New and Improved LOLA DEMs for South Pole Landing Sites

Michael Kenneth Barker



We have derived new high-resolution DEMs of several regions (Sites 1, 4, 7, 11) surrounding high-priority lunar south pole landing sites using exclusively laser altimetry data acquired by LRO-LOLA. By iteratively co-adjusting the LOLA tracks in a self-consistent fashion, we reduced the orbital errors by over a factor of 10 such that the new track geolocation uncertainty is ~10 - 20 cm horizontally and ~2 - 4 cm vertically over each region. The new 5 m/pix LDEMs are substantially more realistic than the previous ones with fewer artifacts due to orbital errors and fewer spurious noise points. While the fraction of interpolated 5-m pixels in these polar LDEMs is necessarily large (~90%) due to LOLA's cross-track and inter-spot spacing, these LDEMs have the advantage of having accurate geodetic control and of being unaffected by shadows, and, thus, will be useful constraints on higher-resolution topographic models derived from imagery.

We developed a method to estimate surface height uncertainty in the new LDEMs that accounts for the reduced orbital errors and interpolation errors. This method circumvents the infeasible computation of the full error-covariance matrix of the LDEM. Instead, we use the fractal nature of lunar topography to build a more computationally manageable statistical ensemble of clones with similar error properties as the data. We show how we use this ensemble to study height and slope uncertainty, as well as the uncertainty in illumination conditions. Such an approach can be useful for a range of other studies, not just pure illumination conditions, when examining the feasibility of potential landing sites.

The LDEM height and slope uncertainties within the RoIs are similar across all the sites with a median RMS Z error ~ 0.30 - 0.50 m and a median RMS slope error ~ 1.5 - 2.5°. Interpolation error depends primarily on gap size, or areal density of the LOLA points, with a secondary dependence on terrain slope that becomes more important over highly sloped terrain. Hence, the interpolation error will be larger at greater distances from the pole for the same pixel scale, because of the lower point density and poorer effective resolution. The illumination conditions within each region of interest (RoI) are impacted by height uncertainties and show more variation that can potentially be used as a guide to rank the different sites and RoIs within a site. Between Jan. 1, 2024 and Jan. 1, 2026, Site 1 has the most area with average illumination > 70% at 1 m above the surface and RoIs 4 - 6 at this site generally have the most favorable illumination conditions even considering their uncertainties. At 5 m above the surface, the effect of LDEM uncertainties on illumination conditions are significantly reduced, and Site 1 has longer continuous illumination periods than the other sites.

Continuing work involves applying these methods to other areas.
Please check back here for updates as new regions are added.


Data

For each site, various products are provided:
- LDEM: surface-interpolated elevation map (pixel values give surface height Z in meters)
- XYZI: final point clouds after track adjustments and cleaning
- LDEC: count map (# of LOLA spots in each pixel)
- Slope map (in degrees)
- Total Z uncertainty in meters (formally, the RMS error)
- Slope uncertainty in degrees (formally, the RMS error)
- 100 clones

All these files (except the XYZI points clouds) are 5 m/pix GeoTiffs with south polar stereographic X/Y coords in meters, MOON_ME reference frame of ephemeris DE421.

See this README file.
The sites are:
- Site01: Connecting ridge
- Site04: Shackleton rim
- Site07: Peak near Shackleton
- Site11: de Gerlache rim
- Site20: Leibnitz beta plateau
- Site23: Malapert massif

Note: Site23 does not yet have error estimates or clones.

The site name numbering follows the Mazarico et al. (2011) paper.





Sites.jpg

Close-ups of each site with Regions of Interest labeled: