Icarus 220 (2012) 144148

Contents lists available at SciVerse ScienceDirect

Icarus
journal homepage: www.elsevier.com/locate/icarus

An impact-induced, stable, runaway climate on Mars

Teresa L. Segura a,, Christopher P. McKay b, Owen B. Toon c
a

Space Systems/Loral, 3825 Fabian Way, G-60, Palo Alto, CA 94303, USA NASA Ames Research Center, MS 245-3, Moffett Field, CA 94035, USA c Department of Atmospheric and Oceanic Sciences & Laboratory for Atmospheric and Space Physics, University of Colorado, Campus Box 600, Boulder, CO 80303, USA
b

a r t i c l e

i n f o

a b s t r a c t
Large asteroid and comet impacts on Mars, such as the one that formed the Argyre basin, delivered considerable amounts of kinetic energy to the planet and raised the surface temperature hundreds of degrees. The impact that formed the Argyre basin occurred 3.83.9 byr ago (Werner, S.C. [2008]. Icarus 195, 4560; Fassett, C.I., Head, J.W. [2011]. Icarus 211, 12041214), during the time of formation of uvial features on the early martian surface, and was capable of causing global-scale precipitation and warming of the surface. Dual solutions to the climate of early Mars, one cold like present Mars and the other in a hot runaway state, exist for the pressure range of 0.0061 bar of CO2, and for water inventories 6.5 bars or greater. A large impact event may have pushed Mars to a long-lasting hot runaway state. The runaway state would persist until escape processes reduced water vapor and forced the planet to return to a cold climate. 2012 Elsevier Inc. All rights reserved.

Article history: Received 9 January 2012 Revised 26 March 2012 Accepted 14 April 2012 Available online 2 May 2012 Keywords: Mars, Climate Mars, Atmosphere Atmospheres, Evolution Terrestrial planets

1. Introduction The largest craters that are still visible on Mars formed about 3.74.1 byr ago, corresponding to the same time frame during which many of the uvial martian Valley Networks formed (Carr and Head, 2010). For example, the Argyre basin is thought to be 3.83.9 byr old (Werner, 2008; Fassett and Head, 2011, respectively) and could have been formed by a 100200 km diameter object, based upon the craters 1140 km diameter and using equations and varying assumptions found in Melosh (1989). The kinetic energy in this object would have been on order of 1026 J. The delivery of this energy to the martian system would have heated the surface and atmosphere signicantly (Segura et al., 2002, 2008). Here we consider whether this change in temperature would lead to a transient alteration of the climate (as described in Segura et al. (2008)) or could have pushed Mars into a long-term stable warm state reminiscent of the runaway state. Ingersoll (1969) and Kobayashi (1967) showed that as the temperature increased at the base of a purely radiative atmosphere with a constant relative humidity of water vapor, a point would be reached where the ux leaving the top of the atmosphere would decline with increasing basal (base of the atmosphere) temperature. They identied this as the point beyond which a runaway greenhouse would occur. Nakajima et al. (1992) further investigated the runaway greenhouse and showed that it also applies to
Corresponding author. Fax: +1 310 813 3395.
E-mail addresses: segura.teresa@ssd.loral.com (T.L. Segura), christopher.p. mckay@ nasa.gov (C.P. McKay), btoon@lasp.colorado.edu (O.B. Toon). 0019-1035/$ - see front matter 2012 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.icarus.2012.04.013

a radiativeconvective atmosphere over a surface. We illustrate this effect in a simplied way with a gray model of a purely radiative atmosphere. Fig. 1 shows computations of FIRtop for a given basal temperature given the following equations:

rT 4 b FIRtop 1=22 3=2s

where FIRtop is outgoing infrared radiative ux at the top of the atmosphere, s is the total gray (wavelength-independent) opacity of all absorbing species in the atmosphere, r is the Stefan Boltzmann constant, and Tb is the basal temperature. If we assume that water vapor is uniformly mixed, that the optical depth is dominated by water vapor, and that sH2O varies with vapor pressure following a simple power law, then:

