Three-dimensional structure and origin of a 1.8 km thick ice dome within Korolev Crater, Mars

Abstract

Korolev is an 80 km diameter impact crater located at 72.7°N, 164.5°E containing a large domed deposit. Perennial water ice is thought to be currently unstable there; however, using a 3-D analysis of the dome's internal radar stratigraphy, we estimate that Korolev's central mound contains between 1400 and 3500 km³ of water ice that is up to 1.8 km thick. Furthermore, the stratigraphic structure of this ice dome is strikingly similar to the north polar layered deposits (NPLD) on Planum Boreum, approximately 600 km to the north. Additionally, our stratigraphic analysis suggests that Korolev ice was not previously part of a once larger polar ice sheet but rather it was deposited locally. We conclude that Korolev Crater's ice likely deposited during the same climate regime as Planum Boreum's NPLD, but independently. This implies that the incorporation of circumpolar ice deposits such as Korolev may prove useful in reconstructing a unique Mars polar climate history.

1 Introduction and Background

The martian north pole is covered by a massive layered deposit of relatively pure water ice [Howard et al., 1982; Malin and Edgett, 2001; Phillips et al., 2008; Grima et al., 2009]. Radar studies have used the stratigraphy of this deposit to further our understanding of its evolution and genesis [Phillips et al., 2008; Putzig et al., 2009; Selvans et al., 2010; Christian et al., 2013]. These studies have revealed that many features of Planum Boreum are constructional, built from largely depositional processes [Holt et al., 2010; Smith and Holt, 2010; Brothers et al., 2013; Smith et al., 2013]. With emphasis on local deposition rather than regional erosion for Mars' northern ice it is necessary to reassess the origin of circumpolar ice deposits such as those found in Korolev and Dokka craters (Figure 1). This work investigates whether circumpolar crater ice is more likely a remnant of a previous geologic epoch with a more extensive ice cap [Fishbaugh and Head, 2000; Garvin, 2000] or a relatively recent feature built by local deposition [Tanaka et al., 2008; Conway et al., 2012]. Each scenario should have a unique stratigraphic signature that can be analyzed with radar sounding.

Figure 1

Topographic map of Korolev Crater, based on Mars Orbiter Laser Altimeter 256 ppd data. The 69 radar observations used in the study are shown here as well as locations for the radargram in Figure 2 (B-B′, red line), the gridded reflector cross section shown in Figure 4 (A-A′, orange line), and the Korolev reflector packages shown in Figure 5 (yellow line). The small inset shows MOLA topography for nearby Planum Boreum and the location for NPLD reflector packages shown in Figure 5 (small gray line on Planum Boreum).

Current climatic conditions permit stable water ice at the surface on Mars' north pole; however, south of 90° ice stability decreases [Levrard et al., 2004; Madeleine et al., 2009]. The latitudinal limits of ice stability have been used as evidence to support an ancient origin for circumpolar ice [Fishbaugh and Head, 2000], which could not have formed in the modern regime if the ice is unstable. Located at 72.7°N, 164.5°E, 600 km south of Planum Boreum's edge, the Korolev impact crater falls nearly on but south of the perennial ice stability line, which, at the longitude of Korolev, is at approximately 74°N [Levrard et al., 2004]. The presence of water ice in Korolev [Kieffer and Titus, 2001; Armstrong et al., 2005] has therefore been regarded as enigmatic. If the ice in Korolev is older than the large shift in obliquity that occurred 5 Myr ago [Laskar et al., 2004], some hypotheses suggest that it was part of a widespread former ice cap [Fishbaugh and Head, 2000], and the internal stratigraphy will reflect that. However, if the ice is modern and therefore much younger than a former ice cap, it should have formed in place, be quasi-stable in the current climate regime, and have stratigraphy consistent with in-place deposition. This study uses orbital sounding radar to map the stratigraphy of the nearly 2 km thick water ice deposit in Korolev Crater [Moore et al., 2012] in order to analyze the stratigraphic architecture of Korolev to help ascertain its origin.

2 Data and Methods

Radar data for this study are from the Shallow Radar (SHARAD) on Mars Reconnaissance Orbiter. SHARAD is centered at 20 MHz frequency with a 10 MHz bandwidth [Seu et al., 2007]. SHARAD penetrates the surface of icy martian deposits, reflecting off subsurface permittivity variations. These reflections create time delay surfaces representative of isochrons. While SHARAD reflectors do not map directly to optical layers, they have been shown to strongly correlate with optical layering [Christian et al., 2013]. The theoretical vertical resolution of SHARAD is 8.4 m in water ice. SHARAD's spatial footprint is 3–6 km across track and compressed to 0.3–1 km in the along-track direction via focusing [Seu et al., 2007]. SHARAD data have been acquired since 2007 with more than 15,000 observations (tracks) recorded. Over 100 orbits have data acquired over Korolev Crater's central mound. Maps for this study were constructed using 69 of the observations crossing Korolev (Figure 1 and Table S1 in the supporting information). Some orbits were excluded from mapping based on data quality and orientation of the observation with respect to surface topography.

