A Noachian proglacial paleolake on Mars: Fluvial activity and lake formation within a closed-source drainage basin crater and implications for early Mars climate

A 54-km diameter Noachian-aged crater in the southern highlands of Mars contains unusually well-preserved inverted fluvial channel networks and lacustrine deposits, all of which formed completely inside the crater. This closed-source drainage basin (CSDB) crater is distinct from previously documented fluvially breached or groundwater-fed crater basin lakes on Mars. We compare our observations to previously established models of crater degradation, fluvial incision, and topographic inversion on Mars to assess the most likely origins of the water that formed the fluvial and lacustrine features. We favor top-down melting of a cold-based glacier as the source of water in the CSDB crater, which would represent the first examples of proglacial fluvial channels and lakes found on Noachian Mars.


INTRODUCTION AND BACKGROUND
Geologic evidence suggests that the ambient climate of early Mars (Noachian; ~4.0-3.6 Ga) was significantly different than today, with compelling evidence for extensive fluvial channels and crater basin lakes, increased erosion rates, and enhanced crater degradation (Craddock & Howard 2002;Howard et al. 2005;Irwin et al. 2005). Indeed, compared to younger craters, most Noachian-aged craters have 1) lowered or completely removed rims; 2) flat, infilled floors; and 3) walls and ejecta deposits modified by fluvial incision and backwasting (Craddock et al. 1997;Craddock & Howard 2002;Mangold et al. 2012). Morphometric studies have interpreted this set of characteristics to indicate degradation by a dual advective-diffusive process (Craddock et al. 1997;Craddock & Howard 2002;Forsberg-Taylor et al. 2004;Matsubara et al. 2018). One such pair of processes involves diffusive rainsplash and advective runoff from rainfall occurring intermittently over tens of millions of years (e.g. Hoke et al. 2011) in a warm and wet Noachian climate (e.g. Craddock & Howard 2002).
Although not favored by Craddock and Howard (2002), they also noted that "it is possible that solifluction generated the diffusional component [of crater degradation] while any surface runoff was the result of snowmelt." This would favor a colder, potentially subfreezing Noachian climate with only transient warming. Some global climate modeling studies support ambient Noachian temperatures well below freezing Wordsworth et al. 2013Wordsworth et al. , 2015 with cold-based glaciation occurring in the southern highlands (Head & Marchant 2014;Fastook & Head 2015). Thus, the nature of the ambient Noachian martian climate is currently debated.
In addition to their degraded morphology, numerous Noachian-aged craters on Mars have been identified as candidate sites of former lakes. Fassett & Head (2008b) identified 210 open-basin lakes (OBLs), craters fed by valley networks that contained outlet valleys indicative of lake filling and overflow. Goudge et al. (2015) identified an additional 205 candidate closed-basin lakes (CBLs), similar to OBLs but lacking outlet channels. Most OBLs and some CBLs are fed by regionally integrated valley networks, suggesting that these paleolake basins formed at a time similar to the peak of valley network activity around the Noachian-Hesperian boundary (Fassett & Head 2008a), and thus may have required sustained warm and wet climate conditions in order to form (e.g. Irwin et al. 2005;Howard 2007; Matsubara et al. 2011). Salese et al. (2019) identified 24 craters near the dichotomy boundary with floor elevations below -4 km that contained crater wall sapping valleys, terraces, and deltas that they 3 interpreted to be sourced from groundwater upwelling; these also appear to have formed around the Noachian-Hesperian boundary.
We report on the geology of a 54-km diameter Noachian-aged crater in the southern highlands (Figs. 1-3) that contains unusually well-preserved inverted fluvial channel networks and lacustrine deposits. Unlike previously described OBLs and CBLs (Fassett & Head 2008b;Goudge et al. 2015), this crater has neither inlet nor outlet channels and shows no evidence of sapping valleys or other associated landforms suggesting groundwater processes (e.g. Salese et al. 2019). On the basis of these unique characteristics, we term this new type of crater a "closed-source drainage basin" (CSDB), distinct from either fluvially breached or groundwater-fed crater basin lakes on Mars. We compare possible water sources for the fluvial and lacustrine features within the CSDB crater with the proposed origins of other crater basin lakes and inverted terrain features on Mars. We favor top-down melting of a cold-based glacier as the source of water in the CSDB crater. This interpretation would provide observational evidence of the cold and icy early Mars climate predicted by certain models and would represent the first examples of proglacial fluvial channels and lakes found on Noachian Mars.

