Magmatic intrusions and a hydrothermal origin for fluvial valleys on Mars

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103, NO. E8, PAGES 19,365-19,387, AUGUST 25, 1998

Magmatic intrusions and a hydrothermal origin for fluvial valleys on Mars

Virginia C. Gulick • Space Science Division, NASA Ames Research Center, Moffett Field, California

Abstract. Numerical models of Martian hydrothermal systems demonstrate that systems associated with magmatic intrusions greater than several hundred cubic kilometers can provide sufficient groundwater outflow to form the observed fluvial valleys, if subsurface permeability exceeds about 1.0 darcy. Groundwater outflow increases with increasing intrusion volume and subsurface permeability and is relatively insensitive to intrusion depth and subsurface porosity within the range considered here. Hydrothermally-derived fluids can melt through 1 to 2 km thick ice-rich permafrost layers in several thousand years. Hydrothermal systems thus provide a viable alternative to rainfall for providing surface water for valley formation. This mechanism can form fluvial valleys not only during the postulated early warm, wet climatic epoch, but also during more recent epochs when atmospheric conditions did not favor atmospheric cycling of water. The clustered distribution of the valley networks on a given geologic surface or terrain unit of Mars may also be more compatible with localized, hydrothermally-driven groundwater outflow than regional rainfall. Hydrothermal centers on Mars may have provided appropriate environments for the initiation of life or final oases for the long-term persistence of life.

1. Introduction

Networks of small valley systems extensively dissect the heavily cratered uplands of Mars [Masursky, 1973; Pieri, 1976; Cart and Clow, 1981]. These valleys vary markedly in morphology and scale, ranging from <1 to 10 km in width and from < 5 to 1000 km in length [Baker et al., 1992]. Unlike Earth, the dominant process of valley formation on Mars seems to have been subsurface outflow and associated sapping erosion, involving groundwater outflow and subsequent surface-water flow [Baker et al., 1990]. In contrast to the Martian outflow channels, which formed episodically by cataclysmic releases of groundwater over weeks to months [Baker et al., 1991], the Martian valley networks formed by prolonged groundwater outflow at modest discharges [Baker, 1982] over periods of several 105 years or more [Gulick and Baker, 1993; Gulick, 1993, 1998].

It is not clear, however, how groundwater was mobilized, cycled, and transported into the surface environment over the period needed to form fluvial valleys, particularly since liquid water is presently unstable on the surface of Mars. The prevailing view of Mars' paleoclimate has held that a dense CO 2 atmosphere existed early in the planet's geologic history. Resulting greenhouse warming of the surface permitted rainfall and the operation of a solar-driven (Earth-like) hydrologic cycle [Pollack et al., 1987]. According to this warm, wet scenario, fluvial valleys on Mars would have formed by rainfall during this early period, much like most fluvial valleys form on Earth today. After this period, fluvial valley formation would have ceased.

•Also at Department of Astronomy, New Mexico State University, Las Cruces

Copyright 1998 by the American Geophysical Union.

Paper number 98JE01321. 0148-0227/98/98JE-01321 $09.00

There are a number of inconsistencies between the warm,

wet Mars scenario and the geologic record. The principal difficulty is that fluvial valleys continued to form in localized regions throughout Mars' geologic history [Gulick and Baker, 1989, 1990; Gulick, 1993, 1998]. In particular, valleys continued to form in isolated regions throughout the Hesperian and Amazonian [Gulick and Baker, 1990], long after the end of the putative early warm, wet climate. Furthermore, recent climate modeling [Kasting, 1991] reveals that even large amounts of CO 2 and H20 vapor in the atmosphere could not have produced enough greenhouse warming to bring surface temperatures up to the freezing point of water early in the planer's history when the solar luminosity was lower.

Moreover, geologic evidence for the operation of an Earth- like (solar-driven, atmospheric) hydrologic cycle on Mars is not compelling. Several characteristics of the valley networks set them apart from similar features on Earth. The principal difficulty lies in the spatial distribution 'of Martian valley networks on a given geologic surface or terrain unit. Martian fluvial valleys seem to have formed either in clusters (Figure 1) or as isolated networks (Figure 2) (e.g., Warrego valles, Nirgal valles) separated by vast expanses (up to several hundreds of kilometers in some cases) of undissected terrain of the same unit [Gulick and Baker, 1993; Gulick, 1993,1998; Gulick et al., 1997]. Second, most Martian valleys have a sapping, not runoff, morphology [Baker et al., 1992], and most sapping valleys are not spatially associated with runoff valleys except perhaps for those on some Martian volcanoes [Gulick, 1993, 1998]. Fluvial valleys with a runoff morphology are rare on Mars.

Terrestrial valleys typically have quite different erosional patterns from those on Mars. Within a fixed climatic setting, terrestrial fluvial valleys are generally distributed uniformly over a given geologic surface and seldom form as isolated systems. This uniform distribution of terrestrial fluvial valleys on a given surface largely reflects their rainfall origin

19,365

19,366 GULICK: MARTIAN HYDROTHERMAL SYSTEMS

........

.....

Figure la. Fluvial valleys in the Margaritifer Sinus region of the Southern Highlands on Mars. This is the most densely valleyed region in the ancient heavily cratered terrains. Valley system in the upper left is Parana Valles; note the degraded headwater tributaries and the pristine lower segments. This system has formed on the dark units of the intercrater plains which may be igneous sill intrusions [Wilhelms and Baldwin, 1989]. Valleys are also located around impact craters; note the asymmetric distribution of valleys on the ejecta blankets. A more uniform distribution of valleys around the impact crater would be expected from rainfall.

GULICK: MARTIAN HYDROTHERMAL SYSTEMS 19,367

Figure lb. Sketch map delineating locations of valleys and impact craters in Figure la.

[Gulick 1993, 1998]. Furthermore, most terrestrial fluvial tributaries draining directly into the main valley are relatively valleys exhibit a runoff morphology with tapered tributary rare. Where present on Earth, however, sapping valleys tend heads, headwater regions that blend in gradually with the to form directly from or along with runoff valleys. Examples surrounding terrain, and a tendency toward forming complex of the interaction between runoff and sapping processes systems of branching, integrated tributary networks. include valleys located in the Colorado Plateau, on the Terrestrial valleys exhibiting a sapping morphology with Hawaiian volcanoes, and in the Canterbury Plains (see stubby, theater-headed tributaries, and many first-order discussion by Gulick [1993, 1998]).


•.. :i -

:-. ......

-': ::'" •" .:--... -• ' 'f•-.."

,,

:."i':' :--.."-'i-:?-':' .. : -. :-....•i:'•, ...". -.'-.:. ':?".:

...... . ß .:::,..:.. ."..;!:---- ß -..:. ...... ,:•::':,: ".: ß ;%'"?•:;-:-•. *•;•::½-',,:::i'-:-

•:'7' ' ;:i;•:•:i • . ::' 7':::.." • '..::-:.':':•-::'•

.. .:-.:.-s:::.- .; •.-:-:,•;½-,

............ ß ::.-:---:..:. ß .... •:,,•.:..•,.,..:.. ...... :"" :';:-'• ' ' -::'

• ....... , ....... , • ........... , -:. ....

*;?. "'7; #i?

:*"-':'::":•--- :-'. --.*.'•.• ..'•:-•' -..::.. ".::% ß .'-:•,,.•:'•.*-'.-::..,•'* :?:.

Figure 2a. Warrego Valles (42.5'S, 92.6'), an integrated fluvial valley system developed along the southern edge of Thaumasia Plateau. Although it is the only such system located along the edge of the plateau for several hundred kilometers in either direction (see inset in Figure 2b), this valley system is often invoked as evidence for rainfall early in Mars' geologic history. If this valley system formed primarily by rainfall, then water should also have collected on adjacent areas, eventually producing erosion and subsequent valley formation in the uplands all along the boundary. Instead these isolated valleys exhibit a radial pattern centered on a region of localized uplift [Gulick, 1993; 1998; Gulick et al., 1997]. The uplift may have been produced by a subsurface magmatic intrusion, which in the presence of groundwater, would have formed a hydrothermal system.

These disparities between valley systems on Earth and Mars reflect the importance of more localized water sources and processes in the formation of Martian fluvial valleys. A localized water source could account for those fluvial valleys which have eroded one area of a geologic surface and have left surrounding areas of the same surface essentially untouched for several hundreds of kilometers or more. This isolated or

clustered distribution of Martian valleys does not seem to reflect a widespread source of water, as would be expected from an Earth-like, global atmospheric rainfall origin [Gulick and Baker, 1993; Gulick, 1993, 1998].

An alternative hypothesis that may be more consistent with the geologic record is that geologic heat sources provided the energy needed to drive groundwater outflow from the planer's extensive subsurface ice and water reserves. Throughout the planer's geologic history, much of the fluvial erosion on Mars is spadally and often temporally associated with periods of magmatic and volcanic activity. For example, the outflow

channels are adjacent to the major volcanic centers, Tharsis and Elysium, with the vast majority of channels located near Tharsis, the largest volcanic construct on Mars. Baker et al. [1991] proposed that the Tharsis volcanism and hypothesized associated massive hydrothermal system may have triggered the cataclysmic outflow of groundwater that formed the outflow channels. Similarly, on a more gradual scale, other geologic heat sources could have played a critical role in the formation of Martian valley networks by initiating and maintaining hydraulic gradients of groundwater flowing in the subsurface for several 105 years or more

On Mars, valleys have formed around some impact craters, on and near dark units in the intercrater plains, and on some volcanoes. These locales are consistent with energy supplied by cooling of impact melt sheets [Newsom, 1980; Brakenridge, et al. 1985; Brakenridge, 1990], formation of igneous sills in the intercrater plains [Wilhelms and Baldwin, 1989], other igneous intrusive events in the heavily cratered


i

100 km

Figure 2b. Sketch map of Figure 2a delineating the valley systems comprising Warrego Valles and the location of the southern boundary of the Thaumasia Plateau. Note the radial drainage pattern. Inset shows a regional view of the Thaumasia plateau and the isolated nature of the valley system.

southern highlands [Brakenridge, 1990, Squyres et al., 1987] and volcano formation [Gulick et al., 1988; Gulick and Baker 1989, 1990]. Such geologic activity would melt ground ice, mobilize extensive groundwater reserves and initiate vigorous hydrothermal cycling of groundwater on Mars.

