Volume 48, Issue 13 e2021GL093880
Research Letter
Free Access

Strong MARSIS Radar Reflections From the Base of Martian South Polar Cap May Be Due to Conductive Ice or Minerals

C. J. Bierson

Corresponding Author

C. J. Bierson

School of Earth and Space Exploration, Arizona State University, Tempe, AZ, USA

Correspondence to:

C. J. Bierson,

[email protected]

Search for more papers by this author
S. Tulaczyk

S. Tulaczyk

Department of Earth and Planetary Science, University of California Santa Cruz, Santa Cruz, CA, USA

Search for more papers by this author
S. W. Courville

S. W. Courville

School of Earth and Space Exploration, Arizona State University, Tempe, AZ, USA

Search for more papers by this author
N. E. Putzig

N. E. Putzig

Plantary Science Institute, Lakewood, CO, USA

Search for more papers by this author
First published: 28 June 2021
Citations: 21


Recent results from the Mars Advanced Radar for Subsurface and Ionosphere Sounding (MARSIS) instrument have been interpreted as evidence of subsurface brine pooled beneath 1.3 km-thick South Polar Layered Deposit (SPLD). This interpretation is based on the assumption that the regionally high strength of MARSIS radar reflections from the base of the ice cap is due to a strong contrast in dielectric permittivity across the basal interface. Here, we demonstrate that the high-power reflections could instead be the result of a contrast in electric conductivity. While not explicitly excluding a liquid brine, our results open new potential explanations for the observed strong radar reflections, some of which do not require liquid brine beneath SPLD. Potential basal materials with suitably high conductivity include clays, metal-bearing minerals, or saline ice.

Key Points

  • Radar reflections can be caused by contrasts in either dielectric permittivity or electric conductivity

  • Reflections observed by Mars Advanced Radar for Subsurface and Ionosphere Sounding (MARSIS) at Mars' south polar region can be explained by contrasts in electric conductivity and do not require liquid water

  • We propose that a polar-ice substrate containing clays, other conductive minerals, or saline ice are the most plausible candidate materials

Plain Language Summary

Previous work reported a regionally strong radar reflection under Mars' south polar ice sheet. Due to its brightness, this radar reflection was interpreted as liquid water (likely with high concentration of dissolved salts). A radar reflection can be bright due to a large contrast in either dielectric permittivity or electric conductivity. Previous work only considered contrasts in dielectric permittivity. We find that contrasts in electric conductivity between materials could also explain the brightness of the reflection. We suggest that this difference could be due to clays, metal-bearing minerals, or saline ice under the polar ice sheet.

1 Introduction

Liquid water is the universal solvent essential to the origin and perpetuation of life on Earth. It has been the central focus of research into the habitability of Mars, both past and present. Most of Mars' water inventory is currently stored within its polar ice caps (Carr & Head, 2015). However, it has long been proposed that given sufficient porosity in the Martian crust, there may also be liquid water 3–8 km below the Martian surface (Clifford, 1987). This aquifer could exist if sufficiently heated by Mars' geothermal heat flux. Flexural constraints, however, indicate a sub-chondritic heat flux that would push this melting point even deeper, to approximately 6–16 km (Phillips et al., 2008).

It was therefore very exciting and somewhat unexpected when the Mars Advanced Radar for Subsurface and Ionosphere Sounding (MARSIS) instrument on Mars Express found strong radar reflections, interpreted as evidence of brines just 1.3 km below the surface of the south polar layered deposits (SPLD) (Lauro et al., 2021; Orosei et al., 2018). This strong reflection was not observed in the higher frequency SHAllow RADar (SHARAD) instrument on the Mars Reconnaissance Orbiter (Orosei et al., 2018). Thermal modeling suggests that, even as a brine, liquid water would require a local heat flux of >70 mW/m2 (Sori & Bramson, 2019). Such a high local heat flux would imply very recent local volcanism (Sori & Bramson, 2019).

The interpretation that the basal return from MARSIS radar is a result of liquid brine is based on the high power of the subsurface signal return when compared to that of the surface return. It was assumed that the strong basal reflection was due to a contrast in dielectric permittivity, which is a common approach in geologic and glaciologic interpretations of radar reflectivity on Mars and Earth. Hence, Orosei et al. (2018) inferred that the dielectric permittivity contrast required to explain the power of the returned signal is so large that it can only be explained by a liquid water reservoir beneath SPLD.

From first principles, there are three material properties whose contrasts can cause a radar reflection from an interface: dielectric permittivity, electric conductivity, and magnetic permeability (Tulaczyk & Foley, 2020). Sometimes, the impact of electric conductivity is parameterized in the imaginary component of the complex permittivity (e.g., Stillman & Olhoeft, 2008). Within this work “dielectric permittivity” will always refer to the real part of the permittivity.

