iCARUS 44, 552--607 (1980)

The Astronomical Theory of Climatic Change on Mars

OWEN B. TOON AND J A M E S B. POLLACK
Theoretical and Planetary Studies Branch, Ames Research Center, NASA, Moffett Field, California 94035

WILLIAM WARD
Jet Propulsion Laboratory, Pasadena, California 91103

JOSEPH A. BURNS
Theoretical and Applied Mechanics Department, Cornell University, Ithaca, New York 14853

AND

KENNETH BILSKI
lnformatics, Incorporated, Palo Alto, California 94303

Received May 8, 1980; revised October 21, 1980

We examine the response of Martian climate to changes in solar energy deposition caused by
variations of the Martian orbit and obliquity. We systematically investigate the seasonal cycles of
carbon dioxide, water, and dust to provide a complete picture of the climate for various orbital
configurations. We find that at low obliquity (15°) the atmospheric pressure will fall below 1 mbar;
dust storms will cease; thick permanent CO2 caps will form; the regolith will release CO2; and H20
polar ice sheets will develop as the permafrost boundaries move poleward. At high obliquity (35°)
the annual average polar temperature will increase by about 10°K, slightly desorbing the polar
regolith and causing the atmospheric pressure to increase by not more than 10 to 20 mbar. Summer
polar ground temperatures as high as 273°K will occur. Water ice caps will be unstable and may
disappear as the equilibrium permafrost boundary moves equatorward. However, at high eccen-
tricity, polar ice sheets will be favored at one pole over the other. At high obliquity dust storms may
occur during summers in both hemispheres, independent of the eccentricity cycle. Eccentricity and
longitude of perihelion are most significant at modest obliquity (25°). At high eccentricity and when
the longitude of perihelion is close to the location of solstice hemispherical asymmetry in dust-
storm generation and in polar ice extent and albedo will occur.
The systematic examination of the relation of climate and planetary orbit provides a new theory
for the formation of the polar laminae. The terraced structure of the polar laminae originates when
eccentricity and/or obliquity variations begin to drive water ice off the dusty permanent I-L20 polar
caps. Then a thin (meters) layer of consolidated dust forms on top of a dirty, slightly thicker (tens of
meters) ice sheet and the composite is preserved as a layer of laminae composed predominately of
water ice. Because of insolation variation on slopes, a series of poleward- and equatorward-facing
scarps are formed where the edges of the laminae are exposed. Independently of orbital variations,
these scarps propagate poleward both by erosion of the equatorward slopes and by deposition on
the poleward slopes. Scarp propagation resurfaces and recycles the laminae forming the distinctive
spiral bands of terraces observed and provides a supply of water to form new permanent ice caps.
The polar laminae boundary marks the furthest equatorward extension of the permanent I-IeO caps
as the orbit varies. The polar debris boundary marks the furthest equatorward extension of the
annual COz caps as the orbit varies.
The Martian regolith is now a significant geochemical sink for carbon dioxide. CO2 has been
irreversibly removed from the atmosphere by carbonate formation. CO~ has also been removed by
regolith adsorption. Polar temperature increases caused by orbital variations are not great enough
552
0019-1035/80/120552-56502.00/0
Copyright© 1980by AcademicPress, Inc.
All rightsof reproductionin any form reserved.
 " MARTIAN CLIMATE 553

to release large quantities of adsorbed COs. Early in Martian history, before the formation of
carbonates and an extensive clay regolith, large quantities of CO2 may have resided in the
atmosphere-polar cap system. Orbital variations could have initiated a high-pressure climate by
driving COs out of early permanent CO2 polar caps. Prior to the formation of high-albedo iron
oxides on the Martian surface, 500 mbar of atmospheric CO2 was enough to maintain the global
average temperature at the freezing point of water. Such high temperatures may have played a role
in creating some of the fluvial features observed on Mars.
For the present orbital configuration and climate of Mars we show that: regolith damping of the
seasonal CO2 cycle is unlikely; the mean atmospheric pressure is probably in equilibrium with the
regolith; the low albedo of the north H20 polar cap can be explained by an admixture of 85% ice
and 15% dust; and the albedo of the polar caps as well as the polar heat budget are very sensitive to
small variations in dust deposition. We suggest that a small variation in dust content of the HzO and
COs caps is mainly responsible for the difference in I-£zO and CO2 deposition in the north and south
polar regions at present.

INTRODUCTION and spin orientation of the planet. Ward

(1973, 1974, 1979) and Murray et al. (1973)
Mariner 9 spacecraft observations of have previously shown that the orbital ele-
Mars during 1971 and 1972, as well as more ments of Mars now undergo very large
extensive Viking observations beginning in changes due to the interaction of the torque
1976, provide evidence that current Mar- of the Sun on the aspherical body of Mars
tian conditions are not typical of all of and the torque of the planets on the inclined
Martian geologic history. The Martian po- and eccentric orbit of Mars. Ward et al.
lar regions are covered with extensive lay- (1979) have shown that geologic processes
ered terrain (Fig. 1), probably composed acting to change the figure of Mars may
mainly of wind-deposited dust and water have modified the solar torque so that the
ice, suggesting cyclic polar processes variations in the orbital elements of Mars
(Cutts et al., 1976). The existence of hun- may once have been somewhat different
dreds of desiccated stream beds indicates from the variations as they are now.
that liquid water once flowed over and The orbital elements of Mars are subject
under the Martian surface even though to much greater variation than are those of
Mars is extremely cold and dry today (Ma- Earth. The idea that terretrial ice ages
sursky et al., 1977; Sharp and Malin, 1975). might be caused by changes in the Earth's
The abundance ratios of nitrogen isotopes orbital elements is very old, although Mi-
and noble gases found in the current Mar- lankovitch (1930) is most widely known for
tian atmosphere require that earlier Mars developing it mathematically. Several em-
had an atmosphere with much greater N2 pirical studies (Hays et al., 1976) have
abundance than at present and that large shown that the periodicity of observed ter-
quantities of CO2 and I-I~O have been out- restrial ice ages is well correlated with the
gassed but now lie in hidden reservoirs periodicity of the Earth's orbital elements.
(McElroy et al., 1977; Owen et al., 1977; There have been few theoretical attempts
Anders and Owen, 1977; Pollack and to model terrestrial climate changes caused
Black, 1979). by orbital element variation. Several inves-
In this paper we explore the details of tigators (Mason, 1976; Pollard, 1978) have
one process that may be important for presented simple theoretical arguments
Martian geologic history---climate changes suggesting that the energy imbalance
caused by variations in the orbit and obli- caused by the changing orbit of the Earth
quity of Mars. Solar energy is the basic was sufficient to cause ice sheet growth, but
factor controlling climate on the terrestrial no model has yet shown an ice age develop-
planets, and its distribution with latitude ment with the time history observed. Prior
and with season is determined by the orbit to the last few million years Earth was
554 TOON ET AL.

FIG. 1. High-resolution picture of laminated terrain in the south polar region of Mars. Terraces are
exposed on steep slopes. Each individual terrace is about 30 m thick and can be traced horizontally for
hundreds of kilometers. The width of the set of terraces exposed in the figure is about 5 km at the
widest. The laminae are thought to contain both dust and H20 ice. Their regular structure and the fact
that they occur in both polar regions indicate that their formation may be controlled by climatic
changes (Mariner 9 picture 4213-21).
 MARTIAN CLIMATE 555

warmer than at present. Large-scale ice is that the Martian climate has been rather
sheets have been generally absent except similar to the present climate since the
during a few glacial ages (Pollack, 1979). formation of large CO2 reservoirs in the
The astronomical theory of ice ages is unable regolith. Early in Martian history, before
to account for such long-time-scale changes the formation of these reservoirs, large
in climate. amounts of COg may have resided in the
There are two difficulties in modeling atmosphere. In Section VI we discuss the
terrestrial ice ages caused by orbital nature of the climate when the atmospheric
changes that do not occur for Mars. First, pressure was greater.
the solar insolation changes caused by the Our discussions of the theory of the
orbital variations for Earth are very small. variations of the volatile, temperature, and
Second, the Earth's climate is extremely dust cycles are summarized at the end of
complex because of latent heat release and each section. In Section VII we present an
the transport of heat by the oceans and by overview of the Martian climate for various
the atmosphere. For Mars the orbital orbital configurations and discuss the ob-
changes are large, oceans are absent, latent servational evidence for the predicted cli-
heat release is unimportant except at the mate changes. In particular a new theory of
poles, and the atmosphere is not dense the formation and evolution of the polar
enough to carry much heat. Thus the theo- laminae is suggested.
retical problem is more tractable for Mars
than for Earth. Unfortunately, our limited I. SOLAR I N S O L A T I O N VARIATION
observational knowledge of Mars does There are three orbital elements which
present problems in assigning boundary affect solar insolation and which are each
conditions and in knowing the importance apt to change significantly: obliquity, 0;
of various processes even on present-day eccentricity, e ; and longitude of perihelion,
Mars. In addition, as for the Earth, varia- to. The obliquity is the angle between a
tions in seasonal cycles are important for planet's spin axis and its orbit normal. As 0
Mars, yet the seasonal cycles are much increases, the planet's pole tilts further
more difficult to model than the annual away from the orbit normal so that the
means. polar energy supply in the summertime is
The Martian climate is primarily defined increased. Changing 0 causes neither varia-
by the seasonal cycles of temperature, CO2, tion in the planetwide energy received nor
I-I~O, and atmospheric dust. We systemati- hemispherical asymmetry. The Earth's ob-
cally explore the variations in these cycles liquity is now 23.5 ° and varies between 22. l
as the orbital elements change. In Section I and 24.5° with a period of about 4 × I04
we review the solar insolation changes years (Berger, 1976; Hays et al., 1976). The
caused by orbital element changes. The obliquity of Mars is now 25.2° and oscillates
temperature and carbon dioxide cycles are by about _+13° with a dominant periodicity
considered in Section II for the simple case of 1.2 × 105 years and a modulation period
of CO2 polar caps containing limited of about 1.3 × 106 years (Ward, 1979). The
amounts of CO2. In Section III we discuss Martian obliquity is forced by torques near
the CO2 cycle in the more complex case a resonance condition so that the mean
when CO2 interacts with the regolith. In obliquity also slightly oscillates. Major geo-
Section IV we discuss the dust cycle which logic events, such as the formation of the
depends upon the CO2 cycle. In Section V Tharsis ridge, may have altered the proxim-
we consider the water vapor cycle and the ity to the resonance condition, and earlier
interactions of water vapor with permafrost. Mars may have had a mean obliquity near
One conclusion from our study of the 32° with maxima and minima of 45 and 25° .
cycles of CO2, ~2 0 , dust, and temperature Even more extreme values are possible
556 TOON ET AL.

