A public, cited comparison of the nine Artemis III candidate regions named by NASA in October 2024. At one published point per region, four readouts, each taken from the same maps with the same measures, an uncertainty where one is published, and a link to its archive. Where a map does not cover a point, the readout says so instead of guessing.
Not a safety assessment, and no site is declared safe. No combined score and no ice-volume estimate. Screening cuts are shown with their source for reference; they are not flight rules. “Earth in view” counts any part of Earth’s disc above the local horizon, which is slightly optimistic compared with a DSN link budget.
Slope in degrees at the 5 m pixel containing the point, from the LOLA product-78 site DEM (Barker et al. 2021). The ± is one published RMS slope error for that pixel. Three regions have no 5 m tile and show no slope; the 20 m mosaic is not substituted. Where the LOLA return count at the pixel is 0, the slope is interpolated.
18.6-year average fraction of hourly steps with any part of the Sun above the horizon, from the 75°S illumination map (Mazarico et al. 2011, 120 m pixel, nearest pixel to the point). It is a long-term average, not the sunlight in a particular week, and has no published error bar.
Same map family and averaging period: the 18.6-year average fraction of hours with any part of Earth’s disc above the horizon. Disc visibility, not a DSN link budget. The hourly timeline on each card is a separate 14-day, terrain-aware product built from that region’s elevation raster, and its window percentages are labelled as such.
Distance from the point to the nearest permanent-shadow pixel (120 m) whose Diviner summer-maximum temperature is below (Williams et al. 2019, 240 m pixel). Searched within 40 km beyond the nine points. The ± is one Diviner pixel, a resolution bound, not a statistical error. Proximity only; no ice-volume estimate.
All nine points (first Wueller et al. 2026 Table A1 point per region) are read by the repository’s python -m atlas.regions and exported by python -m atlas.export_web to web/data/regions.json. Every row uses the same measures: slope and RMS error at the point, 18.6-year sunlight and Earth averages, Diviner summer-maximum temperature, permanent-shadow flag, distance to cold shadow, and LOLA return count. Slope, slope error and return count are not computed for Peak near Cabeus B, Mons Mouton Plateau and Slater Plain because product 78 has no 5 m site DEM containing those points. Each card also has an hourly terrain-aware Sun/Earth visibility strip (2025-01-01 + 14 days, 1 m and 5 m eye height) computed from that region’s elevation raster. Each site zoom is a real hillshade with cast shadows from LOLA elevation: 5 m LOLA elevation (shown at 10 m pixels) for Haworth (HW01), Malapert (MMA01, Site 23), Mons Mouton (MMO01, Site 20), Nobile Rim 1 (NR101, Site 06), Nobile Rim 2 (NR201, DM2) and de Gerlache Rim 2 (dGR101, SL2) from the product-78 site DEMs, and Slater Plain (SP01) from a 20 km window of the product-81 87–90°S 5 m mosaic, which fills gaps by interpolation; and the 20 m LOLA polar DEM (product 90, LDEM_80S_20MPP_ADJ, Barker et al. 2023), a 30 km window at native 20 m pixels, for Peak near Cabeus B (PCB01) and Mons Mouton Plateau (MMP01), which no 5 m LOLA elevation covers. 20 m terrain does not resolve small craters or boulders and its badge says so. The Sun azimuth and elevation come from the DE421 ephemeris. The home Moon is the LOLA average-sunlight map; the Play/Pause control there drives the same DE421 clock as the cards, and its light wash and Sun marker show the real direction of the Sun (its longitude as seen from the pole), not re-rendered terrain. Shadows come only from terrain inside each chip or raster.
Cuts are drawn as dashed lines on the region card and as markers in the comparison table. They are screening values shown with their source, not flight rules, and passing them does not make a site safe.
| Region | Slope at point | Sunlight 18.6-yr avg | Earth in view 18.6-yr avg | Summer max temp. | Nearest cold ground | Shown cuts |
|---|
Slope ± is one RMS slope error (LOLA product 78, 5 m). Sunlight and Earth are 18.6-year averages with no published error bar; Earth is disc visibility, not a link budget. Temperature is the Diviner summer maximum on its own 240 m pixel. Cold ground ± is one Diviner pixel (0.24 km). “Shown cuts” states whether the value falls inside the selected screening cuts. It is a reference marker, not a pass/fail verdict and not a flight rule.