ICARUS 71, 203-224 (1987)

The Case for a Wet, Warm Climate on Early

J. B. AND J. F. KASTING

NASA Ames Research Center, Moffett Field, California 94035

S. M. RICHARDSON

Iowa State University, Ames, Iowa 50011

AND

K. POLIAKOFF

I.M.I. Inc., San Jose, California 95114

Received September 22, 1986; revised March 9, 1987

Theoretical arguments are presented in support of the idea that Mars possessed a dense CO2 atmosphere and a wet, warm climate early in its history. Calculations with a one-dimensional radiative-convective climate model indicate that CO2 pressures between 1 and 5 bars would have been required to keep the surface temperature above the freezing point of water early in the planet's history. The higher value corresponds to globally and orbitally averaged conditions and a 30% reduction in solar luminosity; the lower value corresponds to conditions at the equator during perihelion at times of high orbital eccentricity and the same reduced solar luminosity. The plausibility of such a COz greenhouse is tested by formulating a simple model of the CO2 geochemical cycle on early Mars. By appropriately scaling the rate of silicate weathering on present Earth, we estimate a weathering time constant of the order of several times 107 years for early Mars. Thus, a dense atmosphere could have persisted for a geologically significant time period (-109 years) only if atmospheric COz was being continuously resupplied. The most likely mechanism by which this might have been accomplished is the thermal decomposition of carbonate rocks induced directly and indirectly (through burial) by intense, global-scale volcanism. For plausible values of the early heat flux, the recycling time constant is also of the order of several times 107 years. The amount of COz dissolved in standing bodies of water was probably small; thus, the total surficiai CO2 inventory required to maintain these conditions was approximately 2 to 10 bars. The amount of CO2 in Mars' atmosphere would eventually have dwindled, and the climate cooled, as the planet's internal heat engine ran down. A test for this theory will be provided by spectroscopic searches for carbonates in Mars' crust. © 1987 Academic Press, Inc.

1. INTRODUCTION surface above the temperature at which the planet radiates to space, about 212°K (Pol- Today, Mars is a desert world. The glo- lack 1979). Although temperatures at the bally averaged atmospheric pressure at the surface rise above the freezing point of surface is only about 7 mbar. Hence, the water near the equator during the warmest dominant carbon dioxide atmosphere pro- part of the day, water ice is evaporated so duces only about a 5°K warming of the efficiently that its temperature at the sur- 203 0019-1035/87 $3.00 Copyright © 1987 by Academic Press, Inc. All rights of reproduction in any form reserved. 204 POLLACK ET AL. face, except perhaps for some special widespread volcanism or greenhouse circumstances, cannot reach 273°K (Inger- warming within an exposed ice deposit soll 1970). (Cloy 1984). However, the climate of Mars may not Geomorphic studies of large impact have always been as severe as it is today. basins on Mars indicate that such features Geomorphic evidence and theoretical and their associated secondary craters and arguments suggest that the atmosphere of ejecta were much more heavily eroded in Mars may have been significantly denser the planet's early history--more specifi- early in the planet's history (Walker 1978, cally, prior to the formation of the Argyre Pollack 1979, Pollack and Yung 1980, Cess basin--than at later times (Schultz 1985). If et al. 1980, Toon et al. 1980, Kahn 1985, this interpretation is correct, it implies the Schultz 1985). The resulting greenhouse existence of a much denser atmosphere at warming may have been much larger than these early times. Furthermore, there that at present, perhaps large enough to appears to have been a dramatic decrease permit the common occurrence of liquid in the density of valley networks and a water at the surface (Pollack 1979). This change in drainage style after the Argyre possibility was first indicated by the impact (Schultz 1985). These correlations discovery of fluvially produced channels on are consistent with the existence of a dense images obtained by the 9 orbiting atmosphere in Mars' early history, one spacecraft (Masursky 1973). However, it capable of raising the surface temperature was soon realized that the large outflow above the freezing point of water for an channels might be generated under current extended period of time (-109 years). climate conditions. Large amounts of liquid The most promising gas for creating a water, rapidly released from a near-surface large greenhouse effect on early Mars is reservoir, could flow large distances before carbon dioxide. It is likely that much more totally freezing by forming a top, insulating of it was released into the atmosphere over layer of ice (Peale et al. 1975, Baker 1978). the planet's historywespecially at early This mechanism may be appropriate to times--than resides there today. Further- explain isolated, local channels, but it is more, carbon dioxide is relatively stable harder to make such an argument for the against UV photolysis (Pollack 1979, Pol- fluvial gullies or valley networks. Such fea- lack and Yung 1980, Cess et al. 1980). tures are ubiquitous on the old terrain of the Alternative greenhouse gases, such as southern highlands. Some corresponding ammonia, methane, and sulfur dioxide, are widespread process is needed to generate rapidly photolyzed into gases and/or par- the liquid water that produced them. ticles that either are poor greenhouse gases Careful analyses of and Viking (e.g., nitrogen) or have short residence images of valley networks indicate that times in the atmosphere (e.g., sulfuric acid they were formed chiefly by sapping, rather aerosols) (Pollack and Yung 1980, Cess et than by rainfall (Sharp and Malin 1975, al. 1980). Greenhouse calculations indicate Pieri 1980). This deduction does not pro- that CO2 partial pressures in excess of vide hard evidence that Mars was warmer about 1 bar are required to raise the surface in its early history. On the other hand, there temperature on Mars above the freezing is still a requirement that thermal condi- point of water (Pollack 1979, Cess et al. tions close to the surface permitted the 1980, Toon et al. 1980). Such quantities of occurrence of liquid water over some finite carbon dioxide are consistent with interval of time at many locations in the old estimates of the planet's total CO2 cratered terrain. A warm climate would inventory, as discussed in Section 6 of this represent one such mechanism (Pollack paper. 1979). Alternative possibilities include For the time being, we assume that A WARM AND WET MARS 205 enough carbon dioxide was present in the decomposed by processes associated with early atmosphere of Mars to raise the sur- global-scale volcanism on time scales com- face temperature above the freezing point parable to the weathering time scales. This of water, at least at low latitudes. We first decomposition returns gaseous CO2 to perform greenhouse calculations with the atmosphere. Hence, dense CO2 improved opacities for carbon dioxide and atmospheres can be maintained on early water vapor to define more precisely the Mars with plausible volatile inventories. amount of carbon dioxide needed to - However, at later times, recycling car- tain these conditions. Then, we consider bonate rocks becomes inefficient and so the stability of this dense carbon dioxide liquid water occurs on the surface far less atmosphere against the formation of car- frequently. bonate rocks via the weathering of silicate rocks. On Earth, in the presence of liquid water, chemical weathering proceeds at 2. GREENHOUSE MODELS such a rate that all the atmospheric carbon In this section of the paper, we present dioxide would be eliminated on a time scale estimates of the amount of atmospheric of I0,000 years in the absence of CO2 sour- carbon dioxide needed to raise the surface ces (e.g., volcanic outgassing) (Holland temperature on early Mars above the freez- 1978). We use what is known about the CO2 ing point of water. The greenhouse warm- geochemical cycle on Earth to estimate the ing is produced by the infrared opacity of time scale for CO2 loss on Mars. Next, we both gaseous carbon dioxide and water evaluate the time scale on which the vapor. However, the amount of atmo- atmosphere can be resupplied with gas by spheric water vapor is controlled by juvenile outgassing and volcanically driven its vapor pressure curve, as is the case for recycling of carbonate rocks. Finally, we the present atmosphere of the Earth. determine the total volatile inventory of near-surface carbon dioxide required to 2.1. Procedure maintain a dense carbon dioxide atmo- A one-dimensional radiative-convective sphere over a billion years by a quasi- model was used to determine the steady- steady-state carbon dioxide cycle that state vertical temperature profile in the involves a balance between production and atmosphere and the temperature at the sur- loss. Through such analyses we hope to face. Details of the model have been understand better the plausibility of a published elsewhere (Kasting and Acker- warm, wet climate on Mars in its early man 1986) and will not be repeated here. history. In the final sections of the paper we We assume that the lapse rate in the make such an assessment and propose a troposphere follows a moist adiabat and possible observational test of our scenario. that the tropospheric relative humidity It is perhaps useful to summarize here decreases linearly with pressure, starting the major theses of this paper as a guide to from a value of 0.77 at the surface (Manabe the following sections. Atmospheres con- and Wetherald 1967). This latter formula is taining several bars of CO2 are required to appropriate for Earth, hence its use seems permit the common occurrence of liquid reasonable for modeling an Earth-like early water on early Mars. The chemical wea- Mars. By contrast, Pollack (1979) and Cess thering of surface silicate rocks converts et al. (1980) assumed a constant relative even these amounts of atmospheric CO2 humidity. Clouds are neglected in our cal- into carbonate rocks on time scales much culations; the effect of this assumption is less than a billion years when liquid water is described in Section 2.2. present at the surface. However, on early A key aspect of these new greenhouse Mars, carbonate rocks can be thermally calculations is the inclusion of improved 206 POLLACK ET AL. values for the opacity of gaseous CO2 and averaged value. Furthermore, the orbital HzO in the thermal and solar regions of the eccentricity undergoes quasi-periodic spectrum. For most of this wavelength variations on time scales of 0.1 to 1 million interval, these values were derived from years, with the eccentricity varying bet- line-by-line calculations based on the data ween 0 and 0.14 (Ward 1974). At maximum given on the AFGL tape (McClatchey et al. eccentricity, the insolation at perihelion 1971). In addition, allowance was made for would be 35% larger than its orbitally the pressure-induced transitions of CO2, averaged value. In addition, the equatorial the 8- to 12-/xm continuum of water vapor, regions receive 40% more sunlight than the and the weak lines of both gases in the planet as a whole (Toon et al. 1980). Thus, visible and near-infrared spectral regions the values of S of interest range from about (Roberts et al. 1976, Kasting et al. 1984b, 0.7, corresponding to the orbitally and glo- Moore et al. 1966, Herzberg and Herzberg bally averaged value 4.5 billion years ago, 1953, Kasting 1986). to about 1.3, corresponding to the value at The computed values of surface tempera- the equator and at perihelion at maximum ture, T, depend on the carbon dioxide eccentricity for the same epoch. pressure, P, the surface albedo, As, and the The above discussion contains the impli- diurnally averaged amount of sunlight, S. cit assumptions that the thermal response Here, we define S as the ratio of the actual time of the atmosphere is significantly shor- amount of sunlight at the top of the Martian ter than an orbital period and that atmosphere to the globally and orbitally atmospheric heat transport is relatively inef averaged value for present-day Mars. For a ficient in equalizing temperature differen- given choice of values for As and S, T is ces between low and high latitudes. We computed for a wide range of values of P to now show that these assumptions are rea- determine the value of P at which T equals sonable approximations for the atmo- the freezing point of liquid water. spheric pressures of most interest (1 to A nominal value of 0.215 was selected for 5 bars). the surface albedo, since this yielded a Consider first the thermal response time planetary albedo of 0.212 for the present of the atmosphere. It is given by Martian atmosphere with P = 6 mbar, in good agreement with the value determined cpPST (1) by the Viking IRTM experiment at a time of tr go.T4, low dust loading (Kieffer et al. 1977). However, we also performed calculations where Cp is specific heat at constant press- with A = 0. I to simulate conditions for a ure, 8T is the change of temperature over hypothetical early covered an orbital period, g, is the acceleration of with unoxidized minerals (Toon et al. gravity, o- is the Steffan-Boltzmann con- 1980). stant, and Te is the effective temperature of The value of S depends on the epoch, emission to space. For P = 1 bar and S = 1, latitude, and orbital position (Pollack and our calculations yield Te = 204.2°K. For 6T Yung 1980, Cess et al. 1980). When Mars = 30°K, the response time of the first formed, about 4.5 billion years ago, the atmosphere is 6 × 106 sec or about a factor solar luminosity was about 25 to 30% less of 10 times smaller than the orbital period. than its current value (Newman and Rood Therefore, for the range of parameters of 1977, Gough 1981). It increased by about interest, the atmosphere should respond 5% during the first billion years of the relatively rapidly to changes in insolation planet's history. Mars' orbit is highly over its orbit. eccentric at the present epoch (eccentricity It is more difficult to evaluate the effi- = 0.093). At perihelion, the incident solar ciency of the atmospheric heat transport by flux is 22% larger than the orbitally the paleoatmospheres of interest. Hoffert et A WARM AND WET MARS 207 al. (1981) have presented a parameteri- infrared wavelengths. For the standard zation for heat transport by baroclinic model, a surface pressure of 2.2 bars is eddies that provides an estimate of the needed to raise the surface temperature to temperature difference between midlati- the melting point of water ice. tudes and the poles. Using this parameteri- Figure la also shows the predicted zation for a 1-bar atmosphere, we find a dependence of surface temperature on CO2 temperature difference of about 30°K bet- partial pressure obtained in the earlier cal- ween midlatitudes and the poles or about culations of Pollack (1979) and Cess et al. 50°K between the equator and poles. These (1980). The results of this paper are in differences scale as the inverse of the surprisingly good agreement with those of square root of the atmospheric pressure. Pollack (1979), who simply performed an Thus, atmospheric heat transport will not energy balance calculation at the top of the erase much of the temperature variation atmosphere, neglecting the change of with latitude for the atmospheres of pri- planetary albedo with P, and using much mary interest here. cruder opacity coefficients. This agreement is probably the result of partially com- 2.2. Results pensating effects. On the one hand, Pollack Figures la and lb illustrate the variation (1979) obtained somewhat higher tempera- of surface temperature (standard model) tures for a fixed surface pressure by not and planetary albedo with surface pressure allowing for the influence of Rayleigh scat- for A~ = 0.215 and S = 1. Both quantities tering on the planetary albedo and by using increase slowly at first as P is increased a constant relative humidity throughout the from its present value to several tenths of a troposphere. On the other hand, he bar, and then more rapidly as P is increased underestimated the temperature by not further. The rapid increase in planetary including the pressure-induced transitions albedo at the higher pressures is due to the of CO2. increasing prominence of Rayleigh scat- Our predicted surface temperatures are tering, while the corresponding rapid significantly lower than those obtained by increase in surface temperature at the Cess et al. (1980), who did a full radiative- higher pressures, in spite of the increase in convective calculation, computed the albedo, is a result of the atmosphere planetary albedo, and used apparently becoming optically thick at almost all better opacity coefficients than Pollack