under dynamically allowed motions that of about 0.1 and the other of 9.5 × 104 years
have not been fully explored theoretically with an amplitude of 0.04.
(Ward et al., 1979). The longitude of perihelion is the angle
The eccentricity of a planet's orbit de- between the data of equinox and the time of
scribes the departure from circularity. perihelion passage. For Earth and Mars,
Since the solar flux at a planet depends the equinoxes precess through 360° with
upon the square of the planetary distance periods of about 2.1 × 104 and 1.75 x 104
from the Sun, an eccentricity change alters years (Hays et al., 1976; Ward, 1974). The
the energy received annually by a planet. precession of the equinoxes causes no
However, the magnitude of the alteration is change in the annual average solar insola-
negligible for the climate changes to be tion. However, for an eccentric orbit, both
discussed. On an annual average changes in the diurnal and seasonal average insolation
e do not redistribute energy with latitude are strongly affected by the time of perihe-
and do not create asymmetry between the lion passage so that latitudinally and hemi-
hemispheres. The eccentricity of the spherically asymmetric energy changes oc-
Earth's orbit varies from 0 to 0.06, with a cur.
period near 9.5 x 104 years (Berger, 1976); • Ward (1974) has previously developed
it is presently 0.0167. For Mars the present formulas to calculate the daily and annually
eccentricity is 0.09 and its minima and averaged insolation from the orbital ele-
maxima are 0 and 0.14 (Ward, 1979). The ments. The daily averaged insolation at
oscillation has two superposed periods: one latitude 5, with solar constants So at the
of 1.8 to 2.3 x 10s years with an amplitude mean Mars-Sun distance is

i = So Ill + e cos v ) 2 ] ( 7 / s i n e s i n 8 + sin~9 cos e c o s 8) (1)

(I + e~)2 J \ rr "

Physically, the term in the first bracket insolation than the equator because the
represents the change in solar intensity distance around the latitude circle in the
caused by the r -2 variation in the Mars-Sun polar regions is much less than that in the
distance. The true anomaly of the Sun, v, equatorial regions. The equatorial regions
measures orbital position from perihelion. receive sunlight during the day but not at
The second bracket contains two geometric night, so large daily temperature fluctua-
effects: the length of the Martian day at tions occur. During the polar summer the
latitude 8, as well as the angle between the Sun never sets, giving no daily temperature
local surface normal and the noontime solar fluctuations.
beam. These depend directly on solar lati- Figure 2 presents calculations by Ward
tude ~, and solar longitude 0, and indirectly (1974) of the annual average solar insolation
on 8, 0, v, and to (Ward, 1979). for various obliquities. It is important to
Ward (1974) presented plots of [ for note that the energy supplied to the equator
various latitudes, eccentricities, and obliq- is a weak function of obliquity compared
uities. For obliquities and eccentricities with the energy supplied to the pole. For
within the range of Martian values the obliquities larger than 54° the poles receive
equatorial insolation varies moderately more annual average energy than the equa-
through the year while the polar insola- tor. Even so, the poles only receive the
tion varies from zero to a value larger energy during summer, so the equatorial
than that at the equator (for O > 17°). energy will undergo a smaller seasonal os-
The poles can receive more summertime cillation than does the polar energy. The
 MARTIAN CLIMATE 557

,5 i i form permanent CO2 polar caps under

present Martian conditions.
.4t 90° (a) M o d e l Description
The current Martian atmospheric pres-
v 30. sure is so low that atmospheric heat trans-
port is negligible (Gierasch and Toon,
01/I" , 1973). Hence, the ground temperature at
90 60 30 0
LATITUDE, deg any location is determined primarily by
local radiative and latent heat sources. At
FIG. 2. The latitude dependence of the annual
average solar insolation expressed as a fraction of the
present water does not supply much latent
solar constant at 1.52 AU for various obliquities. Ward heat, nor do globally significant water
(1974) pointed out that for obliquities less than 50° the clouds form. Hence, a model considering
polar insolation is a strong function of obliquity, but only CO2 is a good description for the
the equatorial energy deposition is a weak function of
current climate and for small changes about
obliquity. As obliquity increases, polar insolation in-
creases but equatorial insolation declines. this condition.
In this paper we employ a model that is
essentially identical to those of Leighton
annum average insolation is a very weak and Murray (1966) and Kieffer et al. (1977).
function of e. Our model balances absorbed solar radia-
It should be noted that time series for the tion, infrared radiation from the surface,
insolation reflect the basic periodicities in and heat conduction into the ground. COz
0, e, and to in a complex manner. It is partly condenses whenever the ground tempera-
for this reason that for the Earth compari- ture falls below the temperature at which
sons of simple climate theories with ice age the atmospheric pressure equals the vapor
data have proved elusive (Hays et al., pressure above solid CO2. The heat diffu-
1976). sion equation is solved with a semi-implicit
scheme developed by Cuzzi and Muhleman
II. THE SEASONAL CYCLES OF COz AND (1972). The ground is divided into ten 1.5-
TEMPERATURE WHEN CO2 POLAR DEPOSITS cm layers and ten 15-cm layers; the total
ARE LIMITED
depth is approximately two seasonal ther-
In previous models of climate change on mal wavelengths. The model has 5° latitude
Mars it was assumed that Mars had large grid points and it calculates the energy
permanent CO2 polar caps (Ward et al., balance every half hour, updating the
1974; Gierasch and Toon, 1973; Sagan et changing solar insolation every 5 days. The
al., 1973). Latent heat exchange with solid model was run for 10 years to achieve
COs polar caps would maintain the polar stability, with the lower-layer temperature
regions at a constant temperature through- updated to the annual average after each
out the year. Therefore, previous climate year to avoid a long lag time. The 9th- and
models were based primarily upon annual 10th-year temperatures and pressures were
average calculations. Viking observations, consistent to within 0.01°K and 0.01 mbar.
however, show that the north permanent The model surface albedo is assumed to
polar cap does not contain COs and the be 0.25, within the error bars of the Viking
south permanent polar cap probably con- value of 0.214 _ 0.063 (Kieffer et al., 1977).
tains no more than a small amount of COs, The Viking mean value itself may be some-
if any (Kieffer et al., 1976; James et al., what low because of incomplete coverage
1979). Therefore, we have made new calcu- of Mars by the spacecraft observations. We
lations of the COs and temperature cycles have not treated the albedo of the perma-
assuming that not enough COs exists to nent I-I~O polar caps since they are of small
 558 TOON ET AL.

geographic extent. We a s s u m e that: the depleted in COz and enriched in minor

CO2 frost albedo is 0.65; the soil density is species which would influence the polar
1.6 g cm-3; the thermal inertia is 0.006 cal energy balance. H o w e v e r , the o b s e r v e d
cm -2 sec -1 deg-1; and, the specific heat is low t e m p e r a t u r e s m a y instead reflect low
0.17 cal g-1 deg-1. All these values are close surface emissivity or the p r e s e n c e o f high
to those adopted b y Kieffer et al. (1977). clouds. Until the o b s e r v a t i o n s are m o r e
The t h e r m a l balance o f the a t m o s p h e r e is completely understood they cannot be eas-
not considered in detail, so the energy ily considered in models.
received by the ground from a t m o s p h e r i c We have modeled two orbital con-
infrared radiation cannot be precisely cal- figurations. In the first, 0 = 15, 25, 35,
culated. W e treat the atmospheric radiation and 45 ° for e = 0. T h e s e models have
as 1% of the noon solar radiation. We use a complete s y m m e t r y between the hemi-
ground emissivity of 0.9. Kieffer et al. spheres and the seasonal cycle is indepen-
(1977) assume that the a t m o s p h e r e radiates dent of to. The values of 0 represent the
2% of the noon solar radiation and that the e x t r e m e reasonable values for the current
ground emissivity is 1. T h e s e differences oscillation cycle (15 to 35 °) and the hypo-
between the two models are mutually c o m - thetical oscillation (Ward et al., 1979) prior
pensating and neither affects surface tem- to the formation of the Tharsis ridge (25 to
perature by m o r e than a few degrees. 45°). Values of 0 slightly smaller than 15°
An important assumption of our model is m a y actually be reached (e.g., I 1°), but the
that the total amount of CO2 in the atmo- corresponding climates will p r o b a b l y not be
sphere and seasonal polar caps is about 9 substantially different from those at 15° .
mbar. We chose this value because it The other situation we have modeled is
m a k e s the CO,, caps disappear near solstice the case when e = 0. ! 4, its m a x i m u m value.
under present conditions, so p e r m a n e n t Again we study 0 = 15, 25, 35, and 45 °. We
CO2 polar caps are not found at either pole. have set to = 270 ° so that solstice occurs at
The a t m o s p h e r e is k n o w n to contain about perihelion. Thus, one s u m m e r pole receives
6 m b a r o f CO2 and models of the seasonal the m a x i m u m possible insolation. We have
caps suggest that they contain no m o r e than also p e r f o r m e d one calculation in which 0
a few millibars of CO2 (Briggs, 1974; Davies = 45 ° and to = 0° so that perihelion is at
et al., 1977). equinox, producing the m a x i m u m equato-
There are several features o f the present rial insolation at one equinox and the mini-
Martian climate that are not well treated b y m u m insolation at the other. The equato-
models such as ours. F o r example, the rial, equinoctial insolation m a x i m a are
models are not m e a n t to follow accurately independent of 0.
the details of polar cap recession, so they
do not have albedos which are functions of (b) C a l c u l a t e d T e m p e r a t u r e a n d P r e s s u r e
frost thickness. N e i t h e r do the models con- Cycles, e = 0
sider the polar hood which affects the polar Figure 3 illustrates calculations of daily
radiation balance. Of course, all p a r a m e - average surface t e m p e r a t u r e s as a function
ters, such as thermal inertia and albedo, of latitude and time of y e a r for e = 0 and
vary strongly with location about the m e a n various values of 0. Mars presently has e =
values we use; in addition some soil proper- 0.09, 0 = 25 ° , and to = 250 ° , so it is
ties m a y be functions o f depth. Kieffer et intermediate b e t w e e n the case shown in
al. (1977) report very low polar night tem- Fig. 3B and that shown in Fig. 6A. In Fig.
peratures, well below those that should 3B the s u m m e r t e m p e r a t u r e s are nearly
occur if CO2 latent heating maintains the independent of latitude. The polar annual
t e m p e r a t u r e at the frost point. T h e y suggest cap sublimes a w a y s o m e w h a t after the so-
that the winter polar a t m o s p h e r e b e c o m e s lar energy p e a k s b e c a u s e of the finite en-
 MARTIAN CLIMATE 559

i f i , i i i i i

-- .-- 90 ° LATITUDE WITH A = 0.65 i ~ i i ~ , i i i /

----- 60 ° Pt = 8 . 9 m b a r 90 ° LATITUDE
250
. . . . 30 ° e = 0 o.
o ------ 60° Pt = 8 . 9 m b a r
_____ 0°
90 ° with A = 0.40
8 = 15 °
co = 2 7 0 ° ~ 250 . . . .
------
30°
O°
e = 0
0 = 25 °
-I
-]
~-...~. co = 2 7 0 °
~ ~ ~/~-- "~-

