JOURNAL OF GEOPHYSICAL RESEARCH. VOL. 103, NO. El0, PAGES 22,689-22,694, SEPTEMBER 25, 1998 Atmospheric loss since the onset of the Martian geologic record' Combined role of impact erosion and sputtering David A. Brain and Bruce M. Jakosky Laboratory for Atmosphericand SpacePhysics,Universityof Colorado,Boulder Abstract. Chemicaland geomorphicevidencesuggests that Mars' atmospherehas undergone significantlossor modificationsincethe onsetof the Martian geologic record.Analysesof proposedlossprocesses havebeenunableto individuallyaccount for lossof enoughatmosphereto havesupportedthe presence of liquid surfacewater at the formationtime of the oldestobservedsurfaceunits. Here, we give a simple calculationof the combinedeffectsof major atmospheric lossprocesses. Considering only impact erosionand using resultsfrom an analytical model by Melosh and Vickery[1989],we obtain an expression for surfacepressureas a functionof local crater density. Based on tabulated crater densities,the processof atmospheric crateringcan accountfor a lossof 50-90% of the Martian atmospheresincethe onsetof the geologicrecord. Stable isotopefractionationmeasurements suggest that lossof •090% of atmosphericspeciesto spacehas occurredvia solar wind pick-up-ionsputtering [Jakoskyand Jones,1997]. Combined,•095-99%of Mars' atmospherecouldhavebeen lost to space.Adsorptionof CO2 in the regolithand sequestration in the polesasiceor clathratecouldaccountfor mostof the remaining loss.Thesecalculations suggest that the lossprocesses of impacterosion,sputtering, and sequestrationare together capableof explainingthe inferred transition from an early atmosphereon Mars capable of supporting stable liquid surfacewater to today's climate. 1. Introduction Three main processesare thought to have removed significant amounts of Martian atmosphereover time: Geologicfeatureson the Martian surfaceindicate that ejection of atmospheric gas by impacts, loss to space the climate has changedthrough time. Martian valdue to pick-up-ion sputtering, and sequestration into ley networks, for example, are located predominantly the regolith or polar caps. We examine the potential in old crater-terrain regions,formed during the heavy bombardment [Baker et al., 1992]. Thesevalley net- for a combination of these processesto have removed works are seen on surfacesdating from the Noachian sufficientatmosphereto explain the transition in the climate. First, we considerthe effectsof ejection of gas to epochin Martian geologichistory [Tanaka,1986]. A widely accepted interpretation of the existenceof val- spaceby large impacts on the atmosphericevolution of that there ley networksis that liquid water wasoncestable(or, at Mars. Meloshand Vickery[1989]concluded could have been nearly I bar of primordial atmosphere least,morestable)on the surface [Bakeret al., 1992]. Another argument for climate changecomesfrom the 4.5 Gyr ago on Mars. The geologicevidence for climate changecomesonly from existing geologicfeatures on Mars, however, none of which are likely to predate about 4 Gyr ago. Here, we examinethe role that impact erosion played sincethe time of the onsetof the Martian [CraddockandMaxwell,1993;Carr, 1996;Fanaleet al., geologicrecord about 4 Gyr ago. We do this by using 1992]. The highererosionratesand the evidencefor liquid water are thought to require a warmer climate and the observed cratering record to constrain directly the a thickeratmosphere than is presenttoday [e.g.,Fanale role impact erosion played in depleting the atmosphere et al., 1992]. Severalatmosphericlossprocesses have to its present amount. Next, we combine the loss from been proposedto explain the necessaryclimate change, impact and sputtering processesover Martian history. but none has been able to independently account for Third, we consider loss in the form of sequestrationof CO2 into subsurfaceor polar cap reservoirs,and factor sufficient loss or alteration. the estimated sequesteredinventory into our estimates of total lossand original atmosphericpressure.Finally, Copyright 1998 by the American GeophysicalUnion. we discussimplications of these calculationsfor understanding the possible evolution of the Martian atmoPaper number 98JE02074. 