sH2 O kPn H2 O

where PH2 O , is the vapor pressure of water, computed from the ClausiusClapeyron relation: PH2O = P0eB/Tb, where B = L/R (the latent heat of vaporization over the universal gas constant), Tb is the basal temperature, and P0 is the reference vapor pressure computed at the triple point for water. When k = 0.1 and n = 1, Fig. 1 (when rotated clockwise 90) reproduces Fig. 2 in Nakajima et al. (1992). The gure shows that given unlimited availability of water, for solar uxes below the computed runaway ux of 385 W/m2, two temperature solutions exist for one given planetary ux: one at cool temperature and one at a warm temperature. The warm temperature solution is a non-stable equilibrium state and the temperature will rise until the supply of water on the surface is exhausted by evaporation into the steam atmosphere, hence the term runaway. Thus, when the amount of water available is nite there are three

T.L. Segura et al. / Icarus 220 (2012) 144148

145

Fig. 1. Temperature at the base of a purely radiative atmosphere as a function of the outgoing IR ux at the top of the atmosphere. The solution of Eqs. (1) and (2) shows that for uxes below the Runaway point (marked RA at the turnover point in the curve), two solutions for temperature may exist. However, the upper branch (indicated by the dotted line) is typically considered to be physically unrealistic (Nakajima et al., 1992).

solutions: a cool stable state with water as liquid on the surface (low temperature solutions in Fig. 1), a warm unstable solution (high temperature solutions in Fig. 1), and a hot stable (runaway) state with all the water as vapor in the atmosphere. This hot stable state is not shown in Fig. 1, but will be illustrated in later gures.

2. Numerical computations of uxtemperature space We numerically explored the multiple solutions of the ux temperature phase space, suggested by the gray model equations above, for Mars using a non-gray time-dependent radiative convective 1-D numerical model. The mechanics of the numerical model are described in Section 3 of Segura et al. (2008) and in the Supporting Online Material of Segura et al. (2002). The non-gray radiation code is based on the correlated-k method, with k coefcients derived from line-by-line calculations done at the Ames Research Center for their General Circulation Model (see http:// spacescience.arc.nasa.gov/mars-climate-modeling-group/brief.html for a brief summary) and the radiation code developed by Toon et al. (1989). Radiative heating rates using these coefcients were discussed by Haqq-Misra et al. (2008). Colaprete and Toon (2000, 2003) simplied the k-coefcients to treat overlap more efciently, and we use the same approach for this study. In summary, the modeled atmosphere is divided into 50 layers of equivalent fraction of pressure between adjoining layers, and the model computes the addition (via surface evaporation) and removal (via surface precipitation) of atmospheric water at the surface at each time step. Potential temperature is computed via both radiative transfer and vertical diffusion, and used to compute the atmospheric temperature, pressure, and gas masses at every time step. Water may condense and evaporate within the atmosphere, may evaporate into the atmosphere from the surface, and may precipitate out of the atmosphere when it is saturated. When evaporation occurs, water (as vapor) is added to the atmospheric column. The surface pressure is computed for the added water mass and the pressures at the remaining atmospheric layers are computed as equal fractions of the surface

pressure to preserve hydrostatic equilibrium. The radiative effects of clouds were not included in the model. As initial conditions for this numerical exploration of the multiple solutions shown in Fig. 1, the simulated Mars was assumed to have a surface pressure of 6.1 mbar of CO2 and a total water inventory of 500 mbar (a global equivalent layer of 13.3 m). We determined the equilibrium surface temperature on Mars for a given net incoming solar ux. The incoming solar energy is (1 A)S/4, where A is the albedo (a function of pressure due to Rayleigh scattering) and S is the solar constant at the orbit of Mars (currently 600 W/m2). The quantity [(1 A)S/4], which we call the available incoming solar ux, was varied from 60 to 280 W/m2. The planetary albedo changed as the pressure changed, due to Rayleigh scattering and absorption in the near infrared by the water vapor (Fig. 2). Equilibrium uxtemperature points were found by systematically choosing increasing incoming solar uxes, and next

Fig. 2. Albedo as a function of temperature for model runs presented here.

