues that lunar exploration needs a common Position, Navigation and Timing standard, just as Earth navigation uses standardized terrestrial frames. It recommends adapting a GNSS-style weighted-combination method to the Moon and presents ILuRF2026, the first realization of the International Lunar Reference Frame. ILuRF2026 is a weighted average of three modern lunar ephemerides (DE430, INPOP21a, EPM2021), with weights chosen by a variance component estimator: EPM2021 receives the highest weight. The paper reports internal frame uncertainties of about 17.6 cm over 2010–2030 and Lunar Laser Ranging residuals of 1.7–3.6 cm, matching the best-fitting contributing ephemeris. If adopted, this frame would give spacecraft operators and surface missions a single, redundant, standard reference for lunar navigation.

What carries the argument

The central object is the Variance Component Estimator (VCE) combination, adapted from GNSS orbit combination. It assigns normalized weights to the three contributing ephemerides (0.451 for EPM2021, 0.381 for INPOP21a, 0.168 for DE430) and applies the same weights to origin, orientation, and time-scale parameters. The work of this machinery is to turn mutually disagreeing ephemerides into a single standardized LOOP (Lunar Origin and Orientation Parameters) time series with quantified internal uncertainty, while the frame's materialization is provided by the fixed coordinates of lunar laser retro-reflectors.

What would settle it

Compare ILuRF2026 against an independent tracking data set not used in any of the three ephemerides—for example, Doppler ranging to a lunar orbiter or laser ranging to a new retroreflector in the southern hemisphere. If the differences exceed the claimed 17.6 cm internal uncertainty, the VCE weights and uncertainty estimate are understating the frame's true error.

Watch

Extended reading notes

Core claim

ILuRF2026 defines the lunar reference frame's origin (Moon center of mass), orientation (principal-axis libration angles), materialization (coordinates of lunar laser retro-reflectors), and time scale (TCL–TCB), and delivers these as time series from 1970 to 2050 with interpolation software. The frame's defining choice is the principal-axis (PA) frame, to which tidal and elastic deformations are added as corrections, with Helmert transformation parameters provided for older frames such as DE421 in the mean-Earth frame. The paper's central claim is that a weighted combination of independent ephemerides—rather than reliance on a single provider—produces a standardized frame that is accurate, c

Load-bearing premise

The variance-component weights are derived from the mutual differences of three ephemerides that were all fitted to the same Lunar Laser Ranging data set, and the validation uses that same data type; if that shared data set carries a common bias, the combined frame inherits it.

Editorial extensions

If this is right

  • Any lunar spacecraft or surface asset can use ILuRF2026 as a common reference, with coordinates and time delivered in both TDB and TCL.
  • The frame can be updated as improved ephemerides appear, since the VCE combination is designed for future re-weighting.
  • Helmert transformations allow users of older frames (e.g., DE430, INPOP21a, EPM2021, DE421 ME) to convert coordinates with cm-level consistency, 2–3 cm for PA–PA and up to 14 cm for DE421 ME.
  • The TCL-TCB time series is consistent across independent computations to below current clock stability, supplying a lunar time scale for the frame.
  • New retroreflectors and a southern-hemisphere lunar VLBI station are expected to strengthen the frame definition, particularly the along-track and orientation components.

Reading between the lines

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

  • The frame's uncertainty estimate is only as independent as the data used to build it; a truly external check would require observations not already fitted into the three ephemerides.
  • If ILuRF2026 is accepted as an IERS product, the governance questions of update cadence and inclusion criteria will determine whether the frame can keep pace with accelerating lunar exploration.
  • A southern-hemisphere lunar VLBI station, as suggested in the paper, would likely reduce the along-track and orientation errors that dominate the current 17.6 cm uncertainty.
  • The combination approach could be extended to future lunar ephemerides or to other bodies, where a similar multi-provider weighted frame could support PNT.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper, written on behalf of the IAG/IAU JWG 1.1.3, proposes a definition of the International Lunar Reference System (ILuRS) and presents its first realization, ILuRF2026. The frame is constructed as a variance-component-estimated weighted combination of the lunar solutions DE430, INPOP21a, and EPM2021, with normalized weights 0.451454072137127, 0.380949596178373, and 0.167596331684500, respectively. The delivered products are time series of the lunar origin and orientation (LOOP), LLRRR coordinates, Helmert transformations to other frames, and TCL-TCB time-scale series. The paper reports internal uncertainties of 17.6 cm over 2010–2030 (origin 15.3 cm, orientation 8.6 cm), LLR one-way residual RMS values of 1.7–3.6 cm against ILuRF2026, and independent benchmark agreement for the TCL-TCB computation to below 5×10−17 in drift and 65 ps in detrended residuals.