GEOLOGY OF THE CSDB CRATER
The 54-km crater 20.3°S,42.6°E) is located in Noachianaged highlands terrain (Tanaka et al. 2014) ~800 km northwest of the Hellas basin rim in Terra Sabaea. In a regional study of Terra Sabaea crater floors, Irwin et al. (2018) noted that the interior of this crater, which they designated "B," contained a dark-over-light stratigraphy consisting of a more resistant lower unit underlying a more erodible upper unit outcropping as knobs and mesas. The lower light-toned unit exhibits an unusual spectral signature consistent with Fe 2+ substitution in plagioclase (Wray et al. 2013;Irwin et al. 2018) that sets it apart from more prevalent light-toned units in the region such as phyllosilicates and chlorides. Irwin et al. (2018) identified Al and Fe/Mg smectite in the walls of crater B, but these are both morphologically and spectrally distinct from the light-toned floor. Sinuous ridges preserved in the darktoned upper unit were interpreted by Irwin et al. (2018) as inverted fluvial channels.
On the basis of this previous work, we mapped the rim, walls, and interior of crater B in detail using 5 m/pix Context Camera (CTX, Malin et al. 2007 to the lowest points of the crater ( Fig. 2A), where they terminate in nowdeflated depositional basins. Their proximal morphology is characterized by bifurcated clusters of narrow, subparallel ridges whose heads begin ~1 km below the upslope-facing scarps. The proximal ridge network in the east merges into a wider, flat-topped medial morphology downslope; approximately a dozen medial ridges then coalesce into four larger trunk ridges (Fig. 1B). The smaller network in the south maintains a mostly medial morphology before narrowing into smaller ridges near their termini; one ridge displays an anabranching morphology (Fig. 1C). We interpret these ridges as networks of fluvial channels that have been preserved in inverted relief, in agreement with Irwin et al. (2018).

Depositional Basins
Two depositional basins are located in the lowest parts of the crater floor (Fig. 1A, cyan/green). The larger basin I is outlined by outcrops of the lower light-toned floor unit. Most of the ridges that originate from below the upslope-facing scarps on the eastern crater wall terminate in basin I. Basin II is readily identified by its distinct texture of features we interpret as transverse aeolian ridges (TARs; e.g. Balme et al. 2008) that are more densely spaced than in the surrounding crater floor.
On the basis of topography and morphology (Figs. 1-2), we interpret these two basins as former lakes and depocenters for sediment transported 5 by the inverted fluvial channels. The mouths of the trunk streams flowing into basin I follow a topographic contour at ~2050 m (Fig. 2D), suggesting they were controlled by an equipotential surface. Channels flowing into basin II terminate within meters of the morphological boundary of the basin in several locations, which could indicate rapid dispersal of sediment into a standing body of water (Fig. 1C). Irwin et al. (2018) interpreted the two basins as distinct units, but our analyses indicate that the two basins consist of the same bedrock lithology with varying degrees of aeolian cover. This aeolian sediment is likely to have been reworked from the original lacustrine deposits that filled the two basins and subsequently revealed the underlying light-toned bedrock. The concentration of TARs within basin II in consistent with an increased supply of fine-grained sediments potentially derived from the former lakebed. The lack of typical depositional morphologies such as fans and deltas within either basin suggests that the instantaneous sediment supply into the basins was relatively low, but enough sediment accumulated for it to be reworked or removed by subsequent aeolian activity.
We performed order of magnitude calculations of sediment flux through the channels following the paleohydrologic reconstruction methods of Rosenberg et al. (2019) and Hayden et al. (2019), assuming that the ridges represent channel belt deposits as opposed to single channel fills (Appendix B). Assuming only one channel was active at any given time, our calculations indicate that the paleolake basins could be filled with sediment in less than a single Earth year of constant flow. Given the more likely case of intermittent or seasonal activity, the channels were likely active over a period no greater than ~10 3 years. In the case of inverted fluvial channels in Arabia Terra, Davis et al. (2016Davis et al. ( , 2019 noted that valley to ridge transitions are correlated with slope breaks, indicating that the ridges are likely to have formed out of alluvial channel deposits. Fassett and Head (2007) identified inverted channels in Arabia Terra in association with a Noachian-Hesperian mantling deposit. The characteristics of this mantling unit suggest that volatiles were incorporated either during or immediately following its emplacement. They suggested that the inversion process involved sublimation and volatile loss, combined with aeolian deflation of unconsolidated material, leaving behind indurated fluvial channel deposits.