0148-0227/98/98JE-02074509.00 sphere and climate. paucity of small impact craters on older surfaces,as well as evidencefor erosionof large craters at rates substantially greater than can occur under the current climate 22,689 22,690 BRAIN AND JAKOSKY: CLIMATE CHANGE ON MARS 2. Impact Loss Model where M and P denote atmospheric mass and pressure, respectively, and M0 is the present atmospheric mass and P0 is the present atmosphericsurfacepressure;t is Previous work on the effect of impact erosion has determined that impacts are capable of ejecting some age as in equation(1); b is the slopeof the cumulative fraction of a planer's atmosphereto space,dependent impactor distribution which best predicts the current upon propertiesof both the impactor and the planer's lunar crater distribution and has a value of 0.47; t, is atmosphere [Meloshand Vickery,1989;Walker,1987]. the time required to reducethe atmosphericpressureto Relatively small impactors can remove a mass of at- zero, calculatedto be 63.1 Gyr by assuminga constant mosphereapproximatelyequivalentto the massinter- cratering rate that coincideswith the best fit lunar cra•ceptedby theimpactor asit travels through theat- tering rate; B is 2300 and was estimated by fitting to mosphere[Walker,1987]. Meloshand Vickery[1989] lunar crater densities.The left-handside(s)of (2) give showedthat sufficientlylarge impactorsare capableof removinga more substantialatmosphericmassas the expandingvapor cloud producedby impact dragsambient atmosphereaway from the planet at speedsexceedingthe escapevelocity.The amountof atmosphere ejectedduring somelarge impactscould be equal to the massof atmosphereabovethe plane tangentto the pointof impact [Meloshand Vickery,1989].For the current Martian atmosphere,silicate impactors larger than •3 km and travelingfaster than •14 km/s will removea "tangent-planemass"of atmosphereby this process[Meloshand Vickery,1989]. Meloshand Vickery[1989]examinedthe role of im- algebraic "enhancement"factors for massand pressure as a function of time, indicating the mass of the Martian atmosphereat a given time, relative to the current Martian atmosphere. Meloshand Vickery[1989]usedequation(2) to note that Mars' atmosphere4.5 Gyr ago could have been -• 60-100 times as thick as the current atmosphere,in the vicinity of i bar and that this value might havebeen large enoughto have allowedstable liquid water on the Martian surface. They also recognizedthat the importance of impact erosiondecreasedrapidly with time, so that it was not a significantprocessby the end of the late heavy bombardment3.5 Gyr ago. pact erosionby large projectileson the evolutionof the If we are to understand the transition that occurred Martian atmosphere.Using a modifiedpowerlaw rela- in the Martian climate, however, we are interested in tion for the time-dependentimpactor flux, they derived the atmospheric loss that occurred subsequentto the an analytic expressionfor the densityof Martian craters onset of the geologicrecord, or the earliest time from >4 km as a function of time. The number of craters which we have geologicevidence of a thicker climate. >4 km, denotedN(>4 km), is usedasa statisticallyre- Thus we want to look at impact erosiononly from the liable measureof the number of impactorslarge enough time of the oldest surface features, rather than from 4.5 to erode the entire atmosphereabovethe tangent plane; Gyr ago. To do this usingthe analyticexpressions (1) the function doesnot imply that all cratersgreaterthan and (2) in their presentform wouldrequireknowledge 4 km resultedfrom impacts capableof causingbulk ero- of the agesof the oldest existing Martian terrain, which sion, nor that all 4 km craters will be retained through are not well constrained. However, the existing model time. The lunar crateringrecordand absoluteageswere providestwo expressions,each containing time as a free usedto constrainthis function(Figure la): parameter. We are able to remove time from the two equations, leaving one expressionfor the surface pres- N(> 4km)- 2.68110-a[t+4.57x10-?