high outﬂow channels on mars indicate hesperian recharge at … · 2013-03-26 · high outﬂow...

Icarus 189 (2007) 344–361www.elsevier.com/locate/icarus

High outflow channels on Mars indicate Hesperian recharge at low latitudesand the presence of Canyon Lakes

Neil M. Coleman a,∗, Cynthia L. Dinwiddie b, Kay Casteel c

a US Nuclear Regulatory Commission, Mail Stop T2E26, Washington, DC 20555, USAb Department of Earth, Material, and Planetary Sciences, Southwest Research Institute®, 6220 Culebra Road, San Antonio, TX 78238-5166, USA

c Mercersburg, PA 17236, USA

Received 12 September 2006; revised 8 January 2007

Available online 30 March 2007

Abstract

A number of martian outflow channels were carved by discharges from large dilational fault zones. These channels were sourced by ground-water, not surface water, and when observed on high-standing plateaus they provide indicators of elevated paleo-groundwater levels. We identifythree outflow channels of Hesperian age that issued from a 750-km-long fault zone extending from Candor Chasma to Ganges Chasma. Twoof these channels, Allegheny Vallis and Walla Walla Vallis, have sources >2500 m above the topographic datum, too high to be explained bydischarge from a global aquifer that was recharged solely in the south polar region. The indicated groundwater levels likely required regionalsources of recharge at low latitudes. The floodwaters that erupted from Ophir Cavus to form Allegheny Vallis encountered two ridges that re-stricted the flow, forming temporary lakes. The flow probably breached or overtopped these obstructions quickly, catastrophically draining thelakes and carving several scablands. After the last obstacle had been breached, a single main channel formed that captured all subsequent flow. Weperformed hydrologic analyses of this intermediate phase of the flooding, prior to incision of the channel to its present depth. Using floodwaterdepths of 30–60 m, we calculated flow velocities of 6–15 m s−1 and discharges in the range of 0.7–3 × 106 m3 s−1. Locally higher flow velocitiesand discharges likely occurred when the transient lakes were drained. Variable erosion at the channel and scabland crossing of MOLA pass 10644suggests that the upper 25–30 m may consist of poorly consolidated surface materials underlain by more cohesive bedrock. We infer that anice-covered lake with a surface elevation >2500 m probably existed in eastern Candor Chasma because this canyon is intersected by the OphirCatenae fault system from which Allegheny Vallis and Walla Walla Vallis originated. We introduce a new hydrology concept for Mars in whichthe groundwater system was augmented by recharge from canyon lakes that were formed when water was released by catastrophic melting offormer ice sheets in Tharsis by effusions of flood basalts. This model could help to reconcile the expected presence of a thick cryosphere duringthe Hesperian with the abundant evidence for groundwater as a source for some of the circum-Chryse outflow channels.© 2007 Elsevier Inc. All rights reserved.

Keywords: Mars; Mars, surface; Geological processes; Volcanism

1. Introduction

The transition from the Noachian to the Hesperian era repre-sents a significant shift in surface conditions on Mars. The crustformed during the Noachian and the planet was heavily bom-barded by planetesimals. It is likely that the atmosphere wasthickest and the water inventory highest during this time, be-

* Corresponding author.E-mail addresses: [email protected], [email protected]

(N.M. Coleman).

0019-1035/$ – see front matter © 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.icarus.2007.01.020

cause isotopic evidence shows that the planet has lost large frac-tions of water and other volatile inventories over geologic time(Jakosky and Phillips, 2001). Extensive valley networks devel-oped in the cratered highlands (Carr and Chuang, 1997), andlakes possibly formed in many craters (Cabrol and Grin, 1999).Noachian erosion rates are estimated to have been ∼1000 timesgreater than during later eras (Carr, 1996). By Hesperian time,which began ∼3.5–3.7 Gyr ago (Hartmann and Neukum, 2001),the atmosphere had thinned and a thick, planet-wide cryospheremay have evolved (Clifford, 1993). Immense shield volcanoesbegan forming in Tharsis, and flood basalts inundated large

http://www.elsevier.com/locate/icarus

mailto:[email protected]

mailto:[email protected]

http://dx.doi.org/10.1016/j.icarus.2007.01.020

High channels indicate Hesperian recharge and Canyon Lakes 345

Fig. 1. Regional map showing the Valles Marineris canyon system and the locations of Allegheny Vallis, Walla Walla Vallis, and Elaver Vallis. Floodwaters in thesechannels ultimately drained into the ancestral Ganges Chasma. Viking image draped over MOLA megdr data.

areas. Outflow channel activity peaked during the Hesperian(Tanaka, 1986) when enormous channels were carved and wa-ter bodies may have collected in the northern lowlands (Baker,2001; Clifford and Parker, 2001). The largest channels em-anated from the Valles Marineris canyons and adjacent regionsand discharged their floodwaters to Chryse Planitia and thenorthern plains. In Xanthe Terra, the fluvial erosion that carvedRavi Vallis also triggered the formation of secondary chaoszones deep within the channel system. Analysis of those eventsprovides evidence that the cryosphere was less than one kmthick at the time of the Ravi Vallis flood at that location nearthe equator (Coleman, 2005).

This paper examines how and where water may have ac-cumulated to provide regional sources for the floodwatersthat formed three channels located south and west of GangesChasma (Fig. 1). Two of these channels, Allegheny Vallis andWalla Walla Vallis, have source areas at elevations greater than2500 m, too high to be explained by flow in a global aquiferfrom distant recharge areas in the south polar region (Dinwiddieet al., 2004). We previously concluded (Coleman et al., 2003)that the outflow source for Allegheny Vallis in Ophir Planumwould seemingly have required regional sources of recharge onand east of the Tharsis plateau. Here we examine these channelsystems in more detail, along with the related Elaver Vallis. Byinference, our results provide important clues for the sources ofother Hesperian-aged floodwaters that emerged from the VallesMarineris.

2. Polar recharge model

Extensive groundwater recharge is inconsistent with the ideaof a thick planet-wide cryosphere on Mars, which would in-hibit the migration of water from the surface to undergroundaquifers. To resolve this dilemma, Clifford (1987) and Cliffordand Parker (2001) suggested that Hesperian recharge occurredat the base of the polar layered deposits, and that this watermigrated to low latitudes through a globally connected aquifer

system. Carr (2002) tested this polar recharge model using theMOLA database of surface elevations obtained by Mars GlobalSurveyor. He identified 1500 m as an approximate upper limitfor the elevation at which recharge could efficiently occur bymelting beneath the south polar layered terrains. Carr (2002)concluded that major discharges of groundwater onto the sur-face at elevations above 1500 m are unlikely to have beenderived from recharge sources in the south polar region. Exam-ples of high-standing (>1500 m) water-carved features includevalleys on the flanks of Alba Patera and Ceraunius Tholus.These valleys could readily be explained by local processes,such as erosion by discharges from hydrothermal springs (Carrand Chuang, 1997). Harrison and Grimm (2004) used ground-water modeling results to suggest that recharge occurring onthe Tharsis plateau would have been several times more effi-cient than south polar recharge as a source of groundwater forthe circum-Chryse outflow channels.

Carr (2002) presented several alternative explanations thatcould reconcile high altitude water-carved features with themodel of Clifford and Parker (2001). He suggested that volcanicmelting of ice in the cryosphere could provide the water neededto erode the elevated channels. We agree that this process couldmelt large volumes of ice, but suggest that the interaction offlood basalt lavas with surface ice would far more efficientlyrelease large volumes of water than the melting of ground iceby intrusive or extrusive volcanism. Carr (2002) suggested thatthe melting of ground ice in central Tharsis could have led tothe movement of groundwater along radial fractures into theValles Marineris to form canyon lakes. He also suggested an-other mechanism to account for elevated channels, in which thecrust is ruptured by enhanced hydrostatic pressures created as aconsequence of downward cryosphere expansion. This processwould depend on the degree to which the crust was saturatedby ice and water. Carr (2002) concluded that a coincidence be-tween the ground inventory of water and ice and the holdingcapacity seemed unlikely.

346 N.M. Coleman et al. / Icarus 189 (2007) 344–361

For Allegheny Vallis and Walla Walla Vallis, we find thecryosphere expansion mechanism an unlikely way to rupturethe crust and release floodwaters. Deep canyons of the VallesMarineris almost surround the source areas for Allegheny Val-lis and Walla Walla Vallis. If these deep canyons existed whenthe channels formed, it is more likely that regionally enhancedhydrostatic pressures would have been released where over-burden was minimal and fluid pressures were greatest (i.e., inlow elevation areas of the proximal canyons). The chasmatapresent ideal conditions for the release of regional overpres-sures because the present-day canyon floors lie 4–5 km belowthe martian datum. Analysis by Coleman (2006) showed howthe formation of lakes in the ancestral Valles Marineris wouldhave been likely, perhaps inevitable, as a consequence of therising potentiometric levels of groundwater systems.

3. Outflow channels near Ganges Chasma

We describe three specific channels that reveal new insightsabout Hesperian paleohydrology. These channels provide evi-dence that regional groundwater recharge occurred at low lat-itudes during one or more intervals of Hesperian time. In thefollowing subsections we discuss the outflow nature of thesechannels and infer their relative age of formation.

3.1. Elaver Vallis

Elaver Vallis begins at the eastern rim of Morella, an 80-km-wide crater located south of Ganges Chasma (Fig. 2A).

