Skip to main content
NCBI home page
NASA Author Manuscripts logoLink to NASA Author Manuscripts
. Author manuscript; available in PMC: 2018 Mar 15.
Published in final edited form as: J Geophys Res Planets. 2018 Feb 28;123(2):666–681. doi: 10.1002/2017JE005500

Investigating Mercury’s South Polar Deposits: Arecibo Radar Observations and High-resolution Determination of Illumination Conditions

Nancy L Chabot 1, Evangela E Shread 1, John K Harmon 2
PMCID: PMC5853133  NIHMSID: NIHMS945493  PMID: 29552436

Abstract

There is strong evidence that Mercury’s polar deposits are water ice hosted in permanently shadowed regions. In this study, we present new Arecibo radar observations of Mercury’s south pole, which reveal numerous radar-bright deposits and substantially increase the radar imaging coverage. We also use images from MESSENGER’s full mission to determine the illumination conditions of Mercury’s south polar region at the same spatial resolution as the north polar region, enabling comparisons between the two poles. The area of radar-bright deposits in Mercury’s south is roughly double that found in the north, consistent with the larger permanently shadowed area in the older, cratered terrain at the south relative to the younger smooth plains at the north. Radar-bright features are strongly associated with regions of permanent shadow at both poles, consistent with water ice being the dominant component of the deposits. However, both of Mercury’s polar regions show that roughly 50% of permanently shadowed regions lack radar-bright deposits, despite some of these locations having thermal environments that are conducive to the presence of water ice. The observed uneven distribution of water ice among Mercury’s polar cold traps may suggest that the source of Mercury’s water ice was not a steady, regular process but rather that the source was an episodic event, such as a recent, large impact on the innermost planet.

1 Introduction

In the early 1990s, Earth-based radar observations revealed the first evidence for water ice near Mercury’s poles (Slade et al., 1992; Harmon and Slade, 1992; Butler et al., 1993). Subsequent radar observations greatly increased the spatial resolution of the available radar datasets for both Mercury’s north and south polar regions (Harmon et al., 1994, 2001, 2011; Harcke, 2005; Harmon, 2007). The radar observations of Mercury showed numerous “radar-bright” features, areas with high radar reflectivity and also a high same-sense to opposite-sense circular polarization ratio, distinctive traits associated with ice on the Galilean satellites and the polar caps of Mars (Campbell et al., 1978; Ostro et al., 1980; Hapke, 1990; Hapke and Blewett, 1991; Muhleman et al., 1991; Harmon and Nolan, 2017). The radar-bright features showed a correlation with the interiors of craters, which were suggested to contain permanently shadowed regions with low enough temperatures to sustain the long-term stability of water ice (Paige et al., 1992; Ingersoll et al., 1992; Salvail and Fanale, 1994; Vasavada et al., 1999), though roughly half of Mercury’s surface remained unimaged after the Mariner 10 flybys of 1974–75.

In 2011, the MErcury Surface, Space ENvironment, GEochemistry, and Ranging (MESSENGER) spacecraft became the first to orbit Mercury, and during its just over four-year orbital mission, MESSENGER returned multiple new datasets to investigate Mercury’s radar-bright polar deposits. Images from the Mercury Dual Imaging System (MDIS) (Hawkins et al., 2007) and topographic measurements of Mercury’s north polar region by the Mercury Laser Altimeter (MLA) (Cavanaugh et al., 2007) were used to characterize the geologic setting of radar-bright host craters near Mercury’s polar regions as well as to identify regions of permanent shadow (Chabot et al., 2012; 2013a; Deutsch et al., 2016). Measurements by the Neutron Spectrometer (NS) (Goldsten et al., 2007) detected elevated levels of hydrogen in Mercury’s north polar region, consistent with the radar-bright deposits being composed largely of water ice (Lawrence et al., 2013). Surface reflectance observations by MLA found that radar-bright deposits could have both higher and lower reflectance values than the neighboring sunlit regolith, suggesting the presence of surface water ice as well as the presence of volatiles other than water ice (Neumann et al., 2013). Thermal models based on the MLA-derived topography concluded that water ice and low-reflectance carbon-rich volatiles were stable within permanently shadowed regions near Mercury’s north pole for billions of years (Paige et al., 2013). Direct imaging of the radar-bright deposits by MDIS also revealed both lower and higher reflectance surfaces for the deposits as well as sharp boundaries for the deposits, suggesting that the deposits were relatively young or recently refreshed (Chabot et al., 2014; 2016). Overall, MESSENGER observations provided compelling support in addition to the Earth-based radar observations of extensive water ice deposits within permanently shadowed regions near Mercury’s poles.

MESSENGER studies of Mercury’s polar deposits have largely focused on Mercury’s north polar region. The MESSENGER spacecraft’s highly eccentric orbit about Mercury resulted in the spacecraft passing at low altitudes over the north polar region but with an orbital apoapsis altitude of >10,000 km over the south polar region. At the higher altitudes over the southern region, MESSENGER’s MLA and NS instruments did not obtain any measurements. The resolution of MDIS imaging was substantially lower for the south polar region than for the north polar region, but this lower resolution enabled the MDIS Wide Angle Camera (WAC) to capture the entire south polar region in a single image frame. During MESSENGER’s first year in orbit about Mercury, repeated WAC imaging every other Earth day for an entire Mercury solar day (176 Earth days) produced a map of the illumination conditions of Mercury’s south polar region and identified areas of permanent shadow (Chabot et al., 2012). However, no additional analysis of MESSENGER data focused on Mercury’s south polar region have been conducted since this first study.

The Earth-based radar observations of Mercury’s south polar region have also been more limited than radar observations of Mercury’s north polar region. The highest-resolution radar observations of Mercury’s south pole, obtained by Arecibo in 2005 in S-band (12.6-cm wavelength) with a range resolution of 1.5 km, were limited to two contiguous nights of coverage (Harmon et al., 2011). In contrast, Arecibo radar observations of Mercury’s north pole include 21 nights of coverage obtained over 12 distinct date ranges from 1999–2005, providing radar observations over nearly the full range of longitude aspects (Harmon et al., 2011). Lower resolution Goldstone X-band (3.5-cm wavelength) radar observations of Mercury’s south pole have also been obtained, with a range resolution of 6 km and a viewing direction nearly opposite to that of the 2005 Arecibo observations (Harcke, 2005). All of the existing Arecibo and Goldstone radar-bright features identified in Mercury’s south polar region have been mapped to correspond to areas identified as being in permanent shadow from the MDIS WAC mapping (Chabot et al., 2012).

In this study, we present new results that focus on investigating the water ice deposits near Mercury’s south pole. In particular, we present results from new Arecibo radar observations of Mercury’s south pole, which greatly expand the existing high-resolution radar imaging coverage. We also utilize all available MDIS imaging of Mercury’s south polar region to constrain the illumination conditions and regions of permanent shadow at higher resolution than previous work. Together, these new results enable comparisons to Mercury’s north polar region at a scale and fidelity not previously possible, to better understand the distribution of radar-bright deposits at both of Mercury’s poles, with implications for the source of Mercury’s polar water ice.

2 Arecibo Radar Observations

New Arecibo radar observations were acquired of Mercury’s south polar region over a span of ten days, from March 16–25, 2012. Details of the radar observations are similar to the 2005 Arecibo observations, using the mid-1990s upgraded Arecibo telescope and obtained in 12.6-cm S-band (Harmon et al., 2011). All of the observations were made using binary-phase-coded transmission with a 10-µs baud (synthesized pulse width), which gave a range resolution of 1.5 km. Transmitter power was in the range of 790–846 kW. A circularly polarized wave was transmitted and the echo was received in both circulars. Observations used a long (240-1 element) code, which was effectively non-repeating and thus suitable for delay-Doppler imaging by the long-code method (Harmon, 2002; Harmon et al., 2011). An image was formed using the calibration procedure discussed in Harmon (2002) and by mapping from delay-Doppler space to planetary coordinates. The image pixels were normalized to a dimensionless radar reflectivity, which is the radar cross section per unit surface area. The image processing procedure was applied separately to the signals from the two received polarization channels.

