Massive CO2 Ice Deposits Sequestered in the South Polar Layered Deposits of Mars

See allHide authors and affiliations

Science  13 May 2011:
Vol. 332, Issue 6031, pp. 838-841
DOI: 10.1126/science.1203091


Shallow Radar soundings from the Mars Reconnaissance Orbiter reveal a buried deposit of carbon dioxide (CO2) ice within the south polar layered deposits of Mars with a volume of 9500 to 12,500 cubic kilometers, about 30 times that previously estimated for the south pole residual cap. The deposit occurs within a stratigraphic unit that is uniquely marked by collapse features and other evidence of interior CO2 volatile release. If released into the atmosphere at times of high obliquity, the CO2 reservoir would increase the atmospheric mass by up to 80%, leading to more frequent and intense dust storms and to more regions where liquid water could persist without boiling.

The martian atmosphere is dominated by CO2 with an annual mean pressure currently about 6 mbar (6 hPa) (1), although early in the planet’s history CO2 likely existed at the ~1 bar level. Some of this ancient atmospheric CO2 may be stored in the polar layered deposits (PLD) (2), although, it is now thought, only in modest quantities. The water-ice–dominated southern PLD (SPLD) presently host a small [<5-m thick, ~90,000 km2 (3)] perennial CO2-ice deposit (4) overlying a thin water-ice layer (5), which together compose the south pole residual cap (SPRC). If the SPRC CO2 material were to be released completely into the atmosphere, the increase in pressure would be only a few tenths of a mbar and insufficient to buffer the atmospheric CO2 during changing climatic conditions (5). Here, we use radar-sounder data to show that the volume of sequestered CO2 in the SPLD is substantially larger than previously believed, competing in magnitude with the present atmospheric abundance.

SHARAD (Shallow Radar) is a sounding radar on the Mars Reconnaissance Orbiter (MRO) mission (6), and its results are displayed in radargrams with axes of time delay and orbital position (Fig. 1). Previous mapping of subsurface reflectors by SHARAD in the north PLD (NPLD) revealed a crisp radar stratigraphy to the base of the deposits (7, 8). For the SPLD, the radar signal does not penetrate the deposits as deeply as in the NPLD, and in only a limited number of places is there a well-defined stratigraphy (9). There are some regions with nearly reflection-free subsurface zones (RFZ) extending downward from near the surface to depths approaching 1 km (fig. S1). The RFZs can be subdivided into four distinct types and locations (table S1) on the basis of their radar characteristics; here, we focus on RFZ3, which is spatially coincident with the SPRC. Except for a commonly occurring thin layer that bisects the unit (Fig. 1), RFZ3 is the most reflection-free volume that we have seen on Mars with SHARAD data: the signal level within approaches the background noise. Deeper reflectors passing beneath RFZ3 brighten slightly more than expected on the basis of the change in thickness of typical SPLD material, implying that RFZ3 deposits attenuate a radar signal less severely than these typical regions. Importantly, the low-power RFZ3 radar return is thus not caused by strong scattering or absorption losses within the deposit.

Fig. 1

SHARAD radargram 5968-01 traversing RFZ3 terrain shown in original time-delay format (A) and converted to depth (B) by using the permittivity of water ice. Ground track location is shown in Fig. 3. ORR and LB3 are indicated.

To determine the real permittivity (ε′) of RFZ3, we mapped key SHARAD reflectors for 79 MRO orbits (fig. S2). The lower boundary of RFZ3, LB3, is a highly irregular buried erosional surface that truncates subhorizontal reflectors. Extending several hundred meters beneath RFZ3 is a zone of unorganized, weak radar reflectors that in turn is underlain by a coherent sequence of organized (layered) radar reflectors (ORR). By using the a priori assumption of a bulk water-ice composition (ε′ = 3.15) for the SPLD, we converted the vertical axis of radargrams from time delay to depth. The converted ORR sequence beneath LB3 is typically offset from surrounding regions and exhibits significant topographic variations (Fig. 1B) that are strongly anticorrelated with LB3 (Fig. 2A). This anticorrelation is unexpected because there is very likely no geological link between the earlier deposition of the ORR and the later erosion of the material above it that was subsequently filled with RFZ3 material [see (10) for details]. On the basis of the argument that the anticorrelations are the fortuitous result of an incorrect choice of ε′ for RFZ3, we found for each radargram the ε′ value that gave zero correlation between LB3 and a test reflector (TR) in the ORR sequence (Fig. 2, B and C) (10). A second method (10) sought to minimize topographic perturbations and offsets on the TR by finding the ε′ value that obtained the smallest residuals to a linear regression on this interface (Fig. 2, D and E). Both methods tended to produce a relatively smooth and subhorizontal disposition to the TR, similar in nature to the likely extension of ORR observed by SHARAD in the Promethei Lingula region (9). Forty-one of the 79 radargrams were suitable for quantitative analyses using these procedures, and by using different strategies we found mean values for ε′ of the RFZ3 volume in the range of 2.0 to 2.2, with standard deviations of 0.1 to 0.2 (10).