A fundamental complication with orbital radar sounding data is the presence of off-nadir surface echoes (clutter). Rimmed crater deposits are particularly difficult due to their size and shape. Reflections from the rims and crater edges create surface clutter at time delays coincident with expected subsurface signals. To mitigate this, our study makes heavy use of a coherent echo simulation model developed by the University of Texas at Austin [Holt et al., 2008]. The clutter simulation uses a topographic model to predict surface echoes and create a radargram derived purely from possible surface echoes; hence, the “cluttergram” is void of subsurface signals. In order to map subsurface signals, comparisons between cluttergrams and radargrams were performed for each radar observation (Figure 2). Mapping was conducted in time delay using commercially available seismic software. The seismic software provided the ability to stitch together multiple radargrams into a single image while mapping. This technique allowed consistent radar reflector mapping across all 69 orbits in this study (Figure 3). Data were then exported and depth corrected using a dielectric constant of 3.15, a value appropriate for water ice [Grima et al., 2009] and consistent with the findings of Armstrong et al. [2005]. The data were then positioned using a first return topographic correction algorithm developed by the University of Texas at Austin and converted into a geographic information system (GIS) compatible shapefile. Gridding of the data was performed using Environmental Systems Research Institute's ArcGIS software and the natural neighbor interpolation algorithm. The grid size for all radar-derived rasters presented in this work is 20 m by 20 m.

Figure 2

SHARAD radar observation 2342201000 crossing Korolev. Location is given by the red line in Figure 1. North is to the left. (a) Time delay radargram with bright near-surface reflectors and several deep reflectors. Time is one-way travel. (b) Clutter simulation showing echoes resulting only from surface topography. All reflectors in radargrams matching those in this simulation are not subsurface and were avoided when mapping reflectors. (c) Depth-corrected radargram using the permittivity of water ice (3.15). Notice how the geometry of the radar reflectors at depth drastically changes after conversion to the depth domain. Orange box shows location for Figure 2d. (d) Reflector truncation and a possible pinchout at the intersection with the surface.

Figure 3

An example of combining crossing radargrams to consistently map radar reflectors across the Korolev ice mound. (a) Nadir track of the nine radar orbits used for mapping demonstration is shown in bright yellow on this shaded topography map of Korolev Crater. (b) Radar compilation including the interpretation of both the surface and a subsurface reflector. The gridded version of this reflector is shown in Figure 4. (c) Radar compilation without interpretation. Notice how the clutter does not often align between crossing radargrams, as expected due to changing geometry.

Paleotopography representative of an empty (or nearly empty) Korolev Crater was created from a combination of SHARAD and Mars Orbiter Laser Altimeter (MOLA) data. This topography was used to verify the results of our stepped volumetric calculation as well as to aid visual analysis of reflector mapping results. To construct our paleosurface, SHARAD data were used over Korolev Crater and the deepest visible SHARAD reflector was chosen to represent the crater's base. The GIS data shapefile for Korolev's deepest reflector was combined via nearest neighbor interpolation with MOLA shot point data for the crater rim and surrounding terrain. MOLA shot point data were also included for the crater wall taking precaution to delete any shot point data that were influenced by Korolev's central mound.

3 Results

SHARAD reflector mapping in Korolev Crater has revealed domal reflector geometry (Figure 4). Perhaps more significantly, it has revealed that this geometry has not changed by any appreciable amount through time. From the base upward, the layers have a distinct asymmetry, southward dip, and gradually increasing domal shape. These general mapping results are consistent with, and extend, the findings of Conway et al. [2012] that were derived from optical layer mapping and a single SHARAD radargram. While small-scale fluctuations in morphology do exist, as well as local reflector truncations, the regional reflector trend is consistent.

Figure 4

(a) Transect across the gridded reflectors from A-A′ in Figure 1. The bottom of the crater is a gridded result using our deepest SHARAD reflector but shown with a dashed line to indicate our uncertainty. Notice the reflectors truncating into the surface near 80–90 km transect distance. Additionally, all reflectors are dipping southeast with a maximum elevation shifted north of the crater's center. (b–d) Subsurface reflectors gridded and depth corrected with 30X vertical exaggeration. Each square grid is 10 × 10 km. The depth from the top reflector to the estimated bottom is ~ 1.8 km.

