2.1 Diviner Data and Thermal Modeling

The Diviner Lunar Radiometer Experiment on LRO is an infrared radiometer that has been mapping lunar surface temperatures since 2009 (Paige et al., 2010). Diviner has nine spectral channels, including seven thermal infrared channels that span the wavelength range from 8 to 400 $${\upmu }$$m and are sensitive across the broad range of lunar surface temperatures. We collected all Diviner Reduced Data Records (RDRs) for channel 6–9 available through March 2024 for a $${\sim} 3{}^{\circ}\times 3{}^{\circ}$$ region centered on South Ray crater. Powell et al. (2023) published global midnight bolometric temperature, $${T}_{\mathit{BOL}}$$, maps which have better effective spatial resolution than previously published maps due to a correction for Diviner pointing errors. Additionally, they implement a model which mostly removes the effect of topography on nighttime temperature by accounting for the effect of latitude, local slope, and scattering and emission from surrounding topography. The resulting bolometric temperature anomaly, $${\Delta }{T}_{\mathit{BOL}}$$, maps better reveal faint thermal features which were previously masked by topography. Following the approach described in Powell et al. (2023), we produce $${T}_{\mathit{BOL}}$$ and $${\Delta }{T}_{\mathit{BOL}}$$ maps for the Apollo 16 region gridded at 128 pixels-per-degree (ppd) and in 0.125 lunar hour intervals (Figure S1 in Supporting Information S1).

2.2 Relative Density From Astronaut Footprints

Several investigations of soil mechanics were conducted during the Apollo program (Mitchell et al., 1974), including: (a) core tubes and drill cores which sampled to depths of $${\sim} $$60 cm to $${\sim} $$3 m; (b) penetration resistance tests which probed the upper $${\sim} $$60 cm; and (c) tests of soil compaction by astronaut footprints and vehicle tracks, which probe the upper $${\sim} $$15 cm. Of these, the footprint data are the most relevant for investigating cold spots, as they probe to depths similar to those sensed by nighttime temperatures. Namiq (1970) used laboratory experiments of footprints in regolith simulant and finite element stress-deformation modeling to develop a scaling between footprint depth and the porosity of the upper $${\sim} $$15 cm of regolith. Mitchell et al. (1974) used 776 footprint images across all the Apollo sites to estimate porosities for typical intercrater regolith at each site. In addition, Mitchell et al. (1972) present a breakdown of the average regolith porosity at several research station locations along the Apollo 16 Extravehicular Activity (EVA) traverses.

We follow this prescription and convert the mean and standard deviation porosity values reported in Mitchell et al. (1972, 1974) to relative densities assuming porosity is normally distributed (Table S1 in Supporting Information S1). For each location, the standard deviation and number of samples were used to calculate the standard error of the mean.

3.1 Cold Spot Thermophysical Properties

Figure 1a shows a distinct −1 to −2 K temperature anomaly with ray-like structure surrounding South Ray crater, which we interpret to be a cold spot (herein referred to as South Ray cold spot). The low temperature “rays” of South Ray cold spot match the orientation of optical rays apparent in LRO Camera (LROC) Narrow Angle Camera (NAC) imagery. The continuous portion of the cold spot extends to $${\sim} $$10–20 crater radii, with individual low temperature rays extending $${\sim} $$30–40 crater radii. South Ray cold spot is notably fainter and smaller than other cold spots around similarly sized craters. For example, Figure 1b shows the cold spot associated with Bandfield crater (90.76$${}^{\circ}$$E, −5.39$${}^{\circ}$$N, 898 m) (herein referred to as Bandfield cold spot) which is one of the most prominent cold spots on the Moon and is also significantly younger than South Ray cold spot at 0.23$$\pm $$0.02 Ma (Williams et al., 2018). Bandfield cold spot has a continuous region which extends to $${\sim} $$35 crater radii, within which nighttime temperatures are $${\sim} $$8–10 K cooler than the surrounding regolith. Discontinuous cold spot rays extend to over 200 crater radii, significantly farther than the greatest extent of the South Ray cold spot.

The thermophysical properties of both cold spots can be inferred from their nighttime cooling behavior, shown in Figure 1c. Each curve is fit using a 1-D thermal model (Hayne et al., 2017) where $${\rho }_{s}$$ and $$H$$ are allowed to vary. The best-fit values for typical regolith ($${\rho }_{s}$$ = 1,102 kg/$${\mathrm{m}}^{3}$$ and $$H$$ = 5.37 cm) agree well with previous studies (Bandfield et al., 2014; Hayne et al., 2017; Powell et al., 2023). Bandfield cold spot exhibits temperatures $${\sim} $$8–10 K lower than typical regolith and is clearly distinct from the surrounding regolith almost immediately after sunset. This is best fit with a typical value for surface density and an $$H$$ significantly greater than background ($${\rho }_{s}$$ = 1,097 kg/$${\mathrm{m}}^{3}$$ and $$H$$ = 28.5 cm), resulting in a lower average density over the upper centimeters of regolith (Figure 1d).