~ 200

/ uJ
. . . . . 7..~ _ .~ ~ _ _ ~ - . . . . . -
g /" 200

/' / ~ \'\
~, 150
0 i
130 150
i i i i i i 810 i i
200 240 280 320 O 40 120 160 200 240 280 320 0 0 0 120 160
SOLAR A N O M A L Y , deg S O L A R A N O M A L Y , deg

i , f i i , i i i
, , i q
- - 90 ° LATITUDE
90 ° L A T I T U D E
___-- 60 °
Pt = 8 . 9 m b a r o~ ----- 60 ° Pt = 8 . 9 r n b a r
o 30 °
~z:'250 ~:" 2 5 0 .... 3°° e=O
m~ ---- _ 0°
0= 45 °
, w= 270 °
/
/// //
~ 20o 200 /
// ! ~ \ "'-, ~\ \\\
'\ \~\

150 150 ~
i i i i i / i L i I I I i i I i i /
200 240 280 320 6 40 80 120 160 200 240 280 320 0 40 80 120 160
SOLAR ANOMALY, deg SOLAR ANOMALy. deg

FIG. 3. T h e daily a v e r a g e d g r o u n d t e m p e r a t u r e at v a r i o u s latitudes is illustrated for perihelion

o c c u r r i n g at solstice, a circular orbit, and varying obliquity. All the c a l c u l a t i o n s e m p l o y a CO~ albedo
A = 0.65 and a CO= a b u n d a n c e o f 8.9 m b a r e x c e p t for o n e c a s e in (A) w h e r e the calculation for the
p o l a r t e m p e r a t u r e s w a s run w i t h A = 0.4. T h e s e a s o n a l variation in p o l a r t e m p e r a t u r e s h o w s that l o w
CO albedos will p r e v e n t p e r m a n e n t p o l a r c a p f o r m a t i o n . A n o t h e r case w a s run w i t h A = 0.65 b u t P = 5
m b a r . T h e c o n s t a n t l o w p o l a r t e m p e r a t u r e s h o w s that, if t h e a l b e d o is high, a p e r m a n e n t polar c a p will
f o r m e v e n if the a b u n d a n c e o f CO~ is low. Parts (C) and (D) indicate that high polar s u m m e r
t e m p e r a t u r e s o c c u r at large obliquity.

ergy required to evaporate the high-albedo present conditions low latitudes experience
COz frost. During most of the year CO2 very large excursions of daily temperature,
deposition maintains polar temperatures which often have peaks exceeding 273 ° for
constantly near 150°K. The amplitude of the portions of the day (Kieffer et al., 1977). As
thermal seasonal wave increases with lati- the obliquity increases, the polar CO,, de-
tude. All these results are in basic agree- posits evaporate earlier and earlier since
ment with observations of Mars. the sunlight is more intense because of the
At obliquities greater than 35° the sum- higher sun angle and the greater length of
mer polar temperatures exceed summer the summer day at subpolar latitudes. How-
equatorial ones. In fact, for O = 45 °, sum- ever, polar CO2 ice deposits form further
mer polar ground temperatures exceed equatorward during winter since the length
273°K throughout the entire day for about of the winter day is lessened at higher
45 Martian days. Of course on Mars under obliquity. As the obliquity increases, the
560 TOON ET AL.

magnitude of the annual thermal w a v e

grows at all latitudes. 0 i i i i i i

As the obliquity b e c o m e s less than 25 ° , 5 10

20 ------
15 ° O B L I Q U I T Y

25°
WITH A = 0.65 • = 0

w = 270 °
the poles b e c o m e colder, the e q u a t o r be- '30 --'-- 35° A = 0.65

c o m e s w a r m e r , and the annual thermal ,0 .... .,,, A=0,0t( ----<-.

J" i
w a v e amplitude decreases. The b e h a v i o r of
polar CO., depends upon its albedo. In Fig.
i S0 I - - - - - - ~ "'~,~,

3A we present two alternate models: one

which inhibits polar cap formation due to 90 r "',~
200
,--~/,'l
240 280
i
320
I
0 40
i I
80
k
120
I\
160
very restricted CO., abundance ( P = 5 m b a r SOLAR ANOMALY, deg

and CO., frost albedo Aca2 = 0.65); and one FIG. 5. The seasonal variation of the CO= polar cap
which inhibits polar cap formation due to boundary as a function of obliquity for a circular orbit.
enhanced solar heating ( P = 9 m b a r and The annual cap is larger for larger obliquity.
Aco., = 0.4). P e r m a n e n t polar caps form
only for the case with higher albedo. occur (Ward et al., 1974; Pollack, 1979).
A high polar c a p albedo at low obliquity F u r t h e r m o r e , our model with A = 0.65
is the m o r e likely case. The p e r m a n e n t suggests that 100 cm of polar CO,, will
north polar cap of Mars, c o m p o s e d of H.,O, annually be deposited and r e m o v e d at O =
has an albedo near 0.4 (Kieffer et al., 1976). 15° , cleaning the CO,, each year.
T h e r e are two reasons for the low albedo. Figures 4 and 5 show how atmospheric
Mars is presently a dusty place, and the pressure P and the location of the polar c a p
thickness of w a t e r ice annually sublimed b o u n d a r y v a r y with time of y e a r for various
and redeposited on the poles is small (.~ 1 0. As 0 increases the polar c a p boundary
cm). H e n c e the dust-to-ice ratio is large and m o v e s e q u a t o r w a r d , the annual m e a n at-
the ice is not self-cleaning. F o r CO., perma- mospheric pressure d e c r e a s e s slightly, and
nent deposits neither o f these effects should the amplitude of the seasonal pressure
occur. A high-albedo c a p causes a low w a v e increases. Earlier models with unlim-
atmospheric pressure (Fig. 4) and at low ited CO2 (Ward et al., 1974; Gierasch and
atmospheric pressure dust lifting will not T o o n , 1973) predicted large increases in P
with increasing 0. Ward et al. (1974) also
14 I i J i i i r i i
found, e v e n with unlimited CO2, that an-
13 15"~ O B L I Q U I T Y W I T H A = 6.BS
12 ______26 ° nual polar caps would extend to lower
.~ 11 --.--35 ° • = 0 latitudes w h e n the obliquity increased, as
6 10 -----45° w = 270 °
w" . . . . 15 ° w i t h A = 0 . 4 0 A = 0,65
Fig. 5 indicates. H o w e v e r , with limited
CO2, the seasonal caps contain less CO,,
: ................ -e-.: .................... -7~--( than with unlimited CO2 and do not reach
latitudes as low. The reason that the atmo-
"~ o "-.. "~. '\. "-'I" spheric pressure decreases with increasing
~4
~s obliquity is that greater a m o u n t s of CO2 go
2 into the larger seasonal cap.
I f the polar albedo remains high for per-
o . . 2 4.0 . 2 .8 0
200 320 0 4 , 8'0 '
120 160 ' manent COz caps, then at 15° obliquity the
SOLAR ANOMALY, deg
annual pressure will be quite low, 1 m b a r or
FIG. 4. The seasonal variation of atmospheric pres- less. The precise value of atmospheric pres-
sure is shown for a circular orbit and for various sure at low 0 is sensitive to the choice of the
obliquities. At low obliquity a permanent CO= cap may CO2 albedo. W a r d et al. (1974) chose Aco~
form, decreasing the pressure. Note that the annual
mean pressure declines slightly as the obliquity in- = 0.70 and found =that-at 0 = 15° atmo-
creases above 25° because of the formation of larger spheric pressures were only a few tenths of
annual polar caps. a millibar.
 MARTIAN CLIMATE 561

(c) Calculated Temperature and Pressure 273°K. W h e n O is 45 °, t e m p e r a t u r e s at the

Cycles, e = 0.14 pole reach 300°K, e x c e e d 273°K for about
85 days, and the entire region poleward o f
Figure 6 illustrates the daily average 50 ° has daily average m i d s u m m e r t e m p e r a -
ground t e m p e r a t u r e as a function of 0 for e tures a b o v e 273 ° . The s u m m e r pole in the
= 0.14. The t e m p e r a t u r e s are similar to opposite hemisphere is cooler, and e v e n at
those with e = 0 except that the two 0 = 45 ° ground t e m p e r a t u r e s there n e v e r
hemispheres are no longer identical and the exceed 260 ° . Winter t e m p e r a t u r e s are also
amplitude o f the seasonal thermal w a v e altered. The hemisphere with the w a r m e r
increases. With e = 0.14 and 0 = 35 °, polar s u m m e r pole is colder in winter and the
ground t e m p e r a t u r e s in the hemisphere in seasonal cap reaches lower latitudes.
which perihelion occurs at solstice exceed As the equinoxes precess, the two poles