Hydrothermal systems have long been recognized as being potentially important in the fluvial modification of Mars. Schultz et al. [1982] were among the first to suggest that hydrothermal systems may have played a role in valley formation on Mars. Brakenridge et al. [1985] suggested that impact-induced hydrothermal systems could be responsible for valleys located on ejecta blankets on Mars. Gulick and Baker [1989, 1990] proposed that the discharge of hydrothermal fluids to the surface was an important process for the formation of those valleys which formed on the flanks of Martian volcanos. They noted that hydrothermal systems could circulate water in the subsurface over the millions of years required to form fluvial valleys similar in size and degree of development to those formed in analogous terrestrial volcanic landscapes. After considering model difficulties with the warm, wet Mars scenario, Squyres and Kasting [1994] quoted early numerical modeling results of Gulick and collaborators [Gulick et al., 1991; Gulick and Baker, 1993] to argue that hydrothermal circulation could have indeed have played a more important role than precipitation in Martian valley formation. Most recently, Gulick et al., [1997] proposed that a substantial amount of water could be transported during modest

greenhouse periods from surfaces of frozen bodies of water to higher elevations, despite global temperatures well below freezing (>-240 to 250 K). This water, precipitated as snow, could ultimately form fluvial valleys if deposition sites were located at or near regions of hydrothermal activity. Hydrothermal systems have also received a great deal of attention as agents for producing aqueous alteration of the SNC meteorites [Wentworth and Gooding, 1994], for exchanging deuterium and hydrogen between the crust and atmosphere [Jakosky and Jones, 1994], for providing paleo habitats for life, and for preservation of fossils [Walter and DesMarais, 1993; Farmer and DesMarais, 1994].

This paper presents results of numerical simulations of groundwater cycling and outflow that would be expected from magmatically generated hydrothermal systems in the Martian environment. These modeling results place constraints on the duration, magnitude, and extent of groundwater cycling and subsequent outflow as well as on the subsurface material properties (e.g., permeability) and the intrusion sizes that are compatible with valley formation on Mars.

2. Conceptual Model

The near-surface of Mars probably consists of a blanket of impact ejecta several kilometers thick, interbedded with lava flows, ash beds, and sedimentary material [Squyres et al., 1992]. Underlying this material is a heavily fractured


basement rock that has been intensely bombarded by impacts [Carr, 1979; MacKinnon and Tanaka, 1989]. In many areas, lava flows have subsequently buried this entire sequence, a sequence that is likely similar to the blocky, porous megaregolith that comprises the lunar crust. The permeability of the subsurface is poorly constrained. Permeabilities of geologic materials range from near 0 to over 4000 darcys [Davis and DeWiest, 1966, Carr, 1979]. For example, terrestrial basalts typically have permeabilities ranging from 0.10 to 3000 darcys, with Hawaiian lava flows having some of the highest values [Davis and DeWlest, 1966]. Carr [1979] found that Martian megaregolith permeabilities as large as 3000 darcys were consistent with the discharge rates required for outflow channel formation. Ice likely fills the pores of near-surface materials to a depth ranging from 1 to 3 km at the equator and midlatitudes to a depth of approximately 3 to 8 km at the poles [Fanale, 1976, Rossbacher and Judson, 1981]. Below these regions, groundwater could fill voids to a depth of about 20 km, where the lithostatic pressure will close pores [Squyres et al., 1992]. Throughout Mars' geologic history, the formation of large impacts, volcanoes and igneous intrusions would have repeatedly perturbed the equilibrium thermal structure of the ice-rich permafrost and the underlying groundwater zone.

As an example of such a thermal perturbation, Figure 3 illustrates an idealized conceptual model of groundwater flow associated with a magmatically-generated hydrothermal system as might be produced by the formation of a Martian volcano. Soon after emplacement of the magma, the outer shell of the magma chamber starts to solidify, forming a low

permeability outer shell of hot rock, the thickness of which increases with time [Gasparini and Mantovani, 1984]. Thermal energy is transported primarily by conduction from the magma through the shell, and then primarily by convection into the saturated, permeable country rock. This shell is presumed to prevent groundwater from contacting the intrusion itself. Surrounding groundwater that is heated by the magma forms an upwardly moving buoyant plume of groundwater near the intrusion. Colder, denser groundwater flows in toward the intrusion from surrounding regions, and continues to replace the upwardly moving groundwater as long as a thermal gradient exists. Depending on the size of the intrusion, groundwater within several tens of kilometers or more could flow into the system. Ground ice above and near the intrusion would be melted, locally eliminating or thinning the permafrost zone [Gulick and Baker, 1992; Gulick, 1993]. Groundwater reaches the surface as liquid or vapor or both. The near-surface behavior of the hydrothermal fluids would depend on their temperature and mineral concentration, and on the atmospheric temperature and pressure.

Groundwater which reaches the near-surface environment

can contribute to the geomorphic modification of that surface. If local hydrologic and lithologic conditions permit, water could flow on the surface and re-enter the groundwater system in regions where the rate of infiltration is sufficiently high. However, if the atmospheric temperature and pressure are not favorable for fluid flow, groundwater would initially start to boil and evaporate but then freeze due to the heat liberating process of evaporation. This process would result in the formation of an insulating ice layer beneath which subsequent

Fi'i'•:d ::--'.•)•"nt: l" "X•'•':••Ie•'i'B::•:•:•al:"•.:........••••••.•:.••:.• ..... ::.:,' ...... :.::•.:.•-'-..:•--' '.•••:: ............. '.• ..':.'•i'.::•:: ? .-S'p.'r•:-2•:•:.:•:•:• ..................... :::•....:. ...................... ::::::::::::::::::::::::::::::::: ............................. :•.:•: ......

..........

: .•.

'Ma (:• m a. Cham.;b

'Perm-:-/j. ef-"mst'

Figure 3a. Conceptual model illustrating the groundwater flow field of a vigorous hydrothermal system associated with volcano formation on Mars. Vertical scale is exaggerated by a factor of 4 in order to illustrate details of groundwater flow in the stratigraphic layers of the volcano. Martian volcanoes are unusually large compared to their terrestrial counterparts, ranging on the order of 102 to 103 km in diameters, and those that are dissected by valleys tend to have low aspect ratios with average slopes less than a few degrees. Our numerical model considers hydrothermal systems associated with magmatic intrusions on Mars in general. Topography, multiple intrusions, and boiling are not directly simulated in our model, although their effects on these systems are discussed elsewhere in this paper.


e Zone

Ash Bed

Sapping

Figure 3b. Conceptual model of small valley formation by groundwater outflow on Mars. See text for details.

outflows of hydrothermal water may move as ice-covered rivers [Wallace and Sagan, 1979; Carr, 1983; Brakenridge et al., 1985]. Groundwater that does not intersect with the surface

would help to recharge near-surface aquifers and eventually could outflow to the surface farther away from the intrusion. Whether groundwater remains liquid would depend on the local lithologic conditions, the temperature and mineral concentration of the water, and the atmospheric conditions at the time of groundwater outflow.

While most terrestrial fluvial valleys form mainly by surface runoff processes, some do form by a combination of runoff and sapping. On Mars, the morphology suggests that groundwater sapping was the dominant valley-forming process [Mars Channel Working Group, 1983]. Some valleys, particularly those on the Martian volcanos, also exhibit a runoff morphology component. Despite the presence of some runoff characteristics, it may be possible to form valleys exhibiting such compound morphology solely by groundwater outflow processes. Figure 3 illustrates such a scenario on the slopes of a volcano. In Figure 3a, the volcano's left flank is composed of permeable basalt, so any outflowing groundwater would quickly infiltrate back into the subsurface and recharge near-surface aquifers. This water may again intersect the surface farther down the flank of the volcano, forming a seepage face. At this site, sapping processes (i.e., erosion of the surface by groundwater outflow) may eventually form a valley. On the right flank, the surface is mantled with ash. Water flowing on this ash surface, which is much less permeable than the underlying basalt, infiltrates much more slowly, thereby allowing more water to flow on the surface. With continued surface flow, water will start to erode the

surface and may eventually form a fluvial valley with a dominant runoff morphology.

Processes acting on the right flank are illustrated in more detail in Figure 3b. If the seepage face intersects the surface on a very gently sloping surface (< 5ø), as is common on many Martian volcanoes [R½imers and Komar, 1979; Mouginis- Mark et al., 1992] containing valley forms, and if the surface downslope of the seepage is such that the rate of runoff is greater than the rate of infiltration and is fairly easily eroded

(e.g., ash-mantled), fluvial valleys with a dominant runoff morphology may form. However, if groundwater emerges from springs or seepage faces on a cliff or more steeply sloping surface, valleys with a dominant sapping morphology will probably form. Indeed as a valley which originally exhibited a dominant runoff morphology continues to incise the surface and form steeper slopes, it will eventually tap into deeper aquifers and a sapping morphology will begin to dominate as illustrated. Hence, in this case, groundwater outflow may form fluvial valleys exhibiting both runoff (at least initially) and sapping morphology. This model of valley pattern development and morphology is consistent with those proposed for small valleys in the heavily cratered terrains [Baker and Partridge, 1986] and on the Martian volcanoes [Gulick and Baker, 1989, 1990].

3. Numerical Model

Existing models of terrestrial hydrothermal systems. focus on relatively low-permeability (10 -6 to 10 -3 darcy), high- temperature subsurface environments appropriate for mineral or energy exploitation. However, hydrothermal systems in high-permeability materials (10-1 to 103 darcys) that likely comprise the near-surface of Mars have not been studied. Generally, heat transport in hydrothermal systems characterized by permeabilities above 10 -3 darcy is controlled by advection rather than conduction. High-permeability systems are expected to produce much cooler fluid and wall- rock temperatures and higher fluid velocities than in the well- studied low-permeability environments [Cathies, 1988]. Rather than extrapolating results of relatively low permeability terrestrial hydrothermal system models to Martian conditions, I present here results of my own simulations of Martian hydrothermal systems.

The finite-element, density-dependent groundwater flow simulation computer code SUTRA [Voss, 1984] was used to model groundwater flow and energy transport. SUTRA's simplified treatment of water properties was replaced with an accurate equation of state for H20 [Johnson, 1987]. Details of the numerical model are presented in the appendix. A disk- shaped region surrounding a magmatic intrusion was modeled. As many Martian volcanoes are characterized by exceptionally small slopes [Mouginis-Mark et al., 1992], topography was not directly included in the model. Instead, a more general model of a Martian hydrothermal system surrounding a magmatic intrusion was constructed. For the baseline results reported here, slopes comparable to those on the volcano Alba Patera would project into only a 200 m height difference across the model disk, which is 6 km in depth and 10 km in radius.