The flooding occurred during the mid- to late-Hesperian era be-cause the channel eroded lower Hesperian strata (unit Hpl3) ofthe Plateau Sequence (Scott and Tanaka, 1986; Witbeck et al.,1991). The fact that Elaver Vallis begins at a gap in the wallof Morella Crater indicates that a lake formerly existed in thecrater basin. The surface of this lake would likely have beenfrozen under the cold prevailing surface conditions during theHesperian. The wall of Morella crater was eventually breachedand lake waters catastrophically drained through the gap, carv-ing the Elaver Vallis channel complex. A deep pit known asGanges Cavus resides in the southern part of Morella Crater(Figs. 2A, 3A, and 3B). This feature is indeed a cavus and not acrater. It has no rim or ejecta blanket and is roughly ovate witha major axis length of 42 km and a minor axis length of 33 km.We interpret Ganges Cavus as a subsidence feature producedmainly by dilational faulting that was also partly underminedand excavated by the eruption of confined groundwater. Thenorthern part of the cavus floor forms a 15-km-long linear de-pression that is aligned west to east. The deeper parts of GangesCavus have approximately the same elevation (−4000 m) as thedeepest parts of western Ganges Chasma, located 100 km to thenorth. Groundwater is the only plausible source of the waterthat filled Morella Crater because Elaver Vallis exits Morellaand no inflow channels exist. Thus, Morella Crater representsa Hesperian lake basin that was fed by groundwater, unlike nu-merous other paleolakes in impact structures that appear to havebeen dominantly fed by inflows of surface water (Cabrol andGrin, 1999). The existence of a Hesperian lake in Morella is

Fig. 2. (A) Source areas for Allegheny, Walla Walla, and Elaver Valles. Viking image draped over MOLA megdr data. Allegheny Vallis emerges from Ophir Cavus,a deep pit at 10.0◦ S, 54.5◦ W. The cavus lies at the southern margin of a subdued, unnamed crater. Two chains of pits parallel to Ophir Cavus on both its western andeastern sides suggest these features were produced by a common geologic structure. Elaver Vallis emerges from a gap in the wall of Morella Crater. This morphologysuggests that flows from Ganges Cavus created a lake in Morella Crater until the eastern rim collapsed and the lake partially drained, carving Elaver Vallis. (B) WallaWalla Vallis emerges from unnamed cavi south of Wallula Crater. THEMIS composite image (Christensen et al., http://themis-data.asu.edu/). (C) The depth of OphirCavus is ∼1500 m. From this pit, water discharged onto terrain at an elevation of 2500 m, based on erosional features at the northwest corner of the cavus (seeFig. 4). Location of MOLA profile C–C′ is shown in (B).

http://themis-data.asu.edu/


Fig. 3. Landforms on the floor of Morella Crater. (A) Morella Crater is located south of Ganges Chasma at 9.7◦ S, 51.5◦ W. (B) Ganges Cavus is a 5-km-deeppit at the southern edge of Morella. The irregular, degraded rim of Johnstown Crater [J] appears differentially eroded by the action of water and ice on the craterfloor. Somerset Crater [S] postdates the flooding because its walls are pristine and there is no indication that initial floodwaters diverted around its rim. (C) An earlycrossover channel and the main outflow gap on the eastern rim of Morella Crater (image cut from THEMIS global mosaic). Fluvially-eroded landforms lie just westof the gap near a knob. Arrow points to preserved longitudinal ridges. (D) The eroded rim of the 3.2-km-wide Johnstown Crater. [Credit for all images: Christensenet al., http://themis-data.asu.edu/.]

consistent with the finding of Cabrol and Grin (1999) that pale-olake activity in martian impact craters began in the Noachianand continued through the Hesperian and into the Amazonianera.

It may not be a coincidence that Ganges Cavus formedwithin a sizable crater. The planetesimal impact that formedMorella Crater extensively fractured the country rock, weak-ening it and enhancing its permeability, creating favorable con-ditions for the eventual breakout of confined groundwater whenthe cryosphere was ruptured by faulting. The outflows fromGanges Cavus produced a lake in Morella that rose so highthat the eastern crater wall was eventually breached, leadingto catastrophic release of ponded waters and the carving of theElaver Vallis channel complex. A small crossover channel pre-served on the eastern rim of Morella Crater (Fig. 3C) revealsa high-water mark of 1780 m for the overflow elevation of thepaleolake. Flow in this crossover channel ceased when the over-flow gap to the north captured all the flow. The lake continuedto drain via Elaver Vallis as long as the base of the downward-

eroding gap in the crater wall was less than the stage elevationof the lake. Eventually, flow rates were insufficient to furtherdeepen the gap and the lake level stabilized. THEMIS imagesindicate that the channel floor in the outflow gap (Fig. 3C) hasa minimum width of ∼2.3 km. Gridded MOLA data suggestthat the valley floor in the center of the gap rises to an eleva-tion of 1485 m. However, the presence of a precise curve in theMOLA topography (centered at 9.81◦ S, 50.63◦ W) reveals thehigh ground to be a false artifact caused by smoothing of eleva-tions in an area without MOLA tracks. MOLA data exist within4 km to the east and west, indicating a channel floor elevationof 1230 m on the western side of the gap and 1250 m on theeastern side. We interpret 1250 m as the final outflow elevationfor the lake in Morella Crater. We subtract this elevation fromthe initial crossover elevation (1780 m) and thereby determinethat a lake water column ∼530 m deep was catastrophicallydrained by the flood. Most of the floor of Morella Crater lies atelevations <1250 m. A >50-m-deep lake would have remainedin Morella after flows in Elaver Vallis ceased. The remaining



water column eventually would have frozen to its total depth[e.g., Kreslavsky and Head (2002) calculated that a 500-m-deepocean in the northern plains would have frozen within 2000–7500 years] before undergoing long-term sublimation. Mor-phological evidence suggests the lake depth in Morella Craterwould have been much greater above Ganges Cavus. The cavusis currently ∼3 km deeper than the crater floor, but we cannotbe certain how deep it was when Elaver Vallis formed.

The floor of Morella Crater is relatively flat. Using griddedMOLA data, a 40-km-long profile from west to east across thecenter of the crater shows that elevation varies less than ∼90 m.We checked this result using the higher resolution of an indi-vidual MOLA track. MOLA pass ap10452L crossed easternMorella Crater and indicates a north-to-south elevation changeof <100 m over a 45-km-long distance. This smooth floor mor-phology is consistent with lacustrine deposition or prior resur-facing by lava flows (or both). Using Eq. (1) (Garvin et al.,2000; Barlow and Perez, 2003),

(1)d = 0.19D0.55,

where d = depth (km) of a fresh, non-polar crater of >7 kmdiameter (measured from the top of the rim to the lowest pointwithin the crater), and D = present-day rim diameter of crater(km), we estimate a former depth of 2.1 km for Morella Crater.The present-day depth is ∼0.9 km (mean rim height [∼2050 m]minus mean floor elevation [∼1175 m]). Therefore, we estimatethe crater depth has been reduced 1.2 km as a result of wallcollapse, infilling, and crustal relaxation.

Two craters of similar size occur on Morella’s floor north-east of Ganges Cavus (see Figs. 3A and 3B). Somerset Craterhas a continuous rim, an east-to-west diameter of 4.8 km, a floorelevation of 720 m, and is >400 m deeper than the surround-ing floor of Morella. East of Somerset Crater is JohnstownCrater, which has a similar diameter but a very different ap-pearance (Fig. 3B). The rim of Johnstown (Fig. 3D) is highlydegraded with numerous arcuate embayments where the wallswere eroded. Several of the cove-like openings fully incise thecrater rim to the floor elevation of Morella Crater. The overallappearance of Johnstown Crater is that of a highly eroded craterwith scalloped edges. The lowest part of Johnstown Crater hasan elevation of approximately 1090 m, less than 60 m deeperthan the surrounding floor of Morella Crater. This suggeststhat fluvially transported and lacustrine sediments mostly filledJohnstown crater. Based on its greater depth and pristine ap-pearance, Somerset Crater post-dates the lacustrine event inMorella Crater, whereas the shallow depth and degraded rimof Johnstown Crater provide evidence that it existed before thepaleolake filled Morella Crater.

Various processes may have contributed to the erosion andinfilling of Johnstown Crater. The initial outflows from GangesCavus likely had some erosive effects on the base of John-stown. This early erosion could have ceased quickly, however,in the low-energy environment of lake formation and deepen-ing. Eventually the lake level overtopped the rim of JohnstownCrater, resulting in rapid erosion of the crater walls at inflowpoints and deposition of eroded materials inside the crater.MOLA pass ap10452L crossed the eastern rim of Johnstown

Crater, showing that the residual wall of Johnstown reaches anelevation of ∼1210 m, which is below the final stage eleva-tion of the paleolake (i.e., 1250 m). The formation, breakup,and movement of lake ice during the filling and draining of thepaleolake could have further eroded any high-standing featureswithin Morella Crater, including the walls of interior craters.After the paleolake had partially drained and frozen in place,the subsequent deterioration of surface ice and degradationof ground ice may also have further eroded the crater walls.Other workers have catalogued the characteristics of thousandsof craters on Mars that are larger than Johnstown Crater, i.e.,�5 km in diameter (Barlow, 1988, 2003; Barlow and Bradley,1990). The morphology of Johnstown Crater is unusual amongmartian craters. Other craters may eventually be found that wereeroded and modified by similar processes. We suggest that thescalloped rim of Johnstown Crater is a possible indicator of theerosive processes described above.

Hoffman (2000) proposed that martian outflow channelswere eroded by gas-supported debris flows rather than by wa-ter floods. However, the morphology of Elaver Vallis demon-strates why gas-supported debris could not have formed thischannel complex. If liquid CO2 had erupted from GangesCavus, the resulting debris clouds would have rapidly de-pressurized, dropping their suspended loads within the rim ofMorella Crater. Only a relatively stable and non-compressiblefluid, like liquid water, could have accumulated to form a lakewithin the crater, which eventually breached the eastern craterwall and carved the channel complex. The erosive action of aliquid was required. Carbon dioxide-based models (Hoffman,2000; O’Hanlon, 2001) face additional theoretical difficultiesthat now appear insurmountable (Stewart and Nimmo, 2002;Coleman, 2003; Urquhart and Gulick, 2003).

3.2. Allegheny Vallis

Allegheny Vallis, an outflow channel west of Elaver Vallis,emerges from an elongated pit known as Ophir Cavus (Figs. 2A,2B, and 4). The main channel was nearly imperceptible inViking imagery, but MOLA data and MOC images confirm thatit is a continuous valley more than 190 km long. The plain in-cised by Allegheny Vallis is Noachian in age (units Npl1 andNpl2 of Witbeck et al., 1991). We interpret that its flows mayhave occurred during the same epoch as the flows in ElaverVallis (mid- to late-Hesperian) because both sources originatedfrom the fault system of Ophir Catenae. Erosional features atthe northwest rim of Ophir Cavus show that the water thatcarved Allegheny Vallis erupted from the surface at an eleva-tion of 2500 m (see Fig. 4), far above the 1500 m thresholddescribed by Carr (2002) for contributions from polar basalrecharge. From the source at Ophir Cavus to the terminus atthe rim of Ganges Chasma the channel floor gradually dropsin elevation by 435 m (Fig. 4, inset at lower right). Over this190 km distance, the average slope or gradient of the channelfloor is 0.0023. It is possible that elevations at the cavus sourceand along the channel have changed over time. However, ex-tensional tectonics in the Valles Marineris seems more likelyto have lowered the channel than to have raised it. Therefore,


Fig. 4. Allegheny Vallis imagery and MOLA elevation, from its source at Ophir Cavus to the terminus at Ganges Chasma. Water overflowed the northwest rim ofOphir Cavus, carving erosional features at an elevation of ∼2500 m. Two ridges [R1 and R2] temporarily dammed flow. Two scablands [S1 and S2] were createdalong the channel. Culter Crater (inset at upper left) is centered at 8.85◦ S, 53.93◦ W. [Image credit: THEMIS Daytime IR Valles Marineris Mosaic (Christensenet al., http://themis-data.asu.edu/).]

if the channel source elevation has not changed appreciably,has been lowered since Hesperian time, or if relative elevationdifferences have been preserved, then the outflows must havederived from regional recharge at higher elevations.