(eXt-1)](1) sure enhancement factor as a function of local crater Here, t is the age of the crateredsurface,in Gyr, and density. The resultant expressionis implicit in P, but )• -- 4.53Gyr-1. Numerical valuesin the expressioneasily solvednumerically as shown in Figure lc: comefrom applicationof the revisedSchmidt-Holsapple scalinglaw [Schmidtand Housen,1987]and assump- P0= (1- •,(At, [1-e_•]) P(t) 1 Be -z'6• l/b(3) tions about the Martian cratering rate relative to that of the Moon. The expressionshowsthat crater density on a planetary surfaceevolveslinearly in time, with a superimposedexponential dependencethat decaysas where C1 = 4.57x 10-7 , Co.= 2.68 x 10-5 and all the planet agesand availableimpactorsare usedup. •--W[•C1 ½A[0,+½%]]_/•[C1 -•-•2 ] (4) Fromequation(1), Meloshand Vickery[1989]deter- other variablesare as statedfor equations(1) and (2). mined the flux of impactorslarge enoughfor the vapor cloudto removeatmosphere,and integratedit to obtain an expressionfor the atmosphericmassas a functionof time. Their expressionfor atmosphericmassrelative to today's atmosphericmass is equivalentto the atmosphericsurfacepressurerelative to today's surface pressure(Figure lb): M0 = P(t) P0= (l+t•At, [1-eAt ] (2) Matra(t) t Be -z'6A W denotesthe Lambda function and is given by = Equations(3)-(5) are particularlyrelevantsincewe now need only know the local crater density of a surface region to determine the atmosphericsurfacepressureat the time that region was forming; the precisesurface age is no longer required. In essence,we can directly determine the effectson the atmosphereof the impact craters observed on the oldest surfaces. BRAIN AND JAKOSKY: CLIMATE CHANGE ON MARS 22,691 Crater Density vs Time 10-2 10-3 10-4 10-5 10-6 -4 -3 -2 -1 0 Time From Present (Gyr) Atmospheric Pressure vs Time lOO "' '" N NNH i i ' ' ' I • ' ' ' U H ' ' ' ' ' I' ' ' ' ' ' ' L A ' ' ' • ' ' '' ' ' ' M A ' ' ' I i U A lO , i , I.i ' i . • i -4 -3 -2 , , , , -1 Time From Present (Gyr) Crater Density vs Surface Pressure lOO ' ' ' I ' ' ' I ' ' ' I ' ' ' I ' ' ' I ' ' lO i i •).002 . 0000 ' ' i b, 004 ' '' i 6.006 ' ' i 6,008 ' ' i 6. 0I10i i 6 , 012 MeasuredCumulative CraterDensityN(>4km)(kin-2) Figure 1. (a) Plot of cumulativecrater densityN(>4 km) overthe historyof the Martian surface,as derivedby Meloshand Vickery[1989].The expression (givenin equation(1)) is well matchedto lunar crater densities.(b) Plot of the Martian atmosphericpressureenhancement factor,P(t)/Po, overthe past 4.53 Gyp,as derivedby Meloshand Vickery[1989]and givenin equation(2). The geologicepochsare noted,assuming a-2 powerlaw size-frequency distribution. The boundaries between each age are approximate and are estimated from the work of Tanaka [1986]and appliedto the modelto yield P(t)/Po. (c) Plot of surfacepressureenhancement factor as a functionof surfacecrater densityN(>4 km). The implicit expression is derivedby eliminatingtime as a parameterfrom equations(1) and (2). 22,692 BRAIN AND JAKOSKY: CLIMATE CHANGE ON MARS 3. Results 3.1 Ancient Table1. Estimates ofN (>4km)(inunits ofkm '2) Surface Crater Reference Densities Craddock andMaxwell[1993 ] Scott andTanaka [1981] Hartmann etal. [1981 ] The first step in applying the simplified impact erosion model from above to Mars is to obtain Lower Bound a reliable estimate of the crater density on the oldest surfaces. Three sources are used to determine a range in crater 6.84x 10'4 8.92x 104 7 x 10'4 UpperBound 3 x 10'3 1.43x104 3 x 10'3 densities. From each sourcewe obtain N(>4 kin), or lowing for reasonableuncertainty in