South Ray cold spot is $${\sim} $$1–2 K cooler than typical regolith at its most prominent, late in the night. However, it has similar or slightly higher temperatures early in the night and does not become distinct from its surroundings until $${\sim} $$22:00 local time (Figure S1 in Supporting Information S1). Because nighttime temperatures are typically affected by material at greater depth later in the night, this suggests that the near-surface thermal inertia is similar to or greater than that of typical regolith, while the thermal inertia probed throughout the entire night, indicative of the properties of the upper several centimeters, is lower than that of typical regolith. As a result, the best-fit model for South Ray suggests a vertical layering where the near surface has typical or greater density, and the subsurface has lower density when compared to typical regolith at the same depth ($${\rho }_{s}$$ = 1,183 kg/$${\mathrm{m}}^{3}$$ and $$H$$ = 11.2 cm).

We use the best-fit density profiles to estimate the average density within the upper 15 cm. South Ray cold spot has an average density of $${\sim} $$1,461 kg/$${\mathrm{m}}^{3}$$, slightly lower than typical regolith ($$\bar{\rho }=$$1,566 kg/$${\mathrm{m}}^{3}$$), while Bandfield cold spot requires a much lower density ($$\bar{\rho }=$$1,253 kg/$${\mathrm{m}}^{3}$$). These best-fit values were determined using the thermophysical properties of Hayne et al. (2017), but we find that the relationship between sites persists when using other published thermal model parameters (Figure S2 in Supporting Information S1) (Foote et al., 2020; Martinez & Siegler, 2021; Vasavada et al., 2012). It is important to note that our model assumes a relatively simple vertical density structure, and it is plausible that different or more complex density profiles may yield similar surface temperature trends. However, the inferred values reliably show relative differences in subsurface density between regions.

3.2 Correlation With Other Cold Spots

South Ray cold spot's comparatively smaller size and faint temperature anomaly can likely be explained by fading. We test this hypothesis by comparing the properties of South Ray cold spot to the properties of other cold spots with estimated ages. Williams et al. (2018) dated several large ($${ >} $$800 m) cold spots by counting smaller ($${\sim} $$10 m) craters superposed on their continuous ejecta and found ages ranging from $${\sim} $$220 ka to $${\sim} $$1.3 Ma. South Ray crater is estimated to be $${\sim} 2.08\pm $$0.17 Ma (Arvidson et al., 1975; Drozd et al., 1974; Eugster, 1999). Figure 2a shows the radially-binned median $${\Delta }{T}_{\mathit{BOL}}$$ (Powell et al., 2023) with distance for three example cold spots including South Ray cold spot. $${\Delta }{T}_{\mathit{BOL}}$$ typically peaks at around $${\sim} $$10–20 crater radii and drops off gradually with distance until reaching background levels at $${\sim} $$30–100 crater radii.

We characterize the strength of each cold spot as the peak value of their radially-binned $${\Delta }{T}_{\mathit{BOL}}$$ (Table S2 in Supporting Information S1). Figure 2b shows a strong correlation between peak $${\Delta }{T}_{\mathit{BOL}}$$ and age, with the most prominent cold spots being younger than $${\sim} $$500 ka. The trend for the Williams et al. (2018) dated cold spots is well described by a power-law fit. South Ray cold spot falls along the trailing end of this fit, indicating that its present-day temperature anomaly is consistent with it being an old, faded cold spot. In addition, this validates the crater count-derived model ages estimated by Williams et al. (2018) using cosmic ray exposure ages from Apollo 16 samples. This suggests that the production rate of small ($${\sim} $$10 m) craters over the last $${\sim} $$2 Ma is roughly in agreement with current models of recent crater production (Speyerer et al., 2016; Williams et al., 2014).

Figure 2c shows the extent of each cold spot versus age, where extent is defined as the distance at which the $${\Delta }{T}_{\mathit{BOL}}$$ falls within 0.5 K of background regolith. Cold spot size decreases with age from $${\sim} $$100 crater radii initially to $${\sim} $$20 crater radii for a cold spot the age of South Ray crater. This suggests that the present day South Ray cold spot is a small remnant of a once significantly larger and more prominent cold spot.

It is important to note that these trends may be specific to large cold spots craters. Williams et al. (2018) shows that the global size-frequency distribution (SFD) of cold spot craters is consistent with a retention age of $${\sim} $$150 ka when fit for $${\sim} $$100 m diameter craters. This may suggest that larger cold spots persist longer than smaller cold spots, possibly due to a difference in the initial magnitude of their thermophysical properties.

3.3 In Situ Regolith Properties

South Ray cold spot extends to the location of the Apollo 16 Lunar Module (LM) and Extravehicular Activity (EVA) traverses (Figure 3a). Therefore, the Apollo 16 samples and in situ experiments are representative of faint cold spot properties. Figure 3b shows the mean relative density of the upper $${\sim} $$15 cm of regolith inferred from astronaut footprint depths at each Apollo site, as reported by Mitchell et al. (1974) (Table S1 in Supporting Information S1). The mean relative densities at Apollo 11, 12, 14, 15, and 17 are very similar, while Apollo 16 is $${\sim} $$7 percentage points lower. This difference exceeds the standard error of any of the calculated mean values. A two-sample z-test comparing Apollo 16 to the other sites yields a p-value $$\ll $$0.05, confirming that the Apollo 16 regolith is statistically distinct. This result provides in situ evidence supporting the hypothesis of Bandfield et al. (2014) that the low thermal inertia of cold spots results from a decompaction of the upper several centimeters of regolith.