Fig. 2

(A) Distribution for 41 orbital observations of the correlation coefficients between LB3 and the TR assuming RFZ3 ε′ equals that of water ice, 3.15. (B) For observation 8104-01, the positions of LB3 and TR as a function of RFZ3 ε′. (C) The ε′ values used in (B) and the corresponding correlation coefficients. (D) For observation 8104-01, the positions of the surface, LB3, and TR for RFZ3 ε′ = 3.15. (E) Same as (D) except RFZ3 ε′ = 2.11, which gives the best-fitting linear regression on TR.

These permittivity estimates for RFZ3 are unexpectedly close to a laboratory-measured value of low-porosity CO2 ice of 2.12 ± 0.04 (11), similar to the well-known frequency-independent value of about 2.2 for bulk dry ice (12). The SHARAD-derived permittivity values are substantially lower than those of water ice (3.15) and CO2 clathrate-hydrate ice (~2.85) (13), strongly supporting the hypothesis that RFZ3 is a solid CO2 deposit. An alternative view that RFZ3 is porous water ice can be rejected on the basis of permittivity-thickness relationships (10).

With the permittivities estimated, we converted the time delays through RFZ3 (using ε′ = 2.1) to thicknesses over each of the 79 radar traverses (fig. S3) and by interpolation constructed a continuous thickness distribution. Figure 3 shows this result placed over a geological map showing stratigraphic units in a portion of the SPLD (14, 15). Of interest here are the largely overlapping horizontal extents of the Aa3 unit and the successively overlying water-ice (Aa4a) and CO2-ice (Aa4b) units making up the SPRC. Also shown are the contacts (dashed) for unit Aa3, with the locations constrained well by exposures in troughs and by partial exposures beneath the SPRC. Where SHARAD data are available, there is a remarkable spatial correlation of RFZ3 to the Aa3 unit except for the extremes of northward-extending lobes of the unit (16). Thus, we propose that the Aa3 unit is in fact RFZ3, and its composition is dominated by CO2 ice.

Fig. 3

Polar stereographic map of a portion of the SPLD, showing RFZ3 thickness variations interpolated to a continuous volume for the 79 SHARAD ground tracks where RFZ3 deposits were observed. Bright colors indicate deposit thicknesses calculated by using ε′ = 2.1, and the histogram (inset) provides their relative occurrences. Base map (subdued colors) shows SPLD stratigraphy (14, 15) with geologic units from oldest to youngest: HNu (substrate underlying SPLD); Aa1 (evenly bedded layers, up to 3.5 km thick); Aa2 (evenly bedded layers, <300 m thick); Aa3 (~300 m thick); and Aa4a and Aa4b (water-ice and CO2-ice members, respectively, of the SPRC). The units are separated by unconformities, indicating episodes of erosion between them that resulted in retreat of the original lateral extents of the units and in development of local troughs and depressions. Dashed lines indicate boundaries of unit Aa3 where partially buried. Mars Orbiter Laser Altimeter (MOLA) shaded relief base at 115 m per pixel; because of spacecraft orbital inclinations, no SHARAD or MOLA data are available poleward of ~87°S. Ground track of observation 5968-01 (Fig. 1) is shown.

The Aa3 unit contains a system of large troughs, up to several km wide and typically <100 m deep, that do not cut older units (Fig. 3). In turn, smaller parallel aligned ridges, troughs, and elongate depressions mark some of these large troughs, and in places the depressions appear as coalescing or elongated pits (Fig. 4). Additionally, the westernmost outcrops of unit Aa3 (north of 87°S and near 240° to 270°E) include about 20 rimless circular pits (~300- to ~4000-m diameter), which do not occur in layers underlying unit Aa3 and do not display any rims or ejecta. All of these smaller troughs, depressions, and pits appear to result from erosion and removal of unit Aa3, with a strong component of sublimation and collapse. These features are not found elsewhere in the SPLD, and the CO2-ice layer (Aa4b) of the SPRC is the only other perennial unit in the SPLD that exhibits clear (although different) morphological indicators of sublimation (5). The lack of sublimation features in exposures of the older units Aa1 and Aa2 indicate that CO2, and not H2O, is the sublimating material in the Aa3 unit, as might be expected given their relative volatilities. The Aa3 unit within pits distributed along the linear depressions is covered by a heavily fractured SPRC water-ice layer (Aa4a) that is overlain in places by the sublimating SPRC CO2 layer (Aa4b) that formed after the fracturing (Fig. 4). The fracturing, not found in other SPLD units, may be a response to continuing unit Aa3 sublimation after the pits had first formed. The other three RFZs lack surface expressions of sublimation, but nondetection of sufficiently rugged lower boundaries precluded permittivity estimates.