The hydrothermal system is modeled using a radially symmetric, cylindrical grid with a central cavity for the intrusion (Figure 4a). In the baseline model, a homogeneous, water-saturated medium having a porosity of 0.25 and a permeability of 10 darcys surrounds the intrusion. Similar models with subsurface permeabilities of 0.1 and 1000 darcys are also considered. Heterogeneous models with confining and interbedded layers having a subsurface permeability of 10 -7 that of the background are also explored. Inhomogeneities in the flow field at scales smaller than the grid are accounted for by the dispersivity (set equal to 100 m and assumed isotropic), a measure of the largest obstacle to the flow. Fluid flow boundary conditions are shown in Figure 4b, and discussed in


-- Flow • = 1 bar

I•- ...... Insulating ;No flow

Figure 4a. Numerical grid used in modeling hydrothermal systems. The hydrothermal system is modeled using a radially symmetric, cylindrical grid with a central cavity as illustrated in Figure 4b. Single, cylindrical intrusions of 4 km height are emplaced in the central cavity at a depth of 2 km beneath the Martian surface. The modeled region extends from the outer radius of the intrusion r i to a distance R = 20r i. The 6-km-high grid consists of 440 elements, each 400 m high. Element widths are scaled to the radius of the intrusion and vary from several meters near the intrusion to approximately I km along the outer edge. Each element is 2nr thick, where r is the inner radius of the element.

the figure caption. The background geothermal gradient is neglected.

At time t = 0, the subsurface is saturated with 0øC

groundwater in hydrostatic equilibrium. A single 50 or 500 km 3 cylindrical intrusion of basaltic magma with an initial temperature of 1250øC is then emplaced interior to the grid. Conservative estimates of magma chamber volumes on selected Martian volcanoes range from several 101 to severa'l

102 km 3 [Mouginis-Mark et al., 1982; Wilson and Mouginis- Mark, 1987]. The intrusion is assumed to remain impermeable thereby preventing the magma from coming into direct contact with groundwater. All the thermal energy of the intrusion is assumed to enter the surrounding region; no energy is lost through its roof.

Thermal energy is delivered to the grid by application of a time varying heat flow boundary condition along the inner

z

Figure 4b. system.

THICK- r i i

Radially symmetric, cyclindrical grid with a central cavity used to model the hydrothermal


edge of the grid. This boundary condition simulates the conductive transport of energy from the intrusion into the country rock [Jaeger, 1968]. The temperature along this boundary is not specified since the temperature at the intrusion/wall-rock interface is expected to be significantly cooler than in low permeability systems [Cathles, 1988]. Given the flux, SUTRA computes the resulting temperature at the boundary. Energy is then transported away from the boundary by both convective and conductive processes. Consequently, the resulting temperature gradient is steeper than in a purely conductive system. The actual cooling rate of the intrusion will therefore be somewhat higher than assumed here, resulting in a more intense, shorter-lived (by possibly as much as a factor of 3 to 4) [Norton, 1984] hydrothermal system. This inconsistency has little effect on the hydrothermal system in terms of its duration over time-scales required for valley formation.

4. Model Results

The baseline model is homogeneous with a permeability of 10 darcys, a porosity of 25%, and an intrusion volume of 50 km 3. The effect of variations in the subsurface permeability, porosity, subsurface geology, magma chamber size and depth are explored by changing one variable at a time.

Results for the baseline model are presented in Figure 5. Figures 5a-5c, show temperature contours (5øC contour intervals) of a rising hydrothermal plume at various times after

emplacement of the magmatic intrusion. Peak discharges are reached about 600 years after emplacement of the intrusion, yet groundwater outflow continues for approximately 100,000 years. Figure 5d shows the velocity flow field of the groundwater induced by the thermal anomaly at peak discharge to the surface. The longest vectors represent a velocity of 70 rn/yr. Note that because the subsurface is homogeneous in these models, the discharge is confined to the region near the intrusion.

Figure 6 illustrates a situation where multiple less permeable layers are present. The upper layer extends only a short distance (to radial distance r = 4 km) and impedes direct access of the plume to the surface, while the lower layer extends from 3.3 km to the end of the modeled region. Here, upward moving hydrothermal fluids flow around the upper layer and emerge at the surface at greater distances from the intrusion than in the homogeneous baseline case. Hayba and lngebritsen [1997] also found that caprocks above hydrothermal systems produce lateral flow. They further noted both higher temperatures and more boiling in their models of hydrothermal systems beneath caprocks. If the upper confining layer in our model were ice-rich permafrost, upward moving hydrothermal waters would eventually melt through the layer [Gulick and Baker, 1992; Gulick, 1993] as is discussed below.

In some respects, the effects of an impermeable caprock are similar to those to topography above the intrusion. Numerical modeling of terrestrial hydrothermal systems in the Cascades

lO 5

I I

t = 187 years

I

t = 128,000 years

V=50km2 k=10d

t = 628 years

I I

2 4 • 8 10 2 Radius (km) Radius (km)

/___

.<

<

.<

.<

,

•

Figure 5. Temperature contours (5øC intervals) of a rising hydrothermal plume at (a) 187 years, (b) 628 years, and (c) 128,000 years after emplacement of a 50 km 3 magmatic intrusion. Plot of the groundwater flow field for the system after 628 years is shown in Figure 5d. Vector length is proportional to fluid velocity; the longest vectors represent velocities of approximately 70 rn/yr. The permeability in this system is 10 darcys and is assumed isotropic. The relatively high subsurface permeability used in the model results in higher groundwater velocities and correspondingly lower groundwater temperatures than in low-permeability systems. The low groundwater temperatures combined with a 1 bar surface pressure boundary condition produce an environment in which boiling does not occur. Higher fluid temperatures and/or lower surface pressures would result in near-surface boiling.


2 4 6 8 10 2 4 6 8 10

Radius (krn) Radius (krn)

Figure 6. Similar to the baseline model presented in Figure 5, except that less permeable layers (k = 10 -6 darcys, as shown by shaded areas ih the flow field plots) have been added to the subsurface. As a result of these layers, the plume is hotter and spreads out farther from the intrusion. Fluids find egress to the right of upper layer: (a) t = 300 years, (b) t = 1051 years (peak surface discharge), and (c) t = 139,600 years. Groundwater flow field plots correspond to the temperature plots for (a) t = 300 years, (b) t = 1051 years, and (c) t = 139,600 years. Maximum flow velocity in Figure 6b is 59 m/yr. Boiling is not allowed.

suggests large topography over the intrusion increases the lateral migration of the plume, thereby increasing the distance from the volcanic crest where springs and seeps can emerge [Sammel et al., 1988]. The presence of a relatively impermeable caprock, less permeable interbeds, or topography might account for groundwater outflow at significant distances from an intrusion and why some Martian fluvial valleys begin farther down the flank of a volcano than the homogeneous model would predict (e.g., the Martian volcano Alba Patera).

Figure 7a compares the groundwater discharge to the surface with time for the homogeneous and layered cases. Although peak discharge of fluid to the surface is delayed in the layered case, the total volume of groundwater discharged to the surface is similar because the buoyant plume simply flows around the upper less permeable layer. While these layers do not impede groundwater outflow to the surface, they do hinder surface-water recharge back into the system. Water that enters the numerical model from the surface is not a significant source of recharge into the system. Instead, sufficient groundwater i s drawn into the modeled region along the outer (right) vertical boundary by the hydrothermal system, so that the total volume of water discharged is unaffected.

Figure 7b illustrates the total surficial discharge of groundwater over the lifetime of hydrothermal systems associated with a single 500 km 3 magmatic intrusion for host rock permeabilities of 0.10, 10, and 1000 darcys. The curves in each graph illustrate the dependence of discharge on the permeability of the system; the higher the permeability, the higher the quantity of groundwater outflow from hydrothermal systems associated with a given intrusion volume. However, when comparing total volumes of water discharged from hydrothermal systems associated with different intrusion volumes, the total volume of water discharged is controlled primarily by intrusion volume and to a lesser extent the host rock permeability.

Groundwater outflows produced by the modeled hydrothermal systems can be compared with the amount of surface-water flow produced by rainfall. As shown in Figure 7b, the dashed horizontal lines indicate the rainfall on Hawaii

island over a region equivalent in area to the discharge regions in the numerical models. The average yearly rainfall over the island and the average yearly rainfall on the leeward side (dry side) of the island that would be expected over these regions are shown by horizontal lines in Figure 7b. The lengths of these lines represent the approximate age (400,000 years) of


3,500

3,000

2,500

2,000

1,500

1,000

5OO

Baseline

Two Layers

1 2 3 4 5 6 10 10 10 10 10 10

Time (years)

Figure 7a. Surficial discharge of groundwater outflow with time for hydrothermal systems associated with a 50 km • intrusion. Thick line illustrates baseline model (Figure 5) discharge from a homogeneous subsurface with k = 10 darcy' thin line for layered model shown in Figure 6. Note that while peak surface discharge occurs later in the layered model, total volume of groundwater discharged to the surface is similar for both models.

3,000

2,000

1,000

V- 500 km 3

HI

ave 1000 darcys

10 darcys

O. 1 darcys

HI dry

0 2 4 6 8 10 12 14

Time (10 5 years) Figure 7b. Surficial discharge as a function of time for a hydrothermal system associated with a 500 km * intrusion. Three curves illustrate discharge from a homogeneous subsurface for k = 0.1, 10, and 1000 darcys. The horizontal dotted lines represent the average and below average (dry, leeward side) rainfall per year on the island of Hawaii multiplied by the average area over which the model discharges groundwater to the surface. The lengths of these lines correspond to the approximate age of the fluvial valleys on Kohala volcano, the oldest volcano on the island of Hawaii.


the fluvial valleys on Kohala volcano, the oldest volcano on the island. The volume of groundwater delivered to the surface over the lifetime of a Martian hydrothermal system is comparable to that produced by rainfall on Hawaii, a quantity that is empirically sufficient to form fluvial valleys.