this correction,we the cumulative number of craters with diameter greater estimate N (> 4 km) - 7 x l0-4 - 3 x 10-3 km-2. It than 4 km at the time of formation of the oldest geologic should be noted that the locations of the surface units units on Mars. In some casesthis requires a conversion between incremental and cumulative crater densities. used to obtain crater densities from this reference were unspecified. We also correct for preferentialerosionof small craters A more recent estimate of crater densityon the oldoverthe historyof the Martian surfaceby usinglargedi- est Martian surfaces comes from Craddock and Maxwell ameter crater density estimates and assuminga power law distribution in crater density; the assumedpower [1993].Crater densitiesare givenfor crateredand dissectedplateau sequences,distinguishingbetween fresh law distribution is either a canonical-2 power law disand modifiedcraters and examiningcorrelationbetween tribution or is a distribution in agreementwith figures degradation and elevation of the cratered terrain. The included in the sources. total area investigated is extensive and yields crater Hartmannet al. [1981]obtainedcratersize-frequency distributions for a number of surface units on Mars. Figure2 showsan incrementalcratersize-frequency dis- densities N(>4 kin) = 6.84x 10-4- 3 x 10-3 km-2. This range is obtained using two methods and taking the highest and lowestresulting crater densities.First, tribution for the most heavily crateredregionsstudied. N(>4 kin) is scaledfromvaluesfor N(>16 kin) givenin Convertingto a cumulativedistribution,correctingfor Craddockand Maxwell's Table 1, assuminga-2 power the effectsof removalof small cratersby erosionand allaw distribution; scalingfrom N(>16 kin) shouldnot require a correction for small crater erosion. Second, N(>4 kin) is estimated using Craddockand Maxwell's Figure 1 by correctingfor erosionof craterssmallerthan 20-30 km in diameter. Additional o data on ancient Martian crater densities comefrom Scottand Tanaka[1981]. Basedon counts of Early Noachian craters taken in the Tharsis region -1 of Mars, the oldest units there exhibit crater densities N(>4 kin)= 1.26x 10-3-1.43 x 10-3 km-2. Thisrange %,:, ',• % is obtained by scalingN(>4 kin) from valuesfor "old terra regions"givenin Scottand Tanaka's[1981]Table I for craters of varying size. A -2 powerlaw distribution was assumed,and the highest and lowestresultingdensities were kept. It should be noted that the Tharsis region of Mars in general is relatively young, and the • expanses of ancient cratered terrain are not extensive -5 in area. The crater determined 6 densities on the oldest Martian terrain from the three sources are shown in Table 1. For the remainder of this work we take the low and high values from Table 1 as the range in crater density on the oldestMartian terrain. Thus we usea craterdensity 0.25 I I 0.5 I estimate of N(>4 km) - 6.84x 10-q - 3 x 10-3 km-2. Diameter (km) Figure 2. Incremental size-frequencycrater density 3.2 Atmospheric Loss From Large Impacts The inferred crater densities are used in the simpli- distributionof Hartmann et al. [1981]for the region fiedmodelexpression (equations(3) and(4), andFigure labeled "heavilycrateredplains". To obtain N(>4 kin) lc) to yield P/Po, the enhancementfactor in surface for use in the impact erosionmodel, we correct for preferential removal of small craters by assumingthat the pressure between the current atmosphere and the atpower law distribution associated with larger craters mosphere at the formation time of the oldest surfaces. (N(>32 km)) appliesfor cratersof every size. We es- From Figure lc, the range in crater density corresponds timate from the correcteddistributionthat N(>4 kin) to an enhancement factor in surfacepressureof 2.1-8.7. =7 x 10-4 - 3 x 10-3 km -2 This enhancementfactor suggeststhat the total amount BRAIN AND JAKOSKY: CLIMATE CHANGE ON MARS 22,693 of atmosphereon Mars that was