Thermal Spectroscopy Reveals Pervasive Deposits of Ground Ice in the Southern Polar Region of the Moon

Introduction

Lunar permanent shadowed regions (PSRs) may cold-trap and preserve volatile species for geologic time periods [1, 2]. In the past decades, surface and near-subsurface ice at concentrations of  was mapped on the Moon from orbit, as well as directly observed in the plume excavated by the LCROSS impact [3–7]. The absence of radar-bright features in PSRs further suggest that if any substantial ice persists on the Moon, then it is likely intimately mixed with the regolith [8].

Here, we hypothesize that if sufficiently concentrated subsurface ice persists within lunar cold traps, then it should increase the thermal conductivity of the subsurface – and decrease the amplitude of the diurnal thermal wave observed by Diviner.

Methods

Early observations of the Moon have found thermal emissions from the surface deviate significantly from Lambert scattering, and that this deviation can be well-explained by subpixel surface slopes casting shadows and emitting heat directionally. Under oblique illumination conditions, such as at the lunar poles, this aniostropic emission could increase the temperatures of PSRs through beaming [9, 10].

Previous thermophysical simulations of the Moon have neglected anisotropic scattering, and assumed the lunar surface behaves as a Lambert radiator [2]. This assumption resulted in a systematic underprediction of diurnal and annual maximum PSR temperatures compared to those recorded by Diviner (Figure 1). To address the disagreement found by those studies, we develop a thermophysical model which includes anisotropic (non-Lambert) surface emissions, and use it to re-calculate the surface heat balance in PSRs.

Our new model accepts generalized topography as input, and simulates insolation, scattering and thermal emissions between neighboring slopes using ray casting, as well as subsurface conduction [11]. However, instead of assuming each model facet is isothermal and thus radiates heat isotopically, we assume it is composed of subpixel surface slopes with some temperature distribution . To compute , we adapt a well-established statistical approach [9, 12] to any illumination and observation angles  and .  The model assumes subpixel surface slopes, whose directional components are , are distributed Gaussian, with zero mean and root mean square . This allows computing the radiance emitted by each surface element  as, where  is the Gaussian probability density function,  is the angle between the normal to each facet and the emission direction,  is the Stefan Boltzmann constant, and where both integrations are performed only over the part of the surface visible to the observer. Our anisotropic radiance model was tested and agrees with directional emissions measured by Diviner.

Results

We simulate the temperatures of 25 polar craters (15 southern, 10 northern) at peak southern polar summer, initially assuming radiative equilibrium, and find our new model effectively corrects the previously observed systematic underprediction for non-cold-traps (maximum temperatures > 100 K). However, for cold traps (maximum temperatures < 100 K), we find model-simulated maximum temperatures are warmer than Diviner-measured maximum temperatures, suggesting their thermal conductivities are elevated compared to non-cold-trapping surfaces (Figure 1). Using a 1-D thermal diffusion model with the simulated scattered flux as input, we fit the diurnal amplitude of the thermal wave in each PSR location to estimate its thermal conductivity and potential ice content. We find that the elevated model maximum temperatures are best explained by a volumetric mixture of ice and regolith, with up to 10%wt.

Figure 1.

Conclusion

Here, we apply a new thermal model which accounts for anisotropic emissions from the lunar surface, to resolve previous discrepancies [2] between model-simulated and Diviner-measured temperatures of permanently shadowed regions. We find that when assuming radiative equilibrium, model simulated maximum temperatures of cold-trapping PSRs (maximum measured temperatures < 100 K) are higher than measured temperatures, and hypothesize this disagreement is caused by the presence of ground ice, which increases the thermal conductivity of the surface and decreases the amplitude of the diurnal thermal wave. Using a 1-D thermal diffusion model, we find this decreased thermal wave amplitude is best explained by the presence of up to 10%wt ice, in agreement with previous observations (Figure 2).

Compared to exposed ice, buried ice is significantly more resistant to heating and other destructive mechanisms such as photolysis and impact gardening. As a result, our newly mapped deposits likely preserve water and other volatiles for much longer time periods – and thus offer insight into the historical composition of these substances through time.

Figure 2.

References:

[1]       Watson et al. (1961), J. Geophys. Res. 66
[2]       Paige et al. (2010), science 330
[3]       Hayne et al. (2015), Icarus 255
[4]       Fisher et al. (2017), Icarus 292
[5]       Li et al. (2018), Proc. Natl. Acad. Sci. 115
[6]       Mitrofanov et al. (2010), Science 330
[7]       Colaprete et al. (2010), Science 330
[8]       Campbell et al. (2006), Nature 443
[9]       Smith (1967), J. Geophys. Res. 72
[10]     Rozitis and Green (2011), Mon. Not. R. Astron. Soc. 415
[11]     Rubanenko and Aharonson (2017), Icarus 296
[12]     Bass and Fuks (2013), 93