During the ten days of the 2012 observations, the sub-Earth longitude ranged from 140–193°E and the sub-Earth latitude ranged from 8.0–8.4°S. The specific values were: March 16–18, 2012: 193°E, 8.4°S; March 21–22, 2012: 162°E, 8.4°S; March 24–25, 2012: 140°E, 8.0°S. For the previous Arecibo observations of March 24–25, 2005, the sub-Earth longitude was 11°E and the sub-Earth latitude was 7.5°S. Thus, the new 2012 Arecibo observations are highly complementary to the 2005 results as the viewing directions of the two sets of observations are nearly opposite to each other, greatly expanding the coverage of Mercury’s polar region, as shown in Fig.1. Both the 2005 and 2012 views shown in Fig.1 are weighted averages of the opposite-sense and same-sense polarization images, weighted by the noise variance. The viewing geometry of the 2005 observations resulted in a substantial portion of Mercury’s south polar region near 180°E longitude being beyond the radar horizon of the observations. The 2012 observations reveal many radar-bright features in this region and include all of the radar-bright features identified by Harcke (2005) using Goldstone 3.5-cm X-band observations with a 6-km range resolution. Additionally, for craters viewed by both the 2005 and 2012 observations, the opposite viewing directions resulted in opposite radar shadowing effects; thus, the two views aid in revealing the full extent of radar-bright features, as previously shown from multiple radar viewing geometries available for Mercury’s north polar region (Harmon et al., 2011).

Figure 1.

Figure 1

Open in a new tab

Arecibo radar reflectivity images of Mercury’s south polar region in polar stereographic projection obtained from observations in A) 2005 and B) 2012, produced from weighted averages of the opposite-sense and same-sense polarization images from each set of observations.

For this study, we chose to combine the 2005 and 2012 images shown in Fig.1 to enable comparisons to the image mosaic, permanently shadowed regions, and results from Mercury’s north pole. To produce the combined radar image of Mercury’s south polar region, we first applied a threshold of two noise standard deviations for each image (given in section S1 of the Supporting Information), nulling out pixels that were less than or equal to that value. We explored thresholding at levels from one to four standard deviations, as Harmon et al. (2011) applied both 3-σ and 4-σ thresholds to estimate the areal extent of radar-bright material at Mercury’s north polar region. Applying a 3-σ threshold level and higher resulted in a substantial reduction in the radar-bright signal on the floor of Chao Meng-Fu in the 2005 image, including for areas detected to be radar-bright in the 2012 image even after a 3-σ threshold was applied. Thus, we conservatively selected a 2-σ threshold level for both the 2005 and 2012 images, to preserve any faint, potential signals of radar-bright regions.

However, to further reduce the noise in the radar images, we applied a filter to null out isolated radar-bright clusters composed of three or fewer total pixels in each of the 2005 and 2012 images. Given a range resolution of the radar data of 1.5 km, the application of this filter effectively means that radar-bright regions smaller than ~9 km2 may have been removed, but the scattered noise throughout the images was also greatly diminished by this process. Even after the application of this filter, some areas near the radar horizon in the 2012 image remained noisy, as seen in Fig. 1B, due to these regions being beyond the radar horizon during some of the observing dates and thus being covered by fewer summed images. In preparation for producing one combined radar image of Mercury’s south polar region, we nulled pixels in the 2012 image that were displaying high noise levels due to being beyond the radar horizon during some of the observations.

Prior to combining the 2005 and 2012 images, each was translated slightly, in order to improve the registration to the MESSENGER image mosaic of Mercury’s south pole. The 2005 image was translated by 1 pixel (~1.5 km) along the 0° longitude direction, and the 2012 image was translated by roughly 2 pixels (~4.2 km) along the 135° E longitude direction. These small translations are generally consistent with the roughly 2 km position accuracy given by Harmon et al. (2011) for the 2005 Arecibo image. We chose to combine the images by overlaying the noise-processed 2005 image on the noise-processed 2012 image. The 2005 image more fully captured the higher radar signal along the topographic peaks and crater walls of portions of Chao Meng-Fu, and by placing these pixels over the 2012 image, essentially letting the 2012 image fill in the many null pixels in the 2005 image after our noise-processing procedures, these features were prominently retained in the combined image while also preserving all of the unique radar-bright deposits captured in the 2012 image. Figure 2 shows the combined radar image for Mercury’s south pole produced by these efforts, which we use for comparisons in this study going forward. In total, from this combined image we calculated that the area of Mercury’s south polar region that is radar-bright is 4.4% within 10° latitude of the south pole (~25,000 km2), as listed on Table 1. This is roughly twice the area that is radar-bright within 10° latitude of Mercury’s north pole (Harmon et al., 2011; Deutsch et al., 2016), consistent with the south polar region being composed of older, more cratered terrain, with more topographic features to produce shadows, than Mercury’s north polar region, which has large expanses of younger, smooth plains (Ostrach et al., 2015). Permanently shadowed regions near Mercury’s south pole are examined in detail in the next section.

Figure 2.

Figure 2

Open in a new tab

Radar reflectivity image, produced by combining the 2005 and 2012 images of Fig. 1, overlain on a MESSENGER MDIS mosaic of Mercury’s south polar region. The radar reflectivity values are weighted averages of the same-sense and opposite-sense circular polarization radar images, weighted by the noise variance.

Table 1.

Comparison of radar-bright and permanently shadowed regions near Mercury north and south poles.

80° – 90°
latitude		 85° – 90°
latitude
Mercury North Pole
Radar-bright area1 2.2% 8.4%
Permanently shadowed area2 3.7% 13.4%
Radar-bright deposits in shadow2 78.9% 75.4%
Shadowed terrain that is radar-bright2 46.4% 51.1%
Mercury South Pole
Radar-bright area3 4.4% 10.0%
Permanently shadowed area4 5.7% 13.2%
Radar-bright deposits in shadow5 77.3% 84.9%
Shadowed terrain that is radar-bright6 45.0% 50.2%
Open in a new tab
1

Deutsch et al. (2016), with 4σ threshold applied.

2

Deutsch et al. (2016), determined from MDIS images.

3

Section 2, using noise reduction processing as discussed in the text.

4

Section 3, estimated uncertainty due to thresholding approach is ~20%.

5

Section 4, in or within 1.5 km of a permanently shadowed region.

6

Section 4, as shown in Fig. 7d.

3 Determination of Illumination Conditions

Using WAC images from MESSENGER’s first year in orbit about Mercury, Chabot et al. (2012) determined the illumination conditions of Mercury’s south polar region at a spatial scale of 1.7 km/pixel. In this work, we use similar methods to the study of Chabot et al. (2012) but we utilize MDIS Narrow Angle Camera (NAC) images acquired from MESSENGER’s full orbital mission of just over four years. Images obtained by the NAC have a factor of seven higher spatial resolution than the WAC (Hawkins et al., 2007), enabling a determination of illumination conditions at a scale of 200 m/pixel in this study.

Another key difference between this study and the WAC-based work of Chabot et al. (2012) was that the WAC images used by Chabot et al. (2012) were acquired as part of a dedicated campaign to determine the illumination conditions of Mercury’s south polar region. The WAC images were acquired every other Earth day for a complete Mercury solar day, and this regular imaging cadence captured the changing illumination conditions at Mercury’s polar region. A similar imaging campaign for the NAC was also performed during MESSENGER’s first year in orbit about Mercury (Denevi et al., 2017), with a set of six NAC images acquired every other Earth day for a full Mercury solar day. Due to the NAC’s higher resolution, the set of six NAC images did not cover the full sunlit portion of Mercury’s south polar region but rather imaged a strip of the surface for the longitude near local noon. The sets of six NAC images also had some gaps when mosaicked together, providing incomplete surface coverage even near the subsolar longitude region. Thus, in this study we chose to not limit the images to those acquired during just one MDIS imaging campaign but rather to examine all NAC images acquired during MESSENGER’s entire mission. To determine the overall illumination conditions, we divided the NAC images of Mercury’s south polar region into two broad categories that were based on their exposure times: standard exposure NAC images that were used for mapping the morphology of Mercury’s surface and long-exposure NAC images that resulted in images with a high amount of saturation.

3.1 Standard Exposure NAC Images

We explored determining the south polar illumination conditions with NAC images from the south pole to a latitude of 70°S, but we found insufficient NAC image coverage under the full range of illumination conditions in a Mercury solar day for latitudes equatorward of 80°S. Consequently, we limited our study area to within 10° latitude of Mercury’s south pole, an area which contains numerous radar-bright deposits, as shown in Fig. 2. To include all NAC images that could contribute to mapping the illumination conditions in this area, we included NAC images centered at 77°S and poleward in our dataset. Additionally, for this illumination mapping effort, we limited images to have “standard” exposure durations of <40 ms, as images with longer exposure times were purposefully acquired to have many saturated pixels. We also limited images to have emission angles <60°, as images with higher emission angles did not retain the spatial resolution fidelity to view small scale surface features. In total, we identified 1,090 NAC images meeting these criteria. Each NAC image was calibrated (Denevi et al., 2017) and then orthorectified to a polar stereographic projection with a pixel scale of 200 meters using the MESSENGER MDIS-based global Digital Elevation Model (DEM) and associated smithed kernels (Becker et al., 2016). The global DEM and smithed kernels are available in the MESSENGER PDS archive, and the United States Geological Survey (USGS) Integrated Software for Imagers and Spectrometers (ISIS) software package was used for all of the procedures. We chose not to apply a photometric correction to the images, as the correction is not optimized for the high incidence angles experienced within 10° of the pole (Domingue et al., 2015) and surface coverage near the pole would have been lost in some images if such a correction had been applied.