presentat the onset of the geologicrecordhas beensubsequentlyreducedby a field [Hutchinset al., 1997].In addition,the absolute factor of 2.1-8.7 from large impact erosion. The range in enhancementfactor implies that Mars has lost about 50-90% of its early atmosphereby large impacts. This loss is significantly different from the 98-99.5% loss over 4.5 Gyr suggestedby Melosh and mates of the amount amount of gas lost to space is uncertain. Current estilost differ from earlier estimates in the way the differentisotopicconstraintsare applied [Hutchinsand Jakosky,1996]. We assumehere that the isotopicanalysisprovidesa strong indication only of the fraction of volatiles lost to space and that the Vickery[1989].The craterdensitywouldneedto be > absoluteamount of loss,which is extremely uncertain, 0.01km-2 (almostanorderofmagnitude greater than is sufficientto accountfor muchof the lossrequiredby the highestcrater densityvalueslisted above)in order the calculation below; the uncertaintiesin the absolute for the surface pressure to have been in the vicinity sputtering loss rate allow these assumptions. We take of 1 bar at the time of onset of the geologicrecord. 90% as the nominalvaluefor bulk atmospheric lossby Clearly, a much smaller fraction of Martian atmosphere sputtering. hasbeenremovedby largeimpactssincethe onsetof the Both sputtering and impact erosionact on the atmogeologicrecordthan overthe past4.5 Gyr. This is not a spherein a relative way; the amount removedby either surprisingresult but suggeststhat impact erosionalone processis always dependentupon the total atmospheric is not capable of accounting for the required amount abundance. Bulk atmosphericremoval by impact does of atmospheric loss since the formation of the oldest not preferentially remove light isotopes,while sputterMartian surface units. ing does not affect the impact erosion process. ThereIn addition, we can use the crater densities from fore we assumethat each processacts independentlyof different regions to derive the history of atmospheric the other and that the lossesmultiply rather than add pressuredue to impacts through the Martian geologic (seebelow). Starting from an initial surfacepressure, epochs(Figure lb). Tanaka[1986]givescratercounts sputtering alonewould reducethe pressureby 90%. Imfor surface units formed during the different Martian pact erosionwould remove50-90%of the remainingatepochs. From Figure lb, impact erosionhad a signif- mosphere,leaving1-5% of the originalatmosphere,baricant influence during the Noachian, but its influence ring other processes.Thus 95-99% of the atmosphere rapidly declinedduring the Hesperianand Amazonian would have been lost overall. The uncertainty in the timing of the sputteringloss epochs. Becausethe distinction between the geologic affects the resultsto someextent; here we assumethat epochsis tied closelyto the number of impact craters sputtering lossacts only from the start of the geologic on the surface, this conclusionis independent of the record. A later "start of the clock" for sputteringloss absolute timescale in the cratering history. could occur due to the presenceof an intrinsic Mar3.3 Atmospheric Loss From Impact Erosion tian magnetic field or impact delivery of volatiles, and and Pick-up Ion Sputtering would result in a smallercombinedloss.A strongmagWe have shown that Mars could have lost •050-90% of netic field would cause the solar wind to stand off from its atmosphere by impact erosionsincethe earliest existing surfaceformed; this amount of lossdoesnot indicate the planet and would thereby inhibit sputtering loss the presenceof an early atmospherethick enoughto ac- by measurementsof remanent magnetismmade in the [Hutchinset al., 1997].Sucha fieldis, in fact,suggested count for the abundant evidence for fluvial features seen Martian meteoriteALH84001 [Kirschrinket al., 1997] on the surface[Fanaleet al., 1992].However,whenthe and by direct measurementsof