Fig. 4

MOLA topographic image (A) in the vicinity of 87°S, 268°E, showing linear depressions or troughs in the Aa3 unit. The total elevation range of the image is ~75 m from the lowest (pink) to the highest (green) surface. The troughs are associated with circular pits [(B), part of MRO HiRISE (High Resolution Imaging Science Experiment) image ESP_014342_0930] and are thinly buried by the SPRC (C), with unit AA4b (CO2 ice) displaying sublimation windows into a fractured water-ice unit AA4a beneath (northwestern corner of a pit). The water-ice layer is completely exposed in the northeastern portion of this pit, where intense polygonal fracturing gives way to concentric fracturing on the pit rim (cf).

Because we equate RFZ3 to unit Aa3, we used the areal distribution of the geological unit to extrapolate the RFZ3 volume poleward of ~87°S, achieving a total volume range (17) of ~9500 to 12,500 km3 (10). In contrast, the volume of the CO2-dominated SPRC is less than 380 km3 (3), about 30 times less. The RFZ3 thickness-independent permittivity values (10) imply a density close to that of bulk dry ice, 1500 to 1600 kg m−3 (18), which converts volume to an equivalent atmospheric pressure of 4 to 5 mbar, up to ~80% of the equivalent mass in the current atmosphere. The collapse features in the Aa3 unit suggest that the RFZ3 mass has been waning, and an isolated patch of RFZ3 (at ~345°E in Figs. 1 and 3) appears to be an erosional remnant. This suggests that the atmosphere has contained less than the present ~6 mbar of CO2, hinting at past atmospheric collapse.

The lack of reflections in RFZ3 apart from the bisecting layer can be interpreted as a lack of dust (7). Global climate models (GCMs) suggest (19) that, when the obliquity of Mars drops below a critical value, the atmosphere collapses onto the polar caps. At low obliquities, the ability of the atmosphere to lift dust is greatly diminished (20), possibly providing an explanation for the radar observations. Obviously, the CO2 now buried in RFZ3 was in the atmosphere at some time in the past. A plausible assumption is that the RFZ3 mass was largely in the atmosphere when the insolation at the south pole at summer solstice was at a maximum, which for the past one million years occurred about 600,000 years ago [obliquity = 34.76°, eccentricity = 0.085, longitude of perihelion = 259.4° (21)].

To assess the impact on some first-order climate parameters, we ran a fast version of the NASA/Ames Mars GCM (version 1.7.3) for these orbital conditions with a total exchangeable CO2 inventory (atmosphere plus caps) equal to the present inventory (7.1 mbar) plus 5 mbar. We found that most of the additional 5 mbar of CO2 ended up in the atmosphere. Surface pressures rose uniformly around the planet, with global-mean annually averaged pressures equaling 10.5 mbar. Annual mean cap masses increased by about 0.8 mbar, not accounting for the lost RFZ3 mass. Surface temperatures, however, decreased slightly (~0.7 K) because the CO2 ice was on the ground for a longer period, and this compensated the modest greenhouse effect.

There are two implications of these changes in the climate system. First, the increased CO2 pressure expands the geographic locations where these pressures exceed the triple-point pressure of water, thereby permitting liquid water to persist without boiling (although it may still evaporate, as on Earth) (22). Second, higher surface pressures will lead to higher surface wind stresses, which will loft more dust in the atmosphere, leading to an increase in dust storm frequency and intensity. Given the complex interplay between the dust, water, and CO2 cycles, additional changes in the climate system are very likely.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S5

Tables S1 to S4


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. RFZ3 is seen discontinuously in radargrams here, but key reflectors could not be mapped with high confidence likely because of surface scattering interference, resolution limitations, and lack of coverage.
  3. A lower value of ~4000 to 4500 km3 is obtained with the unlikely assumption that RFZ3 does not extend beyond SHARAD’s data gathering locales, which are limited by MRO’s orbital inclination. See (10).
  4. Acknowledgments: K. Herkenhoff and C. Fortezzo provided useful comments on an earlier version of the paper. Remarks by three anonymous referees were exceedingly helpful. Funding for this work was provided by the NASA MRO project. The radar and imaging data are available through NASA’s Planetary Data System.

Stay Connected to Science

Navigate This Article