146

T.L. Segura et al. / Icarus 220 (2012) 144148

Fig. 3. Comparison of the analytic solutions with the detailed numerical computations. Triangles show the equilibrium temperature and ux computed for a cool start while squares show the equilibrium temperature and ux computed for a warm start. Our choice of water inventory, in this case 500 mbar, sets point L which is the uxtemperature point at which the planet will move to the runaway branch (Curve B see Section 3). Results are irregular near region L because this represents the boundary of the unstable region and therefore may not lie on the stability curve as predicted. The asterisk marks the ux at Mars today, about 105 W/m2.

choosing two different starting temperatures for each ux. One case, referred to as a warm start, approached equilibrium starting with a high initial temperature and the entire 500-mbar water inventory in the atmosphere. This then would simulate Mars cooling from a large temperature perturbation. The other case, referred to as cool start, approached equilibrium starting with a cool initial temperature, and 100% water vapor saturation in the atmosphere. In the cool start, all 500 mbar of the water inventory were assumed to reside in a pool at the surface, available to facilitate evaporation to and precipitation from the atmosphere. When humidity in the atmosphere exceeded 100%, precipitation occurred, and the surface humidity was held at 100% by exchange with the 500-mbar pool of surface water. There was no transfer of CO2 into or out of the atmosphere, and the CO2 pressure was held xed at 6.1 mbar. For each case, the model was run out to equilibrium, which was dened as the point when the net infrared ux at the top of the atmosphere reached 12% of the net solar ux at the top of the atmosphere (planetary ux equilibrium). Since the model was run out to steady state, latent heating in the atmosphere and at the surface was ignored. Model-computed global average temperatures for various values of solar ux are shown in Fig. 3. Warm starts (squares) cooled down, condensing the initially inserted water vapor, until the cool equilibrium temperature state was reached. Cool starts (triangles) warmed to the warm equilibrium temperature state and maintained 100% humidity through exchange of water at the surface. Below 150 W/m2, the cool and warm starts yielded the same equilibrium temperature along Curve A, with the atmosphere at 100% relative humidity. Here, Fig. 3 shows that the outgoing infrared ux increased with surface temperature in a monotonic way. This is reminiscent of the stable lower part of the curve in Fig. 1. Past about 270 W/m2 the warm and cool starts again reach an identical equilibrium temperature along Curve B, and the outgoing infrared ux is monotonic with surface temperature. In the range 150 to 180 W/m2 net ux, the equilibrium temperatures reached by the warm and cool starts begin to diverge. In the range 180250 W/ m2, the warm and cool starts reach completely different equilibrium temperatures. 3. Discussion of non-gray numerical model results Fig. 3 shows the equilibrium points of the warm and cold model starts in uxtemperature space. These points in Fig. 3 exhibit the behavior of a radiativeconvective, non-gray model. Note that the