Alcoves and Upslope-Facing Scarps
Topographic data indicate that the inverted channels in crater B initiate at a slope break at the base of the crater wall ( Figs. 2A, 3), which matches the depositional style inferred by Davis et al. (2016Davis et al. ( , 2019. The thermophysical properties of the channels from THEMIS quantitative thermal inertia data (Fergason et al. 2006) are intermediate between the high-TI, light-toned crater floor bedrock and the relatively 7 low-TI mantling materials on the crater rim and wall. According to Williams et al. (2018), this would suggest that the inverted channels are more competent than the surrounding loose sediment and were formed through induration. However, spectral identifications of primary mafic mineral phases in the inverted channels are consistent with limited postdepositional alteration , and HiRISE images of the channels show active weathering and talus slopes uncharacteristic of indurated materials, so grain armoring cannot be ruled out as an alternative.
We interpret the fluvial channel deposits in crater B to have originally contained alluvial sediment lags derived from mantling deposits within the crater. The nature of the mantling deposit is likely to have been similar to those observed in Arabia Terra, and elsewhere in Terra Sabaea, consisting of airfall deposits from climate-driven dust and volatile deposition, atmospherically transported impact ejecta, or plinian volcanic tephra (Fassett & Head 2007). Fluvial processes concentrated sediments within the fluvial channels and transported suspended loads either into lakes or floodplains, where they were eventually removed or redistributed. Topographic inversion resulted from aeolian deflation, perhaps combined with sublimation of co-deposited volatile layers and cement, as suggested by Fassett and Head (2007)  3.1. Regional Fluvial Processes 8 The distribution of degraded Noachian-aged craters in the southern highlands suggests that the processes responsible for their degradation were widespread (Craddock & Maxwell 1993;Craddock et al. 1997;Irwin et al. 2013). Dawes crater (Fig. 4B) has many of the hallmarks described by Craddock and Howard (2002) and others as typical of Noachian crater degradation: the crater wall has been heavily incised and backwasted by fluvial erosion, and the flat, relatively featureless floor has been infilled presumably by laterally transported sediments. By comparison, the wall and rim of crater B (Fig. 4A) are much more subdued, and no fluvial incision is observed. The floor of crater B contains a dense network of inverted fluvial channels that begin abruptly at the base of the crater wall and do not extend up to or beyond the rim. For these reasons, we believe external drainage from a regionally integrated valley network or inlet channels is the least likely water source in crater B. terrace that, in a groundwater-fed lacustrine environment, would be downcut on the side facing downslope. Finally, the crater B floor elevation (+2 km) is inconsistent with the elevation that groundwater-fed 9 lakes on Mars occupy, usually below -4 km (Salese et al. 2019). Thus, we believe that groundwater sapping is an unlikely water source in crater B.