Ten of Korolev's internal SHARAD reflectors were gridded into 3-D surfaces. While several additional reflectors exist, these 10 were the most distinct and continuous, giving the highest confidence in their reconstructed morphologies (Figure 4). The chosen reflectors range from shallow to deep and are therefore representative of the entire vertical column of ice. Using a stepped volumetric calculation, this study finds a minimum ice volume of ~1400 km³ contained in Korolev. The volume approximation was done by subtracting the elevation value of adjacent reflectors, upper reflector minus lower reflector. This created isopach maps for each reflector set which were then used to calculate volume. As lower reflectors generally span less area and SHARAD does not see to the crater edges, this stepped calculation is underestimating ice volume. An additional ice volume calculation was performed using only the SHARAD-derived Korolev modern surface and our interpolated base. This represents our maximum estimate for ice volume in Korolev Crater and likely overestimates the volume. The result from this calculation was 3400 km³, consistent but significantly larger than our minimum estimate using the stepped calculation. As the minimum estimate includes no material nearer the crater wall than SHARAD data allows, this factor of 2 difference is within reason. Clutter prevents reflector mapping close to the crater wall, and attenuation or scattering of the radar signal prevents definitive mapping of the crater base (discussed below).

While our mapped reflectors do span the entire column of Korolev's ice, reflector spacing is variable. There is a high density of reflectors very near the surface to a depth of approximately 250 m. However, directly below this reflector-dense region is a zone with very few to no reflectors (Figure 2). The thickness of this reflector-free region can reach 550 m. Below the reflector-free zone lies another zone of dense radar reflectors with a thickness greater than 100 m. This general reflector pattern is repeated, creating three to four packages of reflectors with varying thickness.

Mapped reflectors in Korolev have consistent morphology with a dominant southern dip. The magnitude of dip does change, but the orientation remains consistent. In addition, the reflectors have a distinct asymmetry with generally thicker material on the north facing slope and thinner material on the south facing slope (Figure 4). All mapped reflectors, except the hypothesized base, have a roughly domal shape. The deepest mapped reflector is nearly flat with only 0.5° mean slope. This reflector is approximately 1.8 km deep and is used as our approximated crater bottom. Being both discontinuous and faint, this deep reflector has limited mapping coverage and was only identified in 8 of the 69 observations used for this study.

4 Discussion

Reflector geometry can be used to assess the two dominant hypotheses for the origin of ice in Korolev Crater. If the ice was originally part of a regional ice sheet formed during a previous glacial regime, we expect the internal reflectors to either continue laterally across the crater with little dip or possibly be concave upward as ice gradually filled the depression. Concave upward reflectors would be the result of large-scale flow, as is found with terrestrial ice sheets [Robin et al., 1969; Pattyn et al., 2008]. However, the reflector geometry in Korolev neither mimics hypothesized basal topography nor continues unperturbed laterally across the crater. The topography instead forms a 3-D-mounded deposit offset north from the crater's center (Figure 4).

The asymmetrical deposition and domed reflector shape support the second hypothesis that Korolev's central mound was formed by in-place deposition, a finding consistent with optical layer mapping results and a single radar profile [Conway et al., 2012]. However, this alone does not resolve the question of age. While our results help confirm that Korolev's material was not sourced from a regional ice sheet extending from the pole, the timing of localized deposition still remains unknown. While we cannot negate the possibility of ancient in-place deposition preserved through many periods of high obliquity, we can provide observations that support a recent origin.

First, the ice in Korolev is not protected like the ice within midlatitude glaciers [Holt et al., 2008]. We make this assertion on the basis of imagery of exposed layering [Conway et al., 2012], thermal inertia measurements and seasonal observations examined by Armstrong et al. [2005], and radar signal penetration. While an argument for deeper lag deposits could be made, there is no indication that a particular radar reflector is substantially different (either brighter or more diffuse) than the others, as expected from a buried lag deposit. Furthermore, SHARAD is penetrating approximately 1.8 km; similar to the north polar layered deposits (NPLD) where minimal signal attenuation allows such penetration [Grima et al., 2009]. Without a lag deposit, ancient ice would likely not survive obliquity swings. The abundance of layering in imagery [Conway et al., 2012] and SHARAD data, as well as thermal inertia results [Armstrong et al., 2005] and lack of a protective surface lag, are all consistent with the hypothesis that Korolev's ice mound is relatively pure water ice of recent age, similar to that of the NPLD [Laskar et al., 2002; Levrard et al., 2007; Greve et al., 2010].

