Catastrophic floods generated ~3.2 Ga by rapid groundwater evacuation scoured the Solar System’s most voluminous channels, the southern circum-Chryse outflow channels. Based on Viking Orbiter data analysis, it was hypothesized that these outflows emanated from a global Hesperian cryosphere-confined aquifer that was infused by south polar meltwater infiltration into the planet’s upper crust. In this model, the outflow channels formed along zones of superlithostatic pressure generated by pronounced elevation differences around the Highland-Lowland Dichotomy Boundary. However, the restricted geographic location of the channels indicates that these conditions were not uniform. Furthermore, some outflow channel sources are too high to have been fed by south polar basal melting. Using more recent mission data, we argue that during the Late Noachian fluvial and glacial sediments were deposited into a clastic wedge within a paleo-basin located in the southern circum-Chryse region, which at the time was completely submerged under a primordial northern plains ocean. Subsequent Late Hesperian outflow channels were sourced from within these geologic materials and formed by gigantic groundwater outbursts driven by an elevated hydraulic head from the Valles Marineris region. Thus, our findings link the formation of the southern circum-Chryse outflow channels to ancient marine, glacial, and fluvial erosion and sedimentation.
At the end of the Noachian Period Mars experienced an abrupt transition into a climate dominated by extremely cold and dry conditions, which resulted in the subsequent confinement of the planet’s hydrosphere in highly pressurized aquifers beneath thick upper crustal permafrost materials1,2. A popular model proposes that the vast Noachian surface water systems were cold-trapped in the planet’s south polar region to form an ice cap, which gradually infiltrated into a global mega-regolith as a consequence of pressure-induced basal melting1,2. The catastrophic floods that excavated the Late Hesperian southern circum-Chryse outflow channels3 would have occurred in zones where the elevation difference between the Martian south polar region and lower terrains generated superlithostatic pressures1,2.
In order to explain the relative localized occurrence4,5 of catastrophic-flood-formed Martian outflow channels, it was hypothesized that the causative flooding emanated from a compartmentalized global hydrosphere that was also contained within a planetwide megaregolith6. In this model, the existence of high elevation outflow channels in the circum-Chryse region7,8 can be explained by the presence of an aquifer system that extended from the Tharsis Montes to the outflow channel head source regions9. It was subsequently discovered that rather than possessing global megaregolith, much of the Martian upper crust appears to be dominated by Noachian sedimentary deposits of great stratigraphic thicknesses that both infill and bury numerous impact craters10,11. Hence, it was proposed that, instead of a megaregolith-trapped hydrosphere1,2,6, significant portions of these outflow-channel-source (OFCS) aquifers might have consisted of water-ice contained within buried impact craters and impact-fractured rocks12,13. However, because the sedimentary deposits are distributed globally on Mars, the unique conditions leading to the formation of the southern circum-Chryse outflow channel source regions are still poorly understood. Here, we propose a new geologic model linking the development of OFCS aquifers within the highlands of southern circum-Chryse (Fig. 1a) to a stage of Late Noachian large-scale regional marine sedimentation.
Results and interpretative synthesis
The long-held view is that regional groundwater outbursts led to the formation of extensive chaotic terrains (collapsed upper crustal materials) that are the sources of the southern circum-Chryse outflow channels5,14,15,16,17,18,19,20. More recently, zones of surface subsidence have been recognized as including much more extensive areas of adjacent highlands12,13,21. Consequently, the distribution of regional highland surfaces modified by collapse and subsidence were proposed to demark the approximate extent of the OFCS aquifers12,13.