A shortcoming of the numerical model is the assumption that the intrusion remains impermeable throughout the lifetime of the hydrothermal system. In reality, the intrusion will eventually fracture as it continues to cool, allowing water to flow through the intrusion, thereby accelerating the cooling process. Such fracturing does not happen until late in the evolution of the magma chamber, and conduction is the dominant heat transport mechanism for most of the lifetime of the intrusion [Gasparini and Mantovani, 1984]. Furthermore, since the available energy is conserved in our model, the total volume of groundwater discharged from a real system should still be similar to that calculated here. However, the variation

of this outflow with time is likely to be different with a late pulse of water reaching the surface when the intrusion fractures followed by a more rapid decline.

5. Sensitivity to Other Model Parameters

Several other parameters were evaluated for their effect on model results. Subsurface permeability and its distribution and the intrusion volume had the greatest effect on the subsurface

fluid flow and the surface discharge. Variations in intrusion depth, subsurface porosity, and dispersivity were also evaluated for their effect on results. These parameters (and their associated uncertainties) have a much smaller impact on the calculations than those discussed above.

5.1 Intrusion Depth

A "deep" intrusion was considered where all heat was conducted from 4.4 to 6 km beneath the surface, and a "shallow" intrusion was considered where the conductive

region was from 0.8 to 6 km beneath the surface. In each calculation, the total energy entering the modeled region was the same as in the baseline case. Therefore, the differences

between the models are due solely to the depth of energy delivery and not to the total energy input into the model.

The response of the model to these differing conditions is shown in Figures 8 and 9. As a consequence of the assumptions listed above, the deeper intrusion model is characterized by a higher heat flux per unit area than the more shallow model. The peak fluid temperatures and velocities reached in the "deep" model are higher than in the "shallow" model. For example at approximately 1600 years, the peak velocities are 66 and 46 m yr -1, respectively. This 40% velocity difference results in a correspondingly greater dispersion of the flow in the deep case. Therefore the rising plume spreads out over a wider area and, due to mixing, heats a

a t - 63 years

t - 193,000 years

i i i

2 4 6 8 10

Radius (km)

< '<i < <

c

Radius (km)

/_.

/_.

• <

< <

< <

<

•o

2 e

4•

2 e

4•

Figure 8. Similar to the baseline model, except intrusion is more shallow (0.8 to 6 km). Shown are 5 ø C temperature contours and fluid velocity flow field of (a) rising plume at t = 63 years, (b) plume at peak discharge at 317 years, and (c) plume at 193,000 years.


a t = 317 years

• i i

b t = 1564 years

c t - 193,000 years

= 5

2 4 6 8

Radius (km)

•9•AAA <V V V V V I/ V //

iI•A A A /" // // // // // // /"' •

10 2 4 6

Radius (km)

<

<

<

<

<

< ,

<

<

<

10

2 o

4,•

6

2 •

4'•

2 •

4'•

Figure 9. Similar to model in Figure 6, except that the heat source is deeper (4.4 to 6 km). Total volume of intrusion is 50 km 3. Temperature contours at 5 ø C intervals and fluid velocity flow fields are depicted for (a) a rising plume at t = 317 years, (b) plume at peak fluid flow discharge out of the system at t = 1564 years, and (c) plume at 193,000 years after the initiation of hydrothermal activity.

larger volume of water. This may be seen by comparing Figures 8 and 9.

The discharge of the two systems and the baseline model is compared in Figure 10. Because of the mixing, the deep model produces a greater discharge as a function of time than either the baseline or shallow models. The cumulative discharged volume of water to the surface differs between the three models

by only about 10%. The closer proximity of the shallow intrusion to the surface somewhat compensates for its lower fluid velocities. Hence model results are not sensitive to

intrusion depth within the interval considered here. However, very different geometries from those modeled here, i n particular, very shallow (e.g., impact melt sheets) or very deep intrusions would b.e expected to behave quite differently.

5.2 Porosity

The country rock porosity assumed was also varied in some models. Porosity is a measure of the total volume of pore space to the total volume of medium (pores + matrix). Because porosity is not a measure of the degree of interconnected pore space, it is not expected to have as strong a control on groundwater flow as permeability. This was also found to be true in the modeling. In contrast, permeability is a measure of how well the pore space is interconnected within the medium.

Because it reflects the ease with which water travels through the country rock, permeability has a far more significant effect on the groundwater flow.

Temperature contours and velocity vectors are shown for a high-porosity model in Figure 11. In this model, the country rock has a porosity of 35%. The temperature contours and flow characteristics are very similar to the baseline model with a porosity of 25%. The only significant difference is that the fluid velocity is slower than in the baseline model. Not shown are results from a low-porosity model in which the fluid velocity is higher than the baseline model. The buoyancy forces are the same in all three model calculations; only the size of the fluid pathways and therefore the fluid velocity is different. The total discharged volume for all three models (low, high, and baseline porosity) was essentially identical. Therefore the uncertainty in porosity is not a critical element of the models.

Changing the porosity does affect the total quantity of groundwater contained within the finite element grid at any given time. However, since all model cases considered are assumed to be surrounded by an infinite groundwater reservoir, this difference is of little importance. If there were an impermeable boundary adjacent to the modeled region, models with a smaller porosity (and no recharge) would discharge a smaller total quantity of water to the surface.


3,000

2,500

.-.. 2,000

• •,5oo

._

1,000

500

Basehne Shallow

0 • 0 2 4 10 1 10 3 10 10 5 10 6 -15me (years)

Figure 10. Comparison of discharges for baseline, deep, and shallow intrusion models. Differences in first 1000 years are due primarily to selection of different time steps. For lifetime of the hydrothermal system, such differences in intrusion depth do not significantly affect the discharge.

5.3 Dispersivity

SUTRA uses the dispersivity to account for impediments to groundwater flow smaller than the grid's cell size. Maintenance of stability of the numerical calculation requires that (Z•/a) < 4, where AL is the local distance between

element sides along a streamline of flow and a is the dispersivity [Voss, 1984]. This criterion is of special importance when temperatures change along streamlines, as is the case in the hydrothermal simulation. Since the regime of primary interest is the vertical, upward flow where the element spacing is a uniform 400 m, the minimum dispersivity required for numerical stability is 100 m. This was the value used in all calculations except in the following case.

In order to investigate the sensitivity of the model to this parameter, one case with a dispersivity of 500 m was also run. As expected, this model was characterized by greater mixing. Streamlines strayed farther from the intrusion, and a greater volume of cold water was mixed with upward flowing water (Figure 12) which had been heated by the intrusion. The result was a cooler average temperature for the water which reached the surface and a slightly higher discharge (Figure 13), since more water was heated by mixing, albeit to a lower temperature. Larger dispersivities are more accurately analyzed by introducing impermeable layers in the model.

These sensitivity tests suggest that the results of the numerical model are quite robust. Most model parameters do not significantly affect the outcome of the calculation. For a given permeability and energy input, which is determined by the size of the intrusion, the total discharged volume is relatively unaffected by other model details. As long as the fluid has a path to the surface and a recharge source, the plume grows and discharges water to the surface. What happens to this fluid once it flows out onto the surface depends on the

vmx = 86 m/yr

b = Vmax- 77 m/yr

q/- -,•/-- -F-

Vrnax = 2 rn/yr

2 4 6 8 10 2 4 6 8 10

Radius (kin) Radius (kin)

1:3 2 e

4•

1:3 2 e

4'•

Figure 11. Similar to Figure 5, but for a model with the baseline intrusion depth and a homogeneous porosity of 35%.


564 years

I

2 4 6 8

Radius (km)

Figure 12. Temperature contours for baseline-type model with a dispersivity of 500 m at t = 1564 years. Note that while the temperatures are similar to the baseline model, the plume has dispersed over a larger area. Compare to Figure 5.

vagaries of near surface conditions at the time of groundwater outflow.

6. Physical Considerations

The numerical model presented in this paper is by necessity idealized. Here, I briefly consider various physical complexities that are likely present on Mars today.

6.1 Permafrost

The presence of ice-rich permafrost would affect an active hydrothermal system on Mars. There is extensive morphologic evidence for ground ice in the Martian subsurface [Squyres et al., 1992] and numerous estimates for the thickness

of the ice-rich Martian permafrost. Most estimates of ice thickness are based on theoretical considerations of where ice

would be stable, given present atmospheric conditions; some are based on geomorphic studies of landforms. Regardless of the approach used, the thickness of this zone varies with latitude and a common estimate for midlatitude permafrost thickness is 2 km [Squyres et al., 1992].

Upwardly moving hydrothermal fluids must melt through this permafrost to reach the surface. The energy E required to melt through a cubic centimeter of permafrost is given by

E= AT(,OrCp, r(1-œ)+,OiCp, iœ)+œPiL (1) where AT is the difference between the initial temperature of the permafrost and its melting temperature, p is the density of

either rock (r)or ice (i), C? is the heat capacity per gram of rock (r) or ice (i) (assumed •ndependent of temperature), L is the latent heat of melting of ice, and e is the volume fraction of the permafrost that is occupied by ice. Therefore hydrothermal waters impinging on the base of the permafrost must deliver enough energy to raise the temperature of the rock/ice mixture to the melting point of water and also provide the latent heat of melting.

The time required to melt through a 2 km thick permafrost zone can be estimated by dividing the energy required to melt the ice contained within this zone by the energy flux. For the baseline 50 km 3 intrusion, the convective mass flux averaged over the first 10,000 years of hydrothermal activity is 5 x 10 -6 g cm -2 s-1 over an area of approximately 20 km 2. The amount of energy this water delivers to the permafrost can be measured by the amount the water cools upon contact. In Figure 14, the time required to melt through the permafrost is shown as a function of porosity, assuming the water cools by 3.3, 10, and 30øC upon contact with the permafrost. These temperatures correspond to energy fluxes of 0.7, 2.1, and 6.3 W/m 2. The calculation assumes that all the pore space within the soil is filled with ice and that the initial temperature averaged over the thickness of this zone is 240 K. Since the energy necessary to heat both the soil and ice to 273 K must be provided, a finite

3,000

2,500

2,000

1,500

1,000

Baseline

Dispersivity

Porosity

lO 1 10 2 4

Time (years)

Figure 13. Discharge as a function of time for the baseline, high-dispersivity, and high-porosity models. The high-dispersivity model results in greater mixing of warm and cool water and a larger discharge. The high- porosity model produces a curve which is not distinguishable from the baseline model.