Immediately north of Ophir Cavus, Allegheny Vallis crossesa large, subdued, unnamed crater (Fig. 5) that is probablyNoachian in age. The outflows may have formed a tempo-rary lake within the crater walls, but it is unclear whether thedegraded rim of this crater was high enough at the time ofthe flooding to obstruct flow. Continuing outflows from OphirCavus eroded a deep channel across the crater floor (Fig. 5).Multiple channels and a flood-carved island occur northeastof the crater (Fig. 5). The main channel north of the island is∼100 m deeper than the channel south of the island, which is“hanging” at both ends. The top of the island is fringed by a ter-race that shows the floodwaters reached a minimum elevationof 2240 m at this location.

In its western reaches, Allegheny Vallis meanders in a seriesof sweeping, nearly right-angle turns (Fig. 4), which suggeststhat topographic lows in the pre-flood terrain may have beeninfluenced by structural features in the upper crust. Northeastof Ophir Cavus the flows encountered the first of two ridgesthat temporarily dammed the flow (Fig. 4). The floodwaterseventually broke through or overtopped this barrier, eroded amid-channel island at 9.08◦ S, 54.03◦ W, and continued flowingto the northeast where they encountered another topographicobstruction. A temporary lake was created behind this naturaldam, with bypass flow occurring around the obstruction until itgave way (Fig. 6A). Based on the overflow elevation (2032 m)of a bypass channel at 8.745◦ S, 53.602◦ W, we used griddedMOLA data to estimate that the lake had a minimum surface

area of 35 km2. The lake drained catastrophically and flood-waters charged northeastward, carving a spectacular scablandwith anastomosing channels and streamlined erosional features(Fig. 7). MOLA pass 10644 reveals 25–30 m of topographicrelief in this broad scabland. Flow and erosion in the scab-land ceased as an inner main channel deepened and eventuallycaptured most or all of the waning floodwaters. This concen-tration of flow in one channel carved a narrow, well-definedreach that can be traced eastward to the western margin ofGanges Chasma. This main channel cross-cuts a wide, shal-low channel that had exited the scabland to the east, producinga hanging valley (Figs. 4 and 6). The main channel formeda broad, branching, channeled scabland near its terminus thatis truncated by the western rim of Ganges Chasma (Fig. 4).The abrupt termination of this scabland provides evidence thatGanges Chasma continued to expand westward after the fluvialepisode had ended. The southernmost channel at the terminusof Allegheny Vallis disappears into a deep side canyon witha V-shaped cross-section that grew westward from the south-west corner of Ganges Chasma. The channel appears to descend4 km through this side canyon to the floor of Ganges Chasma,but the elevation change is so abrupt that the side canyon morelikely formed after the fluvial episode by faulting and masswasting processes (Fig. 4, inset at lower right).

In addition to the geomorphologic evidence that ground-water carved the channel features, the ejecta pattern of twolocal craters provides evidence for a crust formerly rich involatiles. Water ice was the likely volatile, given the difficul-ties of producing and storing CO2 ice in the crust (Stewart andNimmo, 2002). South of Ophir Cavus at 10.70◦ S, 55.26◦ W isBronkhorst, a crater with a single-layer ejecta morphology and



Fig. 5. (A) Ophir Cavus is the source of Allegheny Vallis. The cavus formed along the southern rim of a large, eroded, unnamed crater. Bronkhorst Crater, centered at10.7◦ S, 55.3◦ W, displays a layered (fluidized) ejecta blanket and central pit that suggest the presence of shallow volatiles in the crust at the time of impact. [Imagecredit: THEMIS Daytime IR Valles Marineris Mosaic (Christensen et al., http://themis-data.asu.edu/).] (B) Where Allegheny Vallis turns eastward, floodwatereroded two channels and carved a central island (Is). The layer that caps the island has an irregular periphery, suggesting it is softer than the underlying layer.Differential erosion of these layers formed a terrace all around the island. Additional terraces are identified on the northern bank of the deeper (main) channel. Twoterraces are continuous over more than 25 km.

a central pit (see Fig. 5A). Craters with this layered ejecta pat-tern are thought to have formed by the vaporization of subsur-face volatiles during crater formation (Barlow and Perez, 2003).Barlow and Bradley (1990) proposed that craters with single-layer ejecta excavated a zone consisting primarily of ice. Thepresence of a central pit in Bronkhorst Crater is also an indicatorthat volatiles were present in the upper crust at the time of im-pact (Barlow and Hillman, 2006). Another crater with layeredejecta, Culter Crater, is superimposed on Allegheny Vallis morethan halfway along its course (Fig. 4) at 8.85◦ S, 53.93◦ W.

We estimate the transient excavation depths of these craters,and thereby estimate the depth of ice reservoirs in the uppercrust at the time of the impacts. Using Eq. (2) (Garvin et al.,2000; Barlow and Perez, 2003),

(2)de = 0.131D0.85a ,

where de = transient crater excavation depth (km), and Da =present-day rim diameter of crater (km), we estimate that the18-km wide Bronkhorst Crater and the 6-km wide Culter Craterhad transient excavation depths of 1.5 and 0.6 km, respectively(note that the diameter of Culter Crater is near the lower limitfor application of Eq. (2)). The transient excavation depths sug-gest that water ice reservoirs were located within ∼1500 m (orless) of the surface at the time the craters formed. No chan-

nels originated at these craters. From this, we infer the depthsof excavation were apparently insufficient to rupture a low-permeability confining layer and generate large-scale flow ofwater from a deep aquifer, or water flowing into the craters frombelow did not overtop the rim, or conditions for such flow didnot exist at that time. Culter Crater is relatively fresh; it over-lies Allegheny Vallis and therefore post-dates the flooding thatformed this channel. We do not know whether the Bronkhorstimpact occurred before or after Allegheny Vallis was formed.The terrain around this crater has been mapped as ridged plainsmaterial of lower Hesperian age (unit Hr of Witbeck et al.,1991). The degraded rim of Bronkhorst and the presence of siz-able craters on its floor and ejecta blanket suggest the crater isHesperian rather than Amazonian in age.

We interpret the following sequence of events in the forma-tion of Allegheny Vallis: (1) gradual development of a ground-water potentiometric surface in Ophir Planum to an elevation>2500 m; (2) tectonic rupture of the cryosphere at the locationof eastern Ophir Catenae, forming Ophir Cavus and dramati-cally releasing confined groundwater onto the surface as flood-water; (3) possible ponding of discharge just north of OphirCavus within the rim of a large, subdued crater; (4) floodwaterincision of multiple channels at 9.43◦ S, 54.63◦ W; (5) pond-ing of water behind a ridge obstruction at 9.08◦ S, 54.20◦ W;



Fig. 6. Location of channeled scabland that formed when a transient lake breached the northernmost topographic barrier. (A) MOLA topography contoured usingGRIDVIEW (GSFC, 2004). Dotted blue line shows the minimum area of a transient lake that formed behind a low ridge. Water release carved a water gap andthe scabland to the north. Arrows delineate a small channel where ponded water apparently bypassed the topographic barrier. (B) Flow calculations in Section 5were performed along the straight reach of channel “A” shown in the dashed box. [Image credit: THEMIS Daytime IR Valles Marineris Mosaic (Christensen et al.,http://themis-data.asu.edu/).] (C) Topography along profile A–A′ (located as shown in panel A), generated with GRIDVIEW. Profile A–A′ coincides with MOLApass 14761. (D) Estimated pre-flood topography measured at 15 locations along the south bank of Allegheny Vallis between longitude 53.493◦ W and 53.126◦ W(located as shown in dashed box in panel B). We interpret locations where the slope of the bank changes abruptly as marking the lateral and vertical limit of fluvialerosion.

Fig. 7. Ridge R2 dammed flow, causing a temporary lake to form in the lower left corner of the image. Bypass flow eroded a shallow channel at lower right, but thisflow ended when the main obstruction to the north was surpassed. Floodwaters rushed to the northeast, carving a water gap, the scabland, and the main channel. Thischanneled scabland formed before incision of the main channel, which ultimately captured the remaining discharge. An erosional scarp marks the western marginof the scabland. MOLA elevations along profile B–B′ (MOLA pass 10644) provide additional evidence for event sequence. Crest to trough relief in the scablandranges 25–30 m. The topography suggests that the upper 25–30 m of material was more easily eroded, with an underlying layer of more resistant material (perhapsintact basalt). [Image credit: THEMIS visible image V06781022 (Christensen et al., http://themis-data.asu.edu/).]




Fig. 8. Walla Walla Vallis (A) Context image. (B) View of Walla Walla Vallis source area and passage through Wallula Crater. (C) A sketch of the river valley isshown offset to the right as a visual cue. The outflow channel emerges from a pit chain south of Wallula Crater; flow entered Wallula Crater, and eroded a channelto the north–northwest. Walla Walla Vallis, which appears to intersect Allegheny Vallis, likely pre-dated it, although concurrent flows are plausible. [Image credit A& C: THEMIS Daytime IR Valles Marineris Mosaic; B: THEMIS V10176001 (Christensen et al., http://themis-data.asu.edu/).] (D) MOLA topography crossing pit
chain, Wallula Crater, and Allegheny Vallis.
(6) erosion or collapse of the ridge obstruction and release ofponded water; (7) flow separation and carving of a mid-channelisland at 9.08◦ S, 54.02◦ W; (8) formation of a large lake be-hind a second ridge obstruction, at 8.6◦ S, 53.65◦ W; (9) ini-tiation of bypass flow around the southern end of the natural“dam,” followed shortly by erosion or collapse of the obstruc-tion; (10) dramatic overland flooding of megaflood proportions,carving a scabland of channels and “islands” in the vicinityof 8.44◦ S, 53.55◦ W; (11) incision of a primary channel thatcaptured all of the flow in Allegheny Vallis, terminating flowin the scabland and producing a “hanging” valley at 8.42◦ S53.15◦ W; (12) carving of a channel that flowed eastward, erod-ing a scabland at 8.46◦ S 52.71◦ W, and ultimately emptyinginto ancestral Ganges Chasma; and (13) ebbing of flow untiltermination of the floodwaters. The final events in the chan-nel system, long after the flooding, included the impact eventthat created Culter Crater, and truncation of the terminus of Al-legheny Vallis by westward growth of Ganges Chasma.

3.3. Walla Walla Vallis

A pit chain (e.g., Wyrick et al., 2004) with margin elevation2525 m is located east of Ophir Cavus. These collapse pits arethe surface expression of the source area for the outflow channelWalla Walla Vallis (Fig. 8), which was not resolved in imageryuntil return of THEMIS data began in late 2002 (Dinwiddie etal., 2004). Walla Walla Vallis floodwaters followed the naturaltopographic relief down into nearby Wallula Crater, located 2km to the north at a base elevation of approximately 2400 m(Fig. 8). Wallula Crater is a subdued 12.2-km-diameter featurelocated near 9.88◦ S, 54.66◦ W. The map by Witbeck et al.(1991) indicates the terrain upon which Walla Walla Vallis andWallula Crater are superimposed consists of the cratered unit ofthe Plateau sequence, unit Npl1. The north–northwest-trendingerosional channel across the center of Wallula Crater is visi-ble in Fig. 8B. Ponded water may have accumulated in WallulaCrater until the water breached or overtopped the northwesternrim of the crater through a weak point or topographic low.