:00";AT,TU' E . . . .
i , , r , r , r
l -:-+ Y
[- . . . . 30° [ ~'~. P, = 8.9 mbar

f
+ 90 ° L A T I T U D E
°~ ~- . . . . 90 ° l \ e : 0.14
- - - - + 30 °
.... 0° Pt = 8.9 mbar
~= 25ol- "-i-~- \ <~ = 27°°
30 ° --- e = 0.14 < i'~. /" "~ .
--'--- 90 ° J~'- "~'~. 0 = 25 °
S"L_ . . . . . . . . F~'~. ,~ : 27oo
#"XJ , ~ ! k---"x -._
i
m 200

>
-/X
L~ \
--i-\ i \
-"-..-- "'-.
= 290v--
/
/
,~
\
\
\
\-.. i
i
~r~
i /
/ \
i
~
\
'\" \
"

<
-I \ i
150 _L...... .L__- , ', 15o . . . . . . . ..~ \ I
i i i i i i i i I i i i i i 410 i i i
200 240 280 320 0 40 80 120 160 200 240 280 320 0 80 120 160
SOLAR A N O M A L Y , deg B SOLAR A N O M A L Y , deg

:d 300 + 90 ° L A T I T U D E .~. 390 , i i

f -i- - - - - + i 30 ° i . / 1
.... oo i \ Pt = 8.9 mbar _ _ + 90 ° L A T I T UDE
e = 0.14 f'~'~, + 30 °
.... 30 ° / '\
o
. . . . 90° i
.
I
"\
\
Pt = 8.9 mbar
e = 0.14
~ o=45 o
o
~=oo![
0=45 ° \
\-:-_:::: \
---- °° r \

25o X ~ r - ~ . \ \ = = 270° ~o
\ /i "q
i
/i ,, }"\ "k. "

Xl ',, 1
,¢,, 200

\\ I / i "q .
", i / i /
150
i~" • , 7 7 7
150
_ ?2i l i i i i i i
. . . . . .

i
290 240 280 320 0 40 80 120 100 200 240 280 320 0 40 80 120 160
C SOLAR A N O M A L Y , deg SOLAR A N O M A L Y , deg
D

FIG. 6. The seasonal variation of g r o u n d temperature at various latitudes is s h o w n for a highly

eccentric orbit. Parts ( A ) - ( C ) illustrate results for various obliquities w h e n perihelion occurs at
s o u t h e r n solstice. The a s y m m e t r y b e t w e e n h e m i s p h e r e s is m o s t noticeable at the poles. Part (D)
illustrates the seasonal temperature variation at 45 ° obliquity w h e n perihelion occurs at an equinox.
T h e t e m p e r a t u r e is m o s t altered at 30 ° latitude.
 562 TOON ET AL.

switch roles. The intermediate case, oJ = 0,

is illustrated in Fig. 6D. H e r e the hemi-
-80 -- , it ' l ' Z ~ %
't --

spheres are nearly symmetric except for -60

,, .,
hysteresis effects. The equatorial tempera- I.-
~,-40
ture shows a strong m a x i m u m at the perihe- ==
-o
lion equinox. >.. - 2 0 -- -----" 25 ° e = 0.14
I . - - 35 ° <~ = 2 7 0 °
Figures 7 and 8, respectively, illustrate ----I 45 ° A = 0.65
~ 0
for e = 0.14 the atmospheric pressure ..... 4 5 ° W I T H co = 0

changes and the modification of the ice cap <~ 20

¢J
edge due to changing 0. T h e s e results are ~C
< 40
similar to those for e = 0 except for an / / /
a s y m m e t r y b e t w e e n the hemispheres. 6O
/ 4
(d) Annual Average Temperatures
, /l , , , \,, ":,','J
200 240 280 320 0 40 80 120 160
Seasonal t e m p e r a t u r e oscillations are SOLAR ANOMALY, deg
quite important for atmospheric pressure
changes and for volatiles stored within the FIG. 8. The seasonal polar cap boundary variation as
a function of obliquity for a high eccentric orbit.
top 100 cm of soils, the u p p e r thermal skin
depth for the annual thermal wave. F o r
m a n y important processes, however, an- increases, as do polar t e m p e r a t u r e s . The
nual average t e m p e r a t u r e s are more insolation at low latitutdes decreases as do
significant than daily average temperatures. equatorial temperatures, but the changes at
Figure 9 illustrates annual average t e m p e r a - low latitudes are smaller than those at high
tures for various 0 and e = 0.0 and 0.14. latitudes. Figure 2 shows that the insolation
E v e n though the annual average insolation near 35 ° latitude is nearly independent of
is nearly independent of e, the annual aver- 0, but Fig. 9 shows that the t e m p e r a t u r e is
age t e m p e r a t u r e does depend on e. The nearly independent of 0 n e a r 50 ° latitude.
greater d e p e n d e n c e of t e m p e r a t u r e on e The latitude shift is due to the nonlinear
occurs because the t e m p e r a t u r e is related dependence of T on insolation through o-T4.
in a nonlinear m a n n e r to the insolation Earlier climate models did not illustrate
through tr T 4. H e n c e , average t e m p e r a t u r e s annual average t e m p e r a t u r e s except at the
are a s y m m e t r i c b e t w e e n the hemispheres poles. Ward et al. (1974) thought polar
and they vary for fixed 0 and different e. t e m p e r a t u r e s would be about 10°K cooler
The t e m p e r a t u r e changes in Fig. 9 approxi- than those shown by Fig. 9 at 0 = 25 and 35 °
mately mimic the annual average insolation because their CO., cap n e v e r sublimed to
variation shown in Fig. 2. As the obliquity expose low-albedo ground, and latent heat
increases, solar insolation at high latitudes kept s u m m e r t e m p e r a t u r e s low. N e v e r t h e -
less, the t e m p e r a t u r e change in going from
9 r r i i
0 = 25 to 35 ° in the model of Ward et al.
, was about 15°K, which is rather close to the
~6 \
\'\.
\
~-~__ i
.//d"
J/,, z
"~-
---
/
value shown in Fig. 9. Gierasch and T o o n
¢j 4
-
-
"\\.. .1:7
,\ . J /
(1973), h o w e v e r , pointed out that with un-
,,=,3
limited CO2 polar t e m p e r a t u r e s would
------ 25 ° OBLIOUITY e = 0.14
--'-- 35 ° w = 270 ~ reach 190°K at 0 = 35 ° because of atmo-
o
----- 45 ° A = 0.65 spheric heat transport from the equator.
<~ 0 i i L i i i i i i
200 240 280 320 0 40 80 120 160
SOLAR ANOMALY, deg (e) Summary
FIG. 7. The seasonal atmospheric pressure variation With limited CO,, available, the Martian
as a function of obliquity for a highly eccentric orbit. climate for various obliquities and eccen-
 MARTIAN CLIMATE 563

230 I i I I i i I ~ I I I i i i i i 1-- III. T H E S E A S O N A L C Y C L E O F COx

I N C L U D I N G CO2 E X C H A N G E W I T H T H E
~ 210 REGOLITH
==.2oo
.
19o In the previous section we discussed
Martian climate changes involving the CO2
in the atmosphere and in the annual polar
~ 160 caps. However, much more CO2 may lie
=" 1,o I hidden on Mars. Here we consider whether
.,3o_8; ~-8'0'_,o -2o ; 2; ,'° ~o I. or not some of this hidden CO2 could partic-
LATITUDE, (leg
ipate in climate changes.
FIG. 9. The annual average temperature at different
latitudes for various obliquities and eccentricities. (a) C02 Reservoirs
Annual average temperatures for all latitudes are given Early analyses of Viking measurements
for an eccentricity of 0.14, while for negative latitudes
of noble gas concentrations suggest that
and 0 = 25 ° and 0 = 45 °, they are given for eccentrici-
ties of both 0.0 and 0.14. The higher polar temperature Mars outgassed 150 to 500 mbar of CO2
at negative latitudes, with 0 = 25 ° and 0 = 45 °, (Anders and Owen, 1977). Since the Pio-
corresponds to e = 0.0. The annual average tempera- neer exploration of Venus, theories of the
ture variation is similar to the insolation variation noble gas distributions on the terrestrial
illustrated in Fig. 2. The temperature is not a linear
planets have been revised, and recent work
function of insolation and is a stronger function of
eccentricity than is implied by the dependence of relating the Viking observations of noble gas
annual average insolation on eccentricity. The polar inventories to the abundances of other vol-
temperatures do not vary greatly between 25 and 35 ° atiles on Mars implies that 1 to 3 bars of
obliquity. Annual average temperatures are important CO2 may have been outgassed during Mar-
because most of the regolith responds to annual aver-
age temperatures rather than daily average tempera-
tian geologic history (Pollack and Black,
tures. 1979). Much of the outgassed CO2 may now
be in the form of carbonates (Toulmin et
al., Anders and Owen, 1977). If 4% of the
soil mass is CO2 in a carbonate phase, and if
tricities is similar to the present climate. As the regolith is 100 m thick, then about 800 g
the obliquity or eccentricity increase, sea- cm -2 of regolith could be CO2. This is
sonal extremes are exaggerated. The major equivalent to about 50 current atmospheres
changes from present-day Martian condi- or about 300 mbar pressure. Unfortunately,
tions are that very high (273°K) summer- CO2 converted to carbonates would be per-
time daily average polar ground tempera- manently prevented from returning to the
tures occur at high obliquity and that atmosphere barring extensive volcanic re-
permanent CO2 pola r caps form at low processing of surface materials.
obliquity if the CO2 albedo remains high, CO2 may also be adsorbed on the rego-
as seems most likely. The annual average lith, as suggested by Davis (1969) and
temperatures depend slightly on the or- Fanale and Cannon (1971, 1974, 1978). Di-
bital parameters. The annual average po- rect evidence that heated Martian soil
lar ground temperature may increase by evolves CO2 was obtained by Viking, but
about 10 to 15°K as the obliquity in- we cannot quantify the fraction desorbed
creases from 25 to 35° . Previous Martian because of the additional possibility of car-
climate models erroneously assumed that bonates breaking down (e.g., Biemanh et
Mars currently has massive COz perma- al., 1977).
nent polar caps. These early models pre- The Martian topsoil may be composed of
dicted large pressure and temperature a combination of clays and iron oxides
changes that are not found in the present (Toon et al., 1976; Toulmin et al., 1977).
limited CO2 model. However, the crust of Mars is most likely
 564 TOON ET AL.