50,000

40,000

30,000

20,000

2 F-O.7Wm

2.1

10,000

0 0.0 0.2 0.4 0.6 0.8

Porosity

Figure 14. Time for an upward fluid flux of 5 x 10 -6 gcm -2 s 'l to melt through a 2-km-thick permafrost layer as a function of permafrost volume fraction of ice. Different lines represent a range of energy fluxes delivered to the ice.

amount of time is required to heat the soil even if ice is not present.

If permafrost on Mars fills a region with a pore space of 1 0 to 35%, then hydrothermal fluids can melt through 2 km thick permafrost in several thousand years. This is a conservative estimate, since in reality, some pore space may also contain gas along with ice. In this case, the amount of energy and time required to melt through the permafrost would be less. However, assuming the conservative scenario, the several thousand year period required to melt through the permafrost is

short compared to the lifetime of the hydrothel•mal system (about 100,000 years for the baseline 50 km" intrusion). Therefore the presence of an ice-rich permafrost zone should have a negligible effect on the lifetime of a hydrothermal system on Mars.

Even directly above a volcanic intrusion, the surface temperature at Mars would be primarily controlled by the balance between absorbed solar and emitted infrared radiation

[Fanale et al., 1992]. The surface temperature will remain below freezing and a residual ice-rich permafrost zone will remain near the surface, except for areas directly above the intrusion or where springs or seeps have formed. Figure 1 5 illustrates the equilibrium thickness for a variety of heat flows and surface temperatures (applying McKay et al. [1985, equation (2)]). A thermal conductivity of 3.3 W m -1 K -1 is assumed. The present-day background geothermal gradient probably provides on average around 0.03-0.04 W/m 2 [Fanale, 1976; ToksOz and Hsui, 1978; Davies and Arvidson,

1981]. However, the average heat flow in the presence of an active terrestrial hydrothermal system can range from 2 to 5 W/m 2. For example, at Wairaki, New Zealand, the regional heat flow averaged over approximately 50 km 2 is 2.1 W/m 2, while fluxes averaged over the more intense regions are of order 500 W/m 2 particularly in localized areas around springs [Elder, 1981]. The graph shows that the equilibrium permafrost thickness above an active hydrothermal system may be less than 100 m. Any inhomogeneities in the

subsurface, such as fractures, would permit egress of hydrothermal water to the surface. As shown in Figure 15, the system can adjust to the new permafrost equilibrium thickness in less than about 10,000 years, a time that is short compared to the lifetime of the hydrothermal system.

6.2. Thermal Vapor Diffusion

Clifford [1991] explored water transport in the subsurface of Mars by the mechanism of thermal vapor diffusion. In the presence of a thermal gradient, water vapor in a porous medium will migrate from warmer (higher vapor pressure) regions to colder (lower vapor pressure) regions, where it will condense. The condensed water may then flow downward back to the water table or, alternatively flow along upper, less permeable layers, if present. The thermally driven vapor flux d depends on the diffusion coefficient D, the thermal gradient a = dT/dz, and the saturated vapor pressure P at the top of the water table at temperature T:

DPo: Ja • (2)

T 3

D is proportional to T 1/2 or T 3/2 depending on whether the pores are small or large compared to the mean free path of the water molecules. For the case examined by Clifford, T = 20øC and cr = 15 K/kin, the flux of vapor reaching 1.33 km above the water table was approximately 104m H20 yr -•, with a factor of 3 uncertainty depending upon the pore size. However, in the presence of an active hydrothermal system, water temperatures may exceed 50øC, and the temperature gradient may exceed 50 K/km in an unsaturated zone directly above the intrusion. Under such conditions the thermal vapor diffusion flux would be 13 to 15 times greater than estimated for background conditions by Clifford. For even hotter water (and

10

:F::O.1

Ja = 180K

Tsurf = 220K

Tsurf =260K

Figure 15. Equilibrium permafrost thickness as a function of geothermal heat flow. Permafrost thickness is shown for three mean surface temperatures. Typical estimates of current Martian heat flow as well as geomorphologic evidence indicate permafrost thicknesses of approximately 2 kin. Higher surface temperatures and interior heat fluxes both produce thinner permafrost layers. Heat flow above an active hydrothermal system may be as large as 3 to 5 W m '2, indicating equilibrium thicknesses of several hundred meters or less.

0.01

0 1 2 3 4 5

Heat flow (watt/rn 2)


higher vapor pressures), the flux could be as much as 20 times greater than the Clifford fluxes, corresponding to several 10 -3 m/yr or 1 km/106 yr.

While thermal vapor diffusion clearly does not provide the large fluxes of water transported by direct fluid flow (which can produce near-surface velocities of tens of meters per year), it does deliver groundwater to the near-surface environment when direct fluid flow cannot. This mechanism may play a role in the long-term replenishing of near-surface, perched aquifers if the condensing water is trapped by a less permeable layer (as in Figure 16) and does not return to the deeper water table [Clifford, 1991]. Over the lifetime of a hydrothermal system, several kilometers of water may be transported by this mechanism.

Early in the lifetime of the system, boiling is likely to be more significant, particularly above shallow intrusions. In terrestrial hydrothermal systems, boiling can occur within 2 km of the surface and can have a complex influence on the fluid flow. Near the critical point, changes in temperature can dramatically alter fluid properties and greatly affect mineral deposition or solution locally [Furlong et al., 1991]. Because boiling effects operate on scales much smaller than model cells, boiling is a difficult problem, even in models of terrestrial hydrothermal systems.

6.3 Impact Melt Sheets

Brakenridge et al. [1985] have proposed that valleys on the heavily cratered terrain formed on the flanks of localized hydrothermal systems driven by impact melt sheets. Scaling relations from studies of terrestrial impact craters and computer models [O'Keefe and Ahrens, 1977] predict that a 100 km diameter Martian crater would produce approximately 760 km 3 of impact melt. However, any resulting hydrothermal systems would be different than those modeled here in several respects. First, the impact melt forms a thin lens relatively close to the surface. Even given the same volume of impact melt as the modeled magmatic intrusions, the impact melt sheet will have a different cooling history and may cycle less water to the

surface. Gulick et al. [1988] estimated that the -700 km 3 impact melt associated with the 65 km diameter Manicouagan crater would cool in -5000 years in association with a hydrothermal system. This is far shorter than the cooling time for a buried magmatic intrusion with the same volume. Gulick et al. [1988] conclude that, in general, only Martian impact craters greater than about 100 km in diameter are likely to support a sufficiently long-lived hydrothermal system to form integrated valley networks. Second, water heated by a near- surface heat source may be more likely to vaporize and be lost to the system. For a deep-seated magmatic intrusion, upwardly moving water has time to cool before reaching the surface. Finally, it may be difficult to form a substantial volume of impact melt in a volatile-rich medium. The vapor cloud resulting from impacts into the volatile-rich subsurface may blow the resulting impact melt from the crater [Kieffer and Simonds, 1980]. Indeed solidified melt pools are found at neither Meteor Crater nor Lonar Lake Crater. Melosh [1989] suggests that the impact melts were thrown from these craters by steam explosions.

7. Implications for Martian Valley Network Formation

Given the volumes of groundwater outflow produced by the hydrothermal systems discussed above, we can estimate the total volume of water required to form individual valley systems on Mars. Previous estimates for the ratio of water to sediment volume based on extrapolations of the sediment- carrying capacity of large terrestrial rivers [Goldspiel and Squyres, 1991] are as low as 2 or 3 to 1. However, by extrapolating terrestrial erosion rates on the Hawaiian volcanoes to Mars, we find that this ratio is probably closer to 1000:1 [Gulick and Baker, 1993; Gulick 1993, 1998]. Valleys on Hawaii are morphologically more similar to Martian valleys than large terrestrial rivers, and the volcanic surfaces on which these valleys formed are likely similar to terrains on Mars.

vapor diffusion/ condensation

Figure 16. Cartoon illustrating two ways in which groundwater may reach the surface above an active hydrothermal system. The local water table is elevated, and if the water table intersects with the surface, groundwater outflow will occur at seeps or springs, as illustrated on the right. If the water table does not intersect with the surface, vapor rising from the elevated water table will diffuse upward into the unsaturated zone. Upon condensing, this water may recharge near surface aquifers and eventually reach the surface, as illustrated on the left of the diagram.

7.1. Erosion Volumes

Although the topographic information from Mars Global Surveyor's, laser altimeter (MOLA) will represent a dramatic improvement, determining valley volumes is difficult because good topographic data are not yet available for Mars. Gulick [1998] discusses this problem in more detail and estimates the range of likely valley volumes. To constrain the amounts of water required to form individual Martian valley systems, the eroded volumes of several Martian valley systems were measured. These include valleys on several volcanoes and two of the best developed valleys in the heavily cratered terrain, Warrego and Parana Valles (Figure 17a).

Eroded volumes for Warrego and Parana Valles are remarkably similar to those estimated for the volcanoes Alba Patera, Hecates Tholus, and Ceraunius Tholus. Hadriaca and

Tyrrhena Paterae have estimated eroded volumes 1 to 2 orders of magnitude larger, possibly owing to more easily erodible surface materials. It is not clear why valley volumes on the volcanoes are so similar to the two valley systems in the heavily cratered terrains. Younger valleys would have had less time to form than the older valleys and should therefore be smaller and less developed than older systems. In addition, Martian valleys that formed during the period of heavy


10 9 Alba Hadriaca Tyrrhena Warrego

Hecates Ceraunius Parana

1014

lO 16

1015•

•o•ø/ Alba Hadriaca

Hecates

5000 km 3

500 km 3

• 5O

Tyrrhena Warrego Ceraunius Parana

km 3

(a) (b)

Figure 17. (a) Total eroded volumes of valleys on six Martian volcanoes and for Warrego and Parana Vailes, two well-developed heavily cratered terrain (HCT) valleys. Volumes for both HCT valleys were comparable. The thickness of each bar indicates the range of uncertainty in eroded volumes assuming valley walls have between 10 ø and 30 ø slopes. (b) Estimated total water volumes required to form these valleys. For each locality, bar thickness is the range of uncertainty in valley volumes based on slopes of valley walls. The lower bar for each locality is the total calculated water volume assuming the 4:1 water to sediment volume from Goldspiel and Squyres [1991]. Note that these volumes represent an extreme lower limit to total water volumes because the volume of water required to erode source material was not considered. The upper bar for each locality is the total water volume based on a 1000:1 water to sediment volume, a ratio derived by Gulick [1993] from estimates of terrestrial fluvial erosion scaled to Mars' gravity. The horizontal lines show the total quantity of groundwater outflow by numerical models of Martian hydrothermal systems associated with magmatic intrusions of 50, 500, and 5000 km •. Note that 500 km • intrusions can provide sufficient groundwater outflow to form most Martian valleys (figures adapted from Gulick [1993]).

bombardment when the climate was thought to have been more Earth-like should have larger eroded volumes (and correspondingly higher degrees of tributary valley development) because erosion rates are assumed to have been much higher during this early period of the planet's geologic history. However, because eroded volumes are similar for both younger and older volcano valley systems as well as for the best developed valley systems located in the heavily cratered terrains, it appears that either climatic controls on valley formation were probably similar throughout Mars' geologic history, or that the control on volumes was not climatic.

