The lunar reference frame: definition and first realization
Key facts
- The paper defines the origin, orientation and scale of the Lunar Reference System and gives the first realization of the ILRF. [1]
- The combined frame carries a mean error of 17.6 cm for 2010–2030: 15.3 cm from the origin and 8.6 cm from the orientation. [1]
- It is built by variance component estimation from three lunar ephemerides: INPOP21a, DE430 and EPM2021. [1]
- It runs from 1970, where the lunar laser ranging data begin, to extrapolated realizations in 2052. [1]
- Our source registry names no journal version of arXiv:2510.15484, so this page describes it as a preprint. [1]
What it proposes
- A definition of the Lunar Reference System: its origin, its orientation and its scale.
- A first realization of that system as the International Lunar Reference Frame, in the Principal Axis system, attached to the surface and co-rotating with the Moon, with its origin at the lunar centre of mass.
- Numerical solutions obtained by variance component estimation from three lunar ephemerides: INPOP21a, DE430 and EPM2021.
- A frame that runs from the start of the lunar laser ranging data in 1970 to extrapolated realizations in 2052.
This page is a reading of the paper, not a review of it. The figures below are model values published by the authors, quoted with the source on every row.
A frame is not a time scale, and this work is about the frame. It defines the origin, the orientation and the scale of the Lunar Reference System, and then gives numbers for a first realization of the International Lunar Reference Frame [1].
What the frame is attached to
The ILRF is defined as the Principal Axis system. It is attached to the surface, it co-rotates with the Moon, and its origin sits at the lunar centre of mass, which the authors call the lunocenter [1].
The realization comes from three lunar ephemerides rather than one. INPOP21a, DE430 and EPM2021 are combined by variance component estimation, for the position of the lunar centre of mass and for the Euler angles of rotation: precession, nutation and proper rotation [1].
How good the first realization is
The combined frame has a mean error of 17.6 centimetres for 2010–2030. Of that, 15.3 centimetres come from the origin and 8.6 centimetres from the orientation [1].
The authors name the reason the origin is the weaker half. The retroreflector network on the Moon has poor geometry, which leaves the scale strongly correlated with the X component of the lunocenter in the Principal Axis system [1]. The measurements themselves are far sharper than the frame: lunar laser ranging post-fit residuals are at 2–3 centimetres in one-way ranges for the best-performing stations [1].
Transformations between realizations are also quantified. Between the ILRF and other realizations in the Principal Axis system the mean error is about 3 centimetres, and to the DE421 Mean Earth frame it is 5 centimetres [1].
Why a timekeeping site cares about centimetres
A clock reading has to be attached to a position, and a position is only meaningful inside a stated frame. Kopeikin and Kaplan put the precision of their lunar time framework at the nanosecond level within Earth’s Hill sphere [2], while this paper puts its own realization at tens of centimetres [1]. The two are quoted in different units, but they are one engineering problem.
That is also where the hardware comes in. The paper names the geometry of the retroreflector network as the limit on the origin, and a new surface station is one thing that would change that geometry. Such a station is proposed in NovaMoon, which would carry laser retroreflectors, atomic clocks and radio links to the lunar south polar region [3].
Where it sits next to the time work
The frame paper and the time papers do not overlap and do not disagree. The rate relations a lunar scale has to carry are published separately, and are read in the lunar reference timescale paper [4]. The transformation between Lunar Coordinate Time and the barycentric scales is computed by the software read in the LTE440 lunar time ephemeris, which is built on the DE440 ephemeris rather than on DE430 [5].
Neither question is settled by a preprint. The International Astronomical Union asked international organizations in 2024 to agree the relations between a lunar reference time scale, a lunar coordinate time and UTC, and the same coordination is what a reference frame needs [6]Official. The orbit proposal that would ride on such a frame is read in two birds with one stone, and every reading we publish is listed in the papers hub.
Key numbers
Every figure below is a model value taken from the source named in its own row. Nothing in this table is our own estimate.
| Quantity | Value | Source |
|---|---|---|
| Mean error of the combined frame, 2010–2030 | 17.6 cm | [1] |
| Part of that error from the origin | 15.3 cm | [1] |
| Part of that error from the orientation | 8.6 cm | [1] |
| Lunar laser ranging post-fit residuals, best stations | 2–3 cm | [1] |
| Mean transformation error to the DE421 Mean Earth frame | 5 cm | [1] |
| Span of the realization | 1970 to 2052 | [1] |
Status
Preprint. arXiv preprint, 17 October 2025[1]. arXiv 2510.15484. The full text is at Definition and Realization of the International Lunar Reference Frame.
Why it matters
A time scale has to be read somewhere, and a place is only defined inside a frame. This paper puts published figures on the frame that lunar positions and clock locations would be referred to, so a lunar position and a lunar time can be quoted in one system rather than two.
Sources
- Definition and Realization of the International Lunar Reference Frame
- Lunar Time in General Relativity
- NovaMoon: A Strategic Lunar Reference Station for Positioning, Timing, and Largely Enhanced Science in the Earth-Moon System
- Lunar reference timescale
- Lunar Time Ephemeris LTE440: definitions, algorithm and performance
- Resolution on the establishment of a coordinated lunar time standard by international agreement
Last verified