some type o f mafic igneous rock such as +

168°K /
FANALE AND CANNON (1978}
195°K
basalt. Whether most of the regolith is • 196OK }
8 273°K FANALE AND CANNON (1971)
composed of clays, like the surface layer. o 302°K
or basalt, like the crust, is unknown. The
Fanale and Cannon (1971, 1974, 1978) ex-
: 160°K
perimental studies of CO2 adsorption on
basalt, pulverized to simulate Martian soil,
suggest that a 1-km-thick basalt regolith
could adsorb 300 to 800 g cm -2 o f COs. In a
basalt-carbonate regolith, carbonate would
contain most of the CO2. Fanale and Can- O~ ~e ~ ~ _ I . . . . J. . . . . . 302°K1
non (1978, 1979) suggested that clay min- 0 5 10 15 20 25 30 35 40 45 50
ATMOSPHERIC PRESSURE, mbar
erals may adsorb 10 times more CO2 than A

basalts. In such a case 300 to 800 g cm -s o f

4O
COs could be contained in a 100 m clay
regolith. A c l a y - c a r b o n a t e regolith would 36 I 160°K
have CO2 equally partitioned between clay , ~SOLID CO2 FORMS

and carbonate. y
/ / o,.,o°,q ':A'A'.E
T o utilize adsorption data it is convenient .~ 28 I- / u19e°KI A N D C A N N O N
to obtain an analytic expression to fit the [ //~ ~ 158°K)(1979)

available data. Figure 10 presents basalt ~ 24 [- ~ 180°K

adsorption data of Fanale and Cannon

(1971, 1974, 1978, 1979) which were ob- d 195° K
tained for different basalt powders with
different measured surface areas (5.8 m s g-1
in 1971 and 4.4 m s g-1 in 1978). Normally
8 sH/ ~ - 23oo,~o
the volume adsorbed is directly propor-
tional to the surface area, so we have
normalized the data to 4.4 m s g-Z which
brings the Fanale and Cannon measure- 0 10 20 30 40 50 60 70
ATMOSPHERIC PRESSURE, mbar
ments at 195 and 196°K into good agree-
ment. FIG. 10. The m a s s o f CO2 absorbed per g r a m o f soil
Unfortunately, there is no universal the- is illustrated as a function o f p r e s s u r e for various
ory o f adsorption. After rejecting several temperatures. In (A) the solid c u r v e s are simple ana-
theoretically justifiable analytic forms, we lytic functional fits to Fanale a n d C a n n o n ' s (1971,
found a good empirical fit to the adsorbed 1974, 1978, 1979) data for basalt. In (B) the solid
curves are functional fits to Fanale and C a n n o n ' s data
mass per gram o f basalt at low temperature for clays.
using
proach 0, Pa approaches 0; Pa is linear in
Pa = p o ( A s / A o ) e x p ( F T ) surface area; and Pa depends on only two
( P / P o ) x g CO2/g soil, (2) constant coefficients, F and X.
Equation (2), as shown in Fig. 10A, over-
where P is the pressure in millimeters o f estimates the COs adsorbed by basalt at
Hg, P0 is 1 mm, T is the temperature, p0 is high temperatures. H o w e v e r , so little COs
1.5 x l0 -s g COs/g soil (equivalent to 7.8 is then adsorbed and Mars is so much
cm s STP g-l), As is the actual adsorption colder than 300°K that this error is not
surface area, ,4o is 4.4 m 2 g-l, X is 0.275, and significant for our work. Two important
F is - 1.5 × 10-S°K. This empirical equation aspects of Fig. 10A are that large tempera-
satisfies three i m p o r t a n t criteria: as P a p - ture changes are needed to r e m o v e substan-
 MARTIAN CLIMATE 565

tial fractions of CO2 from the ground and m deep. However, the new theories of
that the curves have a strong pressure noble gas distributions on the terrestrial
dependence, especially below 2 mbar. planets (Pollack and Black, 1979) would
Fanale and Cannon (1979) recently per- probably considerably raise the predicted
formed clay (nontronite) adsorption mea- abundances of C1 and S and therefore con-
surements, which are illustrated in Fig. 10B siderably increase the estimated regolith
along with an empirical fit using Eq. (2). In depth. Hence, the regolith is estimated to
Eq. (2) we employed the same values of X be between 100 and 1000 m deep with an
and F as for basalt. We used 9 x l0 -a g intermediate value being most likely.
C O J g soil for p0. The value of As at 158°K
is 78 m 2 g-1. Fanale and Cannon (1979) (b) Characteristic Diffusion Times of CO.,
point out that the COs adsorption area for in the Martian Regolith
clays is temperature dependent because CO2 could be driven between atmosphere
I-I~O occupies fewer adsorption sites at and regolith either by temperature or by
lower temperatures leaving more sites pressure changes. In Section II we de-
available for CO2. To simulate this effect scribed typical surface pressure and surface
we introduced additional temperature de- temperature changes caused by orbital ele-
pendence by setting AT = A~58(158/T)2. The ment variations. However, it is necessary
poor fit to the data in Fig. 10B at 195°K to extrapolate these changes into the
illustrates that the adsorption area has a ground. A complete solution requires a
more complex temperature dependence coupled thermal and vapor diffusion model
than this. Further experimental work is at several latitudes which is integrated over
necessary to obtain a meaningful improve- long time periods. Before proceeding to
ment in the analytic fit, however. As for discuss this problem, we consider the ther-
basalt, clay adsorption depends strongly on mal and vapor diffusion time scales, which
pressure for low pressures. Unfortunately are often greatly different.
the available data do not extend to low The thermal skin depth (the depth where
enough pressures to define the trend very A T = e -1 of A T at the surface) is
well. At low temperatures (160°K) clays
release about 10 times more COz than do )k t = (~I /Tr)ll2(I/psCH)
basalts for the same temperature change. "=-- 2 . 2 x 10 -2 c m sec -'/2 (~/~)1/2. (3)
However, near 200°K clays only release a
few times more COz than do basalts for a Here y is the period of the thermal varia-
given temperature change. tion, and we have used typical Martian
The depth of the Martian regolith also surface thermal inertia I (6 × 10-a cal
determines the total CO~ adsorbed. Fanale cm -2 sec -~/z °K), heat capacity CH (0.17 cal
(1976) argued that the regolith may be 1 to 2 g-i OK-i), and soil density Ps (1.6 g cm-a).
km deep based upon photographic indica- The thermal inertia, remotely sensed, is
tions from Mariner 9 and estimated ejecta relevant for the soil surface, but at depth in
volumes from craters. However, Fanale the regolith it could increase because of soil
and Cannon (1979) suggested that some of compaction. The thermal conductivity also
the regolith material may have been lithified depends on pressure when the gas mean
during Martian geologic history and they free path is comparable to the pore size in
therefore adopted a mean regolith depth of the rocks. Thermal inertia increases ap-
a few hundred meters. Anders and Owen proximately as the fourth root of pressure
(1977) suggested, from comparison of pre- in the range from 1 to 100 mbar for coarse
dicted chlorine and sulfur abundances with basalt powders (Wechsler and Glaser,
Viking observations of C1 and S concentra- 1965). For clays with very small pore sizes
tions in the soil, that the regolith is only 70 the pressure dependence might be much
 566 TOON ET AL.

less. Viking observations produced no cor- ever, clays are c o m p o s e d of very thin
relation b e t w e e n surface thermal inertia plates which can be closely packed. F o r
and elevation except for e x t r e m e eleva- clays the adsorption area p o r e size m a y be
tions, suggesting either that the pressure close to the relevant one for diffusion. A
d e p e n d e n c e is low or that some other factor clay with 10 times the adsorption area as
such as particle size is m o r e important. Due basalt has r -~ 5 x 10-a txm. H e n c e the
to these c o m p a c t i o n and pressure effects, possible range of r on Mars is b e t w e e n 5 ×
the thermal skin depth is uncertain by a 10-3 and 1/~m.
factor of several. The time for a thermal The diffusion equation in a porous ad-
w a v e to reach the b o t t o m of a 1-km regolith sorbing m e d i u m is (Barrer, 1967)
is about 10~ years and the time with a 100-m
regolith is about 104 years. A shallow rego- E 00g = E O D g O p g
lith is o p t i m u m for responding to thermal 0t
forcing. o o
The time scale for propagation of a pres- Ot PsPa + Ds ~ p~oa, (4)
sure w a v e depends upon soil porosity, E,
tortuosity, q, pore size, r, and adsorption. where Pg is the mass density of the gas, Dg
Viking m e a s u r e m e n t s (Clark et al., 1977) is the effective volume diffusion coefficient,
suggest that E is 0.60 + 0.15. We adopt a Ps is the soil density, Pa is the adsorbed
value of 0.5 to represent the regolith mass per gram of rock, and Ds is the surface
column. The tortuosity (which represents diffusion coefficient. Physically, the second
the ratio of blind to open paths available to t e r m on the fight-hand side of Eq. (4)
a diffusing molecule) is not known, and we represents the local conversion f r o m ad-
take a value of 5 in agreement with Smolu- sorbed CO., to gaseous CO,, and the final
c h o w s k i ' s ideas (1968). The pore size is the term represents a surface flux which occurs
m o s t uncertain parameter. on adsorbing materials as the molecules
The pore size could be fixed by the grain hop along the surface. We shall ignore the
size of Martian soil, which is thought to be surface flux because its diffusion coefficient
between 10 and 100 /xm (Kieffer et al., is e x t r e m e l y c o m p l e x and cannot be ade-
1977; Shorthill et al., 1976). H o w e v e r , r is quately predicted. F u r t h e r m o r e , its neglect
most likely to be set by the smallest avail- yields a smaller error than is already
able particles. T o o n et al. (1977) and Pol- present because of uncertainties about the
lack et al. (1977) concluded that the size of pore size.
Martian dust storm particles was about 1 We can convert Eq. (4) to a simpler form
p.m, which suggests a pore radius r -< 1 p.m. by using Eq. (2) and the ideal gas equation
In some materials the pore size distribution
is bimodal, so that large pores occur be-
0P _ 1 0 Kd 0 (5)
tween grains and small pores o c c u r on grain P,
0t C 0z 0z
surfaces. Diffusion occurs preferentially in
the larger pores, while adsorption occurs where
mainly in the smaller ones. The radius of
these smaller pores can be determined from
adsorption m e a s u r e m e n t s as r = ½ (pore
v o l u m e / s u r f a c e area). For Fanale and Can- -- - E + XPads°rbed
exp(FT) Poo P - p ,
n o n ' s basalt this calculation yields r -~ 0.05
~ m . F o r basalt, it is unlikely that this pore Kd E(~r rr 2r/h
radius is the relevant one for diffusion cal- =q- X + 4 (1 + 2r/X)
culations b e c a u s e it is m u c h smaller than
the dimensions o f the basalt grains. H o w -
+
(1+ 2r/X))gr (~-~f--)~l~" (6)
I ' 4
 MARTIAN CLIMATE 567

Here Rg is the CO`, gas constant and h is the ~o, I I I I I I I I

mean free path of CO`, ( 6 / z m at P r e f = 6 r = 50/Jm

mm). Here A, = 6 / z m (6 ram/P).