Fig. 9. Parallel pit chains of Ophir Catenae extend ∼750 km from eastern Candor Chasma to near Ganges Chasma. Allegheny, Walla Walla, and Elaver Valles areshown in relation to this interpreted dilational fault system. Average land surface elevations decrease from the rim of Candor Chasma toward Ganges Chasma. Thisregional slope would have favored eastward migration of groundwater along structurally enhanced flowpaths. Viking image draped over MOLA megdr data.

Fig. 10. Schematic illustration of steep dilated faults and their potential relationship to pit chain development (after Fig. 2 of Ferrill et al., 2003). Compare pits inillustration to pits (cavus) and pit chains (cavi) shown in Figs. 2A, 2B, 3A, 3B, 5A, 8B, and 9.

Walla Walla Vallis split into two channels within a smallsemi-circular depression at the northwestern edge of WallulaCrater (Fig. 8C); one channel flowed north–northwestward, ap-pearing to terminate as a hanging valley, while the second chan-nel flowed west before turning to the north (Fig. 8C). THEMISimages suggest the second channel may end as a hanging valleytruncated by Allegheny Vallis. Flows in Walla Walla Vallis thuslikely predate late-stage flows in Allegheny Vallis, althoughconcurrent flows during the early stage formation of Alleghenymay be plausible, and Walla Walla floodwaters may have con-tributed to the total discharge through Allegheny Vallis. Higherresolution imagery is needed to confirm this possibility.

4. Structurally enhanced groundwater flow

The floodwaters that formed Elaver Vallis, Allegheny Val-lis, and Walla Walla Vallis probably discharged during theHesperian era from beneath a thick cryosphere and through adeep-seated dilational fault system. The Ophir Catenae struc-tural complex is a long chain of pits that extends 750 km fromeastern Candor Chasma to western Ganges Chasma (Fig. 9). Weinterpret the en echelon segments of this pit chain as the surfacemanifestation of a dilational fault zone (Wyrick et al., 2004;

Ferrill et al., 2004), i.e., an incipient graben system that waspart of the Valles Marineris structural complex. Collapse pitslikely formed at the source during or after the fluvial episode assurface material dropped into subsurface cavities, was loweredby graben formation, or was carried away with the flow of wa-ter. Fig. 10 (after Ferrill et al., 2003) schematically illustratessteep dilated faults on Mars and their potential relationship topit chain development. This kind of fault system would likelyhave created highly anisotropic conditions in the aquifer sys-tem, greatly enhancing the eastward flow of groundwater. Wepropose that groundwater breakouts coincided with the climac-tic formation of Ophir Catenae faults. Discharges from GangesCavus may indicate the easternmost extent of Ophir Catenae. Atleast two and possibly three channels were carved by aqueousoutflows from the same dilational fault system, which suggeststhat the flows may have been concurrent. More discharge couldhave occurred in Elaver Vallis over a longer duration becauseits outflow zone is ∼1300 m lower in elevation than the sourcesof Allegheny and Walla Walla Valles.

Scott and Wilson (2002) investigated the origin of pit chainsnear the martian volcano Alba Patera. They propose these pitchains formed as a result of volcanic activity above long re-gional dikes. They conclude that smaller pits formed when sur-


face rocks collapsed into spaces vacated by the leakage of gasthat accumulated at the tops of dikes. They suggest that largerpits are produced after explosive Plinian eruptions ejected sub-stantial fractions of magma from dike cores (i.e., the eruptionrates exceeded the supply rate of magma feeding the eruption).Scott and Wilson (2002) suggest that the implosion and collapseof dike walls produced the larger pits. They explain that thereare no visible products of the eruptions because areas near thevents are destroyed by the terrain collapse that formed the pits.Also, the thin martian atmosphere increases the vigor of the ex-plosive eruptions, scattering debris over large areas rather thanconcentrating it near the vents. Wyrick et al. (2004), however,find no direct evidence that pits were formed by intrusive vol-canism. We agree with their conclusion that dilational faultingis more consistent with Earth analogs and provides a simpler ex-planation than the volcanism model of Scott and Wilson (2002).

We can now add Allegheny, Walla Walla, and Elaver Vallesto the group of martian channels carved by groundwater out-flows from large fault zones. Such channels also includeAthabasca Vallis, which emanated from the Cerberus Fos-sae (Burr et al., 2002a, 2002b; Murray et al., 2005), andMangala Valles, which discharged from the Memnonia Fos-sae west of Tharsis (Tanaka and Chapman, 1990; Wilsonand Head, 2004). The observation that some large martianchannels debouched from fault zones supports a groundwatersource and aqueous flood origin, and contradicts other sug-gested mechanisms for outflow channel formation, includingcarving by wind (Whitney, 1979), erosion by gas-supporteddebris flows (Hoffman, 2000), or glacial scour (Lucchitta,1982). Although Lucchitta (1982) proposed glacial scour asone mechanism to erode channels, she also suggested an-other more likely scenario in which ice jams enhanced theerosive power of catastrophic floods. This latter mechanismmay explain why some channel landforms appear to have beensculpted by ice. The absence of distinct terminal morainesnear the mouths of the large outflow channels further supportsa fluvial rather than a glacial origin. The extreme smooth-ness of the northern plains of Mars (Aharonson et al., 1998;Head et al., 1998) can readily be explained by resurfacing ofthese lowlands by aqueous transport and sedimentation.

5. Discharge estimates for Allegheny Vallis

Studies of catastrophic paleofloods on Earth and Mars havetended to be controversial. Over the last 40 years, however,evidence has grown that many megafloods were associatedwith continental ice sheets of Pleistocene age (Baker, 2002).The resulting massive discharges of fresh water are thought tohave significantly influenced ocean circulation and climate. Thestudy of martian megafloods is central to resolving questionsabout the global evolution and fate of water on Mars. It is use-ful to estimate the magnitude of flows that carved AlleghenyVallis to compare with other channel studies on Mars and pre-historic megafloods on Earth. We focus on Allegheny becauseit is an elevated channel (>1500 m) indicative of recharge atlow latitudes. Walla Walla Vallis is also an elevated channel,but available data are not sufficient to define cross-sections for

this small channel. While MOLA and image data are availableto perform detailed analyses of Elaver Vallis paleo-flows, suchanalyses are beyond the scope of this study.

Hydrologic evaluation of the paleoflood in Allegheny Val-lis is challenging because the channel is highly sinuous and,as discussed earlier, there were at least three obstructions thatcould have caused temporary ponding and catastrophic releaseof floodwaters, resulting in erosion of several scablands. Suchflows would have been highly non-uniform. We therefore choseto evaluate flows in the long straight reach of Allegheny Val-lis that occurs beyond the northernmost topographic obstruc-tion (Figs. 6 and 7). The temporary lake that had accumulatedbehind this barrier eventually broke through and eroded the sca-bland shown in Fig. 7. Once the lake had drained, the southern-most channel (henceforth channel “A”) in the scabland capturedall remaining flow in the channel system and continued erod-ing to its present depth until the discharge from Ophir Cavusceased. We surmise that the discharge rate in channel “A” wouldhave become quasi-uniform after the transient lake had drainedand before the channel reached its present depth.

Dingman (1984) makes a fundamental point about open-channel flows: All the flow variables (i.e., slope, width, depth,velocity, and flow resistance) are interdependent and cannot beconsidered as truly independent variables. Four elements areneeded to estimate the physical parameters for the AlleghenyVallis paleofloods: (1) energy slope; (2) width of eroded ter-rain; (3) hydraulic radius of the floodwater channel; and (4) anestimate of the flow resistance for the channel bed (i.e., theChézy or Manning coefficients). The hydraulic radius refers tothe cross-sectional area of flow divided by the wetted perimeter.The wetted perimeter is the distance across a channel measuredalong the channel bottom from the water’s edge on one side tothe other (Dingman, 1984). For flows that are much wider thandeep, the hydraulic radius can be estimated using the averageflow depth. This assumption appears to be reasonable for theintermediate-stage flows we have analyzed here, prior to fullincision of the channel.

The channel is ∼3800 m wide and ∼110 m deep at the lo-cation of MOLA pass 14761, with a floor elevation of 1794 m(MOLA 8.499◦ S, 53.256◦ W) (Fig. 6C). The width and es-timated depth of flooding were used to calculate the cross-sectional area of the flood channel. Wilson et al. (2004) de-scribe a 30 m floodwater depth as a realistic lower limit forchannels on Mars, based on expected rates of evaporation andfreezing that would prevent shallower floodwaters from trav-eling long distances. We apply this lower limit to AlleghenyVallis. Because we evaluated flow conditions for the period af-ter channel “A” captured all the flow, we estimate the elevationof the flood stage using the base level elevation in the chan-neled scabland (i.e., 1895 m) located north of and parallel tochannel “A.” At this intermediate stage of flooding, the scab-land would have drained, but the main channel would not yethave eroded to its full depth. Therefore, we estimated flow ve-locities and discharge rates using a 30–60 m range of floodwaterdepths. A 60 m depth would represent a channel floor elevationof 1835 m, which is ∼40 m less than the present-day channeldepth shown in Fig. 6C. We used Manning coefficients in the


Table 1Hydraulic flow calculations for Allegheny Vallis, Mars

Hydrologic scenario Flow width(m)

Slope Mean flowdepth (m)

Mean flow

velocitya (m s−1)Discharge × 106

(m3 s−1)

Flow in main channel “A”after lake and scablandhad drained (see Fig. 6)

3800b 0.0043 30 6–10 0.7–145 8–13 1–260 10–15 2–3

a Flow velocities calculated using the approach of Komar (1979) and a range of terrestrial Manning n values from 0.04 to 0.06. Values of the Chézy frictioncoefficient (Cf ) were obtained using these n values and the estimated range of flow depths (Carr, 1996). For example, given a flow depth of 30 m and n = 0.05,Cf = 0.008.

b Determined where MOLA pass 14761 crosses Allegheny Vallis (profile A–A′ in Figs. 6A and 6C). MOLA data file (GSFC, 2004) analyzed with GRIDVIEW(2004) software.

range of 0.04 to 0.06 [see discussion of resistance factors byWilson et al. (2004) and Coleman (2005)]. The pre-flood terrainslope was determined along a 23-km-length of the south bankof Allegheny Vallis using slope inflections that mark the lateraland vertical extent of erosion. The elevation change over 23 kmis −100 m (see Fig. 6D) from which we obtain an energy slopeof 0.0043. We used this value (rather than the present-day thal-weg slope) to estimate the energy slope for intermediate-stageflooding prior to full channel incision. Results are presented inTable 1.