As mentioned in section 3, the radar reflectors do not have regular vertical spacing. The reflectors can best be approximated as alternating packages of reflector-dense and reflector-sparse zones. There are three such packages of material that share this pattern with possibly a fourth very near the base of Korolev. This trend of reflectors closely follows the observed reflector density variations for the NPLD (Figure 5) [Phillips et al., 2008].

Figure 5

A comparison of radar reflector packaging in Korolev and Planum Boreum at identical vertical and horizontal scales. (a) Korolev Crater radargram 554201000. Three packages of reflector dense and reflector sparse material are present with a tentative fourth package near the base. These packages are labeled A–D. (b) Radargram 529701000 with NPLD reflector packages as mapped by Phillips et al. [2008]. The location for each radargram can be found in Figure 1.

The reflectors in Korolev share additional similarities to those of the NPLD. The uppermost section of ice has the highest reflector density, and within these dense reflectors exist unconformities. The unconformities are truncations of radar reflectors as pinchouts, downlap, or angular unconformities (Figure 2d). Both Korolev Crater and the NPLD share these shallow unconformities [Tanaka, 2005; Tanaka et al., 2008; Conway et al., 2012]. In addition to similar reflector density and stratigraphic patterns, the overall thickness of Korolev's ice is comparable to that of the NPLD. As measured directly from radargrams as well as calculated from raster grids, maximum thickness of Korolev ice is just over 1.8 km while the maximum thickness of the NPLD is approximately 2.3 km [Brothers et al., 2015]. The average NPLD thickness is only 1.1 km with a standard deviation of 540 m. An average thickness for Korolev would not be meaningful due to the limited extent of the dome. The similarity of each deposit's thickness, reflector packaging, and unconformity relationships supports a shared genetic relationship between NPLD and Korolev water ice. Therefore, it is likely that ice in each was deposited contemporaneously as a result of the same global climate forcing.

While stratigraphic architecture suggests coeval deposition of the deposits in Korolev and the NPLD, it does not address hypotheses [Fishbaugh and Head, 2000; Madeleine et al., 2009] suggesting that ice should be unstable in Korolev and therefore must be a relict deposit. Ice growth modeling has not been performed for Korolev; however, global climate and ice growth models [Levrard et al., 2007; Greve et al., 2010] driven by orbital parameters [Laskar et al., 2004] indicate that the NPLD is likely to be less than 5 Myr old and may only be 4 Myr old. Figure S1 shows a nested, mesoscale atmosphere modeling result for Korolev indicating that at least in the qualitative sense, ice can be stable in the floor of an empty Korolev under current atmospheric conditions. Comprehensive atmospheric modeling including ice accumulation while comparing Korolev to the NPLD may help address the question of stability and age more thoroughly. The three-dimensional structure of Korolev's central mound that we present here provides a new framework with which to conduct and compare modeling results.

5 Conclusions

Korolev Crater's central mound is a 1.8 km thick, domed water ice deposit. Reflector geometry in the deposit has a consistent trend with little change through time. While complexities such as unconformities and reflector pinchouts exist close to the surface, there is no evidence for large-scale erosional events in the radar stratigraphy; therefore, conditions for ice deposition appear to have remained largely stable as the deposit grew in place. This deposit was not part of a regional ice sheet that has since been eroded.

This work also finds that Korolev's central deposit is likely coeval to Planum Boreum's NPLD. Similar thickness, stratigraphic relationships, and radar packaging support our hypothesis that these two deposits are likely genetically related and formed during the same time period. In addition, radar sounding supports optical- and thermal-based imaging studies that do not find evidence for debris cover that could protect Korolev's central mound during periods of high obliquity. The similarity of radar reflectors between Korolev and the NPLD lends credence to early work suggesting that the polar layered deposits record regional climatic conditions [Cutts and Lewis, 1982; Laskar et al., 2002]. This work provides evidence that the climatic signal responsible for the NPLD is more widespread and not limited to Planum Boreum. This work hypothesizes a climatic link between circumpolar depositional features and Mars' NPLD, indicating that the same signal was likely preserved by these deposits separated by 600 km. Thus, we find that martian polar climate (and paleoclimate) can best be deciphered through a unified study of Planum Boreum and circumpolar ice features.

Acknowledgments

This work was supported by NASA grants NNX10AO26G and NNX11AL10G to J.W.H. as well as the MRO Project through a SHARAD Co-I contract to J.W.H. Data used in this study can be found on the Planetary Data System at http://pds-geosciences.wustl.edu/missions/mro/sharad.htm. Both raw data products and the U.S. SHARAD processor products that were used in this study are available there. We would like to thank Aymeric Spiga for his assistance with the LMD mesoscale model and for encouraging us to consider broader implications. We also thank James W. Head III and Susan Conway for their time in reviewing the manuscript.