I i , b .4 300 a' ' ,y //// QO uJ ¢c 280 . CESS et al. (198~L/ ~/ . =, k- < <~ -F~EEZING PO,~ O~W~T~ / --7--/------=> ~ 260 / ./ STANDARD ~.3 ~ 240 .~- ~POLLACK 119791 220 c~

200 i = i L I L 10-2 10-1 1 10 10 -2 10-1 1 10 SURFACE PRESSURE, bars SURFACE PRESSURE, bars

FIG. 1. (a) Surface temperature, Ts, and (b) planetary albedo, Ap, of Mars as the function of the surface pressure of CO2 for the present surface albedo and globally and orbitally averaged solar flux. In (a), the solid curve presents results from this paper, while the other two curves represent results from two earlier calculations. 208 POLLACK ET AL.

(1979). Unfortunately, there appears to about 0.75 to 5 bars are needed to raise the have been a serious coding error in the surface temperature to the melting point of calculations of Cess et al. (1980) that water ice. This critical pressure depends caused them to overestimate the surface most sensitively on the value of S. temperature (V. Ramanathan, private com- The greatest uncertainty in the above munication). Hoffert et al. (1981) per- calculations is undoubtedly caused by our formed a calculation of the CO2 greenhouse neglect of water clouds. Introducing a effect based on flux balances and found cloud layer into the model would not surface temperatures even higher than greatly reduce this uncertainty because one those of Cess et al. (1980) for P -< 1 bar. could not be sure how its height, fractional The reason that their model predicted high area, and optical properties varied with temperatures is not readily apparent, but it surface temperature. Despite these poten- may be caused by their neglect of con- tial problems it is probable that the tem- vection. More recently, Postawko and perature feedbacks caused by clouds would Kuhn (1986) performed a series of Martian be relatively small. The effects of clouds on surface temperature calculations for P = 1 the infrared and solar radiation budgets are and 3 bars; their results are roughly com- to some extent compensating: clouds tend parable to ours. They further suggested that to lower T by reflecting incident sunlight the greenhouse effect on early Mars may back to space, but they also raise T by have been augmented by SO2. But, by enhancing the greenhouse effect (Pollack analogy to conditions in the Earth's tro- 1979, Toon et al. 1980). With this in mind it posphere and stratosphere, SO2 in an early is interesting to note that this same model Martian atmosphere would have been requires only a slight adjustment in surface rapidly oxidized and sedimented out of the albedo (from 0.215 to 0.22) to reproduce the atmosphere as sulfate. observed mean global surface temperature Figure 2 shows the influence of the ass- for a cloud-free Earth. Thus, to the extent umed values of As and S on the computed that a warm early Mars resembled present- surface temperature. According to these day Earth in terms of surface reflectivity results, surface pressures ranging from and cloud properties, our model may do a surprisingly good job of predicting its sur- 290 face temperature.