To investigate the variations in shadowing on the surface throughout the Mercury solar day, we next produced mosaics from NAC images that were grouped according to their subsolar longitude. The previous WAC study of illumination conditions at Mercury’s south pole was based on 89 images covering a full Mercury solar day, resulting in an average sampling of roughly every 4° change in subsolar longitude (Chabot et al., 2012). Applying 4° subsolar longitude bins to the 1,090 NAC images produced incomplete coverage for all subsolar longitude groupings for a full Mercury solar day. Experimenting with the dataset, we found that 7° subsolar longitude bins were the smallest bin size that the NAC dataset could support. This resulted in 50 mosaics grouped into 7° subsolar longitude bins and 1 mosaic grouped into a 10° subsolar longitude bin, which we selected to cover the range of 283.5°–293.5° E as this was one of the least covered areas in the NAC image set. This set of 51 mosaics is comparable in number to the 53 and 59 images used to determine the illumination conditions near the lunar north (Bussey et al., 2005) and south (Bussey et al., 1999) poles respectively.

The resulting 51 subsolar longitude mosaics showed minor mis-registration effects between individual images within a mosaic and also between mosaics. To improve the registration, corrections were applied on an individual, image-by-image basis by translating and, in some cases, warping images to match the larger mosaic or neighboring mosaics, using the “coreg” routine in ISIS. In the most extreme corrections, images were warped or translated by as many as four kilometers at a single control point. Given that a photometric correction was not applied to the images, certain images within mosaics were also noticeably brighter or darker than adjacent images, and these were adjusted using the ISIS “equalizer” routine, to produce a set of 51 mosaics with more uniform brightness. Lastly, unimaged pixels in each mosaic that were located more than 90° of longitude from the subsolar longitude were set to zero, as on the nighttime side of the planet the surface would be in shadow. A movie of the 51 mosaics resulting from this full process is given in the Supporting Information. Averaging the 51 mosaics together produced the south pole basemap mosaic shown in Fig. 3a.

Figure 3.

Figure 3

Open in a new tab

Illumination conditions at Mercury’s south polar region determined from MDIS NAC images. a) Average mosaic produced from the 51 subsolar longitude mosaics used in this study. b) Coverage of the 51 subsolar longitude mosaics for one Mercury solar day. Coverage percentages of 85%, 90%, 95%, and 100% correspond, respectively, to 100%, 85%, 36%, and 11% of the area between 80°–90°S. c) Percent illumination map, produced at 200 m/pixel, shows the percentage of a Mercury solar day for which a given surface is sunlit.

Figure 3b shows the percentage of a Mercury solar day covered by the set of 51 mosaics, taking into account Mercury’s 3:2 spin-orbit resonance and weighting each subsolar longitude mosaic accordingly. In other words, the coverage of each mosaic was weighted according to the percentage of time during a Mercury solar day when the subsolar longitude falls within the corresponding 7° (or 10°) bin. The entire region of 80°S to the pole is covered for ≥85% of a Mercury solar day by the set of 51 mosaics, as shown in Fig. 3b.

To separate sunlit features from shadowed ones, a threshold was applied to each mosaic, producing binary images in which sunlit pixels were set to a value of 1 and shadowed pixels to a value of 0. A unique threshold was defined for each mosaic in the region from 86°S poleward, to provide the best representation of the variable lighting conditions at the floor of Chao Meng-Fu, where grazing sunlight occasionally illuminates the crater’s central peaks and floor. However, if the lower thresholds that exposed the low-intensity light at the floor of Chao Meng-Fu were applied to further equatorward latitudes, we found that these threshold values would underestimate the distribution of shadow in those areas. For this reason, we applied a higher threshold value to the region from 80°S to 86°S. We also found that we could apply the same threshold value to all 51 of the mosaics for regions equatorward of 86°S, as the illumination conditions are comparatively less variable for these areas than they are within Chao Meng-Fu and as the tone brightness adjustment that was applied to the images prior to mosaicking largely corrected for brightness variations between the mosaics. Different threshold values were explored and from defined extremes, it was determined that the extent of shadowed regions varied by ~20%, similar to the results reported in previous MDIS-based studies of Mercury’s polar shadowed regions (Chabot et al., 2012; Chabot et al., 2013a; Deutsch et al., 2016). No changes in the general locations of the permanently shadowed regions were noted with varying threshold levels.

Each threshold mosaic was weighted according to the percentage of the solar day it covered, as was done to determine the image coverage in Fig. 3b. The mosaics were added together to give the weighted sum, yielding the percent illumination of the surface at a pixel scale of 200 m for regions covered in all 51 mosaics. However, additional calculations were needed to address the gaps in coverage shown in Fig. 3b. Since pixels with a value of 0 had been added onto the night side of the terminator in each mosaic, the missing coverage in the dataset was entirely within areas on the dayside. To fill these gaps in coverage, the assumption was made that the missing coverage had on average the same illumination conditions as the existing dayside results. We feel this approximation is appropriate, given the relatively minor adjustment that it introduces to compensate for gaps that are a small fraction (<15%) of a Mercury solar day (Fig. 3b). Figure 3c shows the resulting percent illumination map produced for Mercury’s south polar region.

A qualitative comparison of the new NAC-based illumination map to the lower resolution WAC-based illumination map from Chabot et al. (2012) showed overall highly consistent results; a more quantitative comparison is discussed in Section 3.3. This agreement supports that our approach for the longitudinal bins, the weighting of the mosaics, and the gap filling were appropriate and produced results consistent with those from a set of images that were acquired regularly over one full Mercury solar day. Large permanently shadowed regions identified from the Chabot et al. (2012) WAC-based study are also present in our new NAC-based results of Fig. 3. While the new NAC illumination map is consistent with the previous WAC results, the higher resolution of the new map resulted in small-scale shadowed regions being better defined, particularly in the shadowed, rough terrain within 5° of the south pole.

The higher resolution NAC-based illumination map also enables a more detailed look at the small regions near Mercury’s south pole that are sunlit for the highest percentage of a Mercury solar day. Figure 4 identifies the top five locations, with illumination percentages of 67–74% of a Mercury solar day, which are found along the rim and wall terraces of Chao Meng-Fu as well as near the rims of two other south polar craters, Lovecraft and L’Engle. No location in Mercury’s south polar region is mapped as being permanently sunlit at the 200-m pixel scale of the NAC-based illumination map. Studies of the illumination conditions at the Moon’s poles also have not discovered any permanently illuminated locations but have identified single areas illuminated for comparable fractions of the lunar year (Speyerer and Robinson, 2013) to those shown in Fig. 4. Speyerer and Robinson (2013) also demonstrated the critical influence of spatial resolution in determining the illumination conditions of small regions. For a small area, illuminated surfaces may go undetected if they are considerably below the resolution of the images or simulations, and similarly, higher resolution data may actually result in a lower percent illumination calculation for each surface when it is resolved into more pixels. Given that the highest percent illuminated areas in Fig. 4 are only a few pixels in size, interpretation of the illumination conditions for these small regions should be done with awareness of the limitations of the 200-m pixel scale of the NAC data used in this study.

Figure 4.

Figure 4

Open in a new tab

Locations illuminated for the highest percentage of a Mercury solar day based on the 200 m/pixel illumination map of Fig. 3c. The five highest locations identified in this figure with blue arrows are: 74% (88.1°S, 277.0°E), 73% (88.8°S, 286.6°E), 72% (87.5°S, 70.7°E), 69% (87.9°S, 134.9°E), 67% (86.7°S, 277.9°E).

3.2 Long-Exposure NAC Images

To further investigate the permanently shadowed regions at Mercury’s south pole, we also examined MDIS images purposefully acquired with high levels of saturation. For Mercury’s north pole, images acquired with the WAC broadband filter resulted in sunlit surfaces being saturated while the features of permanently shadowed surfaces were successfully revealed, illuminated by low levels of sunlight scattered off nearby crater walls and other topographic features (Chabot et al., 2014; 2016). Similar WAC broadband filter images were acquired of Mercury’s south pole. However, MESSENGER’s highly elliptical orbit resulted in WAC broadband images that covered extensive areas of Mercury’s south polar region at pixel scales >1.5 km, in contrast to the WAC broadband images obtained for Mercury’s north polar region that focused on specific craters of interest with pixel scales down to 24 m (Chabot et al., 2016). Examination of the WAC broadband images of Mercury’s south polar region did not reveal any features within permanently shadowed regions.