In geologic materials, magnetic permeability varies very little and only notably in metallic ores. In contrast, dielectric permittivity has significant variations between geologic materials like CO2 ice (2.2 (Simpson et al., 1980)), water ice (3.1 (Pettinelli et al., 2015)), silicates (∼10), and liquid water (80) (Table 1). For this reason, it is a common practice within the planetary radar community to interpret strong subsurface radar reflections as being caused by contrasts in the dielectric permittivity.

Table 1. Electrical Properties for Relevant Geologic Materials
Material Relative dielectric permittivity Conductivity (S/m)
Glacier Ice 3.2 (a) 7.00 × 10−5 (a)
Saline Ice 3.4 (a) 10−3–10−2 (b)
Brine 62 (c) 4.8 (d)
Magnetite 2−10 (e) <6 × 10−5 (e)
Hematite 18.1 (f) 0.2 (f)
Ilmenite 0.25–103 (g)
JSC-1 2.8–5.3 (e) <6 × 10−5 (e)
Groundwater 80 (c) 0.37 (h)
Clay 3−5 (i) 0.02–0.002 (i)
  • Note: Values shown are for temperatures between 250 and 290 K because lower temperature data are not available. (a) Christianson et al. (2016) summary of work by Smith and Evans (1972), Glen and Paren (1975), Neal (1979), Peters et al. (2005), and MacGregor et al. (2007); (b) Pettinelli et al. (2015) and Foley et al. (2016); (c) Tulaczyk and Foley (2020) using Lyons et al. (2019) and Buchner et al. (1999); (d) Tulaczyk and Foley (2020), Mikucki et al. (2015); (e) Stillman and Olhoeft (2008); (f) Robinson and Friedman (2003); (g) zero frequency measurement from Nabighian (1988), 0.1 Hz from Zhang et al. (2006); (h) Mikucki et al. 2015; (i) Arcone et al. (2008).

Electric conductivity is somewhat unique among these three properties in that its contribution to reflection strength is inversely dependent on the radar frequency, with lower frequency radar being more sensitive to conductivity contrasts. Electric conductivity is low in most settings, but it can vary by orders of magnitude in some geologic materials including clays, brines, and saline ice (Table 1). In subglacial settings in Antarctica, it has been shown that the conductivity of the basal interface can be particularly important in interpreting <10 MHz radar returns (Tulaczyk & Foley, 2020). In this work, we propose a strong contrast in electrical conductivity as the explanation for the putative lake detection by MARSIS (Lauro et al., 2021; Orosei et al., 2018).

2 Revisiting the Dependence of Radar Reflection Coefficient on Conductivity

To characterize the reflection power that could be due to differences in electric conductivity, we apply a 1D radar propagation model (Courville et al., 2021). In this model, we assume the incident radar signal first encounters the surface ice layer that overlays a homogeneous half-space of sub-ice material. We apply this model using not only MARSIS frequencies but also SHARAD frequencies. For both MARSIS and SHARAD, we simulate the response to each instrument's full pulse (Picardi et al., 2004; Seu et al., 2007). Within each layer, we assume that the material conductivity and permittivity are constant over the range of frequencies investigated. This model assumes that all materials have a relative magnetic permeability of one.

For our nominal case, we assume that the SPLD has a relative permittivity of 3.15 and overlays a unit with a relative permittivity of 5 (full stratigraphy in Table 2). Figure 1 shows how the reflection power varies as a function of electric conductivity in the basal layer. At low basal conductivity, the reflection is dominated by the dielectric contrast and so all frequencies return to nearly the same power. At a basal conductivity of ∼10−3 S/m the returned power matches that of the MARSIS reflection. In this region there is about a 1 dB difference predicted between each of the MARSIS frequencies. This is comparable to the difference observed between the 4 MHZ and 5 MHZ MARSIS channels (Orosei et al., 2018). The median power in the 3 MHZ channel is lower than that predicted by this model, although the coverage at 3 MHZ is significantly less and so many not be representative.

Table 2. Material Parameters Used for the Modeling Shown in Figure 1
Unit thickness (km) Relative dielectric permittivity Conductivity (S/m)
Air (CO2) 300 1.0 0.0
Ice 1.3 3.15 10−8
Lower reflector Inf 5.0 Varied 10−6–102
Details are in the caption following the image

Reflection power as a function of the conductivity of the basal layer. Solid lines are model results for the different Mars Advanced Radar for Subsurface and Ionosphere Sounding (MARSIS) and SHAllow RADar (SHARAD) frequencies. The horizontal dashed lines are the observed median basal reflection strengths as a function of frequency (Orosei et al., 2018). While basal MARSIS reflection power has a wide distribution, the error in the median is ∼0.1 dB. The overlying ice is assumed to be 1.3 km thick and to have a conductivity of 10−6 S/m (Table 1). The boxed region is shown in panel (b) Panel (b) is a zoom in for the region in panel a where the conductivity matches the observed MARSIS return power.