The expression for Kd (Barrer, 1967) in-
~ r= lO.m
cludes the fact that when the pore radius is ~ 1

~
*E
larger than the mean free path for gas ~, '- NO ADSORPTION
molecule collision the CO,, flows out along 170°K

streamlines. When r/A, is less than 1 the ~ r = 1/Jm

CO`, molecules collide with the pore walls ~ 10-,

and diffuse from the ground. In our case we g
expect r/A, ,~ 1 (if r = 1/xm, r/A, = 0.1). In ~
the range 0 < r/A. < 1.5, Kd varies from its r = 0,01 /am

value when r/A, = 0 by only 5%. H e n c e we ~ 10-2

can extract K0 from the derivative in Eq. (5)
f
and define a v a p o r diffusion skin depth as

Xd = (TI~)ll2( KdlC) in = (y/Tr)VZLo. (7) lO" I [ I I I I I I

10 20 30 40 50 60 70 80 90
ATMOSPHERIC PRESSURE, mbar

Values of Ld are given in Fig. 11 for basalt FIG. 11. The diffusion wavelength is illustrated for
various pressures and pore sizes for basalt. For very
powders having various pore sizes and tem-
large pore sizes CO= can rapidly flow through the soil,
peratures. Figure 1I also illustrates the but at small pore sizes CO2 cannot rapidly diffuse
corresponding thermal diffusion term. F o r through the soil. For r = 1 txm we illustrate several
clays at low temperatures, the value of C is cases of interest. We show that diffusion wavelengths
10 times larger than for basalt and the for various temperatures do not differ greatly. We also
diffusion length is 101/2 times smaller than demonstrate a diffusion wavelength curve for a soil
that does not adsorb CO2. The presence of adsorption
those depicted in Fig. 11. O f course, clays severely damps the pressure wavelength. Clays are
are also very likely to have a smaller pore generally 10 times as adsorbing as basalts. Hence the
radius than do basalt soils. diffusion wavelengths for clays are roughly 10t/z times
Figure 11 shows that except for the small smaller than those illustrated here. Also shown is the
thermal diffusion wavelength, which is only weakly
pore sizes typical of clays, the pressure
dependent on pressure. CO2 will normally diffuse
diffusion length exceeds the thermal faster than heat through basalt unless the pore size is
lengths. F o r example, during the time that less than about 0.1 txm. COz will diffuse faster than
an annual thermal wave penetrates 1 m, a heat through clay unless the pore size is less than
pressure wave will penetrate about 5 m in about 0.5 p,m. The quantity depicted is not a proper
wavelength or skin depth until it is multiplied by the
basalt with r = 1 /zm. If adsorption did not square root of the forcing period.
occur the pressure wave could penetrate to
30 m. Adsorption greatly damps diffusion
because part of any v a p o r phase pressure
drop is made by desorption, and much (c) Damping o f the Annual Pressure
more mass resides in the adsorbed phase Wave by the Regolith
than in the vapor phase. Dzurisin and Ingersoll (1975) suggested
It should be recognized that no experi- that the annual CO2 pressure wave (recall
mental studies of diffusion of CO2 through Figs. 4 and 7) might be damped if the
analogs of typical Martian soils are avail- atmosphere were in contact with an isother-
able. Studies of terrestrial analogs would be mal CO2 reservoir such as the regolith.
quite useful to determine more precisely Before considering exchange of CO2 with
the time required for COz to move through the regolith over time scales relevant for
soils. climate, it is first o f interest to consider
568 TOON ET AL.

whether or not the regolith now interacts o f CO2 from the regolith involving the an-
with the annual CO2 pressure wave. Any nual thermal wave. The total COs adsorbed
such observed interaction could provide on a basalt regolith which is 1 annual ther-
experimental guidance for climate prob- mal skin depth deep is only 2% of the
lems. Present observations o f the sea- atmospheric pressure. H o w e v e r , a 1-m clay
sonal atmospheric pressure wave are in regolith may contain 1.5 mbar of COs. Dur-
agreement with simple models not involv- ing 300 days CO2 could penetrate 1 m o f
ing any regolith absorption (Hess et al., clay with a pore radius of 1/~m. H o w e v e r ,
1977, 1979). However, careful compari- the thermal wave probably would not re-
sons o f models o f the annual pressure lease the full 1.5 mbar of CO2 unless r or the
wave with observations have not been adsorption is larger for clays than we ex-
made. pect. A considerable phase lag would exist
The total amplitude of the annual pres- and, in addition, the amplitude would be
sure wave is about 2 mbar. Figure 10 shows damped because adsorption in the winter
that a 2-mbar pressure drop o v e r a basaltic hemisphere would oppose desorption in the
soil will desorb 1 to 2 × 10-4 g o f CO2/g o f summer hemisphere and because the sea-
soil depending on the temperature. Hence, sonal thermal wave is significant only at
to desorb 1 mbar o f CO2, the pressure drop high latitudes.
must occur o v e r a depth of 75 to 150 m. If a
pressure wave is to penetrate 75 m in 300 (d) Atmospheric Pressure Changes Due
days, then Ld from Eq. (7) must be about 2 to Thermal and Pressure Forcing
cm sec -lj2. Figure 11 requires a pore size The characteristic period of the Martian
considerably in excess o f 50/xm for Lo = 2 orbital elements is 105 years, during which
cm sec -1/z. If we assume that the soil is time a thermal wave could propagate 300 m
composed of clays and that the pressure and a pressure wave in r = l-/zm basalt
drop of 2 mbar leads to desorption of 1 to 2 could penetrate 1.5 km. Hence, a basaltic
× 10-3 g of CO2/g o f soil, then we need to regolith could desorb COs essentially instan-
desorb only 7.5 to 15 m of regolith. F o r taneously in response to a slow thermal
such desorption to occur in 300 days a pore wave. F o r clays with r = 0.01 /zm, a
size of about 50/~m would be required. pressure wave could travel only about 40 m
These results suggest that the annual during 105 years. H o w e v e r , a 100-m-thick
pressure wave will not be damped by the clay regolith would contain such a large
regolith unless the pore size is much larger amount of CO2 that desorption over the
than we now suspect or unless surface region affected by the thermal wave would
diffusion is more important than we expect. be similar in amount to that for 1 km o f
Experimental studies of terrestrial analogs basalt. The important point is that pressure
might be useful to gain further insights into diffusion in a basalt or clay regolith on
pore sizes and surface diffusion phenom- climatic time scales probably does not com-
ena. The important point, however, is that pletely prevent the regolith from respond-
the lack of observed interaction between ing to climate changes.
the regolith and the atmosphere on an an- We can now estimate the maximum
nual time scale is due to the difficulty of amount o f CO,, exchanged between the
diffusion through the regolith and it is not atmosphere and regolith caused by the
evidence that Martian soil contains no ad- pressure and temperature changes found in
sorbed CO2. Section II by simply assuming that the full
If future observations do indicate rego- pressure or temperature change at the sur-
lith-atmosphere interaction on a seasonal face propagates through the regohth ex-
time scale, it might prove fruitful to con- changing the maximum possible adsorbed
sider an alternate mechanism for removal CO2 according to Figure 10; and that all the
 MARTIAN CLIMATE 569

COz exchanged either reaches the atmo- temperatures. Our analysis, however,
sphere or is removed from it. We now shows that realistic temperature changes in
perform such a calculation for basalt and equatorial and polar latitude are rather
clay regoliths. small and partially compensating, so that
Figure 9 suggests that, if the obliquity the net pressure change is very small even
changed from 25 to 35° (with e = 0), the when the pressure dependence of the iso-
ground equatorward of 55° latitude would therms is ignored.
cool by 2 to 5°K while the ground poleward Fanale and Cannon (1979) have also eval-
of 55° latitude would warm by 15°K. We uated clay and basalt desorption for a 15°K
can find the maximum possible COo de- polar temperature increase. For clays, they
sorbed by assuming: that the pressure de- suggested that 50 mbar of COz would be
pendence of the adsorption can be ignored; desorbed at high obliquity, but that correct-
that A T occurs over the whole depth of the ing for the pressure dependence of the
regolith; that equatorward of 55° latitude T isotherms would reduce the pressure to
= 220°K and AT = -2°K; and that pole- about 29 mbar and correcting for latitudinal
ward of 55° latitude A T = 15°K while T = dependence of adsorption would further
160°K. Then we find that the equatorward reduce the pressure to 17 mbar. They as-
region adsorbs (over 300 m of regolith) sumed a larger volume of polar regolith, a
about 1.2 g/cm 2 and the polar regions de- higher soil density, and a higher tempera-
sorb about 20 g/cm z. However, only 20% ture dependence of the isotherms at low
of the planet' s area is above 55° latitude, so temperatures than we did, but their final
the planetary average CO2 desorbed is answer is similar to ours. For basalt, Fanale
about 3 g/cm 2. Since the present atmo- and Cannon (1979) found AP = 9 mbar.
spheric pressure on Mars is equivalent to 16 However, they considered a 1-km-thick po-
g/cm 2 the pressure change will be equiva- lar regolith extending over a much larger
lent to only 1.15 mbar. A similar calculation geographic area and a higher soil density (2
for clays suggests that Ap = 20 mbar for a g/cm a compared with our 1.6 g/cm a) than
300-m regolith. we did. The difference between our pre-
If we did not include the equatorial ad- dicted pressure at high obliquity for basalt
sorption, then the polar warming could (7.15 mbar) and theirs (15 mbar) is a mea-
have increased the pressure by about 1.5 sure of the inherent uncertainty of the cal-
mbar for basalt and 25 mbar for clays. The culation.
result is linear with regolith depth, so if the The preceding discussion has centered
thermal wave reaches 1 km rather than 300 on the role of temperature oscillations in
m, then 5 mbar could be released from desorbing CO2 from the Martian regolith.
basalt and 82.5 mbar from clays. For clays 'The effects are small because: (1) tempera-
it is very difficult to justify a 1-km regolith ture changes from orbital perturbations are
which is diffusively coupled to the surface. small and of opposite sign at different lati-
In addition, desorbed CO2 comes from very tudes; (2) the thermal wave cannot propa-
high latitudes where permafrost is likely to gate very far in 10s years; and (3) desorp-
block the pores and further decrease the tion is limited by the pressure dependence
CO2 desorption. of the adsorption isotherms.
Fanale and Cannon (1978) argued that the Orbital element changes cause not only
results of even large (40°C) temperature temperature changes but also pressure vari-
changes over 105 years will be small (30 ations when the permanent polar cap forms
mbar) for basalt because the adsorbed mass at low obliquity. Basalt adsorption is a
depends on pressure as well as temperature strong function of pressure when P is less
so that one cannot move along vertical lines than a few millimeters of Hg. As Fig. 4
in Fig. 10 in studying desorption for varying shows, the atmospheric pressure can fall to
570 TOON ET AL.