the remanentmagnetic results are coupled with the inferred lossfrom pick-up- field overthe ancientsurfaces on Mars [Acu•a et al., ion sputtering, we come closerto reasonablyaccounting 1998]. However,useof the isotoperatiosas an indica- for the climatic transition suggestedby geomorphicev- tor of the efficacyof sputtering losstells us the fraction of gas lost in total, independentof the time of onset of In the sputtering process, photochemical processes sputtering loss. ionize atoms in the upper atmosphere, which are then The assumption that the impact ejection and the accelerateddue to the magnetic field in the solar wind. sputtering lossprocessesof atmosphericremoval are enThese pick-up ions can collide at very high velocities tirely independentis not strictly correct. The minimum with other upper atmospheric species and eject them size of an impactor required to eject gas from the atto space [Luhmannand Kozyra,1991]. This sputter- mospheredependson the mass of atmospherepresent ing process preferentially removes lighter atmospheric [Meloshand Vickery,1989]. Thus atmospheric craterisotopes,which have a larger atmosphericscale height ing will be a more efficientprocessif sputteringlossoc[Jakosky and Jones,1997].By measuring isotoperatios curs prior to it or simultaneouslywith it. In that sense, of atmosphericspeciesan estimate can be made of the our estimates are likely to be a lower bound on the toamount of atmospherethat has been lost by this pro- tal amount of atmospherethat has been lost. If the cess. Observed isotope ratios of Ar, C, H, and N sug- early estimatesregardingthe time history of the maridence. gestthe likely lossof 85-95%of eachgas [Jakoskyand tian magneticfield [Acu•a et al., 1998]hold up, then Jones,1997].The timingof the sputteringlossisuncer- the bulk of the sputtering lossfollowedthe bulk of the tain due to the possibleinfluence of an early magnetic cratering loss;in this case, our assumedmultiplicative 22,694 BRAIN AND JAKOSKY: CLIMATE CHANGE ON MARS behavior is valid, and our estimate of atmosphericloss Maxwell and Michael Carr for acting as Editor for this will be correct. manuscript. Their comments and those of Janet Luhmann and Bob Craddock were appreciated. 4. Discussion and References Conclusions The calculations presentedhere suggestthat a combination of impact erosion and pick-up ion sputtering can remove 95-99% of the atmospherethat was present at the time of the onset of the geologicrecord, about 4 Gyr ago. If no other processeshave acted, then the present 6 mbar pressure would indicate that the Martian surface pressure was in the range 120-600 mbar at the formation time of the oldest surface units. This value is probably not large enoughto support stable liq- Acufia, M.H., et al., Magnetic field and plasmaobservations at Mars: Initial resultsof the Mars Global SurveyorMission, Science, 279, 1676-1680, 1998. Baker, V.R., M.H. Cart, V.C. Gulick, C.R. Williams, and M.S. Marley, Channels and valley networks,in Mars, edited by H.H. Kieffer, et al., pp. 493-522, Univ. of Ariz. Press, Tucson, 1992. Carr, M.H., Water on Mars, Oxford Univ. Press,New York, 1996. Christensen, P.R., et al., Results from the Mars Global Surveyor thermal emissionspectrometer,Science,279, 1692- 1697, 1998. uid surfacewater at any time in Mars' history [Fanale et al., 1992].Alternatively,if we assumethat the atmo- Craddock, R.A., and T.A. Maxwell, Geomorphicevolution sphere was as thick as 3 bar at one time, for example, then 95-99% lossby sputteringand impact would leave 30-150 mbar still in the atmosphere. It is quite plausible that the regolith or polar caps could contain most of this remaining gas. For example, 30-40 mbar of CO2 could be adsorbed in the regolith of the Martian highlands through ancient fluvial processes,J. Geophys.Res., 98, 3453-3468, 1993. Fanale, F.P., S.E. Postawko,J.B. Pollack, M.H. Carr, and R.O. Pepin, Mars: Epochal climate changeand volatile history, in