In addition to the WAC broadband images, a number of NAC images with long exposure times (250–9989 ms) were acquired, focusing on the large permanently shadowed interior of Chao Meng-Fu. Long-exposure NAC imaging in Mercury’s north polar region had also been attempted, but the longer exposure times resulted in considerable smear due to the spacecraft motion, and no features within permanently shadowed regions were successfully identified (Chabot et al., 2013b). Due to MESSENGER’s highly elliptical orbit, similar image smear was not an issue for long-exposure NAC images of Mercury’s south polar region. In total, 67 NAC images with exposure times ≥250 ms were acquired of Mercury’s south polar region, and we examined each of these images individually.

All of the long-exposure NAC images contained high levels of saturation for areas of sunlit surfaces, but by adjusting the brightness and contrast stretch, some of the images revealed details within the floor of Chao Meng-Fu that were not visible with the previous set of NAC images with standard exposure settings. Figure 5 provides a comparison of NAC standard and long-exposure imaging acquired at the same subsolar longitude. Dimly sunlit areas on the floor of Chao Meng-Fu can appear shadowed in NAC images with standard exposure settings (Fig. 5a), though applying a harsh stretch to such an image can reveal some dimly lit areas and long shadows cast by the crater’s central peaks (Fig. 5b). However, application of this stretch to the full image results in other shadowed regions being no longer discernable. A long-exposure (4096 ms) NAC image results in the sunlit surface being highly saturated, and smaller shadowed regions cannot be identified in these locations, but the shadows cast on the floor of Chao Meng-Fu are crisp and clearly distinguished from portions of the lit floor (Fig. 5c). In all of these images, the shadows from the peaks within Chao Meng-Fu are consistent with the subsolar longitude of the images, which suggests that they were produced by direct illumination from the Sun. Shadows from small-scale features on the dimly lit floor of Chao Meng-Fu are also consistent with this direction and with being cast by direct solar illumination. Further stretching of the long-exposure image in Fig. 5c does not reveal any detail within the large shadowed region centered near 180°E. After examination of the full set of 67 long-exposure NAC images, we concluded that the images do not reveal any surface features within permanently shadowed regions but rather such images reveal areas of Chao Meng-Fu’s floor that receive low levels of direct sunlight. Thus, unlike images with high levels of saturation of Mercury’s north polar deposits (Chabot et al., 2014; 2016), these long-exposure NAC images do not provide any constraints on whether ice is exposed on the surface in any of Mercury’s south polar craters or if the radar-bright regions are covered by low-reflectance surfaces.

Figure 5.

Figure 5

Open in a new tab

NAC long-exposure images and revised permanently shadowed region in Chao Meng-Fu. A) The mosaic of standard-exposure NAC images produced from the subsolar longitudinal bin centered at 210°E is shown for the region near Mercury’s south pole, where the 180-km-diameter crater Chao Meng-Fu is the dominant feature. B) The same mosaic as shown in part A. but with the brightness and contrast stretched to display details of the dimly sunlit crater floor. C) A long exposure (4096 ms) NAC image (EN1054440143M) acquired at essentially the same subsolar longitude (209.8°E) as the mosaic shown in parts A. and B. The long-exposure results in sunlit surfaces that are highly saturated but regions of the dimly sunlit floor can be easily identified. Two other long exposure (4096 ms) NAC images acquired during different times of the Mercury solar day reveal different portions of Chao Meng-Fu’s dimly lit floor: D) image (EN1070068490M) at 197.5°E subsolar longitude; E) image (EN1071898378M) at 180.3°E subsolar longitude. F) The region of permanent shadow within Chao Meng-Fu is decreased by 28% when the long exposure NAC images are considered (red) rather than only standard exposure NAC images (yellow).

While it is disappointing that this imaging campaign was unable to reveal details of the permanently shadowed surfaces within Chao Meng-Fu, such images are still useful to more accurately distinguish dimly sunlit and permanently shadowed regions within this crater. Figure 5d and 5e show two additional examples of long-exposure (4096 ms) NAC images acquired at different times throughout the Mercury solar day. From the full set of long-exposure NAC images, we identified four images that provided the best, unique views within Chao Meng-Fu, and these images were thresholded into shadowed and sunlit binary images. The permanently shadowed regions determined from the illumination map in Fig. 3c were compared to these thresholded long-exposure NAC images, and the area mapped as permanently shadowed was refined to only include locations not shown to be dimly sunlit by the long-exposure NAC images. The region mapped as permanently shadowed within Chao Meng-Fu was reduced by 28% as a result of this analysis, as shown in Fig. 5f. This reduced area of the permanently shadowed region in Chao Meng-Fu corresponds strongly to the radar-bright area in this crater, which is shown and discussed in Section 4.

Our conclusion also implies that determining permanently shadowed regions using only standard exposure MDIS images may not provide the sensitivity needed to distinguish between regions dimly lit by low levels of direct sunlight and regions that receive no direct solar illumination. Dimly sunlit regions are likely to be most common and cover the largest area near Mercury’s poles, where Mercury’s low obliquity of 2.04 arc minutes (0.034°) (Margot et al., 2012) results in highly grazing incidence angles throughout the entire Mercury solar day. Thus, for Mercury’s south pole, our modification to the permanently shadowed region within Chao Meng-Fu (Fig. 5f) captures the largest area affected in Mercury’s south polar region. For Mercury’s north polar region, the locations of permanently shadowed regions show good agreement when mapped by either MDIS images or DEM-based modeling (Deutsch et al., 2016), suggesting that there is not a significant impact on the mapping of permanently shadowed regions due to the sensitivity of MDIS to dimly lit areas. However, when specific permanently shadowed regions are examined in detail, the limitation of standard exposure MDIS images to distinguish dimly sunlit surfaces needs to be fully considered, and the extent of the permanently shadowed region associated with the north polar crater Prokofiev (Deutsch et al., 2016) may need to be re-examined given this new finding.

3.3 Combined Illumination Results from All NAC Images

Overall, using our results from both the standard and long exposure NAC images, we estimate that 5.7% of Mercury’s surface within 10° latitude of the south pole is permanently shadowed, and within 5°, that percentage increases to 13.2%, as listed on Table 1. The percentages determined from the lower-resolution WAC study were 7% and 20% for the permanently shadowed area within 10° and 5° of the pole, respectively (Chabot et al., 2012), which are similar but slightly higher than we find in this work, particularly for the region within 5° of the pole that contains Chao Meng-Fu. The higher resolution NAC images allowed small-scale shadowed regions to be better defined and the long-exposure NAC images enabled a detailed look at the shadowing within Chao Meng-Fu, and for these reasons, we conclude that these new NAC results are the current best estimate of the area permanently shadowed in Mercury’s south polar region.

4 Discussion and Implications

Along with providing new results for Mercury’s south polar region, one of the most important contributions of this study is its ability to enable more direct comparisons between the distribution of radar-bright and permanently shadowed regions at both of Mercury’s poles. The study by Deutsch et al. (2016) used the full set of MDIS orbital data to determine permanently shadowed regions near Mercury’s north pole at a pixel scale of 200 m, the same pixel scale used in this study. Similarly, Deutsch et al. (2016) concentrated on investigating the region within 10° latitude of the pole, just as we have done here. Deutsch et al. (2016) also determined permanently shadowed regions by illumination modeling using a DEM derived from MLA measurements, finding very similar results from both the MDIS- and MLA-based approaches. To keep our comparisons as similar as possible, we use the MDIS-based results from Deutsch et al. (2016) for Mercury’s north pole throughout the rest of this study.

Figure 6 shows radar-bright and permanently shadowed regions for both Mercury’s north and south poles. The 180-km-diameter Chao Meng-Fu is the largest crater to host radar-bright deposits in either of the polar regions. The radar coverage and permanently shadowed regions in Fig. 6a of Mercury’s north pole are as discussed in Deutsch et al. (2016). For Mercury’s south polar region in Fig. 6b, the combined Arecibo radar image was used (Section 2, Fig. 2) and the permanently shadowed regions were determined from the combined analysis of both standard- and long-exposure NAC images (Section 3). Mercury’s south polar region has a somewhat larger area that is permanently shadowed within 10° of the pole than the north polar region, 5.7% vs. 3.7% (Table 1). This is consistent with the south polar region consisting of an older, more cratered terrain, which results in more craters and topographic features that create permanently shadowed regions, relative to the younger expansive region of smooth plains located in Mercury’s north (Ostrach et al., 2015). Within 5° of the pole, the permanently shadowed regions are very similar in area for both the north and south polar regions, 13.4% and 13.2% of the area respectively, despite Chao Meng-Fu being a dominant feature near the south pole. In comparison, within ~10° of the lunar poles, the fractional areas in permanent shadow are 4.7% and 5.8% for the lunar north and south poles, respectively (Mazarico et al., 2011), which are similar to the percentages for Mercury. However, within 5° of the lunar poles, the permanently shadowed areas are 6.6% and 9.7% for the north and south, respectively (Mazarico et al., 2011), which are lower than the ~13% area estimated for both of Mercury’s polar regions. The slightly larger percentage of Mercury’s polar regions in permanent shadow within 5° of the pole may be due to Mercury’s smaller spin axis tilt relative to that of the Moon (Margot et al., 2012; Siegler et al., 2011).