For intermediate stage flooding in Allegheny Vallis, our re-sults indicate flow velocities of 6–15 m s−1 and discharges inthe range of 0.7–3 × 106 m3 s−1. We estimate that bottom shearstress was in the range of 400–1000 N m−2, and that flow powerwas in the range of 3000–15,000 W m−2.

5.1. Comparison of terrestrial and martian floods

Table 2 provides estimates of volumetric discharge pub-lished for various paleofloods on Mars and Earth. Also shownfor comparison purposes are discharge estimates for severallarge Earth rivers. The discharges we estimate for AlleghenyVallis are similar in magnitude to those estimated for AthabascaVallis (Burr, 2003), which also discharged from a fault system(i.e., Cerberus Fossae). The Allegheny discharges were simi-lar in magnitude to the Missoula paleofloods on Earth. Non-uniform flows of much greater discharge rate likely occurredin Allegheny Vallis when obstacles to flow were breached byponded floodwaters. The pre-MOLA extreme discharges calcu-lated by Robinson and Tanaka (1990) for Northern Kasei Vallisneed to be reconsidered based on the work of Williams et al.(2000). A pre-MOLA analysis by Komatsu and Baker (1997)reports a large peak discharge for Ares Vallis flooding. Theypresume deep, rim-high flows in a narrow reach of the channel,but do state that the discharge could well have been much lowerthan the estimated peak. MOLA pass 21 crossed their studyarea and verifies a channel depth of ∼1600 m (Smith et al.,1998). This is significantly deeper than estimated by Komatsuand Baker (1997). We agree with Komatsu and Baker (1997)that the peak discharges in Ares Vallis were likely much lowerthan reported in Table 2. By the time the Ares Vallis channelwas fully incised, the floodwater sources were likely to havebeen greatly depleted, resulting in flood stages well below thevalley rims.

5.2. Theoretical floodwater hydrographs

Channels that were carved by groundwater erupting fromlarge fault systems (i.e., Allegheny Vallis, Athabasca Vallis,Walla Walla Vallis, Mangala Valles) likely produced hydro-graphs with a signature that includes a rapid rise followedby a prolonged decline. Confined groundwater was the appar-ent source for the initial outflows that carved Allegheny andWalla Walla Valles. If this confined groundwater were the onlysource, the flow could not have been sustained because confinedaquifers, once released, tend to depressurize rapidly. The dewa-tering of an unconfined aquifer is much more protracted. Thehydrograph for flow released from a confined aquifer shoulddisplay initial rapid release of water, resulting in broad sca-bland erosion. Subsequent drainage under unconfined condi-tions, augmented by distant recharge from an ice-covered lakein ancestral Candor Chasma, would produce a long recession offlood stage. These protracted flows of reduced discharge likelyconcentrated erosion, producing an inner channel in the fluvialsystem. In the case of Allegheny Vallis, channel “A” repre-sents an inner channel that formed from slowly declining floodstages after all flow obstructions had given way. This type ofhydrograph differs from those described by Baker (1978) forthe terrestrial Missoula floods. He concluded that the extra-ordinary preservation of high-stage indicators throughout theChanneled Scabland suggests those floodwaters rose slowly,peaked, and then dropped abruptly, similar to the hydrographscommonly (but not always) seen for Icelandic jökullhlaups(Roberts, 2005). We expect that a similar hydrograph form mayhave represented martian channels carved by flows originatingfrom catastrophic releases from surface impoundments locatedin the deep canyons. The Simud–Tiu Valles system may havebeen formed by such releases.

6. Hesperian recharge zones for high channels

We have identified the region of potential recharge for highchannels like Allegheny Vallis and Walla Walla Vallis, and formost of the large outflow channels that emptied into ChrysePlanitia. Large regions near the Valles Marineris stand higherthan 2500 m (Fig. 11), including the elevated terrain of theTharsis plateau, Olympus Mons, and Alba Patera. Other high-standing terrain includes the Hesperian plains of Syria Planum,Sinai Planum, and Solis Planum. Based on the geologic map-


Table 2Estimated volumetric discharges for martian and terrestrial paleofloods, and for several large present-day rivers

Location Discharge (m3 s−1) × 106 References

Martian floodsMangala Vallis 0.01–40 Komar (1979)

0.3–11 Ghatan et al. (2004)Lunae Planum scabland 0.1–10 Carr (1979)Maja Vallis ∼1 Chapman et al. (2003)Ares Vallis 100–1000 Komatsu and Baker (1997)

5000 Smith et al. (1998)Northern Kasei Vallis 900–1400 Robinson and Tanaka (1990)a

3–20 Williams et al. (2000)b

Kasei Valles 0.08–0.6 Williams et al. (2000)Vedra Vallis 1.3 De Hon et al. (2003)Maumee Vallis 1.9 De Hon et al. (2003)Athabasca Vallis 1–2 Burr (2003)Cerberus Fossae (source of Athabasca Vallis) 1–2 Manga (2004)Ma’adim Vallis inner channel ∼1 Irwin et al. (2004)Ravi Vallis Coleman (2005)

Early overland flow 4–50Late stage flow (in an inner channel) <1

Ravi Vallis Leask et al. (2006)Early stage flow ∼30Late stage flow <10

Allegheny VallisIntermediate stage flow in main channel, “A” 0.7–3 This study

Terrestrial paleofloods and present-day riversLate Pleistocene Missoula Benito and O’Connor (2003)

>25 floods >1>15 floods >3>6 floods >6.51 flood ∼10

Rathdrum (Missoula) 20 Baker and Kale (1998)O’Connor and Baker (1992)

Grand Coulee (Missoula) 5 Baker and Kale (1998)Amazon River 0.3 Baker and Kale (1998)Mississippi River 0.02–0.03 Komar (1979)

a To reproduce the calculations of Robinson and Tanaka (1990), an energy slope of 1.3 × 10−3(2.5/1890) must be used (M. Robinson, personal communication,1999).

a Williams et al. (2000) obtained values two orders of magnitude lower than previous estimates for Northern Kasei Valles. They suggest that the high-water markused by Robinson and Tanaka (1990) lies outside the banks of Northern Kasei Vallis.

ping of Scott and Tanaka (1986), we estimate that over 80percent of the high ground is covered by Amazonian and Hes-perian units. We infer that Hesperian strata occur beneath mostof the Amazonian units because geologic mapping shows onlya few unconformities where Amazonian deposits directly con-tact Noachian units (Scott and Tanaka, 1986). Some Hesperianwater-carved features may have existed in areas now veneeredwith Amazonian strata. In the following paragraphs, we developa theory for regional recharge that could have supplied aquiferssourcing high channels like Allegheny Vallis and Walla WallaVallis, and most other large outflow channels that emptied intoChryse Planitia.

The occurrence of widespread Hesperian volcanism overmost of the potential recharge areas for Allegheny, Walla Walla,and Elaver Valles suggests direct, causal relationships betweenlong-lived volcanism, large-scale genesis of liquid water, andchannel formation. Overall, our understanding of Hesperian re-gional hydrology would improve if we could identify plausiblesources for the water that filled the canyon lakes. The key tothis puzzle may be the accumulation of ice at low latitudes due

to increased obliquity and the dramatic release of water causedby regional volcanism, particularly on the Tharsis plateau.

6.1. Water ice reservoirs and release mechanisms

The present martian climate is dominated by low obliquity.Under these conditions, water ice is unstable at low latitudes.Neutron data from Mars Odyssey demonstrate that rich depositsof ground ice are present poleward of 60◦ latitude (Feldman etal., 2003). The areas with enriched hydrogen generally coincidewith regions where ground ice was predicted to be stable underthe present climate (Mellon and Jakosky, 1995). Jakosky andCarr (1985) explored how ice might accumulate at low latitudesduring times of very high obliquity during the Noachian andlater periods. They concluded this accumulation would havebeen efficient only at extreme obliquities that ensued before theTharsis plateau formed, but may also have occurred during thehighest obliquities over the rest of martian history.

Unlike the Earth, which has an obliquity stabilized by a largeMoon (Laskar et al., 1993), Mars has experienced chaotic obliq-


Fig. 11. Topography for the region south and west of Chryse Planitia. All terrain within the 2500 m contour line stands higher than the outflow locations of Alleghenyand Walla Walla Valles. The terrain east and south of Tharsis Montes that exceeds 2500 m covers an area greater than 1 × 107 km2. Map contoured using MOLAdata grid at 1

4 degree resolution.

uity that varies widely. Laskar and Robutel (1993) performed afrequency analysis of the obliquity of Mars over tens of mil-lions of years. They found the martian obliquity to be chaoticwith possible variations from nearly 0◦ to ∼60◦. Obliquity inthe present epoch is estimated to oscillate between 13◦ and 42◦(Ward, 1992; Carr, 1996). More recent results show that withinthe last 10 Myr the martian obliquity could have exceeded ∼45◦at times (Laskar et al., 2004). Beyond 100 Myr the evolutionof the martian obliquity cannot be evaluated with precision.Nonetheless, Laskar et al. (2004) evaluated the density functionof the obliquity over the age of the Solar System. Their statis-tical results suggest an average obliquity of ∼38◦ over 5 Gyr(standard deviation of ∼14◦) and a maximal value of ∼82◦.Laskar et al. (2004) also note that the present obliquity of ∼25◦is far from the most probable value of ∼42◦ evaluated over4 Gyr. Obliquity variations may have more dramatic effects onthe martian climate than was previously believed (Jakosky etal., 2005).

Mischna et al. (2003) used a general circulation model forMars to investigate the orbital forcing of martian water andCO2 cycles. Their three-dimensional model of the water cy-cle significantly improves upon previous models (e.g., Jakoskyet al., 1993, 1995). Mischna et al. (2003) found that, as obliq-uity increases, abundances of water vapor and water ice cloudsincrease substantially and the region of surface ice stabilitymoves toward the equator. At 45◦ obliquity, water ice is stableonly in the tropics and would preferentially collect in regionsof high topography or high thermal inertia. The highest ele-vations on Mars occur on the volcanic summits of Tharsis.Surfaces on Tharsis have very low thermal inertias, less than∼180 J m−2 s−1/2 K−1 (Jakosky et al., 2000). However, oncesurface ice begins to form, the albedo would increase dramati-cally and positive feedback would induce the further condensa-tion of atmospheric water onto the surface as ice.

Important clues about the possible range of Hesperian cli-matic conditions may be found in the mounting evidence forsignificant variability in Amazonian climates, including recentgeologic time (Baker, 2005). Neukum et al. (2004) reportedevidence that the Tharsis region was volcanically active overhundreds of millions to billions of years. The most recent sum-mit caldera activity on the Tharsis volcanoes was clustered ap-proximately 100–200 Myr ago. Neukum et al. (2004) describeglacial deposits at the base of Olympus Mons as evidence forrepeated phases of ice deposition. Some of the ice deposits maybe as young as 4 Myr. They present morphological evidencethat snow and ice were deposited on Olympus Mons at eleva-tions >7000 m, leading to episodes of glacial activity. Neukumet al. (2004) suggest that buried water ice may be present todayat high altitudes on the edge of the western scarp of OlympusMons.