turnover or runaway point in this curve (marked RA) is about 270 W/m2, rather than 385 W/m2 as was the case in Fig. 1. The inclusion of non-gray effects prevents a simple comparison of the runaway point between Fig. 1, a radiative equilibrium model, and Fig. 3, a radiativeconvective equilibrium model, but a lower value for the radiativeconvective model is expected for two reasons. First, as shown by Nakajima et al. (1992), Section 3b, a radiative convective model gives a lower value of the outgoing infrared ux for a saturated atmosphere. In addition, the ambient greenhouse produced by the CO2 pressure of 6.1 mbar moves the runaway point to the left toward lower ux (Pujol and North, 2003). Within a demonstrated range of ux (180250 W/m2 in the example in Fig. 3), two different stable numerical solutions, one warm and one cool were found. This can be understood by analogy to the three analytical solutions, two stable and one unstable, discussed in the context of Fig. 1. Curves A (solid) and C (dashed) of Fig. 3 represent analogous situations to the solid and dashed curves in Fig. 1. Curve A is the stable, lower part of the curve in Fig. 1 (solid) while Curve C is the unstable upper part of the curve in Fig. 1 (dashed). Curve B in Fig. 3 is dened using Eq. (2) for the xed water inventory of 500 mbar and represents the runaway branch where the entire inventory of water moves into the atmosphere. The dashed Curve C represents an instability boundary above which the system goes to the upper solution (Curve B runaway stable branch) and below which the system returns to the lower solution (Curve A cool stable branch). To test the possibility that Curve C is unstable, we ran a case where the net solar ux was 205 W/m2 and the starting temperature was 320 K. Fig. 3 shows that this point is between Curves A and B, but below Curve C. The equilibrium temperature for this run was 263 K. This is the same equilibrium temperature realized for the cool start run at 205 W/m2 and 230 K (marked by the triangle near 200 W/m2), which illustrates that the simulated Mars, perturbed to 320 K, did indeed move down to the cool stable Curve A. In this case, the temperature perturbation did not lead to a runaway climate because not enough water was available to maintain the runaway. We performed another test where the net solar ux was 235 W/m2 and chose two starting temperatures, one at 340 K and one at 310 K, to straddle Curve C in temperature space. We found that for the warm start the resultant equilibrium temperature was 365 K, and for the cool start the resultant equilibrium temperature was 280 K. This illustrates that Curve C indeed demarks the transition between the two stable climate states. Therefore, a temperature perturbation must increase the temperature past the boundary dened by Curve C to result in an equilibrium temperature on the runaway branch, Curve B. Perturbations that warm the temperature to values below this boundary will eventually cool to the stable cold equilibrium state dened by Curve A.

4. Implications for early Mars 4.1. Varying water inventory In Fig. 3, the asterisk marks the solar ux at Mars today, about 105 W/m2, but lower ux values correspond to the incoming solar ux during the Noachian when it may have been 75% of its present value, or 80 W/m2. This also corresponds to the era when many of the martian Valley Networks were formed (Carr and Head, 2010; Fassett and Head, 2011). The lower ux limit for which a dual stable solution exists depends on the total global water inventory. Fig. 4 shows how the runaway branch (Curve B in Fig. 3) end points can vary for different global inventories of water. From Fig. 4, if the available solar ux was 80 W/m2, the equilibrium temperature was roughly 200 K. If the martian global water inventory was greater