Figure 6.

Figure 6

Open in a new tab

Distribution of permanently shadowed and radar-bright regions for Mercury’s A) north and B) south polar regions. Data for Mercury’s north polar region are from Deutsch et al. (2016) and for Mercury’s south polar region are from this study.

Qualitatively, Fig. 6 shows that for both of Mercury’s polar regions, there is strong agreement between the locations of radar-bright deposits and permanently shadowed regions. Quantitatively, Deutsch et al. (2016) found that 78.9% of the radar-bright features within 10° of Mercury’s north pole mapped directly to regions of permanent shadow, with the remainder found largely either within 5 km of large areas mapped as shadowed or inside small craters below the size limitation of the mapping approach. In this study for Mercury’s south polar region, we find that 58% of radar-bright pixels correspond exactly to pixels mapped as permanently shadowed, for the region of 80°–90°S. However, this percentage increases dramatically to 77% and to 86% for radar-bright areas located in or within one or two pixels, respectively, of a permanently shadowed region, using the 1.5-km pixel scale of the Arecibo data. The remaining 14% of the radar-bright area is located within Chao Meng-Fu or in rough terrain, likely reflecting permanently shadowed regions below the scale captured by our shadow mapping approach. In Table 1, we report the percentage of radar-bright deposits in shadow in Mercury’s south polar region applying the criteria that they be located in or within 1.5 km of a permanently shadowed region, and these results are similar to those for Mercury’s north polar region (Deutsch et al., 2016).

If Mercury’s radar-bright deposits are dominantly composed of water ice, nearly 100% of the radar-bright features should be located within regions of permanent shadow, to experience thermal conditions conducive to the survival of the water ice (Vasavada et al., 1999; Paige et al., 2013). Thus, the results for Mercury’s north and south poles that report only ~80% agreement between permanently shadowed and radar-bright regions would be a challenge to the water ice hypothesis if the absence of total agreement implied by those percentages reflected the amount of radar-bright deposits that are not located in regions of permanent shadow. Figure 7 shows the locations of radar-bright features that do not correspond directly to regions mapped as permanently shadowed for both Mercury’s north and south poles. As seen in Fig. 7, these “radar-bright only” regions are not located as large clusters but rather appear to either border regions mapped as permanently shadowed, correspond to small topographic features, or represent noise in the radar images. Thus, we interpret these “radar-bright only” regions shown in Fig. 7a and 7c to reflect a limitation of the mapping and radar data rather than actual occurrences of radar-bright features being located outside of permanently shadowed regions on Mercury. Overall, we conclude that the distribution of radar-bright features near both of Mercury’s poles is consistent with being located within permanently shadowed regions, as would be necessary for the long-lived thermal stability of water ice deposits.

Figure 7.

Figure 7

Open in a new tab

Distribution of A) radar-bright features near Mercury’s north pole that do not correspond directly to regions mapped as permanently shadowed, and B) permanently shadowed regions near Mercury’s north pole that do not also show radar-bright features. The same figures are shown for Mercury’s south polar region in C) and D). Data for Mercury’s north polar region are from Deutsch et al. (2016) and for Mercury’s south polar region are from this study.

While radar-bright features are highly consistent with being located in permanently shadowed regions, only ~45% of the permanently shadowed regions within 10° of either Mercury’s north (Deutsch et al., 2016) or south poles also host radar-bright deposits (Table 1). Figure 7 shows the locations of permanently shadowed regions that lack radar-bright features. Unlike for the “radar-bright only” regions, “permanent shadow only” regions do appear as highly concentrated areas that do not resemble the patterns expected from mis-registration of the mapping approach, small-scale surface features, or noise within the datasets. This lack of radar-bright deposits in all permanently shadowed regions raises the question of whether all of Mercury’s cold traps are, or are not, occupied with water ice.

In particular, the north polar region has a number of craters near 225°E longitude that host sizable regions of permanent shadow that lack extensive radar-bright deposits (Fig. 7B), as noted and discussed in detail by Deutsch et al. (2016). Deutsch et al. (2016) considered a number of scenarios for these “permanent shadow only” craters: deposits with unusually thick insulating layers of low-reflectance volatiles that mask the water ice from detection by radar observations; limits in the sensitivity or visibility of the radar observations in this region of the north pole; differences in the geologic terrain for the “permanent shadow only” craters versus the rest of the polar craters; differences in the thermal environments within the permanently shadowed regions; and a lack of water ice sufficient for detection by the radar observations in these permanently shadowed regions. Deutsch et al. (2016) found no evidence for the geologic unit or age of the craters to explain the lack of radar-bright signal in these locations, and the thermal environments in the “permanent shadow only” areas are similar to areas that host radar-bright deposits in Mercury’s north polar region (Paige et al., 2013). Modeling of the visibility of the many Arecibo observations that were used to construct the north pole radar image (Harmon et al., 2011) led Deutsch et al. (2016) to conclude that the radar observations were well positioned to detect water ice in these craters near 225°E if water ice was present. The lack of water ice deposits for craters near a longitude of 225°E implies an asymmetrical distribution of deposits that is not directly correlated with Mercury’s hot-pole (0°E, 180°E) or cold-pole (90°E, 270°E) longitudes (Vasavada et al., 1999). If the permanently shadowed regions near 225°E longitude in Mercury’s north polar region do contain unusually thick insulating layers or lack water ice deposits, this spatial distribution pattern is an important clue to the source and evolution of Mercury’s polar water ice deposits.

Our new results for Mercury’s south pole show that the largest “permanent shadow only” areas are gathered in the 0°E longitude direction (Fig. 7D), which is nearly antipodal to 225°E in the north polar region. Similar to mapping done by Deutsch et al. (2016) for Mercury’s north pole, we mapped 388 permanently shadowed craters in Mercury’s south polar region between 80–90°S, down to a diameter of 6 km, which are tabulated and shown in the Supporting Information. Of craters with diameters >10 km, roughly the simple to complex transition diameter for Mercury’s craters (Susorney et al., 2016), nearly 70% host radar-bright features while only about 40% of simple craters with diameters <10 km host radar-bright deposits. These percentages are similar to those found for Mercury’s north polar region (Deutsch et al., 2016). This result could support the conclusion that bowl-shaped simple craters present more challenging thermal environments for the long-term stability of water ice (Vasavada et al., 1999). However, our data processing approach also limits the investigation of these smaller craters, given the 1.5-km spatial scale of the radar data and our data processing procedure of nulling out isolated radar-bright clusters composed of three or fewer total pixels to reduce the noise in the radar images. The crater mapping data show that fewer permanently shadowed craters host radar-bright deposits near 0°E longitude, such as seen in Fig. 7D, and that a similar reduction in radar-bright host craters is not seen near the other hot-pole longitude of 180°E. As such, this distribution is not simply a result of different long-term thermal stability environments available between Mercury’s hot-pole and cold-pole longitudes. Unlike Mercury’s north polar region, there are no MESSENGER-based thermal modeling results for Mercury’s south polar region, and thus the thermal environment in the “permanent shadow only” areas of Fig. 7D cannot be quantitatively evaluated. Qualitatively, there are permanently shadowed craters of similar sizes and latitudes to those that lack radar-bright deposits near 0°E longitude that do host radar-bright deposits at other longitudes, suggesting that the thermal environment differences cannot explain the lack of radar-bright features near 0°E.

The new Arecibo radar observations presented in this work (Section 2) greatly increase the coverage of Mercury’s south polar region. Nevertheless, the radar data for Mercury’s south pole are still only from essentially two viewing geometries (Fig. 1) and are much more limited than the 12 distinct sets of observations from 1999–2005 that were used to identify radar-bright features in Mercury’s north polar region (Harmon et al., 2011). Are limitations in the visibility or sensitivity of the south pole radar observations a possible explanation for the “permanent shadow only” regions shown in Fig. 7D, in that these regions may host water ice that was not detected in the radar observations? For Mercury’s north polar region, Deutsch et al. (2016) used Mercury’s topography to model the visibility within permanently shadowed regions for each of Arecibo’s observations and concluded that radar visibility was not the explanation for the regions that lacked radar-bright features. Similar modeling for Mercury’s south polar region would be worthwhile but the topographic information available for Mercury’s south polar region is much more limited than that for the north, given that MESSENGER’s altitude was too high to acquire MLA measurements of the south polar region. Using >100,000 MDIS images as input, an image-based global DEM of Mercury was produced (Becker et al., 2016), but unlike MLA measurements, the MDIS images used in the global DEM provided no constraints on the topography within permanently shadowed regions. Future work to adapt the MDIS-based global DEM to model the radar visibility and thermal environments of Mercury’s south polar permanently shadowed regions has strong motivation based on our new results.