A s = O.fl"/ ) .j 3. LOSS RATES FOR ATMOSPHERIC CO 2 280 On Earth, CO2 is removed from the c¢ /FREEZ,NGPO,N'"'/ _ __i I_ __ atmosphere/ocean system by the chemical c~ weathering of continental silicate rocks, ~- 270 followed by the deposition of carbonate l,- rocks on the ocean floors (Holland 1978). In uJ o..% detail, carbon dioxide dissolves in rain- 260 @ water and surface water to form a weak acid. This mildly acidic water dissolves silicate rocks, releasing Ca 2+, Mgz+, and 250 /,/ = / /= J t 1 2 3 4 5 6 HCO~ into solution. These dissolved ions SURFACE PRESSURE, bar are transported to the seas by rivers and FIG. 2. Surface temperature as a function of surface eventually are precipitated as calcium and pressure for several values of the surface albedo and magnesium carbonates on the ocean floors. incident solar flux, S. Solid lines refer to results for the current globally averaged albedo of 0.215. S = 1 for Thus, running water, exposed silicate the present globally and orbitally averaged solar flux at rocks, and ocean basins are necessary Mars. ingredients in this type of chemical wea- A WARM AND WET MARS 209 thering. In the following, we scale the wea- where Pe is the partial pressure of carbon thering rates for Earth to estimate those dioxide in the Earth's atmosphere (3.3 x that might apply on an early Mars with a 10 -4 bars). This expression assumes that dense carbon dioxide atmosphere and tem- the high-pressure (2 to 20 bars) laboratory peratures above the melting point of water data can be extrapolated to the relatively ice. The object of these calculations is to low CO2 partial pressure in Earth's estimate the time for the removal of carbon atmosphere. As a consequence of this dioxide by chemical weathering in the relatively weak dependence of the weather- absence of sources of new atmospheric ing rate on pressure, the weathering time CO2. scale increases as the pressure increases (see below). As the temperature increases, water is 3.1. Procedure evaporated at a faster rate from standing The rate of chemical weathering by bodies of water; hence, the precipitation atmospheric COz is a function of the partial rate and the amount of runoff also increase. pressure of CO2 at the surface, P, the Numerical experiments with general cir- surface temperature, T, the cation content culation models indicate that this depen- of the exposed rocks, and the fraction of dence can be expressed as (Manabe and the planet occupied by open bodies of Stouffer 1980, Berner et al. 1983) water, a. On the basis of our present g2 -- exp(c28T), (4) understanding of the silicate weathering rate on Earth, we may formally write the where 8T = T - Te. Te is the Earth's mass loss rate of atmospheric CO2 per unit present mean temperature of 288°K. The surface area of the planet in terms of the constant c2 = 0.038. Here, we use a some- product of a number of scaling factors, gi, what different functional form for g2 than and the present silicate weathering rate for used by Berner et al. (1983) to cover a Earth: wider range of temperature conditions. This exponential increase of the rate of dM dMe precipitation with temperature is ultimately dt = glg2g3g4g5 dt " (2) limited by the amount of sunlight available Here gj incorporates the dependence of the to heat the open bodies of water. Accordin- weathering rate on pressure, gz expresses gly, we demand that g2 be limited by its dependence on the variation of runoff g2 ~ CrnaxS, (5) with temperature, g3 gives its dependence on the variation of the weathering reaction where Cma×is a constant and S is the amount with temperature, g4 is its dependence on of sunlight, normalized to a value of I for the cation content of the rocks, g5 provides the globally and orbitally averaged situation its dependence on the fraction of open for present-day Mars. About half the sun- bodies of water (the ultimate source of the light absorbed by Earth's oceans is used to runoff), and dMJdt is the present weather- supply the latent heat of vaporization (Holl- ing rate of continental silicate rocks on and 1978). This suggests that g2 could not Earth. be any larger than 2 for the present Earth. Results from laboratory experiments Note that since the oceans lose heat to the (Lagache 1965, 1976) indicate that the atmosphere by thermal radiative exchange pressure-dependent factor is given by and boundary layer convective processes, (Walker et al. 1981) consideration of these other processes will not result in an increase in this upper limit, = /p~0.3 unless the atmosphere developed an g' \El ' (3) inversion near the surface. Since S ~ 2 for 210 POLLACK ET AL. the Earth, we select 1 as a nominal value thering rate on the fraction of the planet for Cmax. covered by open bodies of water. Since this Measurements of the latitudinal depen- parameter should be 0 when a equals either dence of the weathering rate on Earth 0 (no water available to evaporate) or 1 (no (Harmon et al. 1975) in conjunction with land available to weather), we have chosen the above expression for gz suggest the the functional form following dependence of the chemical reac- g5 = a(1 - a). (7) tion rate on temperature (Berner et al. 1983): We select a nominal value of 0.05 for a, since there is no evidence for the occur- g3 = exp(c3~T). (6) rence of large oceans on early Mars. The coefficient c3 = 0.049. It should be The chemical weathering rate, if suffi- noted that the expressions for gz and g3 ciently large, is ultimately limited by the were derived for Earth-like conditions, i.e., rate of physical weathering, i.e., the rate at for surface temperatures above the freezing which fresh material is exposed at the sur- point of water. At temperatures below the face. For Earth, the average physical wea- freezing point, chemical weathering rates thering rate is only about 6 times larger than would presumably be much smaller than the average chemical weathering rate (Holl- assumed here; hence, Eq. (2) represents a and 1978). Furthermore, not all elements of strong upper limit on the actual weathering rocks can be chemically weathered; thus, if rates under such conditions. This equation this chemical weathering rate was only a leads to somewhat larger rates than actually factor of several times larger, physical wea- occurred when temperatures were above thering would act to limit it. We express the freezing point at only some locations or this limitation formally as an upper bound times of the year. on the product of scaling factors: On Earth, Ca 2+ and Mg 2+ are the prin- cipal cations that are available to form gjg2g3a <~ 0.71gmax. (8) carbonates. Fe z+ is also capable of forming This limit does not include the g4 factor carbonates, but it is very quickly oxidized since the fraction of rock that is not che- to insoluble ferric iron. Presuming that mically weathered is not directly related to early Mars, like early Earth, had little free the fraction of cations that causes a net atmospheric oxygen, Fe 2+ would also have removal of COz from the atmosphere. It been available for forming carbonates. also does not contain the (1 - a) factor of g5 Therefore, the parameter g4 is obtained by since the constraints involve a comparison estimating the ratio of the Ca, Mg, and Fe of chemical and physical weathering rates content of rocks on early Mars to the Ca per unit area of land. The factor of 0.71 on and Mg content of continental silicate rocks the right-hand side of the equation is the on Earth. Comparison of Viking elemental fraction of Earth's surface area occupied by abundance measurements ( et al. oceans. The present average physical wea- 1977) with the extensive data set for Earth thering rate on Earth is 30 m/106 years (Holland 1978) implies that this ratio is (Holland 1978). It is determined primarily about 4 for the present surface of Mars. by the uplift rate for continents. On early However, since the abundances of Fe and Mars, the physical weathering rate would Mg on Mars' surface may have been sig- probably have been related to the rate of nificantly enhanced subsequent to the early volcanic resurfacing, which should have epochs of interest here (Clark and Baird been about a factor of 10 higher than the 1979), we select a nominal value of 2 for g4. present physical weathering rate for Earth Perhaps the hardest parameter to define (see below). We therefore select 30 as a is gs, which is the dependence of the wea- nominal value for gmax since this parameter A WARM AND WET MARS 211 has a value of about 3 for the present Earth. factor of about two for given values of the This latter value is crudely obtained by relevant parameters. In addition, these taking the product of the ratio of the parameters can assume a range of plausible average physical to chemical weathering values for early Mars. For this reason, we rates (6) and an assumed factor of 0.5 for performed sensitivity studies to elucidate the soluble component of rock, as dis- the impact of these ranges on the values of cussed above. the weathering time scales. Finally, we derived the value for dMe/dt from estimates of the current rates of che- 3.2. Results mical weathering of continental silicate Using the equations given in Section 3.1, rocks (Berner et al. 1983) and the fraction we computed the weathering time scale Zw of continental surfaces occupied by these as a function of P for the nominal parameter rocks. This fraction equals 0.47 (Ehlers and choices as well as for plausible alternative Blatt 1982, Ronov and Yaroshevsky 1969). values. Figure 3a shows this relationship The remainder of the continental area is for the nominal surface temperature of occupied by carbonates, organic deposits, 273°K, along with two higher values. For and evaporites, the weathering of which the nominal temperature, Zw ranges from has no effect on atmospheric CO2. Thus, 1.5 x 10 7 years for a 0.75-bar atmosphere to we find that 6 x 10 7 years for a 5-bar atmosphere. As the temperature is increased by 30°K above dMe= 5.1 x 10 -4 g/cm2/year. (9) the freezing point, the weathering time dt scale drops by about a factor of 10. Thus, in On early Mars, we assume that silicates the absence of a source of fresh CO2, occupy all the land area. This will lead to a surface temperatures on a warm early Mars slight underestimate of the weathering would have fallen rapidly towards the removal time scale. freezing point, as much of the atmosphere The time scale Zw for eliminating was quickly lost by chemical weathering. atmospheric CO2 by chemical weathering is The time scale for this decay is short com- obtained from pared to the geological time scales of M interest (~109 years). (lO) "rw - dM/dt' The sensitivity of the weathering time scale to our choice of gmax is explored in where M is the mass of atmospheric CO2 Fig. 3b. Only if the physical weathering rate per unit area. It is given by was more than 100 times smaller than our nominal choice would it provide a signifi- M = 2.69 × 103p. (11) cant limitation on the chemical weathering P has units of bars in this equation. rate. Thus, it seems to be very unlikely that In the above discussion, we have tried to significantly larger chemical weathering identify the key chemical and physical time scales could be obtained for alterna- processes that determine the rate of che- tive choices of gmax. mical weathering on a planet having liquid Figure 3c shows the dependence of zw on water on its surface. In so doing, we have a. As the fraction of area occupied by made use of the very extensive studies of standing bodies of water increases, rw the Earth's CO2 geochemical cycle. decreases in roughly the same proportion. Nevertheless, a number of the numerical Values of ~'w on the order of 10 9 years could constants in the above expressions are be achieved by selecting a value for a on known only to an accuracy of several tens the order of 0.001. Such a low value for a of percent. Thus, the weathering time seems highly unlikely for a warm, wet Mars scales estimated below are good only to a for which the amount of outgassed CO2 is 212 POLLACK ET AL.