Head et al. (2005) reported evidence for geologically recentand recurring glacial activity at low and middle latitudes onMars. They presented evidence for glaciers veneered with de-bris at the northwest edge of Olympus Mons, extending up to70–120 km from the basal scarp of this volcano. These rockglaciers are interpreted to be just a few million years old, butcurrently inactive. Head and Marchant (2003) interpreted fan-shaped landforms that extend over 350 km from the base of Ar-sia Mons as the deposits and remnants of cold-based mountainglaciers. Shean et al. (2005) concluded that the northwesternflank of each of the Tharsis Montes volcanoes has a fan-shapeddeposit of Amazonian age. These deposits were interpreted asremains of tropical mountain glaciers created by atmosphericdeposition of water ice during periods of high obliquity. High-resolution climate simulations of Mars with an obliquity of 45◦and present-day atmosphere predict ice accumulation by pre-cipitation in regions where glacial remains have been found(Forget et al., 2006).


Head et al. (2003) concluded that Mars is now experiencingan interglacial epoch that has lasted ∼300 kyr, during whichobliquity has remained in the narrow range of 22◦–26◦. Theyinterpret dusty, ice-rich surface deposits at middle to high lati-tudes as forming during a geologically recent ice age from 2.1to 0.4 Myr ago, when obliquity regularly exceeded 30◦. Suchinsolation conditions likely favored transfer of water from thepoles to mid-latitudes, where it was deposited as ice-rich man-tles along with dust eroded from the polar layered terrains. Headet al. (2003) suggest that martian interglacial periods are muchshorter than glacials, preventing the erosion of surface lag de-posits and preserving much of the mantle material.

During the Hesperian when most of the large outflow chan-nels formed, we suggest that large volumes of liquid water wereperiodically produced at or near the martian surface at low lat-itudes. A likely source for liquid water was volcano–ice inter-action (Carr, 1996). We previously suggested that atmosphericconditions in mid- to late-Hesperian, proximal to major vol-canic eruptions, may have permitted precipitation of snow ontothe elevated terrain of the Tharsis plateau and other high terraineast of Tharsis (Coleman et al., 2003). This process may havebeen greatly enhanced at times of high obliquity when surfaceice became stable at low latitudes. Precipitation may also havebeen augmented by gaseous emissions from Tharsis volcan-ism. Ice could have nucleated on volcanic particulate emissionsand other windblown dust, and ice may also have depositeddirectly as frost during the adiabatic cooling of air masses ris-ing onto and traversing Tharsis. Even under the present tenuousatmosphere, orographic formation of water ice clouds is com-mon over the Tharsis rise (Pearl et al., 2001). An increasedatmospheric pressure during Hesperian time compared to thepresent-day (but still much lower than during the Noachian)could have further enhanced cloud formation and snow andice precipitation, especially in association with volcanic erup-tions (Hort and Weitz, 2001). Ice blankets layered with aeoliandust and volcanic ash could have accumulated at rates that out-stripped sublimation, analogous to the deposits described byHead et al. (2003) and Shean et al. (2005). More extensive sur-face ice deposits that may have formed during the Hesperiancould have provided substantial reservoirs of water ice that weresusceptible to catastrophic melting. Eruption of flood basaltscould have periodically melted these reservoirs and providedlarge volumes of water for surface runoff and infiltration. Givena somewhat higher global water inventory during the Hesperian,periods of high obliquity may have resulted in widespread accu-mulation of ice mantles at low latitudes, and not just in Tharsis.However, the locations where these mantles could have beenrapidly converted to meltwaters should have been limited to re-gions of active volcanism, such as the Tharsis plateau.

Highly permeable fractured basalt aquifers could have ef-ficiently transported meltwaters eastward from recharge ar-eas in Tharsis to the ancestral Valles Marineris canyons.The Hesperian precursor of the structural complex of NoctisLabyrinthus (10◦ S, 100◦ W) may have provided a networkof surface and underground pathways that enhanced drainagefrom elevated terrain in central Tharsis. The underlying struc-tural complex is probably larger than mapping indicates (Scott

and Tanaka, 1986), because geologic units of Amazonian andupper Hesperian age appear to have overlapped and concealedpart of its surface expression.

7. Lakes in the Canyons

Coleman (2006) examined the implications of high channelswith respect to the formation of lakes in the ancestral VallesMarineris canyons. Analysis of potential fluid pressures be-neath the canyon floors led to his conclusion that formation oflakes in the ancestral canyons would have been likely, perhapsinevitable, as a consequence of the rising potentiometric levelsof groundwater systems. Conditions appear to have been so fa-vorable for groundwater breakouts that ice-covered lakes likelyexisted in the canyons when the high channels formed.

The former presence of lakes in the Valles Marineris issupported by the discovery of spillover channels that formedwhen lakes overtopped canyon rims (Coleman and Baker, 2007;Coleman et al., 2007). There is also geochemical evidence ofpast aqueous activity in the canyons. The OMEGA hyperspec-tral imager has identified hydrated sulfates in the light-tonedlayered terrains on Mars (Gendrin et al., 2005). Sulfates havebeen identified in all the canyons of Valles Marineris, with thesulfate signatures being mainly found on the flanks of mas-sive deposits and on several isolated interior layered deposits(Quantin et al., 2006). Sulfates have also been detected in Au-reum, Aram, and Iani Chaos (Gendrin et al., 2006). Gendrin etal. (2005) present several interpretations for the origin of thesesulfates, including alteration of mafic minerals by acidic precip-itation, evaporation of standing water bodies, or by the seepageof hydrothermal brines. The sulfates exist over a wide rangeof elevations, which suggests that sulfate sediments may haveformed in shallow water bodies while the canyons continued togrow.

7.1. Alternate recharge model

We have identified plausible locations for Hesperian rechargesources on and east of the Tharsis plateau (Fig. 11). While floodbasalts could have produced large-scale melting of surface icedeposits in this region, given the expected presence of a thickregional cryosphere, it is unclear how large volumes of meltwa-ter could penetrate the surface to reach the aquifer system. Analternate model may be possible for Mars, in which the ances-tral Valles Marineris canyons served as basins that collected theextensive meltwater runoff from volcano–ice interactions. Theancestral Noctis Labyrinthus canyon complex and its underly-ing fault systems could have effectively channeled groundwatereastward from central Tharsis to the main canyons of the VallesMarineris. This process could also have led to groundwaterbreakouts in the canyons (Coleman, 2006) and the gradual ac-cumulation of lake waters beneath ice covers. The cryospherebeneath the canyon floors would eventually disappear underthese conditions and the aquifer system would become directlylinked to the lake water systems. In other words, much of theregional aquifer recharge would have occurred deep within ice-covered lakes rather than on the Tharsis plateau. This alternate


model could help to reconcile the expected presence of a thickcryosphere during the Hesperian with the abundant evidence forgroundwater as a source for some of the circum-Chryse outflowchannels.

8. Implications for life

Analysis of mid- to late-Hesperian thermal conditions sug-gests that ice covers on lakes at low latitudes would havebeen less than 3 km thick (Coleman, 2005). Therefore liq-uid water would have existed beneath the ice covers if lakesin the ancestral canyons were deeper than 3 km. Deep lakesand aquifers would have been ideal refuges for any extant lifeforms that could have evolved earlier in the Noachian. Theseenvirons would have protected organisms from the harsh condi-tions at the present-day surface. The evidence supporting Hes-perian lakes in the canyons and in a large, flat-floored crater(Morella) at 10◦ S, 51◦ W provides potential low-latitude tar-gets for landed science missions. Morella Crater represents apaleolake target analogous to the Noachian-aged Gusev Craterexplored by the Spirit Mars Exploration Rover. The youngerHesperian-aged sediments deposited in Morella Crater (and inthe canyons) may preserve chemical traces of life forms, mi-crofossils, or even spores. Johnstown Crater pre-dates the flu-vial event in Morella Crater, and appears to have been filledby lacustrine materials. This crater therefore represents a sed-iment “trap” that could be targeted for core recovery by alanded mission. As experiments on Earth have shown, spore-forming bacteria 250 million years old can be isolated andgrown (Vreeland et al., 2000). Therefore, the possibility ofrecovering evidence of life, or even rejuvenating ancient mar-tian life, remains an intriguing prospect for future Mars mis-sions.

9. Conclusions

The polar basal recharge model (Clifford, 1987; Cliffordand Parker, 2001) may be a viable mechanism for the creationof Hesperian groundwater recharge on Mars. There are high-standing water-carved features at low latitudes, however, thatcannot be explained by polar recharge and groundwater move-ment in a global aquifer system. Allegheny and Walla WallaValles are high-standing outflow channels that discharged fromterrain >2500 m above the martian datum. The indicated po-tentiometric surface elevations would have required regionalsources of recharge on and east of the Tharsis Plateau. Thefloodwaters that erupted from Ophir Cavus to form AlleghenyVallis encountered topographic obstacles that restricted theflow, forming temporary lakes. These obstructions were prob-ably breached or overtopped quickly, catastrophically drainingthe lakes and carving several scablands. After the last obstaclehad been breached, a single channel formed that captured allsubsequent flow. For this intermediate phase of flooding, we es-timate flow velocities of 6–15 m s−1 and discharges in the rangeof 0.7–3 × 106 m3 s−1. Higher flow velocities and dischargeswould have occurred when the transient lakes were drained.

We also examined the implications of high channels with re-spect to the potential formation of lakes in the ancestral VallesMarineris canyons. Formation of lakes in the ancestral canyonsshould have been likely, perhaps inevitable, as a consequenceof rising groundwater potentiometric levels. Conditions appearto have been so favorable for groundwater breakouts that ice-covered lakes probably already existed in the canyons whenthe high channels formed. In particular, a high-standing, ice-covered lake likely existed in eastern Candor Chasma becausethis canyon is intersected by the Ophir Catenae fault systemfrom which Allegheny Vallis and Walla Walla Vallis originated.

Mars appears to have experienced periods of hydrologic qui-escence with intermittent and pronounced hydrologic activitythat brought about dramatic evolution of landforms during post-Noachian time. The evidence for high-altitude recharge at lowlatitudes, coupled with inferences about ice-covered lakes inthe ancestral canyons, can be used to calibrate models of avolcanic-hydrologic climax during Hesperian time. We also de-scribe an alternate flow model in which groundwater in theValles Marineris region periodically may have been enhancedby the presence of ephemeral ice-covered lakes in the canyons,created by eastward drainage of meltwaters from central Thar-sis via the ancestral Noctis Labyrinthus canyon system. Highlake level elevations could have led to the buildup of pressurein confined aquifers downgradient from canyon lakes, creatingconditions for incipient groundwater breakout. Tectonic activitycould have contributed to the fault-controlled release of ground-water and the formation of chaotic terrain and outflow channels.