7.2. Water Volumes

Although determining the actual volume of individual valley systems is difficult, estimating the volume of water required to erode the valleys is even more uncertain. The erodibility of a given surface and the amount of water required for erosion depends on a variety of factors including the rock types present, slope, degree of weathering, the cohesiveness of surface and subsurface material, in addition to how much,

how often, and how long liquid water is available for erosion. Because the climatic conditions in which these valleys formed is not known, it is difficult to determine the magnitude, duration, and frequency of water flows available for erosion. Nevertheless, using terrestrial analogs, I attempt to estimate the total volume of water required to erode individual valley systems.

Lithologic environments were selected where valleys have formed both on Earth and Mars such as volcanic surfaces and

where the ages of the terrestrial surfaces are known. Martian valley volumes are then combined with terrestrial fluvial erosion rates to constrain the total volumes of water required to

form each set of Martian valleys. Based on these studies of fluvial erosion on terrestrial volcanic landscapes, sediment volume to eroded volume ratios scaled to Mars' gravity are as large as 1000 to 1. The total water volume using each estimated ratio is shown for each valley group in Figure 17b. For each locality, each bar represents the range of possible volumes due to the uncertainty in the slopes of valley walls. The lower bar assumes a water to eroded volume ratio of 3:1,

the upper bar a ratio of 1000:1. As Figure 17b shows, the quantity of water passing through

the two selected valley systems in the heavily cratered terrain does not drastically differ from that on the Martian volcanoes, barring major differences in lithology. The horizontal lines show the total quantity of groundwater outflow by numerical models of Martian hydrothermal systems associated with magmatic intrusions of 50, 500, and 5000 km 3. The modeled 500 km 3 intrusion provides sufficient groundwater outflow to form most Martian valleys (figures from Gulick [1993, 1998]).

8. Discussion

This study demonstrates that groundwater outflow from Martian hydrothermal systems associated with intrusion volumes of at least several 102 km 3 can provide an alternative to rainfall-runoff processes in the formation of fluvial valleys on Mars [Gulick and Baker, 1989, 1990; Gulick, 1993]. If there is an adequate subsurface supply of water, the hydrothermal system need not recycle groundwater back into the system once it has been delivered to the surface. Sufficient quantities of water can flow in from surrounding aquifers beneath the vast undissected regions of the surface and incorporate itself into the hydrothermal system. In this sense,


Martian valley networks, if they are formed by hydrothermally driven groundwater outflow, may share a common genesis to the outflow channels with only the duration and rates of outflow being different. However, a Martian hydrothermal system is clearly capable of melting through existing permafrost, providing a pathway for water to both flow into and out of the subsurface. Thus, all groundwater that flows out to the surface need not be lost to the hydrothermal system.

If a hydrothermal system is to form a valley network on Mars, it must deliver a sufficient quantity of water over a sufficient period of time to form the valleys. Hydrothermal systems, such as those considered in this paper, can indeed deliver sufficient quantities of groundwater to the near surface environment over timescales that are empirically sufficient to form fluvial valleys on Earth. The most important model variables which relate to this conclusion are the subsurface

permeability and the total energy input. If the permeability is too low, then the large quantities of groundwater required to form fluvial valleys do not circulate through the system. Permeabilities associated with basalt (0.1 - 1000 darcys)or fractured megaregolith (approximately 3000 darcys) that likely comprise the near surface of Mars are sufficient. Second, a large energy source is required. Intrusions smaller than several 102 km 3 will not likely discharge sufficient quantities of water to the surface to allow integrated valley systems to form.

All of the models considered here are for single, isolated, magmatic intrusions. On Earth, multiple or composite magmatic intrusions are common and can greatly complicate thermal modeling. This is especially true in the formation of volcanoes where deep seated magma is repeatedly injected into the volcano's magma chamber over its lifetime. On Mars, multiple calderas (and therefore multiple intrusive events) are present on many volcanoes. Although accurate assessment of the timing between such events is a critical factor in modeling, multiple intrusions generally have an additive effect on the thermal history [Brikowski and Norton, 1989]. If an intrusion begins while the host rock is still hot from an earlier event, the reinvigorated hydrothermal system will be hotter and there will be more steam [Hayba and Ingebritsen, 1997].

Localized hydrothermally driven groundwater outflow is consistent with the nonuniform spatial distribution of the valleys, the sapping morphology of the valleys, and the lack of associated runoff valleys. Alternatively, Gulick et al. [1997] have suggested a hybrid concept where atmospheric and hydrothermal/geothermal processes might also produce localized erosion. In this mechanism, water vapor is transported from a frozen body of water to higher elevations where it snows out of the atmosphere. In most regions the snow would slowly sublimate and produce no erosion. In regions of vigorous geothermal activity, the base of accumulating snow would melt and produce runoff or infiltrate back into the groundwater system. Thus erosion would result only in those areas where the geothermal heat flux was high enough to melt through existing permafrost and basal layers of overlying snow. Both the hydrothermal discharge and the atmospheric transport/localized melting mechanisms are consistent with the geologic record. However, the hybrid atmospheric/geothermal concept requires that sizeable ice- covered bodies of water exist at the time during which fluvial valleys formed. Large bodies of water resulting from ponding of outflow channel discharges are presumed to have existed periodically during the middle to late periods of Mars geologic

history [Baker et al., 1991]. Better constraints in surface ages by crater density studies, more detailed geologic mapping and higher resolution images of the surface from upcoming missions may help to answer this question.

The geologic record indicates that fluvial valleys formed throughout Mars' geologic history [Gulick and Baker, 1990; Gulick, 1993, 1998] not just during the early period when Mars was thought to have a warm, wet climate either from greenhouse warming or from a global uniformly higher heat flow. Estimates of the fraction of valley networks formed in the Southern Highlands on geologic surfaces that are younger than Noachian range from 10% [Carr, 1995] to 30% [Scott and Dohm, 1992]. However, when both stratigraphic and morphologic age are considered, Scott and Dohm estimate that approximately 40% of the highland valleys may be younger than Noachian, with most valleys being early Hesperian. In addition, the best developed and most Earth-like fluvial valleys seemed to have formed locally during the middle to late periods of the planet's geologic history well after the postulated early warm, wet period could have persisted. Some SNC meteorites show evidence for aqueous and hydrothermal alteration, less than 170 Myr ago [Swindle et al. 1995; Treiman, 1996]. Thus at least limited fluid circulation was

ongoing very recently on Mars. Therefore mechanism(s) responsible for valley formation must have been present periodically throughout Mars' geologic history. Given the abundant evidence for groundwater on surfaces of all ages, both magmatic and impact generated hydrothermal systems were clearly common throughout Mars' geologic history.

Several orbital and landed missions will carry remote sensing instruments to Mars. Huntington [1996] has discussed some of the mineralogic signatures of hydrothermal systems that are detectable spectroscopically. On Earth, hydrothermal alteration zones are often identifiable by a "bulls-eye" mineral alteration zone pattern surrounding igneous intrusions. This alteration is produced by hydrothermal water circulating through the surrounding country rock. The water temperature, which declines away from the intrusion, produces mineral alteration in concentric zones centered on the intrusion. The state of the water and the

pre-existing mineralogy controls which minerals are produced and in what abundances. While the limited spatial resolution of instruments that are expected to fly in the near future will likely prevent identification of specific thermal spring sites, large, regional zonation patterns of mineral alteration will likely be identifiable. Discovery of such regions will help focus future exploration for groundwater and evidence for past life.

9. Hydrothermal Systems and Life

An interesting question presented by the formation of Martian hydrothermal systems is whether life could have evolved in such environments. There is a growing consensus that all forms of life that exist on Earth today arose from an ancestral sulfur-metabolizing thermophilic organism [Woese, 1987; Lake, 1988]. If this hypothesis is correct, then similar ancestral microorganisms may also have evolved in the sulfur- rich, hot spring environment of a Martian hydrothermal system [Boston et al., 1992]. Alternatively, if Mars had an Earth-like climate early in its geologic history [Pollack et al., 1987], then perhaps life, similar to that of the Archean and early Proterozoic periods on Earth, also evolved on the surface


of Mars. Such life may have eventually migrated to subsurface hydrothermal environments as atmospheric conditions became increasingly inhospitable [McKay et al., 1990]. Recent discoveries of nonphotosynthetic microbial ecosystems in deep-sea hydrothermal vents and in deep subsurface aquifer communities on Earth support the idea that such systems could also have existed in subsurface habitats on Mars [Woese, 1987]. Estimates of the amount of time needed for life on Earth to evolve have been as low as 2.5 Myr [Oberbeck and Fogelman, 1989].

Shock [1997] recently discussed additional remarkable conclusions of biology and geochemistry that demonstrate that life can exist in far more challenging environments than previously supposed. Indeed it is their environment rich in geochemical energy and existing far from thermochemical equilibrium that allows hydrothermal systems to support microorganisms without the need for photosynthesis [Shock, 1997]. As Shock concludes, the importance of this insight is that life could conceivably thrive in hydrothermal systems on a Mars-like planet without an obvious surface expression.

As demonstrated in this paper, hydrothermal systems associated with magmatic intrusions exceeding several 10 2 km 3 can last for several million years; larger ones such as those associated with the Tharsis or Elysium regions might have lasted on the order of a billion years. Therefore hydrothermal systems on Mars, particularly the larger ones, could have easily provided adequate timescales and possibly the necessary environments required for the formation of life. Additionally, regions of past hydrothermal activity would provide the best locations in the search for morphological evidence of microbial fossils on Mars. Hydrothermal systems provide highly concentrated regions suitable for life, and organisms that thrived in such systems would have been readily incorporated into the deposits formed by mineral-rich hydrothermal water and preserved as fossils [Walter and DesMarais, 1993]. Martian hydrothermal systems may thus have provided both the quantities of groundwater outflow needed for valley formation and the final oases for life.