These chaotic terrains have been mapped in detail by previous investigators4,16,22, and here we present the first map showing the full extent of highland subsidence in this region of the planet (Fig. 1b,c). We show that these previously unmapped subsidence zones modify large areas of the highlands and connect eastern Valles Marineris to the initiation zones of numerous outflow channels (Fig. 1b). The subsided terrains include regionally extensive networks of broad valleys (e.g., Fig. 1b,c) that are characterized by faulted margins (Fig. 1c–e) and abrupt breaks-in-slope (Fig. 1c–e) that deform pre-existing landforms (e.g., Fig. 1d,e). These formational structures have been interpreted as synclinal and monoclinal folds and associated ruptures developed by subsidence over caverns that formed by the removal of ice, water, and fluidized sediment12,13.
The southern Margaritifer Terra highlands are extensively dissected by vast canyon systems (Fig. 1b and 2a), which exhibit flanks densely marked by channels (e.g., Fig. 2b)4. The formation of these canyons is thought to have been dominated by Middle and Late Noachian fluvial erosion4. Late Noachian equatorial glaciers may have also contributed to their development23. Within the study region, these canyons converge northwestward into the highlands containing the subsided terrains (Fig. 1a,b). Although the contact between these two terrain types is disrupted by the Margaritifer and Iani chaotic terrains, we have identified a relatively narrow stretch of plateau materials that includes a subset of canyons that exhibit transitional morphologies (Fig. 2a). These hybrid canyons include scarp margins marked by dense channel systems and extensional fractures generated by subsidence (Fig. 2c,d), thus offering a rare insight into the regional upper stratigraphy of the outflow channel source region, consistent with the existence of buried canyons underlying the subsided valleys. Similar hybrid morphologies are observable along the flanks of Ladon basin (Fig. 1a), which constitutes large impact basin that partly captured drainage from the upland fluvial canyons (Fig. 1b).
The terrain contact between the upland and hybrid canyons is defined by elevations ranging between −2050 m and −1900 m (Fig. 3a, label 1), which also mark a section of the dichotomy boundary to the west (Fig. 3a, label 2). The dichotomy boundary corresponds approximately to the margins of water bodies (“Oceanus Borealis”24) that episodically covered the northern plains, most extensively during the Late Noachian24,25,26,27,28 and Late Hesperian25,26,27,28,29. The mean elevation of the Late Noachian ocean shoreline, which is the one of relevance to this study, has been estimated to be approximately −1680 m25. The elevation range also characterizes the inter-crater plains surfaces in western Arabia Terra (Fig. 3a, label 3), a region of the planet which has been interpreted as a frozen, buried remnant of the Late Noachian ocean30, where localized desiccation of the regional hydrosphere generated the widespread evaporite deposits detected by the Opportunity rover31.
We propose the following regional geologic reconstruction leading to the development of the OFCS aquifers: The vast fluvial, or glacial-glaciofluvial, canyon systems of southern Margaritifer Terra4,23 discharged enormous volumes of sediments into southern circum-Chryse, where lower portions of these canyons were completely buried and integrated into the regional upper crustal stratigraphy (Fig. 4a). Coeval fluvial activity through the huge Uzboi-Ladon-Morava channel system, which connects the Argyre basin to the northern plains, is thought to have discharged vast volumes of water generated by the melting of a Late Noachian south polar ice sheet32. This enormous fluvial system could have also contributed to regional sedimentation in southern circum-Chryse2,33. In the proposed geologic scenario the upper extent of regional sedimentation would have been controlled by the elevation of Oceanus Borealis (Fig. 3b, sketches 1 and 2 in Fig. 4a).
We infer a relatively short-lasting duration for the conditions allowing deep upland erosion connected with the formation of the proposed massive clastic sedimentary wedge based on previous investigations that propose a major spike in erosional and sedimentation rates occurring near the Noachian-Hesperian boundary34,35,36,37,38,39,40,41,42,43. This geologic stage likely lasted just a few million years35. To establish a relative chronology for the proposed stage of large-scale sedimentation, we mapped the distribution of (1) collapsed craters, (2) buried craters, (3) flat-floored impact craters infilled up to their rims, and (4) impact craters that retain significant topography (Fig. 5a). Our mapping of buried craters is based on the identification of quasi-circular depressions distributed throughout the regional highlands. These features are thought to have formed by compaction of sediments overlying buried impact craters44. Impact crater statistics yield a Late Noachian age of 3.65 ± 0.01 Ga (Fig. 5b), and point to a major spike in impact cratering rates during this time period. This spike was likely the result of a late phase of the Late Heavy Bombardment thought to have affected Mars up to 3.6 Ga45. Increased global erosional and depositional rates associated with an active surface hydrosphere34,35,36,37,38,39,40,41,42,43 during the Late Heavy Bombardment are consistent with impact-induced climate change as proposed by Segura et al.46.