For comparison, the average densities derived from thermal modeling (Figures 1b and 1d) can be converted to equivalent relative densities using Equation 2. Assuming reasonable values for the minimum and maximum packing density ($${\rho }_{\min }$$ = 1,100 kg/$${\mathrm{m}}^{3}$$ and $${\rho }_{\max }$$ = 1,800 kg/$${\mathrm{m}}^{3}$$), our thermal model results predict relative densities, $${D}_{r}$$, of $${\sim} $$76.5% and $${\sim} $$63.5% for background regolith and South Ray cold spot, respectively. This corresponds to a difference of $${\sim} $$13 percentage points. This is a larger difference than we infer from footprint depths, which may result from uncertainties in thermal model parameters or the laboratory testing of footprint depths. However, both methods see a roughly similar decrease in relative density on the order of $${\sim} $$10 percentage points.

One complicating factor is that, unlike most Apollo sites, Apollo 16 landed in highlands terrain far from mare deposits (Spudis, 1984), so its lower relative density may be intrinsic to highlands material and not specific to South Ray cold spot. The Fra Mauro Formation explored by Apollo 14 is also located in the highlands. However, geologic mapping (Iqbal et al., 2023) and impact breccia sampling (Merle et al., 2014; Nemchin et al., 2009; Snape et al., 2016) suggest it is primarily composed of Imbrium ejecta, which may not be representative of typical highlands. The similarity in relative density between Apollo 14 and the other sites suggests near-surface regolith properties are broadly consistent across major terrain types, excluding cold spots and other local anomalies. Lunokhod 2 landed in the mare but traversed into nearby highlands and observed similar tire track depths, suggesting similar geotechnical properties (Basilevsky et al., 2021). Diviner nighttime temperatures between mare and highlands are also typically similar, implying comparable thermophysical properties (Bandfield et al., 2011; Hayne et al., 2017; Powell et al., 2023). In contrast, South Ray cold spot shows a distinct thermophysical signature compared to surrounding highlands terrain. We investigate this further by examining spatial variations in footprint depth relative to the cold spot's structure and extent.

Figure 3a shows the locations of several Apollo 16 research stations where footprint depths were measured (Mitchell et al., 1972). The Lunar Module (LM), Apollo Lunar Surface Experiments Package (ALSEP), and Station 4, 5, and 10 are comfortably within the cold spot region. Stations 1 and 8 are within the cold spot extent, but lie near boundaries with slightly warmer terrain. Station 11 is near the rim of North Ray crater, well outside the cold spot. Figure 3b shows relative densities at these locations (Mitchell et al., 1972) with distance from South Ray crater. The research stations within the cold spot region have low relative densities with a slight trend of increasing relative density with distance from South Ray crater. Stations 1 and 8, located near the cold spot boundary, have slightly higher relative densities than stations at similar distances but located clearly within the cold spot. This suggests that the Apollo footprint depths are spatially correlated with the observed cold spot thermophysical properties.

One exception to this trend is the Lunar Module (LM) location, which does not have a low relative density. This is likely due to modification by LM exhaust during descent. The photometric anomaly associated with the Apollo 16 blast zone extends $${\sim} $$100 m from the LM (Clegg & Jolliff, 2014; Clegg et al., 2012). The ALSEP site and station 10, located just outside the blast zone, both show lower relative densities than the LM. Additionally, Lunakhod 1 and 2 rover tire tracks were shallower near their landing platforms than farther away, which Basilevsky et al. (2021) attributed to modification by exhaust. We assert that the LM data point likely does not represent the pre-mission properties at that location.

The higher relative density of Station 11, the only location outside the cold spot extent, suggests that non-cold spot regolith in the Descartes highlands is similar to the other Apollo sites. Station 11 is located near the rim of North Ray crater, a region which appears warm in Diviner due to large boulders. While the presence of rocks does not necessarily indicate that the inter-rock regolith is anomalous, this raises uncertainty about whether Station 11 is representative of typical, non-cold spot terrain. North Ray crater has an estimated age of $${\sim} 50\pm $$1.4 Ma (Arvidson et al., 1975; Maurer et al., 1978), significantly older than South Ray crater. One interpretation of the $${\sim} $$Ma fading timescale of cold spots is that the density structure of the upper tens of centimeters returns to “normal” on roughly this timescale, suggesting that Station 11 has had sufficient time to achieve a typical highlands density. Future landed missions, such as Artemis, planned for the South Polar highlands, could provide these data. Overall, our retrospective investigation of Apollo 16 footprint depths indicates that the regolith is lower density than at the other Apollo sites, with low-density regions spatially well-correlated with the structure and extent of South Ray cold spot.