Acknowledgments

We thank Nadine Barlow (Northern Arizona University),Devon Burr (SETI Institute), and Gary Walter, Danielle Wyrick,and Wes Patrick (all with Southwest Research Institute) fortheir thoughtful reviews, which helped to substantially im-prove this manuscript. Additional thanks to Shannon Coltonand Danielle Wyrick for their assistance with the data process-ing that led to preparation of Figs. 1, 2A, and 9. This articlewas prepared, in part, by an employee of the US Nuclear Reg-ulatory Commission (NRC) on his own time apart from regularduties. NRC has neither approved nor disapproved its technicalcontent.

References

Aharonson, O., Zuber, M.T., Neumann, G.A., 1998. Mars: Northern hemisphereslopes and slope distributions. Geophys. Res. Lett. 25, 4413–4416.

Baker, V.R., 1978. Paleohydraulics and hydrodynamics of scabland floods. In:Baker, V.R., Nummedal, D. (Eds.), The Channeled Scabland. NASA Officeof Space Science, Planetary Geology Program, Washington, DC, 186 pp.

Baker, V.R., 2001. Water and the martian landscape. Nature 412, 228–236.Baker, V.R., 2002. The study of superfloods. Science 295, 2379–2380.Baker, V.R., 2005. Picturing a recently active Mars. Nature 434, 280–283.Baker, V.R., Kale, V.S., 1998. The role of extreme floods in shaping bedrock

channels. In: Rivers Over Rock: Fluvial Processes in Bedrock Channels. In:Geophysical Monographs, vol. 107. American Geophysical Union, Wash-ington, DC, pp. 153–165.

Barlow, N.G., 1988. Crater size-frequency distributions and a revised martianrelative chronology. Icarus 75, 285–305.


Barlow, N.G., 2003. Revision of the “Catalog of large martian impact craters.”In: Sixth Int. Conf. on Mars. Abstract 3073.

Barlow, N., Bradley, T., 1990. Martian impact craters: Correlations of ejectaand interior morphologies with diameter, latitude, and terrain. Icarus 87,156–179.

Barlow, N.G., Hillman, E., 2006. Distributions and characteristics of martiancentral pit craters. Lunar Planet. Sci. XXXVII. Abstract 1253.

Barlow, N.G., 2003. Perez, C.B., Martian impact crater ejecta morphologies asindicators of the distribution of subsurface volatiles. J. Geophys. Res. 108,doi:10.1029/2002JE002036.

Benito, G., O’Connor, J.E., 2003. Number and size of last-glacial Missoulafloods in the Columbia River valley between the Pasco Basin, Washington,and Portland, Oregon. GSA Bull. 115, 624–638.

Burr, D.M., 2003. Hydraulic modeling of Athabasca Vallis, Mars. Hydrol.Sci. 48, 655–664.

Burr, D.M., McEwen, A.S., Sakimoto, S.E.H., 2002a. Recent aqueous floodsfrom the Cerberus Fossae, Mars. Geophys. Res. Lett. 29, doi:10.1029/2001GL013345.

Burr, D.M., Grier, J.A., McEwen, A.S., Keszthelyi, L.P., 2002b. Repeated aque-ous flooding from the Cerberus Fossae: Evidence for very recently extant,deep groundwater on Mars. Icarus 159, 53–73.

Cabrol, N.A., Grin, E.A., 1999. Distribution, classification, and ages of martianimpact crater lakes. Icarus 142, 160–172.

Carr, M.H., 1979. Formation of martian flood features by release of water fromconfined aquifers. J. Geophys. Res. 84, 2995–3007.

Carr, M.H., 1996. Water on Mars. Oxford Univ. Press, New York.Carr, M.H., 2002. Elevations of water-worn features on Mars: Implica-

tions for circulation of groundwater. J. Geophys. Res. 107, doi:10.1029/2002JE001845.

Carr, M.H., Chuang, F.C., 1997. Martian drainage densities. J. Geophys.Res. 102, 9145–9152.

Chapman, M.G., Gudmundsson, M.T., Russell, A.J., Hare, T.M., 2003.Possible Juventae Chasma subice volcanic eruptions and Maja Vallesice outburst floods on Mars: Implications of Mars Global Surveyorcrater densities, geomorphology, and topography. J. Geophys. Res. 108,doi:10.1029/2002JE002009.

Christensen, P.R., Gorelick, N.S., Mehall, G.L., Murray, K.C. THEMIS Pub-lic Data Releases, Planetary Data System Node. Arizona State University,Tempe, AZ, http://themis-data.asu.edu.

Clifford, S.M., 1987. Polar basal melting on Mars. J. Geophys. Res. 92, 9135–9152.

Clifford, S.M., 1993. A model for the hydrologic and climatic behavior of wateron Mars. J. Geophys. Res. 98, 10973–11016.

Clifford, S.M., Parker, T.J., 2001. The evolution of the martian hydrosphere:Implications for the fate of a primordial ocean and the current state of thenorthern plains. Icarus 154, 40–79.

Coleman, N.M., 2003. Aqueous flows carved the outflow channels on Mars.J. Geophys. Res. 108 (E5), doi:10.1029/2002JE001940. 5039.

Coleman, N.M., 2005. Martian megaflood triggered chaos formation, revealinggroundwater depth, cryosphere thickness, and crustal heat flux. J. Geophys.Res. 110, doi:10.1029/2005JE002419. E12520.

Coleman, N.M., 2006. Formation of lakes in the ancestral Valles Marineris inthe epoch of Allegheny Vallis, Mars. Lunar Planet. Sci. XXXVII. Abstract1879.

Coleman, N.M., Baker, V.R., 2007. Evidence that a paleolake overflowed therim of Juventae Chasma, Mars. Lunar Planet. Sci. XXXVIII. Abstract 1046.

Coleman, N.M., Dinwiddie, C.L., Casteel, K., 2003. High channels on Marsindicate Hesperian recharge at low latitudes. In: Sixth Int. Conf. on Mars.Abstract 3071.

Coleman, N.M., Dinwiddie, C.L., Baker, V.R., 2007. Evidence that floodwa-ters filled and overflowed Capri Chasma, Mars. Geophys. Res. Lett. 34,doi:10.1029/2006GL028872. L07201.

De Hon, R.A., Washington, P.A., Thibodeaux, C.J., 2003. Excavation time forthe Vedra and Maumee channels (Mars) by application of equilibrium sed-iment transport theory. Lunar Planet. Sci. XXXIV. Abstract 1178.

Dingman, S.L., 1984. Fluvial Hydrology. Freeman, New York.Dinwiddie, C.L., Coleman, N.M., Necsoiu, M., 2004. Walla Walla Vallis and

Wallula Crater: Two recently discovered martian features record aqueoushistory. Lunar Planet. Sci. XXXV. Abstract 1316.

Feldman, W.C., Prettyman, T.H., Boynton, W.V., Squyres, S.W., Bish, D.L., El-phic, R.C., Funsten, H.O., Lawrence, D.J., Maurice, S., Moore, K.R., Tokar,R.L., Vaniman, D.T., 2003. The global distribution of near-surface hydro-gen on Mars. In: Sixth Int. Conf. on Mars. Abstract 3218.

Ferrill, D.A., Morris, A.P., Waiting, D.J., Franklin, N.M., Sims, D.W., 2003.Influence of gravity on the geometry of martian normal faults. Lunar Planet.Sci. XXXIV. Abstract 2050.

Ferrill, D.A., Wyrick, D.Y., Morris, A.P., Sims, D.W., Franklin, N.M., 2004.Dilational fault slip and pit chain formation on Mars. GSA Today 14 (10),4–12.

Forget, F., Haberle, R.M., Montmessin, F., Levrard, B., Head, J.W., 2006. For-mation of glaciers on Mars by atmospheric precipitation at high obliquity.Science 311, 368–371.

Garvin, J.B., Sakimoto, S.E.H., Frawley, J.J., Schnetzler, C., 2000. North po-lar region craterforms on Mars: Geometric characteristics from the MarsOrbiter Laser Altimeter. Icarus 144, 329–352.

Gendrin, A., Mangold, N., Bibring, J.P., Langevin, Y., Gondet, B., Poulet, F.,Bonello, G., Quantin, C., Mustard, J., Arvidson, R., LeMouelic, S., 2005.Sulfates in martian layered terrains: The OMEGA/Mars Express View. Sci-ence 307, 1587–1591.

Gendrin, A., Bibring, J.P., Quantin, C., Mangold, N., LeMouélic, S., Hauber,E., Gondet, B., Langevin, Y., Poulet, F., Arvidson, R., and the OMEGAteam, 2006. Two years of sulfate mapping in Valles Marineris and TerraMeridiani as seen by OMEGA/Mars Express. Lunar Planet. Sci. XXXVII.Abstract 1872.

Ghatan, G.J., Head, J.W., Wilson, L., Leask, H.J., 2004. Mangala Valles, Mars:Investigations of the source of flood water and early stages of flooding.Lunar Planet. Sci. XXXVII. Abstract 1147.

GRIDVIEW, 2004. An interactive graphics tool for analysis and imaging ofMOLA and other gridded data, available at: http://core2.gsfc.nasa.gov/mola_pub/gridview/gridview.html from the Geodynamics Branch, GoddardSpace Flight Center, 2004.

GSFC, 2004. High-resolution MOLA grid in GMT format (data file 30s_0_300e_330e.grd) is available at http://core2.gsfc.nasa.gov/gridview/gridview_grid.html from the Planetary Geodynamics Laboratory, GoddardSpace Flight Center, Greenbelt, MD.

Harrison, K.P., Grimm, R.E., 2004. Tharsis recharge: A source of ground-water for Martian outflow channels. Geophys. Res. Lett. 31, doi:10.1029/2004GL020502. L14703.

Hartmann, W.K., Neukum, G., 2001. Cratering chronology and the evolution ofMars. Space Sci. Rev. 96, 165–194.

Head, J.W., Marchant, D.R., 2003. Cold-based mountain glaciers on Mars:Western Arsia Mons. Geology 31, 641–644.

Head, J.W., Kreslavsky, M., Hiesinger, H., Ivanov, M., Pratt, S., Seibert, N.,Smith, D.E., Zuber, M.T., 1998. Oceans in the past history of Mars: Tests fortheir presence using Mars Orbiter Laser Altimeter (MOLA) data. Geophys.Res. Lett. 25, 4401–4404.

Head, J.W., Mustard, J.F., Kreslavsky, M.A., Milliken, R.E., Marchant, D.R.,2003. Recent ice ages on Mars. Nature 426, 797–802.

Head, J.W., Neukum, G., Jaumann, R., Hiesinger, H., Hauber, E., Carr,M., Masson, P., Foing, B., Hoffmann, H., Kreslavsky, M., Werner, S.,Milkovich, S., van Gasselt, S., HRSC Co-Investigator Team, 2005. Tropi-cal to mid-latitude snow and ice accumulation, flow and glaciation on Mars.Nature 434, 346–351.