Appendix

A1. Mathematical Background to Numerical Model

Transport of fluid and heat in the subsurface is modeled numerically using equations that describe time-dependent flow through saturated porous, permeable media. Conservation of fluid mass can be expressed at each point by

• = -V.(epS)+Qp (3)

where e is the porosity,/2 is the density of the fluid, v is the

fluid velocity, t is time, and Qp is the fluid mass source or sink (kg/m3s). Conservation of fluid momentum is expressed by a general form of Darcy's law:

• = - ß (Vp- p•,) (4)

Double underscores indicate tensors. The permeability tensor k is assumed isotropic. Fluid viscosity is /.t, p is the pressure of the fluid, and g is the gravitational acceleration, which for

Mars is 3.71 m/s 2. Conservation of energy, or the balance of changes in stored energy with various energy fluxes, is represented as

c•[ePew +(1-e)Pses ] =-V.(ePewS)+V.(,,t,I.VT ) at = (5)

The energy per unit mass of the solid and fluid are e s and e w ,• is the bulk thermal conductivity of the solid matrix plus fluid, I is the unit matrix or identity tensor (2x2, ones on diagonal:, zeroes elsewhere), T is the temperature of the fluid, D is the dispersion tensor (m2/s ß 2x2 matrix), T s is the temperature of the source fluid (0øC), and c w is the specific heat of water. The first term to the right of the equal sign in equation (5) is the convective energy transported by the average-uniform flowing fluid, the second term is the energy transported by heat conduction through the solid matrix and fluid, the third term expresses the dispersive energy transported by irregular flows and mixing not represented in the first term, and the fourth term expresses the energy gain or lost at fluid sources or sinks. Remaining intrinsic material parameters are presented in Table I.

A2. Model Setup

The finite-element, density-dependent groundwater flow simulation model SUTRA (Saturated-Unsaturated Transport) [Voss, 1984] employs equations (3) to (5), to model groundwater flow and energy transport. SUTRA's simplified treatment of the water properties has been replaced with an accurate equation of state for H20 [Johnson, 1987]. Stability of the numerical solution is preserved by restricting the size of the time step At to less than the size of the cell Az divided by the maximum velocity:

At _< ZXZ /I v max I (6) The hydrothermal system is modeled using a radially

symmetric, cylindrical grid with a central cavity for the intrusion (Figure 4). Single, cylindrical intrusions of height 4 km emplaced 2 km beneath the Martian surface are considered. The modeled region extends from the outer radius of the intrusion r i to a distance R = 20r i. The 6-km-high grid consists of 440 elements; each element is 400 m high. Element widths vary from several meters near the intrusion to approximately 1 km along the outer edge; widths are scaled to the radius of the intrusion. The thickness of each element is

2n:r, where r is the inner radius of the element. Several model cases are considered. In the baseline model, a

homogeneous, water-saturated medium having a porosity, e, of 0.25 and a permeability, k, of 10 darcys surrounds the intrusion. Two other homogeneous cases are modeled where subsurface permeabilities are set at 0.1 and 1000 darcys, respectively. In the heterogeneous model, confining and interbedded layers are assigned a subsurface permeability of 10 -7 that of the background. Inhomogeneities in the flow field at scales smaller than the grid are accounted for by the dispersivity (assumed isotropic), a measure of the largest obstacle to the flow. The models assume a dispersivity of 100 m.

A3. Boundary Conditions

Water (at 0øC) is allowed to flow into or out of the system along the outer vertical (right) boundary and along the top


Table 1. Summary •f..intrinsic Material Properties Used in Models Material Property Symbol Value Unit

All solids specific heat capacity c s 8.4x 102 thermal conductivity ;k s 3.5 density Ps 2650 porosity œ 0.25

Magma intrusion temperature Tin t 1250 density Pm 3000 thermal diffusivity K m I x 10 -6 specific heat capacity c m 8.4x 102

Country rock permeability k b 10- 1000 Layered units permeability k a 10-7kb

joules/(kg øC) joules/(s _m øC)

kg/m 3

øC

kg/jn 3 m2/s

joules/(kg øC) darcy darcy

boundary. The boundaries along the base of the grid and along the inner vertical (left) edge are impermeable to fluid flow. Groundwater is held at hydrostatic pressure along the outer vertical boundary at r = R. Surface pressure is set at 1 bar along the top of the modeled region in order to maintain a numerically stable system by keeping hydrothermal fluids circulating near this boundary in the liquid state. However, in the actual Martian environment, the near-surface behavior of

the discharged water will depend on the surface temperature and pressure conditions.

Thermal energy enters the model by conduction through the portion of the left boundary that is directly next to the intrusion (i.e., at depths below 2 km). All other boundaries are thermally insulated, although energy can be transported out of the modeled region along the top and right boundary by discharging hydrothermal fluids. The conductive heat flux at the inner boundary at depths below 2 km is modeled as a time varying energy source applied at the appropriate nodes.

The magnitude and time variation of this flux are calculated from equations that describe the temperature distribution within a conductively cooling cylindrical magma intrusion. At 40 separate dimensionless times r, where r = Kt/a 2, K is the thermal diffusivity and a is the cylinder radius, I numerically solved for the energy content of the cylinder and its cooling rate [Jaeger, 1968]. An expression was then fit to the calculated dimensionless heat flow as a function of time:

A(•) =

40.0-1768•' r < 0.01

3.7( •' ) -ø'6 0.01<r<0.2772 k,O.20J

4.5 0.2772 < •:

(7)

The actual energy flux is given by KHT o rCpA(v), where H is the height of the intrusion and T O is the initial temperature. Radius a only enters this expression through •', since the heat flow is proportional to the cylinder volume divided by the surface area of its sides. No heat flows through the top or bottom of the intrusion in this model. Equation (7) is accurate to within 5% for all r < 1.3, is most accurate for r < 0.8, and the integral of the expression over time is equal to the original heat content of the intrusion. The resulting expression is used as the time-varying boundary condition in SUTRA.

A single, cylindrical intrusion of magma with a volume of 50 km 3 is used in the baseline case; however, an intrusion volume of 500 km 3 is also modeled. An initial magma temperature of 1250øC, typical for basaltic magmas, is assumed. The intrusion is emplaced at time t = 0 along the

grid's inner or left side at a depth of 2 km. The intrusion is assumed to remain impermeable in order to prevent the magma from coming into direct contact with groundwater. A background geothermal gradient is neglected, and an initial, uniform groundwater temperature of 0øC is assumed within the modeled region. All the thermal energy of the intrusion is assumed to enter the numerical model; no energy is lost through its roof. In addition, the method used for solving the cooling of the cylindrical intrusion is not strictly self- consistent with the numerical model. The heat loss calculation

assumes that energy is conductively transported away from the boundary into the country rock. However, in the numerical model, energy is transported away from the boundary by both convective and conductive processes. While this affects neither the total amount of thermal energy entering the system nor the total volume of groundwater outflow, it does affect the rate at which energy is delivered to the actual hydrothermal system. Energy would be transported away from the boundary at a faster rate, thus steepening the magma temperature gradient and resulting in a more intense, though somewhat shorter (possibly as much as a factor of 2 to 3) [Norton, 1984] hydrothermal system lifetime. However, this would have little effect on the hydrothermal system in terms of its duration over timescales required for valley formation.

Acknowledgments. This research was supported by a NASA Graduate Student Research Fellowship and later by a National Research Council Postdoctoral Fellowship. I thank Vic Baker for insightful reviews and discussions. This paper benefited from helpful reviews by Horton Newsom, Bob Brakenridge, and Mark Marley.

References

Baker, V.R., The Channels of Mars , Univ. of Texas Press, Austin, 1982.

Baker, V.R., and J. Partridge. Small Martian valleys: Pristine and degraded morphology. J. Geophys. Res.: 91, 3561- 3572, 1986.

Baker, V.R., R.C. Kochel, J.E. Laity, A.D. Howard, Spring sapping and valley network development, in Groundwater Geomorphology, edited by C.G. Higgins and D.R. Coates (Spec. Pap. Geol. Soc. Am. 252), 235-265, 1990.

Baker, V.R., R.G. Strom, V.C. Gulick, J.S. Kargel, G. Komatsu, and V.S. Kale, Ancient oceans, ice sheets and the hydrological cycle on Mars, Nature, 352, 589-594, 1991.

Baker, V.R., M.H. Carr, V.C. Gulick, C.R. Williams, and M.S. Marley, Channels and valley networks, in Mars, edited by H.H. Kieffer, B.M. Jakosky, C.W. Snyder, M.S. Matthews, pp. 493-522, Univ. of Ariz. Press, Tucson, 1992.


Boston, P.J., M.V. Ivanov, and C.P. McKay, On the possibility of chemosynthetic ecosystems in the subsurface habitats on Mars, Icarus, 95, 300-308, 1992.

Brakenridge, G.R., The origin of fluvial valleys and early geologic history, Aeolis quadrangle, Mars, J. Geophys. Res., 95, 17,289-17,308, 1990.

Brakenridge, G.R., H.E. Newsom, and V.R. Baker, Ancient hot springs on Mars: Origins and paleoenvironmental significance of small Martian valleys, Geology, 13, 859- 862, 1985.

Brikowski, T. and D. Norton, Influence of magma chamber geometry on hydrothermal activity at mid-ocean ridges, Earth Planet, Sci. Lett., 93, 241-255, 1989.

Carr, M.H., Formation of Martian flood features by release of water from confined aquifers, J. Geophys. Res., 84, 2995- 3007, 1979.

Carr, M.H., Stability of streams and lakes on Mars, Icarus, 56, 476-495, 1983.

Carr, M.H., The Martian drainage system and the originofnetworks and fretted channels, J. Geophys. Res., 100, 7479-7507, 1995

Carr, M.H. and G.D. Clow, Martian channels and valleys: Their characteristics, distribution, and age, Icarus 48, 91- 117, 1981.

Cathies, L.M., Economic Geology, 75th Annv. Vol., 424- 457, 1988.

Clifford, S.W., The role of thermal vapor diffusion in the subsurface hydrologic evolution of Mars, Geophys. Res. Lett., 18, 2055-2058, 1991.