T.L. Segura et al. / Icarus 220 (2012) 144148

147

with 1 bar of CO2 and 6.5 bars of water, and shows that for a solar ux of 80 W/m2, Mars would have been in the region where dual solutions exist. Hence Figs. 4 and 5 indicate that dual solutions to the climate of Mars exists for the range from 0.006 to 1 bar of CO2, for water inventories 6.5 bars or greater in the Noachian. Comparing Figs. 4 and 5 illustrates that the amount of water needed on Mars to have the warm temperature solution is not a strong function of the CO2 abundance. 5. Impact-induced temperature perturbation As shown previously, if the system is perturbed in temperature to a value above the dashed curves in Figs. 4 and 5, it will end up on the upper (runaway) stable solution branch and will stay on this upper branch until water loss (for example, hydrodynamic escape to space) reduces the water inventory such that the upper solution is no longer viable. The system would then collapse to the lower stable solution, albeit with much less water than it started. Temperature perturbations caused by impact events such as Argyre would have been capable of pushing Mars to the upper solution because of the large delivery of kinetic energy and the probable large inventory of water on Mars. Fig. 5 shows that a perturbation equivalent to 150 K would be sufcient to increase the temperature past the dashed curve, and the model results presented in Fig. 3 indicate Mars would warm to the upper runaway state. The energy required for this temperature perturbation may be computed by solving for the energy balance: Ereq = MatmCp dT + L dMwater, which is dominated by the second term. If dT is 150 K, Cp is 8.8 102 m2/s2/K, and L is 2.83 106 J/kg then using this equation for 6.5 bars of water, the impact event would need to provide at least 1026 J, if the impactor operated at roughly 25% efciency (Sleep et al., 1989). This corresponds to an impacting object of about 170 km in diameter, or the size of the object that created the Argyre basin, depending on assumptions for impactor/target density, and impactor velocity (Melosh, 1989). 6. Conclusions The results presented here show that impacts like the one that created the Argyre basin could have perturbed Mars to a long-term greenhouse state where most of the water is in the atmosphere and the temperature was very warm but stable (Curve B in Fig. 3). While liquid water would not be present on the surface in such a state, because it would evaporate into the atmosphere and add to the runaway, liquid water could have been present in the subsurface if its transport to the atmosphere was slow. The investigation here focused on the feasibility of an impactinduced runaway rather than attempting to predict the duration of the greenhouse. The duration may have been limited by the escape rate of hydrogen. However, our use of a radiativeconvective model is a rst step in understanding the possible runaway greenhouse. Of course there are many limitations to radiativeconvective models (Pierrehumbert, 2010), and more sophisticated three-dimensional simulations would be valuable. It is possible that the duration of the state, and the amount of water needed to maintain it, depend on dynamical effects that a radiativeconvective model cannot consider. In addition, we did not consider the radiative effects of clouds and dust, which could be treated in a more sophisticated modeling framework. As the greenhouse collapsed, perhaps by hydrodynamic escape of hydrogen, and Mars cooled to the lower branch, precipitation would have occurred and could have formed the uvial features imaged on Mars today (Segura et al., 2008). Or they may have been formed as below ground water warmed, and then owed across the surface before evaporating into the runaway atmosphere.

Fig. 4. Fluxtemperature space for varying water inventories on Mars assuming an atmospheric inventory of 6 mbar of CO2. The lower limit of incoming solar ux for which a double solution exists depends on the total water inventory.

than 500 mbar, the ux point at which dual solutions exist moves to the left. Indeed, if the early martian global water inventory was about 9 bars, then for an available solar ux of 80 W/m2, the simulated Mars would be in the ux range where dual stable solutions exist. From Fig. 4, the two solutions are the cold equilibrium state of 200 K, and the warm equilibrium state (runaway) of about 440 K. If the global martian water inventory was less than 9 bars with the available ux of 80 W/m2, Mars would have been in the ux range where only the cold equilibrium state solution exists. Should Mars have been perturbed in temperature here, it would have cooled to the cold equilibrium state of 200 K. Mars must be both in the range where dual solutions exist, and be cooling from a temperature perturbation greater than the Curve C instability boundary to end up on the runaway branch Curve B.

4.2. Varying CO2 inventory The atmospheric inventory of CO2 on early Mars was probably greater than it is today, and the total inventory of water is certainly greater than 500 mbar. Clifford and Parker (2001) estimate that the Noachian cryosphere could have stored the equivalent of <85170 m of water as ground ice, with a total global water inventory of 5501400 m. If we take the value of 170 m of available subsurface ice computed by Clifford and Parker (2001) as the inventory of water for early Mars, this is the equivalent of 6.5 bars, given martian values of surface area and gravity. If we assume early Mars had 1 bar of CO2, and 6.5 bars of water, the runaway point would occur at a lower solar ux (Pujol and North, 2003). Fig. 5 illustrates the uxtemperature space for Noachian Mars

Fig. 5. The uxtemperature space representative of Noachian Mars, with 1 bar CO2 and 6.5 bars of water. For a Noachian Mars ux [=(1 A)S/4] value of 80 W/m2 (indicated by the asterisk), the planet is in the region where dual solutions exist and, if perturbed enough in temperature, could be pushed to the upper solution. In this computation the RA point has moved to lower available uxes than in Fig. 4 because of the assumed 1 bar of CO2.