from their nominal values. Sample cal- ~ 10 E a T,K -- 273 culations also show that ~w changes very ,~ 10~ ----- 28~ --- 303 ~,.t little as the solar flux parameter S varies f- from 0.7 to 1.3 and that it increases by no ~ 10 7 more than a factor of several as Cma× is ~ 106 ,< decreased by a factor of several below our 105 nominal value of 1 (see Eq. (5)). ,<.J The above discussion indicates that ~w is , , ,,i , . ..i .... = .... , ..... , •~ 10-3 10-2 10-1 1 101 102 of the order of 107 years for CO2 pressures CO2 PRESSURE, bar near 1 bar and that plausible changes in parameter choices could not increase ~w to 101( ~ b near 109 years. Thus, if a wet, warm climate gmax ...... 0.1 occurred on early Mars for a lengthy period ~_ 109 ~, 1. ....'"""" u. of time and if it were produced by a CO2 ~ 108 greenhouse, then the loss of CO2 by che- ~ 107 mical weathering of surface silicate rocks z ~ 10E uJ must have been balanced by a sufficiently

,~ 10E vigorous CO2 source. In Section 4 we con-

,,=, ~Q4 ...... ,,I I , ,=[ , , ,=1 , , ,,I sider two such possible sources. Z 10 10 -2 10 -1 1 101 102 O CO 2 PRESSURE, bar

4. SOURCES OF ATMOSPHERIC CO2 ~'10 TM : c The most obvious sources of CO2 are the u3 a ...... 0.01 outgassing of juvenile CO2 from the ~ 109 -- 0.05 ...... u. --- 0.25 "~ 10 8 planet's interior and the recycling of car- bonate rocks by intense, global-scale vol- ~ 107 canism. In the latter case, carbonate rocks ~ 106 could have been thermally decomposed by ,~ 105 contact with hot lava near the surface or by

, , ...... , ...... ,i , ,,,,,,,i , ,,,,,.i ...... i ~ 10.3 10-2 10-1 1 101 102 burial to a depth sufficient to reach tem- CO 2 PRESSURE, bar peratures above their decomposition point. FIG. 3. Chemical weathering time scale as a function We formulate both problems in terms of of the surface pressure of CO2. Results are shown for characteristic time scales for comparison three different choices of (a) surface temperature, T; with the weathering time scales of the last (b) upper bound on the chemical weathering rate section. imposed by the physical weathering rate, gmax; and (c) fraction of the surface occupied by open bodies of water, a. 4.1. Procedure Since the Moon experienced extensive global-scale volcanism during the first not greatly different from that expected for several hundred million years of its history an Earth analog (and hence a nontrivial (Wasserburg et al. 1977), it is reasonable to amount of water would have also been suppose that such a situation also prevailed outgassed). on early Mars. Early intense volcanism Plausible variations in any of the could have played a key role in the CO2 parameters other than the ones already cycle by providing an effective mechanism discussed do not alter the computed values for releasing juvenile CO2 stored in the of ~'w by more than a factor of several. For planet's mantle and by bringing lava to the example, the cation factor g4 and hence 7w surface to recycle carbonate rocks. Here, might plausibly vary by a factor of 2 or so we present a simple calculation of the time A WARM AND WET MARS 213 required for each of these sources to re- ture. In our calculations we use represen- plenish the CO2 in Mars' atmosphere. tative values for basaltic lavas of 4 × 109 The surface heat flux, Ft, on early Mars ergs/g for L, 1.2 × 107 ergs/g/°K for Cp, and would have been determined by the com- 1550°K for Trn (Basaltic Volcanism Study bination of heat delivered to the surface by Project 1981). We use Eq. (15) to determine thermal conduction, Fc, and heat generated the value of dm/dt for specified values of Ft by the cooling of lava at the surface, F~. Let and a. a represent the fraction of Ft carried by Given a knowledge of dm/dt, we can heat conduction. The conductive heat flux readily calculate the rate at which the sur- is given by face is covered over by lava extrusions: dT dz 1 dm - (17) Fc = c~Ft = k d-z' (12) dt pdt' where k is the thermal conductivity and where p is the density of lava, here taken to dT/dz is the rate of increase of temperature be 3 g/cc. This equation is used to deter- with depth. Heat conduction results from a mine the time scales for resupplying the combination of lattice conductivity, kl, and atmosphere with CO2 by both juvenile out- radiation transport, kr. For basaltic rock (a gassing, ~'j, and recycling of carbonate reasonable analog for the surfaces of rocks, ~'r. interest) these components of the thermal Time scale rj is related to dz/dt by conductivity may be written in terms of the subsurface temperature, Ti, according to = ' (18) (Basaltic Volcanism Study Project 1981)

10 7 where Dm is the depth of the mantle that is kl - (13) participating in the outgassing and Pm is the 7.4 + 0.05T~' amount of CO2 contained in this region of with kl having units of ergs/cm/sec/°K. The the mantle, expressed in terms of an radiative conductivity is equivalent atmospheric pressure. We take a value of 500 km for Dm, based on thermal k~ = 2.3(T~ - 500) × 10z (14) history models of Mars (Basaltic Volcanism when T~ is greater than 500°K and is other- Study Project 1981). wise zero. Time scale zr is obtained from The component of the surface heat flux (.z), due to extruded lavas may be written as rr = Dd \~-] , (19) dm F 1 = (1 - ot)F t = (L + CpST) d-T' (15) where Dd is the depth to which carbonate rocks need to be buried to thermally where L is the latent heat released by the decompose into silicates and CO2 gas. This lava freezing, Cp is its specific heat, dm/dt depth can be calculated from the finite is the mass of lava extruded per unit area difference form of Eq. (12) and a knowledge per unit time, and 8T is given by of the minimum temperature needed for decomposition, Td: 6T = Tm - T, (16) k(Td- T) with Tm being the initial temperature of the Dd = (20) otgt lava when it reaches the surface and T is its final temperature after cooling. We assume Equation (19) for ~'r neglects thermal that Tm equals the melt temperature and T decomposition that occurs when hot lava equals the unperturbed surface tempera- comes in contact with carbonate rocks. 214 POLLACK ET AL.

Thus, our computed values for rr will be mW/m2 ( and Arvidson 1981), which somewhat larger than what they would may be compared with a value of about 100 actually be for the scenario under discus- mW/m2 for the average heat flux from the sion. At the depths where decomposition Earth's ocean basins. During the early his- takes place (5-20 km), a nominal value of tory of Mars, Ft was probably larger than 950°K is appropriate for Td (Harker and its current value by a factor of 3 to I0 Tuttle 1956). The parameter k in Eq. (20) is because of a larger release of heat by long- evaluated at a depth halfway between the lived radioactive elements, the formation of surface and the depth of decomposition. the Martian core, and a larger residual heat For interpreting the results presented from planetary accretion (Davies and below, it is useful to write Eq. (19) in aform Arvidson 1981, Stacey 1980). that displays more explicitly the depen- Figure 4a presents our estimate of ~-j as a dence of 7r on a and Ft. Combining Eqs. function of Ft for three alternative values of (15), (17), (19), and (20), we obtain P/em, the ratio of the amount of CO2 in the atmosphere to that contained in the portion kp(Td -- r)(L + CoaT) of the mantle undergoing outgassing. Note "rr = o~(1 - o0FZt (21) that this time scale does not depend on the value of P by itself, but merely upon this 4.2. Results ratio. According to Fig. 4a, small values of In this section, we present estimates of the pressure ratio are required for rj to be the time scales, ~-j and ~'r, for resupplying comparable to the weathering time scales, the atmosphere with CO2 by juvenile out- ~'w, presented in the previous section. For gassing and by volcanic recycling of car- example, ~w equals approximately 3 × 10 7 bonate rocks. The results depend primarily years for P = 2 bars and our nominal upon the surface heat flux, Ft. Ft is useful in parameter choices. For rj to match this this role since it, in effect, provides an value at reasonable heat fluxes, the ratio energy constraint on both processes and P/Pm must be less than 0.07. This requires a since one can place good constraints on its total CO2 inventory in excess of 30 bars. value during the early history of Mars. While such a large value cannot be totally Thermal history models for Mars predict excluded, it is higher than seems likely, as that Ft for the present epoch equals 30 discussed more fully in a later section of

1011 b

10 TM :a 9010( ® 1011 ~ 10 9 2 1010 10 5

109 107 • 0.1 P/Pm "-- o 0.25 m 10 8 ..... 0.1 106 -- 0.50 ----;o ...... --- 0.75 10 ! i ~ i i i i llJ , i , , ~ i ,Jr 10710 102 103 10 102 103 SURFACE HEAT FLUX, mW/m 2 SURFACE HEAT FLUX, mW/m 2