Significance. If the frame is adopted, ILuRF2026 would be a practically useful, standardized product for lunar PNT, providing a single realization from multiple established ephemerides and the first such product under the new IAU LCRS/TCL resolutions. The paper's strengths include the open delivery of time series and software, the TCL-TCB benchmark using the TEMPUS library (Table 2), and the analysis of Helmert transformation parameters, including the 0.97 X-translation/scale correlation. However, the central quantitative claim—an absolute accuracy of 17.6 cm supported by LLR validation—is not established: the VCE weights and the validation both draw on the same LLR data that constrain all three input ephemerides, so the reported numbers measure internal agreement, not external accuracy. The significance of the paper therefore rests on its value as a standards and consistency product rather than on a demonstrably improved absolute frame.

major comments (4)
  1. [§3, Definition of the Origin and Orientation (footnote 1)] The WMSE is defined as Σ w_i(y_i − ŷ_i)/Σ w_i, which is a weighted mean difference, not a weighted mean square error. Since the headline internal uncertainty of 17.6 cm is this quantity, the formula must be corrected and the exact definition supplied. More fundamentally, this statistic measures dispersion among three LLR-constrained ephemerides; common-mode errors from LLR station biases, limited Earth–Moon geometry, and shared dynamical assumptions are absent by construction. I recommend that the text explicitly label this as internal dispersion rather than an accuracy estimate.
  2. [§3, Validation with LLR (Table 1)] The LLR residual validation is not independent of the construction of the frame. Each input ephemeris was fitted to LLR data, the VCE weights are estimated from the mutual differences of those same solutions, and the validation in Table 1 reuses LLR observations. That ILuRF2026's RMS closely matches EPM2021 is expected from EPM2021's highest weight (0.451454...) and does not constitute confirmation. 'Independent software packages' reproduce numeric residuals but do not introduce an independent data type. Please reframe this part as an internal-consistency check and, if possible, add a genuinely out-of-sample test (e.g., post-2025 NGLR-1 data, withheld normal points, or VLBI observations), or explicitly disclaim absolute validation.
  3. [§3, Validation with LLR (Table 1 and Table 3)] The residual analysis is not reproducible from the paper: no normal-point counts, per-station time spans, weighting/editing criteria, or formal uncertainties for the RMS values are given. The empirical corrections A1, A2, A3 are listed in Table 3 but are not defined, and the paper does not state whether they were adjusted in the residual computation. Without these details, differences such as 2.5 cm versus 2.6 cm between ILuRF2026 and EPM2021 at Wettzell are not meaningful, and the table cannot support the precision implied in the text.
  4. [§3, Method] The VCE combination is described only by a one-sentence summary and a reference to [14], which is listed as submitted. Because the weights and the 17.6 cm uncertainty estimate are load-bearing results, the paper should include the covariance model, the definition of the parameter vector, the treatment of correlations among the three ephemerides, and the exact VCE equations. As written, an interested reader cannot assess whether the reported weights are stable, overfit to the particular three solutions, or sensitive to the chosen parameterization.
minor comments (6)
  1. [Abstract and text] There are repeated typos: 'the the', 'temporay', 'Celectial', 'webiste', and 'and and'. Please copy-edit carefully.
  2. [§3, Validation with LLR, and Definition of Origin] The text gives two different end dates for the DE430 data fit: 'up to 2014' in one place and 'ending 2012' in the validation paragraph. Reconcile these values.
  3. [§3, Validation with LLR] The sentence 'Independent groups are currently working on reproducing these findings with success' is not documented. Either provide a reference or remove it from the scientific argument.
  4. [§3, Delivery and associated parameters] The delivery is on a 'temporary' website and a future ESA URL. For a publishable product, a permanent DOI or archive (e.g., Zenodo) should be assigned so the frame and interpolation software are citable and stable.
  5. [§3, Method] The normalized weights are quoted to 15 significant figures, which implies a precision inconsistent with the stated 17.6 cm uncertainty and the simplified VCE. Rounding to a meaningful precision would be clearer.
  6. [§3, Transformation to other frames] The sentence about ILuRS becoming part of the IERS Product Center ('accepted May 2026') lacks a reference or official notice; if this is a factual claim, it should be supported.