The rapid groundwater evacuation required to generate the Late Hesperian catastrophic floods would have required groundwater migration via extraordinarily permeable structures, most likely including extensive systems of large interconnected caverns filled with water and/or unconsolidated fluidizable sediment12,13. We estimate that ~2.8 × 105 km3 of groundwater and eroded or fluidized sediments were rapidly evacuated from the hypothesized caverns. This value was calculated as the total area of subsided highlands (~4 × 105 km2) times an average negative relief of ~700 m [−1200 m minus −1900 m, respectively, elevation averages of undeformed highland plateaus and adjoining subsided terrains (Supplement)].
The distribution of the subsided valleys, here proposed to have been largely controlled by the distribution of the ancient buried troughs, indicates that the hypothesized caverns formed a direct connection between eastern Valles Marineris and a lower-lying equatorial belt of chaotic terrains located at the upstream portions of numerous outflow channels (dashed yellow line in Fig. 1a). These chaotic terrains occur within highland surfaces located within a narrow range of surface elevations (approx. ~500 m to −1000 m), which is consistent with groundwater outbursts produced by rapid release from an initially confined, underground body of water conforming to a potentiometric surface defined by the upper boundary of groundwater saturated crust in eastern Valles Marineris. Vast aquifers confined beneath a seal of ice-cemented permafrost are proposed to have existed within eastern Valles Marineris and nearby plateau regions during the Late Hesperian47,48,49. These aquifers are thought to have reached superlithostatic pressures as a consequence of ~3.5 to ~4.5 kilometers of hydraulic head generated due to the warping of regional upper crustal rocks by the Tharis uplift19.
Elaver Valles (Fig. 1a), located at an elevation of ~2000 m, is the highest outflow channel in eastern Valles Marineris. In contrast, the northern margin of the subsided terrain adjoining upper Ares Valles exhibits elevation ranges between approximately −1500 and −2500 m (Supplement). Moreover, elevation profiles from nearby undeformed highland surfaces show that the pre-subsidence topography was likely closer to approximately −1200 m, ranging between approximately −500 m and −1500 m (Supplement). These relief estimates provide a basis to quantify relevant hydraulic pressures. Prior to the catastrophic outflows, hydrostatic pressures within the confined, water-filled caverns could have attained hwgρ between ~13 and 16.5 MPa (or 11 MPa, if corrected to pre-subsidence topography hw (~3000 m)), where hw is the relief in meters (~3500 m to ~4500 m), g is Martian surface gravity (3.711 m/s2), ρ is water density (1000 kg/m3). These estimates of regional groundwater pressurization approximate or exceed the upper estimate predicted to have caused outflow channel activity in circum-Chryse [10 MPa]50. The pressure may have been even greater if the groundwater was highly saline, saturated in CO2, or contained large volumes of fine-grained sediments. Following the groundwater outbursts, the hydraulic head within the previously confined cavernous water systems would have rapidly diminished due to drawdown of the regional water table, however, continued exsolution of CO2 could have maintained the outflow channel discharges.
Our proposed scenario implies that the sedimentary deposits that buried the troughs were highly porous and likely included large volumes of hydrated clays and perhaps glacal ice. As Oceanus Borealis receded, these water-saturated deposits would have progressively frozen into extensive areas of ice-rich permafrost (Fig. 4a, sketch 2), which remained stable for a few hundred million years.
Subsequent Late Hesperian groundwater eruptions and subsidence would have been controlled by the distribution of the buried troughs (Fig. 4a, sketches 3–5). Consequently, the water-filled caverns that produced the immense outflow channel discharges and related subsidence12,13 must have been excavated within the troughs’ permafrost or water-saturated infill. The generation and integration of these caverns have been attributed to localized permafrost melting by magmatic dikes in a way that large groundwater bodies were generated and confined by surrounding icy Late Noachian strata12,13,21, which likely contained significant volumes of less permeable phyllosilicates23. The integration of these caverns and their connection to a kilometers-deep aquifer system in Valles Marineris47,48,49 would have allowed for the role of superlithostatic pressures (Fig. 4b) in driving immense discharges to the southern Chryse basin. An alternative scenario might also involve the pressurization of meltwater beneath the Valles Marineris ice sheet51.
We find that Southern circum-Chryse had a unique geologic history that combined large-scale fluvial sedimentation along the margin of an ancient Late Noachian ocean; conduit development within the sedimentary materials infilling vast systems of buried troughs; and the development of superlithostatic pressures within the lower reaches of these conduits, perhaps driven by elevated potentiometric surfaces transferred from eastern Valles Marineris. High hydraulic pressure gradients across large regions of water-filled caverns permitted the rapid drainage of the immense groundwater systems. This region of the planet contains the largest outflow channel system on Mars, so our hypothesis, including the precursory erosional and sedimentation processes, centers on one of the most dramatic hydrogeologic stories in the Solar System.
On Mars there is evidence for the long-term retention of large groundwater volumes within buried impact craters12,52 and within thick upper crustal hydrated mineral deposits53. Clastic wedges containing large amounts of relic massive ice have been recognized within relatively young Earth deposits emplaced during recent glacial periods. These include glacial and glaciomarine sedimentary sequences54,55,56,57 produced during high sea-level stands58 that infilled and buried older valleys excavated by glacial and glaciofluvial erosion54,55,56,57. However, frequent warm climatic excursions on Earth imply that these ice masses can only remain stable over relatively short geologic periods. On the other hand, the colder climate and lower geothermal heat flow on Mars could have allowed Martian clastic wedges to retain vast volumes of ice during billions of years. This model and our new observations provide support for the Martian primeval ocean24,25,26,27,28 and link the formation of the vast outflow channel source region aquifers to the emplacement of thick water-rich upper crustal sedimentary deposits during a spike in global hydrologic activity34,35,36,37,38,39,40,41,42,43. The spike was perhaps triggered by impact-induced climate change46 during a late phase of the Late Heavy Bombardment45. If these extensive Late Noachian aquifer systems formed globally, then their development could have significantly depleted the surface hydrosphere, thereby playing a potentially significant role in triggering the abrupt transition into the colder and drier conditions that characterized most of the post-Noachian climatic regimes1. Thus, global Late Noachian sedimentary infill of buried canyons and craters might still retain much of the planet’s hydrosphere.
How to cite this article: Rodriguez, J. A. P. et al. Martian outflow channels: How did their source aquifers form, and why did they drain so rapidly? Sci. Rep. 5, 13404; doi: 10.1038/srep13404 (2015).
We are grateful to James Head III for his insightful comments and suggestions, which significantly improved the quality of this manuscript. Funding provided by NASA’s NPP program to J. Alexis P. Rodriguez and by MRO HiRISE Co-Investigator funds to V.C. Gulick. A. G. Fairén was supported by the Project “icyMARS”, funded by the European Research Council, Starting Grant no 307496. Hideaki Miyamoto was supported by KAKENHI 23340126. We would also like to express our gratitude to Alexander Cox for his helpful suggestions regarding the editing of this paper. HiRISE images were analyzed using HiView developed by the Lunar and Planetary Laboratory at the University of Arizona.
About this article
Rights and permissions
This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/