FIG. 4. Time scale for resupplying the atmosphere with CO2 as a function of the surface heat flux from the interior. Results are shown for (a) outgassing of juvenile CO2 from the planet's interior (the three curves correspond to alternative choices for the ratio of CO2 in the atmosphere to that in the interior) and (b) a recycling mechanism, for which carbonate rocks formed by weathering are thermally decomposed after being buried by lava to a sufficient depth. The curves differ in the choice of the fraction of the heat flux carried by thermal conduction, a. A WARM AND WET MARS 215 this paper. Such a large requirement on the On Earth the amount of CO2 dissolved in volatile inventory is an almost inevitable the oceans exceeds the amount in the consequence of requiring that the atmosphere by a factor of ---60 (Holland atmosphere be resupplied with juvenile 1978). Were this same ratio to apply to COz every few times 107 years for an early Mars, this would place unrealistic extended period on the order of 109 years. demands on Mars' total CO2 inventory; Figure 4b illustrates the dependence of e.g., a 1-bar CO2 atmosphere would require the volcanic recycling time scale, ~'r, on a total COz endowment in excess of 60 bars. surface heat flux for several choices of o~, The partitioning of CO2 between the fraction of the heat flux carried by atmosphere and hydrosphere depends on conduction. "rr does not vary very much as the temperature, size, pH, and cation con- changes over the range used for this tent of the hypothesized early ocean. While figure; in fact, it has almost the same value we cannot hope to estimate these quantities for a = 0.25, 0.5, and 0.75. This is because with great precision for early Mars, it is of two compensating effects: On the one possible to make a simple model to bracket hand, increasing a decreases the amount of some of the uncertainties. Here, we assume extruded lava, but on the other hand it that the Martian ocean was isothermal (5°C) decreases the depth at which the sub- and had a salinity of 35 parts per thousand, surface temperature equals the decompo- like Earth's oceans. We further assume that sition temperature of carbonates. Both the ocean was saturated with respect to these effects operate in the same direction calcite (CaCO3) and that the dissolved Ca 2+ as Ft is varied; thus, ~r depends quadra- concentration was 1.03 × 10 -2 M (moles/ tically on Ft, as shown in Eq. (21). liter), the same as for Earth. The relevant For Ft = 100 mW/m 2, rr ranges between 1 carbonate equilibria can be written as and 3 x 108 years, while for Ft = 300 [H2CO3] = HP (22) mW/m 2, it lies between 1 and 3 × 107 years. Furthermore, as mentioned above, these [H+][HCO~-] values for ~'r represent slight underestimates K1 - [HzCO3] (23) in that thermal decomposition of car- [H+I[CO 2-1 bonates by contact with hot surface lavas K2 - [HCOj-] (24) has been neglected. Thus, the recycling time scales are comparable to the weather- Ksp =- [Ca2+][CO~-]. (25) ing time scales and offer a plausible means for maintaining a dense COz atmosphere on Equations (22)-(25) represent, respec- the planet over an extended period of time tively, CO2 solubility equilibrium ('s in its early history. law), the first and second dissociations of carbonic acid, and dissolution of calcite. 5. PARTITIONING OF CO2 BETWEEN THE The relevant constants at the specified tem- ATMOSPHERE AND HYDROSPHERE perature and salinity are (Broecker and In the previous two sections we have Peng 1982, pp. 63,151): H = 5.21 x 10-ZM argued that a dense CO2 atmosphere could atm -l, K1 = 7.17 × 10 -7 M, K2 = 4.04 × have existed on early Mars for about a 10 -l° M, and Ksp = 4.8 × 10 -7 M 2. billion years, provided that recycling of Given the CO2 pressure in the carbonate rocks was sufficiently rapid to atmosphere and the Ca 2+ content of the balance weathering of silicates. To this ocean, the hydrogen ion concentration can point, however, we have neglected the fact be derived by combining all four equations that some CO2 should have been dissolved to yield in the early oceans (or lakes) as carbonic [H+]2 HK|K2P[Ca2+] = (26) acid and as carbonate and bicarbonate ions. Ksp 216 POLLACK ET AL.

Once [H +] is known, the total dissolved The reasons are that Earth's oceans are 6 carbon concentration in the ocean (~CO 2 = times deeper than the (generous) Martian [H2CO3] + [HCO~] + [CO2+])may be paleoocean considered here and Earth's obtained from Eqs. (22)-(24): gravity is 3 times stronger. For climatically reasonable CO2 pressures of 1 to 5 bars (see ~C02 = HP 1 + [-~ 1 + Section 2), the model predicts that Mars' atmospheric CO2 reservoir is larger than (27) the oceanic reservoir by a factor of 10 to 15. This estimate is, of course, subject to con- The CO2 content of the atmosphere in siderable uncertainty because one does not moles cm -2 is given by really know the Ca 2+ concentration in Matm = 1.013 x 106p/44g, (28) Mars' early oceans. If [Ca 2+] was 10 times lower than assumed, the CO2 carrying where P is the pressure in atmospheres and capacity of the oceans would increase by g is the surface gravity (=373 cm sec -2 for about a factor of 3. The bulk of the availa- Mars). The total CO2 content of the oceans ble CO2 would, however, still remain in is given by the atmosphere. If the amount of water Moc = 102]~CO2 D, (29) on early Mars was smaller than Carr's estimate, the oceanic CO2 reservoir would where D is the ocean depth in km. As a be correspondingly less important. reasonable upper limit on ocean depth we Thus, despite the uncertainties in the use the value of 0.5 km estimated by Carr actual physical conditions on early Mars, (1987 this volume). we feel confident in asserting that most of Using the above equations and parameter the CO2 that was not bound up in carbonate choices, we have calculated several rocks was indeed present in the atmo- parameters characterizing the hypothetical sphere. Martian oceans for different atmospheric COz pressures. These results are summa- 6. DISCUSSION rized in Table I. The result of direct interest In this section, we assess the potential here is the ratio of atmospheric to dissolved likelihood and longevity of a wet, warm CO2, which is given in the last column. For period in Mars' early history, consider the an Earth-like CO2 pressure of 3.4 x 10 -4 possible time evolution of Mars' climate, atm, the model predicts an ocean pH of 8.0 compare our treatment of the CO; cycle (compared to 8.3 for Earth's oceans) and a with that of earlier papers, and present ratio matm/moc = 0.33. This ratio is smaller observational tests of the concepts put for- than that for Earth by a factor of about 20. ward in this paper.

TABLE I

PARTITIONING OF CO2 BETWEEN THE ATMOSPHERE AND HYDROSPHERE FOR A 0.5-km-DEEP OCEAN

pCO2 [H +] pH ~CO 2 Matm Moc Matm/Moc (atm) (M) (M) (moles cm 2) (moles cm 2)

3.4(-4) 1.05(-8) 7.98 1.27(-3) 2.10(-2) 6.35(-2) 0.331 0.1 1.80(-7) 6.74 2.60(-2) 6.17 1.30 4.75 1 5.69(-7) 6.24 0.118 61.7 5.90 10.5 2 8.05(-7) 6.09 0.197 123.4 9.85 12.5 5 1.27(-6) 5.90 0.408 308.6 20.4 15.1 A WARM AND WET MARS 217