Mars, edited by H.H. Kieffer, et al., pp. 11351179, Univ. of Ariz. Press, Tucson, 1992. Hartmann, W.K., et al., Chronologyof planetaryvolcanism by comparativestudiesof planetary cratering,in Basaltic [Zent and Quinn,1995],and as muchas 100-200mbar Volcanismon the Terrestrial Planets, Basaltic Volcanism could be sequesteredas ice or clathrate in the poles Study Proj., Pergamon, Tarrytown, N.Y., 1981. (though a few tens of mbar may be a more plausible Hutchins, K.S., and B.M. Jakosky, Evolution of Martian atmosphericargon: Implications for sourcesof volatiles, upper limit) [Mellon, 1996]. Although an unknown J. Geophys.Res., 101, 14,933-14,949, 1996. amount of CO2 may be present as carbonate minerals Hutchins,K.S., B.M. Jakosky,and J.G. Luhmann,Impact of in the crust, a substantial carbonate reservoir is not a paleomagneticfield on sputtering lossof Martian atmoneeded. sphericargonand neon, J. Geophys.Res., 102, 9183-9189, 1997. These calculations imply that impact erosionwas an Jakosky, B.M., and J.H. Jones, The history of Martian important processearly in Mars' geologichistory but volatiles, Rev. Geophys.,35, 1-16, 1997. that it cannot account for enough atmosphericlossto Kirschvink, J.L., A.T.' Maine, and H. Vali, Paleomagnetic have allowed the formation of the observedvalley netEvidence of a low-temperature origin of carbonatein the works on Mars' oldest surfaces.Similarly, sputtering is Martian meteorite ALH84001, Science, 275, 1629-1633, a processthat probably has acted throughout most of Martian history and has profoundly affectedthe isotopic composition of Mars' atmosphere, but it is not efficient enough acting alone to have produced the inferred climate changethrough atmosphericloss. Only when we combine the three lossprocessesof large impacts, pickup ion sputtering, and sequestration into the surface are we able to identify a plausible means of removing several bars of early atmosphereon Mars. The present analysis is not unique, and there are considerableuncertainties in quantifying all of the processesby which gas can be removed from the atmosphere. However, it does suggestthat plausible mechanisms exist that can provide for a transition from a thick, warm, early climate to the later cold, dry climate. Interestingly, we can do this without requiring that large surface or subsurface deposits of carbonate minerals contain the CO2 from the early atmosphere; the absence of detectable carbonate spectral features [e.g.,Christensen et al., 1998]thereforedoesnot present a significant problem. Acknowledgments. This research was supported in part by the NASA Planetary Atmospheresprogramthrough grant NAGW-3995 and the Exobiology program through grant NAG5-4530. B.M.J. thanks Associate Editors Ted 1997. Luhmann, J.G., and J.U. Kozyra, Dayside pickup oxygen ion precipitation at Venus and Mars: Spatial distributions, energy deposition and consequences,J. Geophys. Res., 96, 5457-5467, 1991. Mellon, M.T., Limits on the CO2 content of the Martian polar deposits, Icarus, 12•, 268-279, 1996. Melosh, H.J., and A.M.Vickery, Impact erosionof the pri- mordial atmosphereof Mars, Nature, 338, 487-489, 1989. Schmidt, R.M., and K.R. Housen, Int. J. Impact Eng., 5, 543-560, 1987. Scott, D.H., and K.L. Tanaka, Mars: Paleostratigraphic restoration of buried surfacesin Tharsis Montes, Icarus, •5, 304-319, 1981. Tanaka, K.L., The stratigraphy of Mars, J. Geophys.Res., 91, E139-E158, 1986. Walker, J.G.C., Impact erosion of planetary atmospheres, Icarus, 68, 87-98, 1987. Zent,A.P.,andR.C.Quinn, Simultaneous adsorption of•O2 and H20 under Mars-like conditions and application to the evolution of the Martian climate, J. Geophys.Res., 100, 5341-5349, 1995. David A. Brain and Bruce M. Jakosky, Laboratory for Atmospheric and Space Physics, University of Colorado, Campus Box 392, Boulder, CO 80303-0392(e-maih [email protected]) (Received April 30, 1997; revisedMay 30, 1998; acceptedJune 16, 1998.)
© Copyright 2026 Paperzz