While we have not quantitatively modeled the radar visibility for the 2005 and 2012 south polar Arecibo observation dates, we do note some qualitative traits that support the “permanent shadow only” regions in Mercury’s south polar region as being due to a lack of extensive water ice deposits in these locations. First, both the 2005 and 2012 radar observations consistently show a dearth of radar-bright features near 0° E longitude for sizable craters between 80–85°S (Figs. 6B, 7D), which is a latitude range covered by both the 2005 and 2012 Arecibo observations (Fig. 1). The 2005 and 2012 radar observations are from nearly opposite viewing directions, providing highly complementary opportunities to detect any radar-bright features if such features were present. Second, in notable contrast, the region near 180° E longitude shows that nearly all of the sizable permanently shadowed regions between 80–85°S have associated radar-bright features (Figs. 6B, 7D). Lastly, while the radar horizon limits the surface covered in each radar image, many radar-bright features are still definitively detected that are located near the radar horizons of the 2005 and 2012 observations (Fig. 1). For example, radar-bright features near 81°S, 155°E in the 2005 radar image (Fig. 1A) and 77°S, 290°E in the 2012 radar image (Fig. 1B) are strongly detected and more closely located to the radar horizon than the extensive region near 0°E where the “permanent shadow only” features are found (Fig. 6B, 7D). Thus, these observations support that the lack of radar-bright features near 0°E longitude for Mercury’s south polar region are not caused by a limitation of the radar observations but rather reflect the uneven distribution of water ice in Mercury’s south polar region. These “permanent shadow only” regions may lack water ice completely or may have unusually thick (>2 m to attenuate an S-band return by >80%, Harmon et al., 2001) insulating layers that hide any water ice beneath them from detection by the radar observations.

What are the implications of the observed distribution of permanently shadowed regions that appear to be largely devoid of water ice in Mercury’s north (Fig. 7b) and south (Fig. 7d) polar regions? A non-uniform distribution of water ice that is not tied to the age of the surface or the thermal environment of the available cold traps could be inconsistent with a steady process as the source of the water ice. Possible steady, regular sources of water on airless rocky bodies in the Solar System include solar wind interactions, planetary outgassing, and delivery via micrometeorites. Such steady processes could be expected to occupy all viable cold traps and an uneven distribution of water ice in Mercury’s polar regions could be difficult to explain by such processes. In contrast, models of cometary impacts on the Moon have shown that while a single large impact event can deliver water to both polar regions, the resulting distribution of the water among the available cold traps can be uneven, depending on the specific impact conditions and transport timescales (Stewart et al., 2011; Prem et al., 2015). Thus, delivery of Mercury’s water ice by a large impact event could potentially result in an uneven distribution of water ice in Mercury’s polar cold traps, though no models similar to those conducted for the Moon have been run for an impact on Mercury. An episodic delivery of water ice to Mercury, such as a large impact event, may also be more consistent with the high purity (<5% silicates by volume) inferred for Mercury’s water ice deposits from early radar observations (Butler et al., 1993); revisiting the constraint on the purity of Mercury’s polar water ice deposits, given the increased spatial resolution of the radar datasets and the new insights provided by MESSENGER observations, would be highly worthwhile. Additionally, a recent, large impact event could simultaneously deliver water ice as well as other volatiles, such as complex organic compounds that have been suggested to produce the low-reflectance surfaces of many of Mercury’s polar deposits (Neumann et al., 2013; Paige et al., 2013), resulting in deposits with distinct surface reflectance properties with sharp boundaries (Chabot et al., 2014; 2016). Overall, the distribution of Mercury’s water ice deposits is an important constraint on the source of Mercury’s water, especially now that we have comparable views of both Mercury’s north and south polar regions. The observed non-uniform distribution of water ice at both of Mercury’s polar regions is additional support for a recent, large impact event as the source of Mercury’s polar water ice deposits.

5 Conclusions

Our new Arecibo radar observations of Mercury’s south polar region reveal numerous radar-bright features not previously detected from prior radar observations, due to the limitations of the viewing geometries and resolution of previous radar datasets. When combined with the Arecibo radar imaging results obtained in 2005 (Harmon et al., 2011), the coverage of Mercury’s south polar region is substantially increased, enabling a more complete examination of the radar-bright features in this area of the planet. Mercury’s south polar region is found to have roughly twice the radar-bright area of Mercury’s north polar region (Deutsch et al., 2016), at 4.4% vs. 2.2% of the area within 10° latitude of the pole. This is consistent with the older, more cratered terrain found in Mercury’s south polar region providing more shadowed regions to host radar-bright deposits than the younger, smooth plains terrain located in Mercury’s north polar region.

Utilizing all of the NAC images obtained from MESSENGER’s full mission duration, 1,090 NAC images of Mercury’s south polar region were mosaicked into 51 mosaics covering different subsolar longitude bins. Thresholding of the mosaics into sunlit and shadowed surfaces enabled the first illumination map of Mercury’s south polar region to be produced at the same spatial resolution as studies of permanently shadowed regions in Mercury’s north. In addition, 67 long-exposure NAC images, which had considerable amounts of saturation, were examined to further identify dimly sunlit areas on the floor of the 180-km-diameter crater Chao Meng-Fu, centered very near Mercury’s south pole, and to refine the total area mapped as permanently shadowed. No permanently sunlit regions are identified in Mercury’s south polar region, but small locations on the rims and terraces of craters near the pole are estimated to be sunlit for up to ~70% of a Mercury solar day. Permanently shadowed areas make up 5.7% of the old, heavily cratered area within 10° of Mercury’s south pole, which as expected is slightly higher than the value of 3.7% determined for the comparable region near Mercury’s north pole (Deutsch et al., 2016) that is dominated by younger smooth plains.

Overall, our new results from both the Arecibo radar observations and the high-resolution illumination study using MESSENGER images support the conclusion that Mercury’s radar-bright features are dominantly composed of water ice. Of the radar-bright features near Mercury’s south pole, 77% are located within 1.5 km of a permanently shadowed region, equivalent to the pixel scale of the Arecibo radar data. This percentage is similar to that determined previously for Mercury’s north polar region (Deutsch et al., 2016). The remaining fraction, or radar-bright pixels not in or within 1.5 km of a permanently shadowed region near Mercury’s south pole, are not located in large clusters but rather as isolated regions or rimming regions of permanent shadow. Thus, this slight discrepancy is attributed to limitations in our methods, such as misregistration or small-scale features below the resolution of our mapping capabilities. We conclude that the observed locations of the radar-bright features in Mercury’s south polar region are consistent with being found in permanently shadowed regions, as would be expected for deposits of water ice.

While radar-bright deposits are located in regions of permanent shadow, roughly half of the permanently shadowed area in Mercury’s south polar region is not mapped as being radar-bright, a percentage that is very similar to that found for Mercury’s north polar region (Deutsch et al., 2016). One explanation is that these permanently shadowed regions do contain water ice, but the water ice was not detected by the radar observations, due to limitations in the radar viewing geometries or the sensitivity of the radar observations. While radar observations of Mercury’s south polar region are limited to essentially two distinct viewing geometries, the 2005 and 2012 observations provide nearly opposite viewing directions and both views consistently lack or detect radar-bright features for the polar region. Additionally, the north polar Arecibo radar observations cover an extensive range of viewing directions, and modeling suggests that radar visibility limitations cannot explain the lack of radar-bright features associated with some sizable permanently shadowed regions in Mercury’s north (Deutsch et al., 2016). Thus, we favor the alternate explanation that these permanently shadowed locations lack water ice deposits or that the water ice deposits are more deeply buried in these locations to not be detected by the radar observations. Either scenario of this explanation would imply that water ice is not distributed evenly among viable cold traps near Mercury’s poles.