Circularity Check

3 steps flagged · score 6.0 of 10

ILuRF2026's uncertainty and 'validation' are internal consistency checks: the VCE WMSE measures scatter among three LLR-fitted ephemerides and the LLR residual table is in-sample for those same inputs.

  1. fitted input called prediction [Section 3, 'Validation with LLR' (Table 1)]
    "ILuRF2026 was validated against LLR data using multiple independent software packages, fixing its LOOP, LLRRR coordinates, Love number h2, and empirical corrections to compute residuals without adjusting other parameters. As shown in [14] and Table 1, ILuRF2026 residuals are consistent with those of other ephemerides, confirming its robustness."

    The three constituent ephemerides DE430, INPOP21a and EPM2021 were each fitted to LLR normal points, as the paper itself acknowledges in the abstract ('a single data type'). The one-way LLR residuals in Table 1 are therefore in-sample for all three inputs, and ILuRF2026 is a weighted average of those inputs. Matching EPM2021's RMS is forced by the largest VCE weight (0.451454...). Thus the 'validation' confirms internal consistency, not agreement with data independent of the construction; any LLR systematic error common to the three solutions is inherited by the combination and absent from the residual test.

  2. self definitional [Section 3, 'Definition of the Origin and Orientation' and footnote 1]
    "Internal ILuRF uncertainties which are the VCE Weighted Mean Square Error (WMSE)1 are 17.6 cm over 2010–2030 (origin 15.3 cm, orientation 8.6 cm) and 31.6 cm over the full period [14]."

    The footnote defines WMSE as the weighted residual of the SOA ephemerides relative to their own weighted combination, i.e., the scatter of the inputs. The VCE weights themselves are estimated from the mutual differences of the same three ephemerides. Hence the advertised uncertainty is, by construction, a within-sample dispersion statistic, not an estimate of absolute error. If DE430, INPOP21a and EPM2021 share common LLR-related errors (same data type, station biases, Earth-Moon geometry), those common errors cancel in the differences and are invisible in the WMSE.

1 more flagged steps
  1. self citation load bearing [Section 3, opening: 'Context' and throughout]
    "Most of the details of this section can be found in [14]. Here, we summarize only some important aspects."

    The construction method, the VCE weights, the 17.6 cm uncertainty, the LLR residual table and the Helmert transformations are all attributed to [14] (arXiv:2510.15484), a preprint from the same IAG JWG 1.1.3 group whose members are credited in the acknowledgments for 'their innovative work on ILuRF construction.' These are load-bearing numerical claims for the first-realization result, and they are imported from the group's own submitted work rather than demonstrated here or anchored to an externally verified/independent source. This elevates the citation from normal background reference to a circular support for the paper's central claim.

full rationale

The paper's core construction is a weighted average of three ephemerides, each independently developed but all fitted to Lunar Laser Ranging data. The VCE weights are estimated from the mutual differences of these same three solutions, so the weighted mean and its WMSE are descriptive statistics of the input set. The 'internal ILuRF uncertainties' (17.6 cm/15.3 cm/8.6 cm) are the WMSE, i.e., the weighted scatter of the inputs around their own combination; they cannot capture common-mode errors. The Table 1 validation is computed on LLR one-way residuals, the same observable and largely the same normal points used in fitting DE430, INPOP21a and EPM2021. Hence it is an in-sample consistency check, not an independent confirmation. The close agreement with EPM2021 is expected since EPM2021 received weight 0.451. The paper's own abstract flags the limitation: 'a single data type.' Additionally, most construction details and numerical results are delegated to [14], a same-group preprint (arXiv:2510.15484), making the first-realization claim dependent on that citation rather than on independent evidence presented here. However, the proposal also includes a TCL time-scale computation benchmarked against external published time series, and the Helmert transformation and delivery infrastructure have independent content; these prevent the whole paper from being purely circular. Therefore score 6, not higher.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The central product is a weighted average of existing ephemerides; the only numbers introduced here are the VCE weights and empirical corrections. No new physical entities are posited. The main epistemic load is the assumption that the three ephemerides form a reliable, semi-independent ensemble and that LLR residual agreement is a meaningful validation given that LLR data were used to build the individual ephemerides.