For the curves in Figure 1, the conductivity of the SPLD itself is assumed to be 10−8 S/m, nearly an order of magnitude lower than the nominal value of 8x10−6 S/m in Table 1. The value in Table 1, however, is for ice near the melting pointing (∼250 K). The conductivity of water ice is expected to drop well below 10−8 S/m at temperatures below 200 K (Figure S1, Matzler, 1998). If the SPLD ice was near 10−5 S/m, the loss within the SPLD would prevent a strong basal reflection from being observed regardless of the basal properties (Figure S2). As long as the attenuation within the ice is weak, the best-fit conductivity of the underlying materials changes very little (Figure S3). This value is also not sensitive to the overlying ice thickness (Figures S4 and S5).

The modeling shown in Figure 1 assumes that the basal reflector is an infinite half-space. This assumption raises the question of what thickness of deposit is actually required to produce the observed response. The wavelength of MARSIS within the SPLD is roughly 200 m, and generally a contrast in permittivity would need to be on that scale to fully determine the reflection strength (Church et al., 2020, Figure 9). However, for a conductive interface, the thickness required is set by the conductive skin depth (Tulaczyk & Foley, 2020). For MARSIS frequencies and electrical conductivity of 10−3 S/m, this thickness is ∼10 m. Thus, for a conductive material to be responsible for the reflection, it would only need to be a very thin layer at the base of the SPLD.

The multiple frequency modes used by MARSIS to observe the putative lake region allow us to test if this reflection is due to a contrast in conductivity or dielectric permittivity. If it is a contrast in dielectric permittivity, all frequency modes would have the same reflection power (left most region on Figure 1). However, if it is due to a conductivity contrast, our modeling suggests that that the 5 MHz return should be 1 dB weaker than the 4 MHz return. This 1 dB difference matches the difference between the median reflection strengths measured by Orosei et al. (2018), Figure 1.

Another process that could also produce a frequency-dependent return strength without a difference in the basal reflector, is attenuation within the SPLD itself. Attenuation does preferentially reduce the strength of higher frequencies, consistent with the trend in the MARSIS data. The main problem with this explanation is that if the attention is strong enough to produce a 1 dB difference between the 4 and 5 MHz modes, it would be too strong for any basal reflection to be brighter than the surface reflection (Figure S2). Similarly, materials with a frequency-dependent permittivity can also produce a frequency-dependent reflection strength. However, across the narrow range of frequencies used by MARSIS, plausible materials for the basal reflector have a very weak frequency dependence (Stillman & Olhoeft, 2008).

In addition to comparing different MARSIS frequency modes, we extend this analysis to SHARAD which sweeps frequencies between 15 and 25 MHz. Orosei et al. (2018) note that the putative lake was not detected by SHARAD, and our modeling suggests that the higher frequency of SHARAD would cause the reflection strength at the base to be ∼8 dB lower than the 4 MHz return (depending on the basal permittivity contrast). Given this prediction, it is important to consider how significant the lack of SHARAD detection is.

In the region of this bright reflector, there are SHARAD reflections within the SPLD but not down to the base of the SPLD (Orosei et al., (2018); Figure S1). Unfortunately, the region of the putative lake corresponds to an area where the SHARAD data has been described as containing radar “fog” (Whitten & Campbell, 2018), which is a diffuse echo in the subsurface. It has been suggested that this “fog” could be caused either by surface roughness at the scale of the radar wavelength or internal scattering (Whitten & Campbell, 2018). The preferred interpretation of Whitten and Campbell (2018) is scattered from buried lag deposits within the SPLD. Regardless of its cause, the presence of this radar “fog” makes it impossible to use the lack of SHARAD detection of the basal reflector to either disprove or confirm our conjecture that electrical conductivity contrasts are at least partly responsible for the high basal reflectivity observed by MARSIS beneath parts of SPLD.

3 Proposed New Interpretations

We now consider what electrically conductive sub-ice materials can be responsible for the strong reflection from the base of the SPLD. Table 1 shows the dielectric permittivity and conductivity for a range of geologic materials that may be present. We can roughly sort these into three candidate reflectors; liquid water, saline ice, and conductive minerals.

If liquid water is present, it would have to be a highly concentrated brine (Orosei et al., 2018), and it would also be highly electrically conductive. Hence, the observed frequency dependence of the measured reflection strengths is consistent with the presence of a brine layer. Because a brine would also exhibit an extremely large dielectric discontinuity, under ideal circumstances this brine would also be detectable by SHARAD. However, there is a total lack of reflections from SHARAD below 1 km depth in this region, which can be explained by attenuation of these higher frequencies in the water ice, thereby masking this signal.