The distribution of water ice deposits near Mercury’s poles and whether it is evenly distributed or not has important implications for the source of the water ice. A steady, regular process, such as water delivery by solar wind generation, planetary outgassing, or micrometeorite delivery may be expected to result in water ice being distributed to all viable cold traps. In contrast, delivery of water ice to Mercury by a large impact event could potentially produce an uneven distribution of water ice among the polar cold traps, though such a possibility is based on results from models focused on the Moon. Studies specifically examining the retention and transport of water from a large impact event on Mercury are currently lacking but would be highly worthwhile, as would studies examining the migration and timescales of ice survivability on Mercury in general from any of these processes. The observed non-uniform distribution of water ice at both of Mercury’s polar regions, the suggested presence of low-reflectance organic-rich volatile compounds, the inferred purity of the water ice, and the observations of the deposits’ distinct surface reflectance properties and sharp boundaries are all suggestive of a recent, large impact event as the source of Mercury’s water ice. The large rayed 97-km-diameter crater Hokusai is an intriguing possibility for resulting from the same cometary impact event that produced Mercury’s polar deposits. (Ernst et al., 2016).

Finally, the south polar maps produced in this study, and similar ones available for Mercury’s north pole, will be valuable resources to guide further exploration of Mercury’s polar water ice deposits by the joint ESA-JAXA BepiColombo mission (Benkhoff et al., 2010), scheduled to orbit Mercury in 2025. BepiColombo’s planned orbit will bring it much closer to Mercury’s south pole than MESSENGER’s highly eccentric orbit about Mercury, positioning BepiColombo to make new measurements, and new discoveries, related to Mercury’s south polar deposits.

Supplementary Material

S-Fig2_mosaic in ISIS cub format
S_Fig2_radar in ISIS cub format
S_Fig3_coverage in ISIS cub format
S_Fig3_illumination in ISIS cub format
S_Fig6_PSR in ISIS cub format
S-dataset in excel table format
S_movie in .mp4 format
Download video file (3.8MB, mp4)
S_text in Word docx format

Key Points.

  • New Arecibo radar observations of Mercury’s south pole reveal numerous radar-bright deposits and substantially increase coverage

  • Mercury’s south polar deposits are shown to be located in regions of permanent shadow, consistent with being water ice

  • The observed uneven distribution of water ice at both of Mercury’s poles may suggest a large impact event as the source of the water

Plain Language Summary.

Even though Mercury is the planet closest to the Sun, there are places at its poles that never receive sunlight and are very cold – cold enough to hold water. Our new results from the Arecibo radar observatory reveal numerous locations of water ice near Mercury’s south pole. Using images from NASA’s MESSENGER mission, we mapped how much sunlight Mercury’s south pole receives over one complete day and identified locations that are always in shadow. The permanently shadowed locations match the features seen in the Arecibo images, as expected for water ice. However, about 50% of the permanently shadowed locations lack water ice, and we find that this is similar for both Mercury’s north and south poles. What would cause this uneven distribution of water on Mercury? We conclude that this result is most consistent with Mercury’s water coming from a large, recent impact of a comet on the planet.

Acknowledgments

Acknowledgments and Data

This work was supported by NASA Discovery Data Analysis Program grant NNX15AK89G to N. L. Chabot. This research made use of the Integrated Software for Imagers and Spectrometers of the U.S. Geological Survey. All MESSENGER data analyzed in this paper are archived at the NASA Planetary Data System. Arecibo radar images used in this paper are publicly available at: (http://www.naic.edu/~radarusr/Mercpole/), with the 2005 radar image of Fig. 1a labeled as “Image7a” and the 2012 radar image of Fig. 1b as “Image8” in this directory. Derived data products produced by this work are available at the APL Data Archive (http://lib.jhuapl.edu/), specifically at: (http://lib.jhuapl.edu/papers/investigating-mercurys-south-polar-deposits-arecib/).We thank the APL ASPIRE and NASA/APL intern programs for supporting E. E. Shread’s contributions on this work. J. K. Harmon wishes to thank the staff of Arecibo Observatory for their support of the radar observations and would specifically like to thank D. Campbell and Cornell University Consulting Contract #56324 for support of the 2012 observations and analysis. We appreciate comments and reviews by S. Hauck, W. Fa, M. Slade, G. Neumann, and an anonymous reviewer, which led to improvements in the final paper. No conflicts of interest are noted for any of the authors for this work.