6.1. Mars' Volatile Inventory For example, the amount of outgassed CO2 for Venus and Earth are very similar, yet In this section, we first define our current the abundance of several rare gases in their knowledge about the amount of CO2 that atmospheres differs by two orders of mag- was available for forming and maintaining a nitude. Such differences may result from dense CO2 atmosphere in the early history fractionation of rare gases during an early of Mars and then compare these estimates episode of hydrodynamic escape (Hunten with the requirements obtained in earlier et al. 1987). It is also possible to try to sections of this paper for sustaining a wet, reconstruct early abundances by using the warm climate. Unfortunately, the total observed enhancement of the 15N to 14N inventory of volatile elements on Mars is ratio for Mars' present atmosphere not well constrained. A straightforward, (McElroy et al. 1977), but such reconstruc- but possibly naive, estimate can be tions depend on identifying and quantifying obtained by scaling the Earth's and Venus' all relevant atmospheric loss processes, inventories, which are much better deter- including the formation of minerals through mined. Approximately 60 and 90 bars of surface weathering processes. CO2 have been outgassed over the lifetimes Mars' outgassed CO2 inventory can also of Earth and Venus, respectively (Holland be estimated in other, indirect ways. If one 1978, Pollack and Yung 1980). Much of the knew the amount of outgassed water and outgassing is likely to have occurred during assumed a terrrestrial ratio for the relative these planets' very early history. If we proportions of outgassed CO2 and H20, assume that Mars had the same volatile then one could readily obtain the desired content per unit mass of planet as did Earth value. Carr (1986) has estimated that the and Venus and that it outgassed the same water equivalent of at least a 0.5-l-km- fraction in its early history as did its neigh- deep ocean has been outgassed on the basis bors, then the amount of COz present at of geomorphic arguments. If his estimate is Mars' surface would have been about an accepted, then the corresponding amount order of magnitude smaller than for its of outgassed CO2 is about 3-6 bars. bigger neighbors. When we account for Comparisons of the abundances of differences among the three planets in sur- elements detected by the Viking X-ray face area and gravity, we infer that the experiment and the weight of the analyzed pressure equivalent of outgassed CO2 for samples showed that 6 to 8% of their con- early Mars would have been on the order stituents were not detected. This deficit can of 10 bars. A similar value for the inven- be accounted for in part, although not tory is obtained if one assumes that im- uniquely, by presuming that all the detected pacts with volatile-rich comets and aster- Ca (5% by weight) was in the form of oids served as the primary source of CO2 carbonates (Toulmin et al. 1977). If one for the terrestrial planets (Pollack and grants this and assumes that the Viking Black 1982). results are representative of the top 1 km of There have been a number of attempts to the soil, then the equivalent of 1.2 bars of define the volatile inventory of outgassed COz would be present in this reservoir CO2 for Mars by scaling the observed (Toon et al. 1980). Presumably, even more amount of rare gases in its atmosphere by carbonate would be present at greater the amounts in the atmospheres of Venus depths. and Earth. (For a summary, see Pollack In summary, the outgassed C02 volatile and Yung 1980.) Typical results are in the inventory for early Mars is not well defined, range from 0.1 to several bars. Unfortu- but values of about several to 10 bars seem nately, the assumptions or unknowns in all to be consistent with present knowledge. these estimates can be called into question. Below, we compare these values with the 218 POLLACK ET AL. abundances needed to sustain a warm, wet wet climate on early Mars could have been climate on early Mars. sustained by the volcanic recycling mech- According to the results of Section 2, anism for an extended period of time. C02 partial pressures of 0.75 to 5 bars are Furthermore, it should be remembered that needed in the early Martian atmosphere to our recycling time scales are overestimates raise the surface temperature to the melting in that they neglect the thermal decompo- point of water ice. The lower value cor- sition of carbonates by hot lavas, and our responds to conditions at the equator and at weathering time scales are underestimates perihelion for the maximum orbital when the temperature is above the freezing eccentricity and a surface albedo of 0.1. point of water in only some places or only The higher value corresponds to a globally at some times of the year. Both these and orbitally averaged situation with a sur- effects make it even easier to sustain a wet, face albedo of 0.215. Consider first a partial warm climate. pressure of 0.75 bar. As argued in Section 5, a much smaller amount of C02 was 6.2. C02 Geochemical Cycle: Other dissolved in the paleooceans and so con- Studies sideration of this second reservoir does not The possible occurrence of carbonate significantly enhance the amount of C02 rocks and their relevance for buffering the present in the combined reservoirs. For our atmospheric abundance of CO2 on Mars nominal parameter choices, the weathering was first pointed out in Harold Urey's time scale equals 1.5 x 107 years, according classical book on planets (Urey 1952). In to Section 3. If a juvenile mantle source his discussion of this problem, Urey con- of CO2 is required to replenish the sidered only the thermodynamic equilib- atmospheric COR over the first 109 years, rium situation, whereas we now suspect then a total inventory of 50 bars is implied. that exchange of gases between surface Such a requirement seems to be too rocks and the atmosphere is dominated by demanding. A juvenile source could, kinetics at the low temperatures of interest however, have sustained the required (Walker 1977, Holland 1978). Walker (1978) atmosphere for perhaps one to two hundred presented a discussion of the geochemical million years. Essentially identical results cycle of volcanism, weathering, and burial are obtained when a 5-bar atmosphere is that controls the abundance of volatiles at considered. the surface of a planet and considered the Next consider the requirements placed application of this cycle to Venus, Earth, on the volatile inventory by the volcanic and Mars. He emphasized the role of tec- recycling mechanism. For the nominal tonism in controlling the abundance of sur- parameter choices of Section 4, recycling face volatiles, such as carbon dioxide, and times of about 2 x l0 7 and 2 x 108 years are suggested that they were much more obtained for internal heat fluxes of 300 and abundant at earlier times on Mars than 100 mW/m z, respectively. Comparing these today as a result of a hotter interior in the values with the weathering time scale for a past. 0.75-bar CO2 atmosphere, we infer total Pollack and Yung (1980) discussed the inventories of about 2 and 11 bars, for the modern geochemical cycle of terrestrial higher and lower heat flux values, respec- CO2 and considered its possible application tively. The corresponding inventories for a to Mars. They, as well as Toon et al. (1980), 5-bar atmosphere are about 7 and 23 bars. suggested that a dense CO2 atmosphere Thus, the requirements on the outgassed may have existed in Mars' early history due CO2 volatile inventory placed by the vol- to a much higher outgassing rate then, but canic recycling mechanism do not seem to that the dense atmosphere would have been be excessive and it is possible that a warm, gradually eroded away by the formation of A WARM AND WET MARS 219 carbonate rocks. Pollack and Yung (1980) and presumably on early Mars, chemically also pointed out that significant additional weathered carbonates simply serve as outgassing may have taken place at later sources of new carbonate rocks, with no times in association with major volcanic net production of atmospheric CO2 (Walker events. However, both Pollack and Yung 1977, Holland 1978). He also pointed out (1980) and Toon et al. (1980) tended to the possible importance of episodic out- consider carbonate rock formation as an gassing events at later times on Mars for irreversible process for Mars, although temporally increasing the partial pressure Toon et al. (1980) pointed out the possi- of CO2. bility of extensive volcanic reprocessing of The calculations reported in this paper surface materials. represent a further step in defining the CO2 Fanale et al. (1982) were the first to geochemical cycle on Mars by providing estimate the time scale for the elimination quantitative estimates of the time scales of a CO2 atmosphere of 1-2 bars by the associated with major sources and sinks of weathering of surface rocks. Using a simple atmospheric CO2. Thus, it is possible to scaling argument based strictly on the mass assess the longevity of a dense CO2 of CO2 in the atmosphere (equivalent to atmosphere in the early history of Mars. setting our various g factors equal to 1), Fanale et al. (1982) obtained a weathering time scale of about 10 7 years, in surpris- 6.3. Climate Evolution ingly good agreement with the results of Let us assume that enough CO2 was this paper. They concluded that a dense outgassed into the early atmosphere to CO2 atmosphere would not have survived raise the surface temperature above the for an extended period of time. However, melting point of water ice over large por- as shown here, there may have been effec- tions of the planet and that this amount of tive means of resupplying the atmospheric outgassed CO2 was large enough to sustain CO2 during Mars' early history. the greenhouse warming by a volcanic Kahn (1985) proposed that the formation recycling of carbonate rocks, despite of carbonate rocks over much of the whatever losses of atmospheric gases lifetime of the planet may have led to a occurred during early times. Accepting progressive thinning of the atmosphere these assumptions for the time being, we with time and that this process may have wish to discuss the possible long-term acted as the ultimate determiner of the evolution of the planet's climate. current partial pressure of atmospheric As suggested by Walker (1978), the key CO2. He invokes transitory pockets of drive for the evolution of the climate from liquid water to effect the continual chemical the hypothesized early wet, warm climate weathering of surface rocks, after the to the dry, cold conditions of today would period when the atmospheric pressure of have been the thermal evolution of the CO2 was high enough to create an effective planet's interior. A combination of the greenhouse warming of the surface. Rates decay of long-lived radioactive isotopes, an of chemical weathering for later times on initial hot interior due to accretion, and Mars warrant further scrutiny and quan- possibly early core formation could have tification. sustained a value for the surface heat flux, Kahn (1985) also discussed the CO2 geo- Ft, above 100 mW/m2 for a time scale on the chemical cycle on Mars and suggested that order of 10 9 years after the formation of recycling of carbonate rocks might be pos- Mars (Davies and Arvidson 1981). During sible. Unfortunately, he emphasized the this epoch the recycling time scale may not weathering of carbonate rocks as a means have been much longer than the weathering of accomplishing the recycling. On Earth, time scale and hence a warm, wet climate 220 POLLACK ET AL. may have been maintained, as discussed water environments (Kahn 1985), and the above. loss of carbon and oxygen atoms to space. Ultimately, the decline of F, with time Although the climate of Mars probably would have caused the recycling process to became progressively drier and colder with become progressively less effective, time, there may have been some short-lived because of an increasing recycling time episodes of a warmer climate. These tran- scale (which varies inversely as Ft2 in our sient warming epochs could have occurred model) and because of the increasing local in association with times of intense tec- character of the volcanism. The latter con- tonism, volcanism, and outgassing. Large sideration implies that carbonate rocks amounts of CO2 may have been temporally formed at some locales would not be introduced into the atmosphere from both recycled effectively. Thus, at very early mantle sources and the thermal decompo- times, temperatures in excess of 273°K may sition of carbonate rocks that were present have occurred over much or all of the in the active regions of the planet. Quite planet during most or all seasons of the modest amounts of atmospheric CO2-- year. At somewhat later times, this benign several tens to several hundreds of mil- climate may have prevailed over more re- libars-could have sufficed to raise the stricted spatial and temporal locations, as a temperature in local regions, especially the greater fraction of the volatile inventory polar regions, above the freezing point of was tied up in carbonate rocks. These water ice (Pollack 1979, Toon et al. 1980). locations would have included lower- The requirement on the amount of latitude regions, at first during all seasons atmospheric CO2 at these later times is and later only near perihelion when the eased by the higher solar luminosity then as orbital eccentricity was large. It might have compared to its value at the early times of also included polar regions during summer primary interest for this paper. Carbonate solstice, especially at times of high orbital rock formation as well as other removal obliquity (Toon et al. 1980). Still later, processes would have depleted the at- these favored locations would have dis- mospheric CO2, ultimately eliminating appeared almost entirely, as even more the benign climate conditions. At these atmospheric CO2 was removed to the rock later times, recycling of CO2 was improb- reservoir. At this point there would have able because the geophysically active been a balance between a very sluggish rate regions encompassed only limited areas, in of carbonate decomposition and an equally contrast to their more global extent during sluggish rate of formation of new carbonate the planet's early history. rocks, because of the very restricted con- If the above scenario is correct, it has ditions under which temperatures above important implications for the possible the melting point of water ice were occurrence of . It is almost achieved. Essentially no carbonate rocks universally accepted that the exobiology should have been formed when the tem- experiments on Viking provided strong, perature fell below 273°K. Finally, other perhaps conclusive evidence against the atmospheric CO2 removal mechanisms habitation of living organisms on Mars at would have caused enough of a decline of the present epoch. We suggest that much the atmospheric reservoir to entirely elimi- more favorable climate conditions for life nate the circumstances under which tem- may have existed for an extended period of peratures above the melting point of water time during the planet's early history. If so, ice could be achieved. These include the it seems premature to abandon the exo- incorporation of adsorbed COz into a grow- biological investigation of Mars and par- ing regolith (Fanale et al. 1982), continued ticularly premature to assume that life formation of carbonates in transitory liquid never existed on this planet. A WARM AND WET MARS 221

6.4. Observational Tests Siderite is not present in large amounts in the Earth's geological record, chiefly Geomorphic studies and mineralogical because the Fe 2+ that is leached from determinations offer two potential means silicate rocks by chemical weathering is for assessing whether or not early Mars quickly converted to insoluble ferric iron ever had a wet, warm climate, as proposed by atmospheric oxygen. in this paper. Schultz's (1985) analyses of Near- and mid-infrared reflection spec- large impact basins have provided support troscopy offers a promising means for for the occurrence of a much denser detecting carbonate minerals on the surface atmosphere on early Mars (see Section 1). of Mars. Carbonates have a relatively weak Further work on the erosional history of the absorption band near 2.3-/xm wavelength Martian surface would be most useful for and a much stronger signature near 4 and 7 testing our proposal, as would searches for /zm (T. Roush, private communication). ancient lake and ocean deposits and Searches for these features could be con- boundaries. ducted during the 1988 opposition with A number of minerals would be produced Mars from the ground at coarse spatial by chemical weathering processes on a wet, resolution and at much better spatial warm Mars. These include clays, evapo- resolution by the VIMS and TES rites (e.g., gypsum), and, of course, car- instruments that will be part of the payload bonates. Kahn (1985) has previously of the forthcoming Mars Observer space- suggested that the search for the presence craft mission. of carbonate rocks would provide a means of testing his scenario for the evolution of 7. CONCLUSIONS atmospheric CO2 on Mars. Strong support In this paper, we have attempted to for our hypothesis could be provided by the define the conditions under which a wet, unambiguous detection of the above warm climate could have existed for an minerals and by the demonstration that extended period of time during the early they could not be produced by alterative history of Mars. Using improved opacity mechanisms. Firm mineralogical identifi- coefficients in a one-dimensional radiative- cations and the association of the minerals convective model, we find that CO2 partial with specific landforms, such as the old pressures ranging from 0.75 to 5 bars would cratered terrain, may provide a basis for have been needed to raise the surface tem- making this judgment. The association of perature to the melting point of water ice. carbonates with terrain of intermediate or This range of pressures reflects different young age would indicate that carbonates choices of surface albedo, latitude, and continued to be formed at later times on orbital position. In all cases, a 30% lower Mars, in accord with Kahn's suggestions solar luminosity, appropriate for early and the discussion of the previous sub- Mars, is used. Low values of the required section. We note that there is already pressure correspond to equatorial latitudes, limited evidence for the occurrence of all perihelion for large values of orbital three classes of minerals (e.g., Toulmin et eccentricity, and low surface albedo. The al. 1977). chief uncertainty in these calculations is In addition to calcite (CaCO3) and caused by the neglect of the radiative dolomite (MgCa(CO3)2), the chief car- effects of water condensation clouds. bonate minerals present in the Earth's geo- Atmospheric carbon dioxide would have logical record, we would also expect that been removed relatively quickly by the abundant quantities of siderite (FeCO3) and formation of carbonate rocks. Allowing for perhaps magnesite (MgCO3) would be pro- the dependence of the weathering rate duced when Mars had a wet, warm climate. on temperature, pressure, rock cation 222 POLLACK ET AL. abundance, and the fraction of surface area Note added in prooJ~ J. Gooding has apparently occupied by standing bodies of water, we detected trace amounts of calcium carbonate in SNC estimate that the weathering time scale for meteorites, which are believed to have originated on Mars. (This result was reported at the 1987 Lunar and the elimination of a dense (several-bar) CO2 Planetary Science Conference.) atmosphere is several times 107 years for a surface temperature of 273°K. This time scale could have varied by about an order of magnitude for plausible changes in press- REFERENCES ure, temperature, and areal coverage by ACKERMAN, T. P., K. N. Llou, AND C. B. LEOVY water. Thus, in the absence of CO2 sources, 1976. Infrared radiative transfer in polluted it would have been very difficult to sustain atmospheres. J. Appl. Meteorol. 15, 28-35. BAKER, V. R. 1978. The Spokane flood controversy a wet, warm climate for a period on the and the Martian . Science 202, order of l09 years. 1249-1256. Outgassing of juvenile CO2 from the Basaltic Volcanism Study Project 1981. Basaltic Vol- interior and thermal decomposition of car- canism on the Terrestrial Planets. Pergamon, New bonate minerals represent two possible York. BERNER, R. A., A. C. LASAGA, AND R. M. GARRELS sources of atmospheric CO2 on early Mars. 1983. The carbonate-silicate geochemical cycle and The recycling mechanism is more pro- its effect on carbon dioxide over the past 100 million mising because it imposes less of a demand years. Amer. J. Sci, 283, 641-683. on Mars' outgassed COz volatile inventory. BROECKER, W. S., AND T.-H. PENG 1982. Tracers in Global-scale volcanism could have acted as the Sea. Dougherty Geological Observa- tory, Palisades, NY. the driver for such recycling by directly BURCH, D. E., D. GRYVNAK, E. B. SINGLETON, W. L. decomposing carbonates through contact FRANCE, AND D. WILLIAMS 1962. Infrared with hot lava or by burying carbonate Absorption by Carbon Dioxide, Water Vapor, and sediments to a depth where they could have Minor Atmospheric Constituents. Air Force Cam- been thermally decomposed. The time bridge Research Laboratory Report 62-698. CARR, M. H. 1986. Mars: A water-rich planet. In scale for this latter process can be MECA Symposium Abstracts. July 17-19, 1986, estimated by relating the rate of volcanism Washington, D.C.; h'arns 68, 187-216. to the surface heat flux from the interior. CESS, R. D., V. RAMANATHAN, AND T. OWEN 1980. This time scale is insensitive to the par- The Martian paleoclimate and enhanced carbon titioning of the surface heat flux between dioxide. Icarus 41, 159-165. CLARK, B. C., AND A. K. BAIRD 1979. IS the Martian conductive and volcanic transport, but lithosphere sulphur rich? J. Geophys. Res. 84, varies inversely as the square of the total 8395-8403. heat flux. For values of the heat flux lying CLARK, B. C., et al. 1977. The Viking X-ray between 100 and 300 mW/m 2, values fluorescence experiment, analytical methods and typical of early Mars, the recycling time early results. J. Geophys. Res. 82, 4577-4594. CLOY, G. D. 1984. Radiation-dominated snowmelt on scale lies between about 1 × l0 7 and 3 × Mars. In MECA Workshop on , pp. 108 years. Under such circumstances, 15-16. recycling could have maintained a dense DAVIES, G. F., AND R. E. ARVIDSON 1981. Martian CO2 atmosphere if the available volatile thermal history, core segregation, and tectonics. inventory was between 2 and 20 bars. Icarus 45~ 339-346. EHLERS, E., AND H. BLATT 1982. Petrology: Igneous, Continued studies of the geomorphology Sedimentary, and Metamorphic. Freeman, San of the old cratered terrain and spectrosco- Francisco. pic searches for carbonate minerals offer FANALE, F. P., J. R. SALVAIL, W. B. BANERDT, AND means of testing the hypotheses of this R. S. SAUNDERS 1982. Mars: The regolith- paper. If Mars did have a wet, warm cli- atmosphere-cap system and climate change. Icarus 50, 381-407. mate during the first billion or so years, the FEES, S. B. 1979. Simple strategies for the inclusion of possibility that life arose during this time Voight effects in infrared cooling calculations. Appl. needs to be seriously considered. Optics 18, 2634-2637. A WARM AND WET MARS 223

GOUGH, D. O. 1981. Solar interior structure and LACIS, A. A., AND J. E. HANSEN 1974. A parameteri- luminosity variations. Solar Phys. 74, 21-34. zation for the absorption of solar radiation in the HARKER, R. I., AND O. F. TUTTLE 1956. Experimental earth's atmosphere. J. Atmos. Sci. 31, 118-133. data on the PcO2-T curve for the reaction: calcite + LAGACHE, M. 1965. Contribution a letude de l'al- quartz ~ wollastonite + carbon dioxide. Amer. J. teration des feldspaths, dans l'eau, entre 100 et Sci. 254, 239. 200°C, sous diverses pressions de CO2, et appli- HARMON, R. S., W. B. WHITE, J. J. DRAKE, AND J. cation a la synthese des mineraux argileux. Bull. Fr. W. HESS 1975. Regional hydrochemistry of North Miner. Crist. 88, 223-253. America carbonate terrains. Water Resources Res. LAGACHE, M. 1976. New data on the kinetics of the 11, 963-967. dissolution of alkali feldspars at 200°C in CO: char- HERZBERG, G., AND L. HERZBERG 1953. Rotation- ged water. Geochim. Cosmochim. Acta 40, vibration spectra of diatomic and simple polyatomic 157-161. molecules with long absorbing paths. XI. The spec- MANABE, S., AND R. J. STOUFFER 1980. Sensitivity of trum of carbon dioxide (CO:) below 1.25 txm. J. Opt. a global climate model to an increase of CO2 conce- Soc. Amer. 43, 1037-1044. ntration in the atmosphere. J. Geophys. Res. 85, HOFFERT, M. I., A. J. CALLEGARI, C. T. HS1EH, AND 5529-5554. W. ZIEGLER 1981. Liquid water on Mars: An energy MANABE, S., AND R. T. WETHERALD 1967. Thermal balance climate model for C02/H20 atmospheres. equilibrium of the atmosphere with a given distri- Icarus 47, 112-129. bution of relative humidity. J. Atmos. Sci. 24, HOLLAND, H. D. 1978. The Chemistry of the 241-259. Atmospheres and Oceans. Wiley-Interscience, MASURSKY, H. 1973. An overview of geological results New York. from Mariner 9. J. Geophys. Res. 78, 4009-4030. HUNTEN, D. M., R. O. PEPIN, AND J. C. G. WALKER MCCLATCHEY, R. A., R. W. FENN, J. E. A. SELBY, F. 1987. Mass fractionation in hydrodynamic escape. E. VOLZ, AND J. S. GARING 1971. Optical Proper- Icarus 69, 532-549. ties of the Atmosphere. AFCRL Report 71-0279, INGERSOLL, A. P. 1969. The runaway greenhouse: A Hanscom AFB, MA. history of water on Venus. J. Atmos. Sci. 26, MCELROY, M. B., T. Y. KONG, AND Y. L. YUNG 1191-1198. 1977. Photochemistry and evolution of Mars' INGERSOLL, A. P. 1970. Mars: Occurrence of liquid atmosphere: A Viking perspective. J. Geophys. water. Science 168, 972-973. Res. 82, 4379-4388. IZATT, J. R., H. SAKAMI, AND W. S. BENEDICT 1969. MEADOR, W. E., AND W. R. WEAVER 1980. Two- Positions, intensities, and widths of water vapor stream approximations to radiative transfer in lines between 475 and 892 cm -I. J. Opt. Soc. Amer. planetary atmospheres: A unified description of 59, 19-26. existing methods and a new improvement. J. Atmos. KAHN, R. 1985. The evolution of C02 on Mars. Icarus Sci. 37, 630-643. 62, 175-190. MOORE, C. E., M. G. J. MINNAERT, AND J. HOUT- KASTING, J. F. 1987. Runaway and moist greenhouse GAST 1966. The Solar Spectrum 2935 ,~ to 8770 A; atmospheres and the evolution of Earth and Venus. Second Revision of Rowland's Preliminary Table of Submitted for publication. Solar Spectrum Wavelengths. Nat. Bur. Stand. KASTING, J. F., AND T. P. ACKERMAN 1986. Climatic Monograph 61. U.S. Govt. Printing Office, consequences of very high C02 levels in Earth's Washington, D.C. early atmosphere. Science 234, 1383-1385. NEWMAN, M. J., AND R. T. ROOD 1977. Implications KASTING, J. F., J. B. POLLACK, AND T. P. ACKERMAN of solar evolution for the Earth's early atmosphere. 1984a. Response of Earth's atmosphere to increases Science 198, 1035-1037. in solar flux and implications for loss of water from PEALE, S. J., G. SCHUBERT, AND R. E. LINGEN- Venus. Icarus 57, 335-355. FELTER 1975. Origin of Martian channels: Clathrates KASTING, J. F., J. B. POLLACK, AND D. CRISP 1984b. and water. Science 187, 273-274. Effects of high CO: levels on surface temperature PEPIN, R. O. 1985. Volatile inventory of Mars. In and atmospheric oxidation state of the early Earth. MECA Workshop on the Evolution of the Martian J. Atmos. Chem. 1, 403-428. Atmosphere, pp. 20-21. KASTING, J. F., S. M. RICHARDSON, J. B. POLLACK, PIERI, D. C. 1980. Martian valleys: morphology, dis- AND O. B. TOON 1986. A hybrid model of the C02 tributions, age, and origin. Science 210, 895-897. geochemical cycle and its application to large impact POLLACK, J. B. 1979. Climatic change on the ter- events. Amer. J. Sci. 286, 361-389. restrial planets. Icarus 37, 479-553. KIEFFER, H. H., T. Z. MARTIN, A. R. PETERFREUND, POLLACK, J. B., AND D. C. BLACK 1979. Implications B. M. JAKOSKY, E. D. MINER, AND F. D. PALLU- of the gas composition measurements of Pioneer CONI 1977. Thermal and albedo mapping of Mars Venus for the origin of planetary atmospheres. during the Viking primary mission. J. Geophys. Res. Science 205, 56-59. 82, 4249-4291. POLLACK, J. B., AND D. C. BLACK 1982. Noble gases 224 POLLACK ET AL.

in planetary atmospheres: Implications for the origin UREY, H. C. 1952. The Planets, Their Origin and and evolution of atmospheres. Icarus 51, 169-198. Development. Yale Univ. Press, New Haven, CT. POLLACK, J. B., AND Y. L. YUNG 1980. Origin and VARANASI, P. 1971. Line widths and intensities in evolution of planetary atmospheres. Annu. Rev. H20-COz mixtures. II. High resolution mea- Earth Plan. Sci. 8, 425-487. surements on the u2-fundamental of water vapor. J. POLLACg, J. B., O. B. TOON, AND R. BOESE 1980. Quant. Spectrosc. Radiat. Transfer 11, 223-230. Greenhouse models of Venus' high surface tempera- ture, as constrained by Pioneer Venus mea- VARDAVAS, I. M., AND J. H. CARVER 1984. Solar surements. J. Geophys. Res. 85, 8223-8231. and terrestrial parameterizations for radiative- POSTAWKO, S. E., AND W. R. KUHN 1986. Effect of convective models. Plan. Space Sci. 32, 1307-1325. the greenhouse gases (COz, H20, SO2) on Martian WALKER, J. C. G. 1977. Evolution of the Atmosphere. paleoclimate. Proc. Lunar Planet. Sci. Conf. 16th, Macmillan Co., New York. J. Geophys. Res. 91, D431-D438. WALKER, J. C. G. 1978. Atmospheric evolution on the ROBERTS, R. E., J. E. A. SELBY, AND L. M. RIBER- inner planets. In Comparative Planetology, pp. MAN 1976. Infrared continuum absorption by 141-163. Academic Press, New York. atmospheric water vapor in the 8-12/zm window. WALKER, J. C. G., P. B. HAYS, AND J. F. KASTING Appl. Opt. 15, 2085-2090. 1981. A negative feedback mechanism for the long RoNov, A. B., AND A. A. YAROSHEVSKY 1969. Che- term stabilization of Earth's surface temperature. J. mical composition of the Earth's crust. In The Geophys. Res. 86, 9776-9782. Earth's Crust and Upper Mantle (P. J. Hart, Ed.). WANKE, H., AND G. DREIBUS 1985. Can a high Amer. Geophys. Union Monograph 13. abundance of moderately volatile and even some SCHULTZ, P. H. 1985. The Martian atmosphere before highly volatile elements on Mars be reconciled with and after the Argyre impact. In MECA Workshop on its low abundance of primordial rare gases? In the Evolution of the Martian Atmosphere, pp. MECA Workshop on the Evolution of the Martian 22-23, Atmosphere, pp. 27-28. SHARP, R. P., AND M. C. MAL1N 1975. Channels on Mars. Geol. Soc. Amer. Bull. 86, 593-609. WARD, W. R. 1974. Climatic variations on Mars. 1. STACEV, F. D. 1980. The cooling Earth: a reappraisal. Astronomical theory of insolation. J. Geophys. Res. Phys. Earth Planet. Inter. 22, 89-96. 79, 3375-3386. TOON, O. B., J. B. POLLACK, W. WARD, J. A. BURNS, WASSERBURG, G. J., D. A. PAPANASTASSIOU, F. AND K. BILSKI 1980. The astronomical theory of TERA, AND J. C. HUNEKE 1977. Outline of a lunar climatic change on Mars. Icarus 44, 552-607. chronology. In The Moon, A New Appraisal, pp. TOULMIN, P., III, A. K. BAIRD, B. C. CLARK, K. 7-22. Royal Society, London. KEIL, H. J. ROSE, JR., R. P. CHRISTIAN, P. H. WATKINS, G. H., AND J. S. LEWIS 1985. Evolution of EVANS, AND W. C. KELLIrtER 1977. Geochemical the atmosphere of Mars as the result of asteroidal and minerological interpretation of the Viking and cometary impacts. In MECA Workshop on inorganic chemical results. J. Geophys. Res. 82, the Eoolution of the Martian Atmosphere, pp. 4625-4634. 29-30.