148

T.L. Segura et al. / Icarus 220 (2012) 144148 Ingersoll, A.P., 1969. The runaway greenhouse: A history of water on Venus. J. Atmos. Sci. 26, 11911198. Kobayashi, M., 1967. Discrete equilibrium temperatures of a hypothetical planet with the atmosphere and the hydrosphere of one component-two phase system under constant solar radiation. J. Meteorit. Soc. Jpn. 45, 137139. Melosh, H.J., 1989. Scaling of crater dimensions. In: Impact Cratering A Geologic Process. Oxford University Press, New York, pp. 112125. Nakajima, S., Hayashi, Y.-Y., Abe, Y., 1992. A study on the runaway greenhouse effect with a one dimensional radiativeconvective equilibrium model. J. Atmos. Sci. 49, 22562266. Pierrehumbert, R.T., 2010. Principles of Planetary Climate. Cambridge University Press, Cambridge, p. 652. Pujol, T., North, G.R., 2003. Analytical investigation of the atmospheric radiation limits in semigray atmospheres in radiative equilibrium. Tellus A 55 (4), 328 337. Segura, T.L., Toon, O.B., Colaprete, A., Zahnle, K.J., 2002. Environmental effects of large impacts on Mars. Science 298, 19771980. Segura, T.L., Toon, O.B., Colaprete, A., 2008. Modeling the environmental effects of moderate-sized impacts on Mars. J. Geophys. Res. 113, E11007. http:// dx.doi.org/10.1029/2008JE003147. Sleep, N.H., Zahnle, K.J., Kasting, J.F., Morowitz, H.J., 1989. Annihilation of ecosystems by large asteroid impacts on the early Earth. Nature 342, 139142. Toon, O.B., McKay, C.P., Ackerman, T.P., Santhanam, K., 1989. Rapid calculation of radiative heating rates and photodissociation rates in inhomogeneous multiple scattering atmospheres. J. Geophys. Res. 94, 1628716301. Werner, S.C., 2008. The early martian evolutionConstraints from basin formation ages. Icarus 195, 4560.

Related large impact events such as Isidis and Hellas could also have caused this perturbation to the greenhouse climate state. In this case the Argyre impact event might have ended the runaway climate state by eroding away the water by impact erosion. Once enough water was lost by impact erosion or other escape mechanisms, such that Mars could no longer remain on the runaway branch, the global inventory of martian water accessible to the atmosphere would be set for the rest of time. References
Carr, M.H., Head, J.W., 2010. Geologic history of Mars. Earth Planet. Sci. Lett. 294, 185203. Clifford, S.M., Parker, T.J., 2001. The evolution of the martian hydrosphere: Implications for the fate of a primordial ocean and the current state of the northern plains. Icarus 154, 4079. Colaprete, A., Toon, O.B., 2000. The radiative effects of martian water ice clouds on the local atmospheric temperature prole. Icarus 145, 524532. Colaprete, A., Toon, O.B., 2003. Carbon dioxide clouds in an early dense martian atmosphere. J. Geophys. Res. 108 (E4), 5025. http://dx.doi.org/10.1029/ 2002JE001967. Fassett, C.I., Head, J.W., 2011. Sequence and timing of conditions on early Mars. Icarus 211, 12041214. Haqq-Misra, J.D., Domagal-Goldman, S.D., Kasting, J.F., Kasting, P., 2008. A revised, hazy methane greenhouse for the Archean Earth. Astrobiology 8. http:// dx.doi.org/10.1089/ast.2007.0197.