1 mm or less at low 0. Such low pressures sure, is a very strong function of surface
could lead to nearly complete desorption of albedo, which is not well known, so that the
a basaltic, r = 1/xm, Martian regolith. (The equality o f the vapor pressure and the ob-
temperature changes due to the altered 0 served pressure is not very exact. Viking
have negligible effects on adsorption by results (Kieffer et al., 1976) have now
comparison.) For clays with r -~ 0.01 /zm shown that the north polar permanent cap
perhaps only 40 m could be desorbed. F o r of Mars is not solid CO2, although the south
clays we are not certain that the isotherms polar cap may contain some solid CO,,.
are strongly sloping with pressure at 1 mm, Since the behavior o f the south polar cap
but a strong slope must develop at some during Viking observations seemed anoma-
low pressure. In either case, permanent lous (James et al., 1979), it is likely that the
CO., caps may form not only from the south polar cap does not contain a substan-
present atmospheric mass but also from the tial CO,, reservoir. Even if the entire cap
adsorbed mass. Assuming that the perma- were CO,, it would only contain the equiva-
nent caps occupy 1% o f the area of Mars, lent of 6 mbar o f CO,, (Fanale and Cannon,
the thickness of a permanent cap formed 1979). H e n c e , the atmospheric pressure is
from the present atmosphere would be probably not the vapor pressure of polar
about 10 m. If there is even as little as 30 CO,, deposits. This means that the observed
times as much CO,, mass in the regolith as pressure is either a nonequilibrium tran-
in the current atmosphere, the polar caps at sient or that it is in equilibrium with the
low obliquity could reach a depth of 300 m. regolith as previously suggested by Fanale
It is interesting to speculate concerning and Cannon (1978).
possible hysteresis effects in the regolith- Figure 11 emphasizes that for moderate
polar cap system. F o r example, suppose pore sizes even a 1-km regolith would be in
declining obliquity reduces the atmospheric equilibrium with the atmosphere on time
pressure to a very low value and slowly scales longer than about 1@ years, which is
desorbs the regolith. If some other process, much shorter than reasonable time scales
such as an episode of volcanic activity, for outgassing, or carbonate formation. In
occurred to blacken the polar cap, then the addition, the annual mean pressure is not a
CO,, could be driven into the atmosphere. strong function of obliquity except for ob-
The CO,, might not return to the cap if liquities less than 20° , or greater than 30° ,
feedback effects maintained high polar tem- and Ward's (1979) calculations suggest that
peratures (Gierasch and T o o n , 1973) and the Martian obliquity has not been lower
the CO2 could not reenter the regolith im- than 20° nor higher than 30° for nearly 5 ×
mediately because of diffusion-time con- 105 years. H e n c e , it is likely that the current
straints. pressure is in equilibrium with the regolith.
As previously discussed, the thermal wave
(e) Regolith Maintenance of Mean due to the obliquity oscillation should pro-
Atmospheric Pressure over Geologic duce only small pressure changes.
Time Scales If the atmospheric pressure is controlled
The remaining question concerning rego- by regolith adsorption, then the present
lith adsorbed CO2 is its role in maintaining atmospheric pressure on Mars is the result
the mean Martian atmospheric pressure. of the total CO2 outgassed, the present
Leighton and Murray (1966) noted that the regolith depth, and the present conversion
Martian atmospheric pressure is very close of CO2 into nonadsorbed forms. The pres-
to the vapor pressure of CO2 at reasonable sure is exceedingly sensitive to any
Martian polar temperatures. Ingersoll modification of these parameters. F o r ex-
(1974) pointed out, however, that the polar ample, if the depth of a basalt regolith were
temperature, and therefore the vapor pres- increased by 20% the adsorbed CO,, per
 MARTIAN CLIMATE 571

gram would decrease by 20% and Fig. 10 was distributed quite differently than now.
indicates that the atmospheric pressure Perhaps early in Martian geologic history
would fall by about 50%. Likewise, CO2 is CO2 had been outgassed, but no regolith or
probably converted to carbonate on Mars, only a thin basalt regolith had formed. Then
although we have no direct evidence of the CO2 would remain in the atmosphere-polar
carbonate abundance (Gooding, 1978). If cap system. Alternatively, suppose that the
4% of the regolith is CO2 tied up in carbon- regolith on Mars is not distributed uni-
ates, then approximately 20 times more formly with latitude. At present, extensive
CO2 is stored as carbonate than as adsorbed dust laminae of total thickness between 1
CO~ on basalt. Figure l0 shows that in- and 6 km extend poleward of 80° at both
creasing the adsorbed COs by a factor of 20 Martian poles (Dzurisin and Blasius, 1975;
by converting the carbonates to CO2 would Cutts et al., 1976). In addition, high Mar-
lead to an immense increase in the atmo- tian latitudes are covered by a deep mantle
spheric pressure. The precise pressure can- presumably of loose, wind-eroded debris
not be calculated because the adsorption at (Soderblom et al., 1973). At present T =
high pressures has not been measured and 170°K for these materials, but suppose that
many adsorption isotherms have greatly in the past they were all at the equator
increased dependence of adsorption on where T = 220°K and that they then cov-
pressure at higher pressures (e.g., Bru- ered one-half the planet's surface to a depth
nauer et al., 1967). However, for a current of 500 m. The resulting amount of desorbed
regolith containing 50 mbar of adsorbed COe, if the regolith were basalt, would have
CO2, it would certainly seem possible for been about 15 mbar. If the regolith were
the atmospheric pressure to exceed 150 clay, then 180 mbar of CO,, could have been
mbar if all the hypothetical carbonates were desorbed although the pressure depen-
converted to adsorbed CO2. If the regolith dence of the isotherms would damp some-
is composed to clays, then carbonates may what the net CO2 entering the atmosphere.
contain only 50% of the total CO2, and
converting the carbonates to CO2 would (f) Summary
only result in a modest pressure increase. As much as 1 to 3 bars of CO2 may be
Over geologic history various processes, adsorbed on the Martian regolith or perma-
such as greenhouse warming, might have nently tied up as carbonates in the regolith.
maintained the Martian surface tempera- Diffusion of CO2 through the Martian soil is
ture at much higher temperatures than at slow, preventing the adsorbed CO2 from
present. The curves in Fig. 10 show that at affecting the annual CO~ pressure cycle. On
low temperatures basalt adsorption is a climatic time scales diffusion cannot be
strong function of pressure. For example, ignored, but it probably does not prevent
all the CO2 desorbed by a sample of 6 mbar, the regolith from responding to climate
if warmed from 160 to 195°K, would be changes in the surface pressure and temper-
readsorbed before the pressure reached 25 ature. In agreement with Fanale and Can-
mbar. However, experimental data suggest non (1979) we find, even given the most
that at high temperatures (273°K) the ad- favorable assumptions, that polar ground
sorption is a weak function of pressure. If temperature changes are too small to de-
some processes (greenhouse warming in an sorb large quantities of CO2 at high obliqui-
early atmosphere) considerably warmed ties. The pressure at 0 = 35° is likely to be
Mars for a long period, then little CO,, no greater than 20 mbar for either a clay
would be adsorbed on the regolith. regolith or a basalt regolith. At low obliqui-
There may have been some period when ties, the low atmospheric pressure caused
Mars had no regolith, had only a weakly by the formation of a permanent COz polar
adsorbing regolith, or had a regolith that cap may strongly desorb the regolith and
 572 TOON ET AL.