Glacial Meltwater Processes
Thus far, we find that our observations in crater B do not match craters modified by rainfall-derived fluvial breaching or groundwater sapping. As noted by Craddock and Howard (2002), "…it is possible that…any surface runoff was the result of snowmelt." Snowmelt may refer to any kind of melting resulting from the heating of surficial water in its solid phase, and we thus investigate this option as a water source in crater B.  (Fastook & Head 2014) down to the exposed base of crater walls, forming terminal and lateral moraines whose boundaries are preserved as upslope-facing scarps (Berman et al. 2005;Jawin et al. 2018;Fig. 4F). In larger craters with shallower wall slopes, numerous glaciofluvial valley networks formed as a result of glacial meltwater drainage below the scarps (Berman et al. 2009;Fassett et al. 2010). Atkins 2013) as the result of stranded ice-cored ground exposed by glacial retreat; this process is distinct from the topographic inversion that we interpret for the formation of the ridges in crater B.
We conclude that episodic top-down melting of a cold-based crater wall glacier within crater B represents a plausible water source to form the observed fluvial and lacustrine features. This CSDB water source differs from previously considered hypotheses for the formation of other types of paleolakes on Mars, which call on water sources external to the crater (Fassett & Head 2008b;Goudge et al. 2015;Salese et al. 2019). In addition, ambient cold climates would favor the emplacement of climatedriven snow, ice and dust mantles (e.g. Head et al. 2003). The loss of such volatile components by ablation could significantly assist in the topographic inversion and mantle loss processes in crater B, as also envisioned in Arabia Terra (Fassett & Head 2007).

CONCLUSIONS
We describe a Noachian closed-source drainage basin (CSDB), a new type of paleolake on Mars with neither inlet nor outlet channels. An ensemble of ridges, basins, alcoves, and upslope-facing scarps found within the crater provides evidence for possible drainage sources and mechanisms leading to the formation of inverted fluvial channels and lacustrine deposits on the crater floor (Figs. 1-3). We find that the features within the crater are unlikely to have formed in a warm and wet early Mars climate dominated by fluvial valley network incision and crater breaching. The occurrence of inverted channels within the crater differs from descriptions of large, regionally integrated inverted channels that appear to be the depositional counterparts to traditional valley networks 11 (Davis et al. 2016(Davis et al. , 2019Dickson et al. 2020). The crater also does not share many of the characteristics of groundwater-fed crater basin lakes (fans, deltas, shorelines; Salese et al. 2019), and we thus consider groundwater to be an unlikely source of water in the CSDB crater.
Alternatively, we have explored the suggestion of Craddock and Howard (2002) Table A1. The CTX stereo DEM was generated with the Ames Stereo Pipeline (Beyer et al. 2018). Topographic profiles were measured from the DEM in ArcMap using the Interpolate Line tool. The global CTX image mosaic is described by Dickson et al. (2018).

Figure
Instrument Image Where Cf is the coefficient of friction, R is the submerged specific gravity of sediment, g is the gravitational acceleration, v is the kinematic viscosity of water, D50 is the median grain size, and d is the paleochannel depth. Hayden et al. (2019) specifically noted that the measured caprock thickness is likely to be ~1.5x the original paleochannel depth d. They also noted that the current slope of the geomorphic surface is likely not the original bed slope; instead, they inferred the bed slope S from the bankfull Shields stress * expressed as a function of the particle Reynolds number Rep: Caprocks are not distinguishable at the scale of our DEM, so for channel depth d, we conservatively assume a caprock thickness that is 10% the total ridge height, and use the minimum and maximum measured ridge heights for the inverted channels in crater B to bracket a possible range of d = 0.7-1.3 m. We use a median grain size of medium-coarse sand, D50 = 0.5 mm.
For the above values, we find stream velocities U = 2.5-2.7 m/s and bankfull discharges Q = 2,000-4,400 m 3 /s per channel for the trunk streams that flow into the basins.
We then calculate the total time that would be required to fill the two basins in crater B with a sediment cover 50 m thick, which is approximately the difference in elevation between the deepest part of the basins and the channel termination points. This results in a combined sediment volume of ~1.7x10 10 m 3 .
In order to determine the amount of sediment that flowed through the channels for a given discharge, we use the fluid-sediment ratio relation