The next plausible material we consider is saline basal ice. On Earth, saline basal ice forms in cold-based glaciers in contact with saline brines such as beneath the Taylor Glacier in Antarctica (Montross et al., 2014). On Mars, such saline ice could have formed during the Amazonian period if basal melting occurred in this region or from freezing of upwelling deep saline groundwater. During refreezing of a brine, salts are concentrated in the remaining liquid trapped between growing ice crystals until the eutectic freezing point. Thus, the last liquid to freeze would be extremely saline and conductive. A significant caveat to this explanation is that the electric conductivity of saline ice (Table 1) has only been measured at temperatures around 250 K, whereas the annual mean surface temperature of the putative lake region on Mars is 163 K. With a nominal heat flux of 30 mW/m2, the basal temperatures would be <180 K (Sori & Bramson, 2019). Colder temperatures may restrict the mobility of ions, reducing the conductivity, although it is unclear by what magnitude without laboratory studies at relevant temperatures.

The final plausible group of sub-ice materials we consider are conductive or semiconductive minerals that may be responsible for the basal reflection. These not only include metallic ores like hematite, but also weathered clays and hydrated minerals like jarosite. Hydrous minerals have been found throughout the Martian southern highland terrains using the CRISM spectrometer (Carter et al., 2013; Wray et al., 2009). Surveys of these minerals, however, rarely extend poleward of 60° latitude due to masking by surface frosts. Geologic mapping of the polar region suggests that the SPLD overlay Noachian sedimentary deposits (Kolb & Tanaka, 2001) similar to those which are observed to have hydrous mineral (clay) exposures farther north in the southern highlands. Clays are highly conductive in subglacial environments on Earth (Tulaczyk & Foley, 2020). Much of this conductivity, however, is facilitated by liquid water being present in pores or at least adhering to grain surfaces. No laboratory studies have examined the high-frequency conductivity of clays under Martian conditions. Similarly, there are no conductivity measurements for dry jarosite at Martian temperatures. However, thermodynamic considerations suggest that it is more plausible to have wetted clays or hydrated minerals at the base of the southern polar ice caps than to maintain a large body of liquid brine.

While the available radar observations may not allow for an unambiguous identification of the reflecting material, we can weigh the relative plausibility of these three candidate groups. As was noted by Sori and Bramson (2019), for even a hypersaline brine to be present, there must be a localized and recent heat source. While not excluded by any existing data, there are also no other observations suggesting that this heat source is present. Saline basal ice only requires that basal melt was present at some point in the time since the cap formed, which may be tens of millions of years or more. However, as noted previously, it is possible that even very saline ice at these low temperatures may not be conductive enough to explain the observed reflection power.

Our preferred explanation is the mineralogical one. Conductive, hydrous minerals are known to be present within the Noachian southern highlands that the SPLD overlays. If these hydrous mineralogical phases are present under the SPLD, it does imply that liquid water was present near the South pole, at least transiently, during the Noachian period.

The interpretation presented here that the basal reflector is due to a conductivity contrast is nonunique. If the reflection is due purely to a difference in permittivity, the difference in power between the MARSIS observing frequencies could still be explained by frequency-dependent attenuation in the SPLD. We note however, that if the basal reflector is a liquid brine, it would be expected to have a strong reflection due to conductivity (Table 1). Thus, incorporating the role of conductivity remains important to interpreting the strong basal reflections observed in MARSIS data. What our analysis does do, is to demonstrate that it is not necessary to assume that these reflections can only be explained by the presence of a liquid body of brine beneath SPLD.

At present, this area of bright subsurface reflectors under the SPLD appears unique. This may in part be due to most areas in the SLPD having thicker ice, which causes more attenuation and thereby hides signals that may otherwise be similar. This work demonstrates, however, that many common Martian materials are conductive enough to produce strong radar reflections of the MARSIS signal but not the SHARAD signal due to the difference in the radar frequencies. This contrast provides an opportunity for future work to potentially constrain the conductivity of subsurface reflectors by comparing these two data sets in other settings. Similar constraints may be possible with the dual frequency radar instrument REASON aboard Europa Clipper (Bayer et al., 2019).


The authors thank two anonymous reviewers for the helpful comments. ST's contribution to this project was supported by NSF grant #1644187.

    Data Availability Statement

    The base software used for this work, RadSPy is publicly available at https://github.com/scourvil/RadSPy.git. All scripts and data files needed to reproduce Figure 1 and all supplementary figures are available at https://doi.org/10.5281/zenodo.4677654.