Hoffman, N., 2000. White Mars: A New Model for Mars’ surface and at-mosphere based on CO2. Icarus 146, 326–342.

Hort, M., Weitz, C.M., 2001. Theoretical modeling of eruption plumes on Marsunder current and past climates. J. Geophys. Res. 106, 20547–20562.

Irwin, R.P., Howard, A.D., Maxwell, T.A., 2004. Geomorphology and hy-draulics of Ma’adim Vallis, Mars, during a Noachian/Hesperian boundarypaleoflood. Lunar Planet. Sci. XXXV. Abstract 1852.

Jakosky, B.M., Carr, M.H., 1985. Possible precipitation of ice at low latitudesof Mars during periods of high obliquity. Nature 315, 559–561.

Jakosky, B.M., Phillips, R.J., 2001. Mars’ volatile and climate history. Na-ture 412, 237–244.

Jakosky, B.M., Henderson, B.G., Mellon, M.T., 1993. The Mars water cy-cle at other epochs: Recent history of the polar caps and layered terrain.Icarus 102, 286–297.

http://dx.doi.org/10.1029/2002JE002036

http://dx.doi.org/10.1029/penalty z@ 2001GL013345

http://dx.doi.org/10.1029/penalty z@ 2002JE001845

http://dx.doi.org/10.1029/2002JE002009

http://themis-data.asu.edu

http://dx.doi.org/10.1029/2002JE001940

http://dx.doi.org/10.1029/2005JE002419

http://dx.doi.org/10.1029/2006GL028872

http://core2.gsfc.nasa.gov/mola_pub/gridview/gridview.html

http://core2.gsfc.nasa.gov/mola_pub/gridview/gridview.html

http://core2.gsfc.nasa.gov/gridview/gridview_grid.html

http://core2.gsfc.nasa.gov/gridview/gridview_grid.html






Jakosky, B.M., Henderson, B.G., Mellon, M.T., 1995. Chaotic obliquity and thenature of the martian climate. J. Geophys. Res. 100, 1579–1584.

Jakosky, B.M., Mellon, M.T., Kieffer, H.H., Christensen, P.R., Varnes, E.S.,Lee, S.W., 2000. The thermal inertia of Mars from the Mars Global Sur-veyor Thermal Emission Spectrometer. J. Geophys. Res. 105, 9643–9652.

Jakosky, B.M., Haberle, R.H., Arvidson, R.E., 2005. The changing picture ofvolatiles and climate on Mars. Science 310, 1439–1440.

Komar, P.D., 1979. Comparisons of the hydraulics of water flows in martianoutflow channels with flows of similar scale on Earth. Icarus 37, 156–181.

Komatsu, G., Baker, V.R., 1997. Paleohydrology and flood geomorphology ofAres Vallis. J. Geophys. Res. 102, 4151–4160.

Kreslavsky, M.A., Head, J.W., 2002. Fate of outflow channel effluents in thenorthern lowlands of Mars: The Vastitas Borealis Formation as a sublima-tion residue from frozen ponded bodies of water. J. Geophys. Res. 107,doi:10.1029/2001JE001831.

Laskar, J., Robutel, P., 1993. The chaotic obliquity of the planets. Nature 361,608–612.

Laskar, J., Joutel, F., Robutel, P., 1993. Stabilization of the Earth’s obliquity bythe Moon. Nature 361, 615–617.

Laskar, J., Correia, A.C.M., Gastineau, M., Joutel, F., Levrard, B., Robutel, P.,2004. Long term evolution and chaotic diffusion of the insolation quantitiesof Mars. Icarus 170, 343–364.

Leask, H.J., Wilson, L., Mitchell, K.L., 2006. Formation of Ravi Vallis outflowchannel, Mars: Morphological development, water discharge, and durationestimates. J. Geophys. Res. 111, doi:10.1029/2005JE002550.

Lucchitta, B.K., 1982. Ice sculpture in the martian outflow channels. J. Geo-phys. Res. 87, 9951–9973.

Manga, M., 2004. Martian floods at Cerberus Fossae can be pro-duced by groundwater discharge. Geophys. Res. Lett. 31, doi:10.1029/2003GL018958.

Mellon, M.T., Jakosky, B.M., 1995. The distribution and behavior of martianground ice during past and present epochs. J. Geophys. Res. 100, 11781–11799.

Mischna, M.A., Richardson, M.I., Wilson, R.J., McCleese, D.J., 2003. On theorbital forcing of martian water and CO2 cycles: A general circulationmodel study with simplified volatile schemes. J. Geophys. Res. 108, doi:10.1029/2003JE002051.

Murray, J.B., Muller, J.P., Neukum, G., Werner, S.C., van Gasselt, S., Hauber,E., Markiewicz, W.J., Head, J.W., Foing, B.H., Page, D., Mitchell, K.L.,Portyankina, G., HRSC Co-Investigator Team, 2005. Evidence from theMars Express high resolution stereo camera for a frozen sea close to Mars’equator. Nature 434, 352–356.

Neukum, G., Jaumann, R., Hoffmann, H., Hauber, E., Head, J.W., Basilevsky,A.T., Ivanov, B.A., Werner, S.C., van Gasselt, S., Murray, J.B., McCord,T., HRSC Co-Investigator Team, 2004. Recent and episodic volcanic andglacial activity on Mars revealed by the High Resolution Stereo Camera.Nature 432, 971–979.

O’Connor, J.E., Baker, V.R., 1992. Magnitudes and implications of peak dis-charges from glacial Lake Missoula. GSA Bull. 104, 267–279.

O’Hanlon, L., 2001. The outrageous hypothesis. Nature 413, 664–666.Pearl, J.C., Smith, M.D., Conrath, B.J., Bandfield, J.L., Christensen, P.R., 2001.

Observations of martian ice clouds by the Mars Global Surveyor Ther-mal Emission Spectrometer: The first martian year. J. Geophys. Res. 106,12325–12338.

Quantin, C., Gendrin, A., Mangold, N., Bibring, J.-P., Hauber, E., Allemand,P., OMEGA Team, 2006. Stratigraphy and elevation distribution of sulfatedeposits in Valles Marineris. Lunar Planet. Sci. XXXVII. Abstract 2046.

Roberts, M.J., 2005. Jökulhlaups: A reassessment of floodwater flow throughglaciers. Rev. Geophys. 43, doi:10.1029/2003RG000147. RG1002.

Robinson, M.S., Tanaka, K.L., 1990. Magnitude of a catastrophic flood eventin Kasei Valles, Mars. Geology 18, 902–905.

Scott, D.H., Tanaka, K.L., 1986. Geologic map of the western equatorial regionof Mars. US Geol. Survey Geol. Invest. Series Map I-1802-A.

Scott, E.D., Wilson, L., 2002. Plinian eruptions and passive collapse eventsas mechanisms of formation for martian pit chain craters. J. Geophys.Res. 107, doi:10.1029/2000JE001432.

Shean, D.E., Head, J.W., Marchant, D.R., 2005. Origin and evolution of a cold-based tropical mountain glacier on Mars: The Pavonis Mons fan-shapeddeposit. J. Geophys. Res. 110, doi:10.1029/2004JE002360.

Smith, D.E., Zuber, M.T., Frey, H.V., Garvin, J.B., Head, J.W., Muhlerman,D.O., Pettengill, G.H., Phillips, R.J., Solomon, S.C., Zwally, H.J., Banerdt,W.B., Duxbury, T.C., 1998. Topography of the northern hemisphere of Marsfrom the Mars Orbiter Laser Altimeter. Science 279, 1686–1692.

Stewart, S.T., Nimmo, F., 2002. Surface runoff features on Mars: Testing thecarbon dioxide formation hypothesis. J. Geophys. Res. 107, doi:10.1029/2000JE001465. 5069.

Tanaka, K.L., 1986. The stratigraphy of Mars. In: Proc. 17th Lunar Planet. Sci.Conf. J. Geophys. Res. 91, E139–E158.

Tanaka, K.L., Chapman, M.G., 1990. The relation of catastrophic flooding ofMangala Valles, Mars, to faulting of Memnonia Fossae and Tharsis volcan-ism. J. Geophys. Res. 95, 14,315–14,323.

Urquhart, M.L., Gulick, V.C., 2003. Plausibility of the “White Mars” hypothe-sis based upon the thermal nature of the martian subsurface. Geophys. Rev.Lett. 30, doi:10.1029/2002GL016158.

Vreeland, R.H., Rosenzweig, W.D., Powers, D.W., 2000. Isolation of a 250million-year-old halotolerant bacterium from a primary salt crystal. Na-ture 407, 897–900.

Ward, W.R., 1992. Long term orbital and spin dynamics of Mars. In: Kieffer,H.H., Jakosky, B.M., Snyder, C., Matthews, M.S. (Eds.), Mars. Univ. ofArizona Press, Tucson, pp. 298–320.

Whitney, M.I., 1979. Aerodynamic and vorticity erosion of Mars. Part I. Theformation of channels. GSA Bull. 90, 1111–1127.

Williams, R.W., Phillips, R.J., Malin, M.C., 2000. Flow rates and durationwithin Kasei Valles, Mars: Implications for the formation of a martianocean. Geophys. Res. Lett. 27, 1073–1076.

Wilson, L., Head, J.W., 2004. Evidence for a massive phreatomagmatic eruptionin the initial stages of formation of the Mangala Valles outflow channel,Mars. Geophys. Res. Lett. 31, doi:10.1029/2004GL020322.

Wilson, L., Ghatan, G.J., Head III, J.W., Mitchell, K.L., 2004. Mars outflowchannels: A reappraisal of the estimation of water flow velocities from waterdepths, regional slopes, and channel floor properties. J. Geophys. Res. 109,doi:10.1029/2004JE002281. E09003.

Witbeck, N.E., Tanaka, K.L., Scott, D.H., 1991. Geologic map of the VallesMarineris region, Mars. US Geol. Survey Miscellaneous Invest. Series MapI-2010.

Wyrick, D.Y., Ferrill, D.A., Morris, A.P., Colton, S.L., Sims, D.W., 2004. Dis-tribution, morphology, and origins of martian pit crater chains. J. Geophys.Res. 109, doi:10.1029/2004JE002240. E06005.

http://dx.doi.org/10.1029/2001JE001831

http://dx.doi.org/10.1029/2005JE002550


http://dx.doi.org/10.penalty z@ 1029/2003JE002051

http://dx.doi.org/10.1029/2003RG000147

http://dx.doi.org/10.1029/2000JE001432

http://dx.doi.org/10.1029/2004JE002360


http://dx.doi.org/10.1029/2002GL016158

http://dx.doi.org/10.1029/2004GL020322

http://dx.doi.org/10.1029/2004JE002281

http://dx.doi.org/10.1029/2004JE002240


http://dx.doi.org/10.penalty z@ 1029/2003JE002051


high outﬂow channels on mars indicate hesperian recharge at … · 2013-03-26 · high outﬂow...

Documents