free parameters (4)
  • VCE weight for EPM2021 = 0.451454072137127
    Estimated from mutual ephemeris differences via VCE; not from independent ground truth.
  • VCE weight for INPOP21a = 0.380949596178373
    Same VCE estimation using the three ephemerides' differences.
  • VCE weight for DE430 = 0.167596331684500
    Same VCE estimation.
  • Empirical corrections A1, A2, A3 = 4.4, 1.6, 1.2 mas
    Ad hoc corrections used in LLR residual computation; they are borrowed from prior model fits, and their values directly affect the reported residual RMS.
assumptions (4)
  • domain assumption The three selected ephemerides (DE430, INPOP21a, EPM2021) are accurate representations of lunar motion and sufficiently independent for a VCE combination.
    The entire frame is a weighted mean of these models; if they share common systematic errors (e.g., from the same LLR data type), the combined frame inherits them.
  • domain assumption The GNSS orbit-combination VCE framework applies to lunar ephemerides.
    The paper adapts [16] as a 'simplified VCE' without justifying that the statistical model of GNSS orbit combination transfers to lunar dynamics and LLR-derived orientation.
  • domain assumption Lunar Laser Ranging retro-reflectors provide a stable materialization of the lunar surface.
    The materialization of ILuRF2026 uses LLRRR coordinates relative to the Moon's center of mass; no new control points are added, and all reflectors are in the northern hemisphere.
  • standard math Relativistic time scales TCB/TCL are correctly implemented in TEMPUS following IAU resolutions.
    TCL-TCB time series are computed with TEMPUS and benchmarked to [6] and [9]; no independent verification is provided here beyond those comparisons.

how reviews work

0 comments
Cite this review

Pith. "Pith review of The International Lunar Reference System." pith.science (2026). https://pith.science/paper/BCRPOL6N

@misc{pith2026260727762,
  author       = {Pith},
  title        = {Pith review of: The International Lunar Reference System},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BCRPOL6N}},
  note         = {Machine review of arXiv:2607.27762}
}
read the original abstract

As the exploration of the Moon accelerates in the coming years, there is already an urgent need to standardise the Position, Navigation and Timing of spacecraft, and consequently for the definition of the lunar reference system and frame. This document gathers the most recent recommendations of the the International Association of Geodesy (IAG) / International Astronomical Union (IAU) Joint Working Group (WG) 1.1.3 "Lunar Reference frames" on the topic. Despite the challenges for producing a Lunar Reference System and Frame due to the limited number of control points and a single data type, the WG proposes applying an approach developed for Global Navigation Satellite Systems (GNSS). This supports an accurate and reliable reference system, consistent with International Earth Rotation and Reference Systems Service (IERS) standards and, based on redundant sources, ensures resilience for the lunar Position, Navigation and Timing (PNT) framework. The first realization of the International Lunar Reference Frame 2026 (ILuRF2026), ILuRF2026, is delivered on the temporay website https://ilurs-6e772d.gitlab.io/ and on the future http://ilurf.gssc.esa.int together with associated software.

Figures

Figures reproduced from arXiv: 2607.27762 by the authors.

Figure 1
Figure 1. Transformations from the International Celestial Reference System (ICRS), International Terrestrial Reference System (ITRS), Lunar Celestial Reference System (LCRS) and the International Lunar Reference System (ILuRS). The dashed zones correspond to new definition and transformation proposed by the WG 1.1.3. ’LT’ flags a required Lorentz transformation due to differing metrics between reference systems. The ICRF-to-… view at source ↗
Figure 2
Figure 2. compares SOA ephemerides with ILuRF2026 in origin (left) and orien￾tation (right). For origin, along-track differences reach up to 4 m in the prediction period, cross-track is stable at ±1 m (well-defined orbital plane), and radial dif￾ferences are at centimeter level, consistent with LLR constraints. For orientation, precession angle ϕ shows the largest deviation, with DE430 differences systemat￾ically larger due t… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 7 canonical work pages

  1. [14]

    (2025) Definition and Realization of the Interna- tional Lunar Reference Frame.Astronomy & Astrophysics, submitted, https://doi.org/10.48550/arXiv.2510.15484,

    Sośnica, K, et al. (2025) Definition and Realization of the Interna- tional Lunar Reference Frame.Astronomy & Astrophysics, submitted, https://doi.org/10.48550/arXiv.2510.15484,

  2. [1]

    Beutler, G., et al. (1995). Combining the orbits of the IGS Analysis Centers.Bulletin Géodésique, 69, 200–222. https://doi.org/10.1007/BF00806733

