An international geodesy working group has released the first realization of a standardized lunar reference frame, a common coordinate system meant to support position, navigation, and timing for spacecraft operating around the Moon. The frame, designated ILuRF2026, combines three independent lunar ephemerides using a weighting method adapted from satellite navigation, and is described in a proceedings paper posted to arXiv on July 30 by Agnes Fienga of the Observatoire de la Cote d’Azur on behalf of the joint working group.
The paper is the latest step in an effort that began in 2022, when the International Association of Geodesy and the International Astronomical Union established their joint working group on lunar reference frames. The motivation is practical: between 6 and 10 crewed missions to the Moon are planned in the coming years, and with multiple space agencies and private companies involved, the group argues that spacecraft need a common definition of where the Moon is, how it is oriented, and what time it is there. Since 2025, the United Nations Office for Outer Space Affairs has organized an annual cislunar positioning, navigation, and timing workshop to address the same problem.
The system, called the International Lunar Reference System, follows the criteria used for terrestrial reference systems but adapted to the Moon. It adopts the principal axis frame, defined by the diagonalization of the Moon’s inertia matrix, rather than the mean Earth frame based on the averaged Earth-Moon orientation. The working group chose the principal axis frame because it is defined directly by lunar ephemeris Euler angles, while the mean Earth frame requires additional averaging that can introduce misinterpretation. The frame co-rotates with the Moon and is anchored to an undistorted reference Moon, an idealized undeformed body to which tidal, rotational, and elastic deformations are added as corrections.
The realization of the frame, ILuRF2026, is built from three state-of-the-art ephemerides of comparable accuracy: DE430 from JPL, INPOP21a from the IMCCE, and EPM2021 from the Russian Institute of Applied Astronomy. Rather than rely on any single provider, the group combines them with a simplified variance component estimator, an approach close to the one used for combining satellite navigation orbits. The normalized weights are 0.451 for EPM2021, 0.381 for INPOP21a, and 0.168 for DE430. The combination is materialized through the coordinates of lunar laser ranging retro-reflectors relative to the Moon’s center of mass, and time series of the frame’s origin, orientation, and time scale from 1970 to 2050 are distributed with interpolation software.
The internal uncertainties of the frame, computed as the weighted mean square error of the combination, are 17.6 centimeters (6.9 inches) over 2010 to 2030, split between 15.3 centimeters (6.0 inches) in origin and 8.6 centimeters (3.4 inches) in orientation, and 31.6 centimeters (12.4 inches) over the full period. These errors stem mainly from the northern-hemisphere distribution of the retro-reflectors and limited Earth-Moon geometry. The frame was validated against lunar laser ranging data using multiple independent software packages, and its residuals closely match those of EPM2021, the ephemeris that received the highest weight.
The frame’s orientation and origin parameters, called LOOP, show differences from the contributing ephemerides that reach up to 4 meters (13 feet) along-track in the prediction period, with cross-track differences stable at about 1 meter (3.3 feet) and radial differences at the centimeter level. The group also provides 7-parameter Helmert transformations between ILuRF2026 and other reference frames, including mean Earth versions of older ephemerides. Errors between principal axis frames are 2 to 3 centimeters (0.8 to 1.2 inches), while transformations to the mean Earth DE421 frame reach 10 centimeters (3.9 inches) with the full 7-parameter transformation and 14 centimeters (5.5 inches) with a reduced 3-parameter version, which is why the full transformation is recommended.
For the time dimension, the working group relies on Lunar Coordinate Time, computed with an open-source Python library called TEMPUS. The differences between independent computations of the lunar time offset are below 5 x 10^-17 in drift and 65 picoseconds in remaining residuals, smaller than current clock stability and accuracy. Over 25 years, the time transformation introduces a periodic trend of about 2.7 milliarcseconds on the precession angle and an additional drift of 0.032 milliarcseconds per day in the rotation angle.
ILuRF2026 is delivered through the temporary website ilurs-6e772d.gitlab.io and the future site ilurf.gssc.esa.int, which the European Space Agency’s navigation support center plans to open in fall 2026. The frame has been proposed to become part of the International Earth Rotation and Reference Systems Product Center, with acceptance reported in May 2026, though future inclusion criteria, update cadence, and governance remain to be defined.
The group expects improvements from new lunar surface observations, notably the European Space Agency’s NovaMoon station on the Argonaut missions planned for 2029, which will add very long baseline interferometry, radio transponders, and two compact atomic clocks alongside retro-reflectors. A southern-hemisphere VLBI station is identified as the most efficient way to address the geometry limitation, improving the definition of the LOOP parameters by an estimated 75 percent. The paper does not establish when those improvements will be incorporated into the frame, or how the governance of the product center will handle future updates.
Sources
1. Fienga, A., “The International Lunar Reference System,” arXiv:2607.27762: https://arxiv.org/abs/2607.27762
2. International Lunar Reference System ILuRS, ILuRF2026 delivery page: https://ilurs-6e772d.gitlab.io/ILuRS2026/ILuRF_2026.html