References

  1. Becker KJ, Robinson MS, Becker TL, Weller LA, Edmundson KE, Neumann GA, Perry ME, Solomon SC. First global digital elevation of Mercury. Lunar Planet. Sci. 2016;47 abstract 2959. [Google Scholar]
  2. Benkhoff J, van Casteren J, Hayakawa H, Fujimoto M, Laakso H, Novara M, Ferri P, Middleton HR, Ziethe R. BepiColombo – Comprehensive exploration of Mercury: Mission overview and science goals. Planet. Space Sci. 2010;58:2–20. [Google Scholar]
  3. Bussey DBJ, Spudis PD, Robinson MS. Illumination conditions at the lunar south pole. Geophys. Res. Lett. 1999;26:1187–1190. [Google Scholar]
  4. Bussey DBJ, Fristad KE, Schenk PM, Robinson MS, Spudis PD. Constant illumination at the lunar north pole. Nature. 2005;434:842. doi: 10.1038/434842a. [DOI] [PubMed] [Google Scholar]
  5. Butler BJ, Muhleman DO, Slade MA. Mercury: Full-disk radar images and the detection and stability of ice at the north pole. J. Geophys. Res. 1993;98:15,003–15,023. [Google Scholar]
  6. Campbell DB, Chandler JF, Ostro SJ, Pettengill GH, Shapiro II. Galilean satellites: 1976 radar results. Icarus. 1978;34:254–267. [Google Scholar]
  7. Cavanaugh JF, Smith JC, Sun X, Bartels AE, Ramos-Izquierdo L, Krebs DJ, McGarry JF, Trunzo R, Novo-Gradac AM, Britt JL, Karsh J, Katz RB, Lukemire AT, Szymkiewicz R, Berry DL, Swinski JP, Neumann GA, Zuber MT, Smith DE. The Mercury Laser Altimeter instrument for the MESSENGER mission. Space Sci. Rev. 2007;131:451–480. [Google Scholar]
  8. Chabot NL, Ernst CM, Denevi BW, Harmon JK, Murchie SL, Blewett DT, Solomon SC, Zhong ED. Areas of permanent shadow in Mercury’s south polar region ascertained by MESSENGER orbital imaging. Geophys. Res. Lett. 2012;39:L09204. doi: 10.1029/2012gl051526. [DOI] [Google Scholar]
  9. Chabot NL, Ernst CM, Harmon JK, Murchie SL, Solomon SC, Blewett DT, Denevi BW. Craters hosting radar-bright deposits in Mercury’s north polar region: Areas of persistent shadow determined from MESSENGER images. J. Geophys. Res. Planets. 2013a;118:26–36. [Google Scholar]
  10. Chabot NL, Ernst CM, Denevi BW, Nair H, Murchie SL, Blewett DT, Head JW, Harmon JK, Solomon SC. Imaging inside Mercury’s permanently shadowed craters: First images from MESSENGER. Lunar and Planetary Science Conference. 2013b;44 abstract 1693. [Google Scholar]
  11. Chabot NL, Ernst CM, Denevi BW, Nair H, Deutsch AN, Blewett DT, Murchie SL, Neumann GA, Mazarico E, Paige DA, Harmon JK, Head JW, Solomon SC. Images of surface volatiles in Mercury’s polar craters acquired by the MESSENGER spacecraft. Geology. 2014;42:1051–1054. [Google Scholar]
  12. Chabot NL, Ernst CM, Paige DA, Nair H, Denevi BW, Blewett DT, Murchie SL, Deutsch AN, Head JW, Solomon SC. Imaging Mercury’s polar deposits during MESSENGER’s low-altitude campaign. Geophys. Res. Lett. 2016;43:9461–9468. doi: 10.1002/2016GL070403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Denevi BW, Chabot NL, Murchie SL, Becker KJ, Blewett DT, Domingue DL, Ernst CM, Hash CD, Hawkins SE, III, Keller MR, Laslo NR, Nair H, Robinson MS, Seelos FP, Stephens GK, Turner FS, Solomon SC. Calibration, Projection, and Final Image Products of MESSENGER’s Mercury Dual Imaging System. Space Science Reviews. 2017 doi: 10.1007/s11214-017-0440-y. [DOI] [Google Scholar]
  14. Deutsch AN, Chabot NL, Mazarico E, Ernst CM, Head JW, Neumann GA, Solomon SC. Comparison of areas in shadow in the north polar region of Mercury from imaging and altimetry, with implications for polar ice deposits. Icarus. 2016;280:158–171. doi: 10.1016/j.icarus.2016.06.015. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Domingue DL, Murchie SL, Denevi BW, Ernst CM, Chabot NL. Mercury’s global color mosaic: An update from MESSENGER’s orbital observations. Icarus. 2015;257:477–488. [Google Scholar]
  16. Ernst CM, Chabot NL, Barnouin OS. Examining the potential contributions of the Hokusai impact to water ice on Mercury. Lunar Planet. Sci. 2016;47 abstract 1374. [Google Scholar]
  17. Goldsten JO, Rhodes EA, Boynton WV, Feldman WC, Lawrence DJ, Trombka JI, Smith DM, Evans LG, White J, Madden NW, Berg PC, Murphy GA, Gurnee RS, Strohbehn K, Williams BD, Schaefer ED, Monaco CA, Cork CP, Del Eckels J, Miller WO, Burks MT, Hagler LB, DeTeresa SJ, Witte MC. The MESSENGER Gamma-Ray and Neutron Spectrometer. Space Sci. Rev. 2007;131:339–391. [Google Scholar]
  18. Hapke B. Coherent backscatter and the radar characteristics of outer planet satellites. Icarus. 1990;88:407–17. [Google Scholar]
  19. Hapke B, Blewett D. Coherent backscatter model for the unusual radar reflectivity of icy satellites. Nature. 1991;352:46–47. [Google Scholar]
  20. Harcke LJ. Radar imaging of solar system ices. Ph.D. thesis. Stanford, CA: Stanford University; 2005. p. 201. [Google Scholar]
  21. Harmon JK. Planetary delay-Doppler radar and the long-code method. IEEE Trans. Geosci. Remote Sens. 2002;40:1904–1616. [Google Scholar]
  22. Harmon JK. Radar imaging of Mercury. Space Sci. Rev. 2007;132:307–349. [Google Scholar]
  23. Harmon JK, Nolan MC. Arecibo radar imagery of Mars: II. Chryse–Xanthe, polar caps, and other regions. Icarus. 2017;281:162–199. [Google Scholar]
  24. Harmon JK, Slade MA. Radar mapping of Mercury: Full-disk images and polar anomalies. Science. 1992;258:640–643. doi: 10.1126/science.258.5082.640. [DOI] [PubMed] [Google Scholar]
  25. Harmon JK, Perillat PJ, Slade MA. High-resolution radar imaging of Mercury’s north pole. Icarus. 2001;149:1–15. [Google Scholar]
  26. Harmon JK, Slade MA, Vélez RA, Crespo A, Dryer MJ, Johnson JM. Radar mapping of Mercury’s polar anomalies. Nature. 1994;369:213–215. [Google Scholar]
  27. Harmon JK, Slade MA, Rice MS. Radar imagery of Mercury’s putative polar ice: 1999–2005 Arecibo results. Icarus. 2011;211:37–50. [Google Scholar]
  28. Hawkins SE, III, Boldt JD, Darlington EH, Espiritu R, Gold RE, Gotwols B, Grey MP, Hash CD, Hayes JR, Jaskulek SE, Kardian CJ, Jr, Keller MR, Malaret ER, Murchie SL, Murphy PK, Peacock K, Prockter LM, Reiter RA, Robinson MS, Schaefer ED, Shelton RG, Sterner RE, II, Taylor HW, Watters TR, Williams BD. The Mercury Dual Imaging System on the MESSENGER spacecraft. Space Sci. Rev. 2007;131:247–338. [Google Scholar]
  29. Ingersoll AP, Svitek T, Murray BC. Stability of polar frosts in spherical bowl-shaped craters on Moon, Mercury, and Mars. Icarus. 1992;100:40–47. [Google Scholar]
  30. Lawrence DJ, Feldman WC, Goldsten JO, Maurice S, Peplowski PN, Anderson BJ, Bazell D, McNutt RL, Nittler LR, Prettyman TH, Rodgers DJ, Solomon SC, Weider SZ. Evidence for water ice near Mercury’s north pole from MESSENGER Neutron Spectrometer measurements. Science. 2013;339:292–296. doi: 10.1126/science.1229953. [DOI] [PubMed] [Google Scholar]
  31. Margot J-L, Peale SJ, Solomon SC, Hauck SA, II, Ghigo FD, Jurgens RF, Yseboodt M, Giorgini JD, Padovan S, Campbell DB. Mercury’s moment of inertia from spin and gravity data. J. Geophys. Res. 2012;117:E00L09. doi: 10.1029/2012JE004161. [DOI] [Google Scholar]
  32. Mazarico E, Neumann GA, Smith DE, Zuber MT, Torrence MH. Illuminations conditions of the lunar polar regions using LOLA topography. Icarus. 2011;211:1066–1081. [Google Scholar]
  33. Muhleman DO, Butler BJ, Grossman AW, Slade MA. Radar images of Mars. Science. 1991;253:1508–1513. doi: 10.1126/science.253.5027.1508. [DOI] [PubMed] [Google Scholar]
  34. Neumann GA, Cavanaugh JF, Sun X, Mazarico E, Smith DE, Zuber MT, Mao D, Paige DA, Solomon SC, Ernst CM, Barnouin OS. Bright and dark polar deposits on Mercury: Evidence for surface volatiles. Science. 2013;339:296–300. doi: 10.1126/science.1229764. [DOI] [PubMed] [Google Scholar]
  35. Ostrach LR, Robinson MS, Whitten JL, Fassett CI, Strom RG, Head JW, Solomon SC. Extent, age, and resurfacing history of the northern smooth plains on Mercury from MESSENGER observations. Icarus. 2015;250:602–622. [Google Scholar]
  36. Ostro SJ, Campbell DB, Pettengill GH, Shapiro II. Radar observations of the icy Galilean satellites. Icarus. 1980;44:431–440. [Google Scholar]
  37. Paige DA, Wood SE, Vasavada AR. The thermal stability of water ice at the poles of Mercury. Science. 1992;258:643–646. doi: 10.1126/science.258.5082.643. [DOI] [PubMed] [Google Scholar]
  38. Paige DA, Siegler MA, Harmon JK, Neumann GA, Mazarico EM, Smith DE, Zuber MT, Harju E, Delitsky ML, Solomon SC. Thermal stability of volatiles in the north polar region of Mercury. Science. 2013;339:300–303. doi: 10.1126/science.1231106. [DOI] [PubMed] [Google Scholar]
  39. Prem P, Artemieva NA, Goldstein DB, Varghese PL, Trafton LM. Transport of water in a transient impact-generated lunar atmosphere. Icarus. 2015;255:148–158. doi: 10.1016/j.icarus.2014.10.017. [DOI] [Google Scholar]
  40. Salvail JR, Fanale FP. Near-surface ice on Mercury and the Moon: A topographic thermal model. Icarus. 1994;111:441–455. [Google Scholar]
  41. Siegler MA, Bills BG, Paige DA. Effects of orbital evolution on lunar ice stability. J. Geophys. Res. 2011;116:E03010. doi: 10.1029/2010JE003652. [DOI] [Google Scholar]
  42. Slade MA, Butler BJ, Muhleman DO. Mercury radar imaging: Evidence for polar ice. Science. 1992;258:635–640. doi: 10.1126/science.258.5082.635. [DOI] [PubMed] [Google Scholar]
  43. Speyerer EJ, Robinson MS. Persistently illuminated regions at the lunar poles: Ideal sites for future exploration. Icarus. 2013;222:122–136. [Google Scholar]
  44. Stewart BD, Pierazzo E, Goldstein DB, Varghese PL, Trafton LM. Simulations of a comet impact on the Moon and associated ice deposition in polar cold traps. Icarus. 2011;215:1–16. [Google Scholar]
  45. Susorney HCM, Barnouin OS, Ernst CM, Johnson CL. Morphometry of impact craters on Mercury from MESSENGER altimetry and imaging. Icarus. 2016;271:180–193. [Google Scholar]
  46. Vasavada AR, Paige DA, Wood SE. Near-surface temperatures on Mercury and the Moon and the stability of polar ice deposits. Icarus. 1999;141:179–193. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

S-Fig2_mosaic in ISIS cub format
NIHMS945493-supplement-S_Fig2_mosaic.cub (72.3MB, cub)
S_Fig2_radar in ISIS cub format
NIHMS945493-supplement-S_Fig2_radar.cub (4MB, cub)
S_Fig3_coverage in ISIS cub format
NIHMS945493-supplement-S_Fig3_coverage.cub (69.6MB, cub)
S_Fig3_illumination in ISIS cub format
NIHMS945493-supplement-S_Fig3_illumination.cub (69.6MB, cub)
S_Fig6_PSR in ISIS cub format
NIHMS945493-supplement-S_Fig6.cub (72.3MB, cub)
S-dataset in excel table format
NIHMS945493-supplement-S_dataset.xlsx (65.9KB, xlsx)
S_movie in .mp4 format
Download video file (3.8MB, mp4)
S_text in Word docx format
NIHMS945493-supplement-S_text.pdf (355.4KB, pdf)

RESOURCES