  3. [2]

    Fienga A, et al. (2021) INPOP21a planetary ephemerides.Notes Scientifiques et Techniques de l’Institut de Mecanique Celeste, 110,https://www.imcce.fr/ content/medias/recherche/equipes/asd/inpop/inpop21a.pdf

  4. [3]

    Lunar References Systems, Frames and Time-scales in the context of the ESA Programme Moonlight

    Fienga, A., et al. (2024) Lunar References Systems, Frames and Time-scales in the context of the ESA Programme Moonlight,arXiv e-prints, arXiv:2409.10043, https://doi.org/10.48550/arXiv.2409.10043

  5. [4]

    M., et al

    Folkner, W. M., et al. (2014). The planetary and lunar ephemerides DE430 and DE431.Interplanetary Network Progress Report, 196(1), 42–196.https://ilrs. cddis.eosdis.nasa.gov/docs/2014/196C.pdf

  6. [5]

    IAU (2024) Resolution II: To establish a standard Lunar Celestial Refer- ence System (LCRS) and Lunar Coordinate Time (TCL).Resolutions adopted at the XXXII IAU General Assembly.https://drive.google.com/file/d/ 1-Ps-YyCjmjzg2NGi0A-VE0kWqr8w7LnE/view

  7. [6]

    Relativistic time scales in the Solar system

    Klioner, S. (2025) Relativistic time scales in the Solar system IAU colloq. 401: Advancing Reference Systems, Ephemeris, and Standards: From the Earth and the Moon to Solar System Bodies.Proceedings of the International Astronomical Union,https://doi.org/10.48550/arXiv.2604.16006 8 A. Fienga et al

  8. [7]

    Kopeikin, S. M. et al. (2024). Lunar Time in general relativity.Physical Review D, 110, 084047, https://doi.org/10.1103/PhysRevD.110.084047

Show all 16 references
  1. [8]

    & Mueller, I

    Kovalevsky, J. & Mueller, I. I. (1981) in Astrophysics and Space Science Library, Vol. 86, IAU Colloq. 56: Reference Coordinate Systems for Earth Dynamics, ed. E. M. Gaposchkin & B. Kolaczek, 375

  2. [9]

    (2025) Lunar time ephemeris LTE440: Definitions, algorithm, and performance.Astronomy & Astrophysics, 704:A76, December 2025

    Lu, X., et al. (2025) Lunar time ephemeris LTE440: Definitions, algorithm, and performance.Astronomy & Astrophysics, 704:A76, December 2025

  3. [10]

    Molli S, et al. (2026) NovaMoon: A Strategic Lunar Reference Station for Posi- tioning, Timing, and Largely Enhanced Science in the Earth-Moon System.arXiv e-prints, arXiv:2602.08432, https://doi.org/10.48550/arXiv.2602.08432,

  4. [11]

    Pitjeva, E., et al. (2021). Planetary and lunar ephemeris EPM2021 and its sig- nificance for Solar system research.Proceedings of the International Astronomical Union, 15(S364), 220-225. https://doi.org/10.1017/S1743921321001447

  5. [12]

    (2026) Lunar reference systems and their realisa- tions using INPOP ephemerides.Astronomy & Astrophysics, 705, A88, https://doi.org/10.1051/0004-6361/202555962

    Rambaux, N., et al. (2026) Lunar reference systems and their realisa- tions using INPOP ephemerides.Astronomy & Astrophysics, 705, A88, https://doi.org/10.1051/0004-6361/202555962

  6. [13]

    Soffel, M., et al. (2003) The iau 2000 resolutions for astrometry, celestial mechan- ics, and metrology in the relativistic framework: Explanatory supplement.The Astronomical Journal, 126(6), 2687–2706, https://doi.org/10.1086/378162

  7. [15]

    G., et al

    Williams, J. G., et al. (2008). DE421 Lunar Orbit, Physical Librations, and Surface Coordinates.JPL Inter Office Memorendum, 335-JW,DB,WF-20080314-001, March 14, 2008.https://naif.jpl.nasa.gov/pub/naif/generic_kernels/spk/planets/ a_old_versions/de421_lunar_ephemeris_and_orien...

  8. [16]

    Zajdel, R., et al. (2025). Advancing Multi-GNSS Orbit Combination in the Variance Component Estimation Framework.Journal of Geodesy, 99, 90, https://doi.org/10.1007/s00190-025-02005-w

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.