Davies, G.F. and R.E. Arvidson, Martian thermal history, core segregation, and tectonics, Icarus, 45, 339-346, 1981.

Davis, S.N. and R.J.M. DeWiest, Hydrogeology, John Wiley, New York, 1966.

Elder, J., Geothermal Systems, Academic, San Diego, Calif., 1981.

Fanale, F.P., Martian volatiles: Their degassing history and geochemical fate, Icarus, 28, 179-202, 1976.

Fanale F.P., and S.E. Postawko, J.B. Pollack, M.H. Carr, and

R.O. Pepin, Mars- Epochal climate change and volatile history, in Mars, edited by H.H. Kieffer, B.M. Jakosky, C.W. Snyder, M.S. Matthews, pp. 1135-1179, Univ. of Ariz. Press, Tucson, 1992

Farmer, J., and D. DesMarais, Exopaleontology and the search for a fossil record on Mars (abstract), Lunar Planet. Sci., XXV, 367-368, 1994.

Furlong, K.P., R.B. Hanson, and J.R. Bowers, Modeling thermal regimes, in Contact Metamorphism, edited by D.Kerrick, pp. 437-506, Mineralogical Soc. Am., 1991.

Gasparini, P., and M.S.M. Mantovani, Heat transfer in geothermal areas, in Geophysics of Geothermal Areas: State of the Art and Futu')e Development, edited by A. Rapolla and G.V. Keller, pp. 9-40, Colo. Sch. of Mines Press, Golden, 1984.

Goldspiel, J.M. and S. W. Squyres, Ancient aqueous sedimentation on Mars, Icarus, 89, 392-410, 1991.

Gulick, V.C., Magmatic intrusions and hydrothermal systems: Implications for the formation of Martian fluvial valleys, Ph.D. thesis, Univ. of Ariz., Tucson, 1993.

Gulick, V.C., Origin of the valley networks on Mars: A hydrological perspective, Geomorphology, in press, 1998.

Gulick, V.C. and V.R. Baker, Fluvial valleys and Martian paleoclimates, Nature, 341, 514-516, 1989.

Gulick, V.C. and V.R. Baker, Origin and evolution of valleys

on Martian volcanoes, J. Geophys. Res., 95, 14,325- 14,344, 1990. .

Gulick, V.C. and V.R. Baker, Martian hydrothermal systems: some physical considerations (abstract), Lunar Planet. Sci. Conf, XXIII, 463-464, 1992.

Gulick, V.C. and V.R. Baker, Fluvial erosion on Mars:

Implications for paleoclimatic change (abstract), Lunar Planet. Sci. Conf XXIV, 587-588, 1993.

Gulick, V.C., M.S. Marley, and V.R. Baker, Hydrothermally supplied groundwater: A mechanism for the formation of small Martian valleys. Lunar Planet. Sci. Conf., XIX, 441- 442, 1988.

Gulick, V.C., M.S. Marley, and V.R. Baker, Numerical modeling of hydrothermal systems on Martian volcanoes: Preliminary results. Lunar Planet. Sci. Conf, XXII, 509- 510, 1991.

Gulick, V.C., D. Tyler, C.P. McKay, and R.M. Haberle, Episodic ocean-induced CO2 pulses on Mars: Implications for fluvial valley formation, Icarus, 130, 68-86, 1997.

Hayba, D.O. and S.E. Ingebritsen, Multiphase groundwater- flow near cooling plutons, J. Geophys. Res., 102, 12,235- 12,252, 1997.

Huntington, J.F., The role of remote sensing in finding hydrothermal mineral deposits on Earth, in Evolution of Hydrothermal Ecosystems on Earth (and Mars?), 214-235, John Wiley, New York, 1996.

Jaeger, J.C., Cooling and solidification of igneous rocks, in Basalts vol 2., edited by H.H. Hess and A. Poldervaart, 503- 536, John Wiley, New York, 1968.

Jakosky, B. and J. Jones, Evolution of water on Mars, Nature 370, 328-329, 1994.

Johnson, J.W., Critical phenomena in hydrothermal systems: State, thermodynamic, and transport properties of H20 in the critical region, Ph.D. thesis, Univ. of Ariz., Tucson, 1987.

Kasting, J.F., CO2 condensation and the climate of early Mars, Icarus, 94, 1-13, 1991.

Kieffer, S.W. and C.H. Simonds, The role of volatiles and

lithology in the impact cratering process, Rev. Geophys. and Space Phys. 18, 143-181, 1980.

Lake, J.A., Origin of the eukaryotic nucleus determined by rate-invarient analysis of rRNA sequences, Nature, 331, 184-186, 1988.

MacKinnon, D.J. and K.L. Tanaka, The impacted Martian crust: Structure, hydrology, and some geologic implications, J. Geophys. Res. 94, 17,359-17,370, 1989.

Mars Channel Working Group, Channels and valleys on Mars, Geol. Soc. Am. Bull., 94, 1065-1054, 1983.

Masursky, H., An overview of geological results from Mariner 9, J. Geophys. Res. 78, 4009-4030, 1973.

McKay, C.P., G.D. Clow, R.A. Wharton, and S.W. Squyres, Thickness of ice on perennially frozen lakes on Mars, Nature, 313, 561-562, 1985

McKay, C.P., E.I. Friedman, R.A. Wharton, and W.L. Davis, History of water on Mars: A biological perspective, Adv. Space Res., 12, 231-238, 1990.

Melosh, H.J. Impact Cratering. A geologic process, Oxford Univ. Press, New York, 1989

Mouginis-Mark, P.J., L. Wilson, and J.W. Head III, Explosive volcanism on Hecates Tholus, Mars: Investigation of eruption conditions, J. Geophys. Res., 87, 9890-9904, 1982.

Mouginis-Mark, P.J., L. Wilson, and M.T. Zuber, The


physical volcanology of Mars, in Mars, edited by H.H. Kieffer, B.M. Jakosky, C.W. Snyder, M.S. Matthews, pp. 424-452, Univ. of Ariz. Press, Tucson, 1992.

Newsom, H.E., Hydrothermal alteration of impact melt sheets with implications for Mars, Icarus, 44, 207 -216, 1980.

Norton, D., The theory of hydrothermal systems, Annu. Rev. Earth Planet. Sci., 12, 155-177, 1984.

Oberbeck, V.R., and G. Fogelman, Estimates of the maximum time required for origination of life, Origins of Life and Evolution of the Biosphere, 19, 549-560, 1989.

O'Keefe, J.D. and T.J. Ahrens, Meteorite impact ejecta:Dependence of mass and energy lost onplanetary escape velocity, Science, 198, 1249-1258, 1977.

Pieri, D., Martian channels: Distribution of small channels in the Martian surface, Icarus, 27, 25-50, 1976.

Pollack, J.B., J.F. Kasting, S.M. Richardson, and K. Poltakoff, The case for a wet, warm climate on early Mars, Icarus, 71, 203-223, 1987.

Reimers, C.E. and P.D. Komar, Evidence for explosive volcanic density currents on certain Martian volcanos, Icarus, 39, 88-110, 1979.

Rossbacher, L.A., and S. Judson, Ground ice on Mars:

Inventory, distribution, and resulting landforms, Icarus , 45, 39-59, 1981.

Sammel, E.A, S.E. Ingebritsen, and R.H. Mariner, The hydrothermal system at Newberry Volcano, Oregon, J. Geophys. Res. 93, 10,149-10,162, 1988.

Schultz, P., R. Schultz, and J.Rogers, The structure and evolution of ancient impact basins on Mars. J. Geophys. Res. 87, 9803-9820, 1982.

Shock, E., High-temperature life without photosynthesis as a model for Mars, J. Geophys. Res., 102, 23,687, 1997.

Squyres, S.W., and J.F. Kasting, Early Mars- how warm and how wet?, Science, 265, 744-747, 1994.

Squyres, S.W., D.E. Wilhelms, and A.C. Moosman, Large- scale volcano-ground ice interactions on Mars, Icarus, 70, 385-408, 1987.

Squyres, S.W., S.M. Clifford, R.O. Kuzmin, J.R. Zimbleman, and F.M. Costard, Ice in the Martian regolith, in Mars, edited by H.H. Kieffer, B.M. Jakosky, C.W. Snyder, and

M.S. Matthews, pp. 523-556, Univ. of Ariz. Press, Tucson, 1992.

Swindle, T.D., M.K. Burkland, J.A. Grier, D.L. Lindstrom, and A.H. Treiman, Noble gas analysis and INAA of aqueous alteration products from the Lafayette meteorite: Liquid water on Mars < 350 Ma ago (abstract), Lunar Planet. Sci. Conf. XXVI, 1385-1386, 1995.

ToksCz, M.N. and A.T. Hsui, Thermal history and evolution of Mars, Icarus, 34, 537-547, 1978.

Treiman, A., To see a world in 80 kilograms of rock, Science, 272, 1447-1448, 1996.

Voss, C.I., SUTRA, U.S. Geol. Surv. Water Resour. Inv. Rep. 84-4369, 1984.

Wallace, D. and C. Sagan, Evaporation of ice in planetary atmospheres: Ice-covered rivers on Mars, Icarus, 39, 385- 400, 1979.

Walter, M.R. and D.J. DesMarais, Preservation of biological information in thermal spring deposits: Developing a strategy for the search for fossil life on Mars, Icarus, 101, 129-143, 1993.

Wentworth, S.J. and J.L. Gooding, Carbonates and sulfates in the Chassigny meteorite: Further evidence for aqueous chemistry on the SNC parent planet, Meteoritics 29, 860- 863, 1994.

Wilhelms, D.E. and R.J. Baldwin, The role of igneous sills in

sha[•ing the Martian uplands, Proc. Lunar and Planet. Conf. 19 ttt, 355-365, 1989.

Wilson, L. and P.J. Mouginis-Mark, Volcanic input to the atmosphere from Alba Patera on Mars, Nature, 330, 354- 357, 1987.

Woese, C.R., Bacterial evolution, Microbiol. Rev., 51, 221- 271, 1987.

V.C. Gulick, Space Science Division, MS 245-3, NASA-Ames Research Center, Moffett Field, CA 94035. (e-mail: v gulick @ mail.arc.nasa.go v)

(Received October 10, 1996; revised April 10• 1998; accepted April 21, 1998.)

Magmatic intrusions and a hydrothermal origin for fluvial valleys on Mars

Documents