large quantities of CO2 from the regolith hances dust-lifting winds over a large area
may be deposited on the polar caps. The of the planet (Pollack et al., 1979). The
present atmospheric pressure is likely to be wind speed required to lift dust depends
in equilibrium with the regolith as sug- upon the pressure, while the wind veloci-
gested by Fanale and Cannon (1978). Over ties themselves depend upon horizontal
geologic time the atmospheric pressure has thermal gradients.
probably greatly decreased as carbonates Wind tunnel studies (Pollack et al., 1976)
have formed, as highly adsorbing clays have shown that the wind speed, u, at the
have formed, as the regolith has become top of the boundary layer required to lift
deeper, and as the regolith has been prefer- dust varies with pressure as u a P -°.35. At
entially deposited in polar regions. During a present, u is about 95 m/sec, but ifP were 1
possible early epoch with high surface tem- mbar, u would be 230 m/sec. I f P were 25
peratures, adsorption would not have been mbar, u would be about 40 m/sec. Since
an important sink for COz. Martian winds now are barely sufficient to
lift dust, it is clear that at lower obliquities
IV. THE SEASONALCYCLE OF and lower atmospheric pressures dust lift-
ATMOSPHERIC DUST ing will not occur (Ward et al., 1974; Pol-
Global-scale dust storms, which now oc- lack, 1979). At slightly higher pressures,_
cur every Martian year, are an important which could be reached at high obliquity if
component of Martian meteorology and cli- the regolith is composed of highly adsorb-
matology. The atmospheric thermal struc- ing clays, dust lifting should occur much
ture is stablized by solar energy deposition more easily than it does today (Pollack,
in suspended dust grains. Particulate back- 1979).
scattering of solar radiation may sig- At the present time, global dust storms
nificantly reduce the solar energy reach- occur preferentially during southern hemi-
ing the summer polar cap and alter its heat sphere summer, presumably because wind
budget. Dust grains deposited in the perma- speeds reach their peak at that time as the
nent and seasonal ice caps substantially solar insolation is maximized at perihelion
modify the ice albedo which also alters the passage. At slightly higher pressures, dust
polar heat budget. storm generation may occur during the
As we have discussed, variations in the summers of both hemispheres because of
orbital elements can modify the atmo- the less restrictive wind speed require-
spheric pressure, the temperature, and the ments. Hence, at both larger and smaller
seasonal polar cap boundary. Below we obliquities than those of today, dust storm
review ideas about how these modifications generation may be independent of the ec-
may alter the dust storm cycle and then we centricity and precessional cycles.
discuss how changes in the dust storm The mechanisms responsible for creating
cycle can "feedback" to modify pressure, winds strong enough to lift dust are not
temperature, and polar cap boundaries. known well enough, nor are they simple
enough, to allow more than qualitiative
(a) Effects o f Climate Changes on Dust extrapolations to other climates. Pollack et
Storm Generation and Decay al. (1979) discuss the possible importance
The generation of global dust storms on of polar cap, tidal, topographical, frontal,
Mars requires local winds at the high veloc- and seasonal winds for dust storm genera-
ity end of the Martian wind spectrum. tion at present.
These winds lift dust locally. Then a posi- Polar cap winds are driven by the thermal
tive feedback occurs between radiant en- contrast between frost-free and frost-cov-
ergy deposition in the suspended dust ered terrain and by the CO2 mass flux into
grains and the global winds; this then en- and out of the polar caps. Peterfreund and
 MARTIAN CLIMATE 573

Kieffer (1979) reported that polar cap winds winds that drive the dust-lifting processes
generate many local dust storms. The ther- (Pollack et al., 1979). For this reason,
mal contrast winds will be unaffected by global dust storms during a single season
changes in the orbital elements, but the CO2 may not be more frequent at high obliquity
mass flux will be enhanced and the edge of than they are at present during southern
the polar cap will move equatorward at high hemisphere summer. The decay of the dust
obliquity because of the larger seasonal storms should be nearly independent of any
exchange of CO2 as seen in Figs. 4 and 5. orbital changes. The removal of the dust
Topographical winds are driven by large- from the atmosphere apparently occurs
scale elevation differences, while tidal preferentially in polar regions as CO2
winds are driven by the diurnal movement of " s n o w " forms around the dust grains and
the subsolar point. The winds should be makes them large enough to fall out of the
affected in three ways by orbital element atmosphere (Pollack et al., 1979). Jones et
changes. First, both wind systems depend al. (1979) observed from Viking that even
upon the small value of the radiative relaxa- nighttime CO2 frost formation is sufficient
tion time of the Martian atmosphere. At to remove dust from the atmosphere. Polar
higher obliquities, and hence higher pres- CO2 deposits will cover a larger geographic
sures, these wind systems could be damped area at higher obliquity and more mass will
because the radiative relaxation time will be exchanged between the atmosphere and
become longer. Second, at higher eccen- poles at higher obliquity. Hence, dust may
tricities the diurnal temperature gradients be removed from the atmosphere slightly
will be accentuated near perihelion, and faster at high obliquity than at low obli-
this may increase the amplitude of tidal and quity. Pollack (1979) has pointed out that
topographic winds. Finally, the time of year the polar debris mantles probably mark the
of perihelion passage and the obliquity may limits of the CO2 annual cap at high obli-
affect the phase, amplitude, and latitude of quity.
wind speed maxima. These changes could
be important if tidal or topographic winds (b) Effects of Changes in Dust Storm
interact either with each other or with sea- Activity on the C02 Cycle
sonal winds to produce additive effects. The CO2 cycle on Mars is controlled by
Seasonal and frontal winds will be altered the polar energy budget which is a strong
by thermal gradient modifications due to function of the orbital parameters and of
obliquity changes. The frequency of frontal the polar cap albedo. As the obliquity
systems may increase and fronts may be varies from 15 to 35° the annual average
concentrated in lower latitudes at higher solar energy available at the pole changes
obliquity because midlatitude temperature by nearly a factor of 2. If the COz frost
gradients will decrease while lower latitude albedo were to change from 0.6 to 0.8, or if
gradients will increase (Figs. 3 and 6). In the permanent I-~20 ice cap albedo were to
addition, at high obliquity, summertime change from 0.4 to 0.7, the energy budget
temperature gradients as large as those would also change by a factor of 2. Below
during winter, but of opposite sign, will we shall show that changes in the dust-to-
occur between the summer pole and equa- ice mixing ratio can cause such changes in
tor. Strong summertime easterlies would be the ice albedo so that modifications of dust
expected with possible summertime frontal storm activity due to orbital element
activity. change may be just as important to the
Global dust storms start to decay as soon polar energy budget as are variations in
as the storms are generated because radia- solar insolation caused directly by orbital
tive heating in an optically thick dust layer element variations.
stabilizes the atmosphere and shuts off the The albedo of the Martian polar caps is
574 TOON ET AL.

significantly less than unity. The north sum- ~o

0.99999 0.9999 0.999 0.99 0.9 0
mer water ice cap is observed to have an 1"0 1~ " ~ [ ' [ ' ' ' 4

albedo near 0.40 (Kieffer et al., 1976). The

albedo of the Martian southern permanent
ice cap is not known, but F a r m e r et al.
,~ .0
- - g=o.e9 "~\
t
-~
(1979) state that it is >0.55. The albedo of
the seasonal CO2 caps has not been mea- ~ .2
sured in detail, although James et al. (1979)
0 , I i I L I , ,"~"1
report values of 0.73 in the red and 0.36 in 10 -5 10 -4 10 -3 10 -2 10 -1 1
the violet. A spectrally averaged albedo of (1 -~o)
0.65 +_ 0.05 has long been used in models of FIG. 12. The spherical albedo of a dirty ice cap
the seasonal polar cap recession (e.g., Kief- varies greatly with the single-scattering albedo of the
fer et al., 1977; Davies et al., 1977). F o r ice-dust grains c o m p o s i n g the ice, but it is only a
comparisons, the albedo of the Earth's Ant- moderate function o f the a s y m m e t r y p a r a m e t e r of the
grains.
arctic ice cap is about 0.82 in summer and
0.88 in winter (Hummel and Reck, 1979);
terrestrial snows on open tundra have al- stream approximation (Irvine, 1975) as a
bedos near 0.8. function o f the single-scattering albedo t~0
The Martian ice caps could have low of the ice-dust grains for a s y m m e t r y factors
albedos for three reasons. The first, that the g of 0.88 and 0.80. Pollack et al. (1979) find
ice may be optically thin, is unlikely. Ter- t h a t g for Martian dust is about 0.8 and &0 is
restrial observations show that only a few about 0.86. The calculated Martian surface
centimeters of snow is enough to com- albedo from Fig. 12 is thus about 0.20,
pletely mask the ground. In addition, sim- which is in agreement with the observed
ple two-stream radiative transfer calcula- albedo of 0.214 _+ 0.063 (Kieffer et al.,
tions of the albedo of CO~ frost, overlying 1977), confirming that the two-stream rep-
ground with typical Martian soil albedo, resentation gives a reasonable approxima-
show that only 1 cm of compacted CO2 is tion to the albedo.
enough to raise the albedo to 95%, Both the The surface albedo is a strong function of
CO~ seasonal caps and the H~O permanent t~0, but it is only a weak function o f g when
caps are much thicker than a few centi- in the range of 0.8 to 0.88. According to
meters except perhaps in isolated regions, Hunt (1973), a value of 0.85 to 0.88 f o r g at
or during the early stages of ice deposition. visible wavelengths is representative of ice
A second possible cause of low ice al- particles for sizes between 4 and 50 /zm.
bedos is exposed patches of bare ground. Since the refractive index o f CO2 is close to
The albedo of broken snow is A = f A i e e + that of H,O ice, a similar value o f g applies
(l - f ) Aground, w h e r e f is the fraction of ice to CO2 ice (Thompson et al., 1973).
cover. For an ice albedo o f u n i t y , f must be Figure 12 shows that a polar cap albedo
0.53 to obtain an albedo of 0.65, a n d f must of 0.4 requires that &0 = 0.97, while an
be 0.2 to obtain an albedo of 0.4. Although albedo of 0.65 requires that &0 = 0.994. The
exposed ground may help lower the polar single-scattering albedo (t~0) of a mixture of
albedo, the bare ground fraction suggested ice having unit albedo with dust o f albedo
by these numbers is so large that some &0d can be related to the volume mixing
other process must be mainly responsible ratio of dust (V) by assuming that the ice
for the low albedo. and dust particles are individual grains and
The final possible cause of a low polar that the optical depths of the dust and of the
albedo is that dust is admixed with the ice. ice grains are given by the geometric cross
Figure 12 shows the spherical albedo of a sections, a good assumption for the particle
semi-infinite layer found from the two- sizes of interest. Then it is easy to show
 MARTIAN CLIMATE 575

that the N~/Nd values just found. For twice the

dust input o f Pollack et al. (1979), the polar
V = (N,/Nd)~/2[(I - &o)/(tho - &od)]3/2 (8)
albedo would decrease from 0.65 to 0.55.
where N i / N d is the ratio of the number of For one-half this dust input, the albedo
ice particles to the number of dust particles. would increase to 0.7, and for one-tenth
Pollack et al. (1979) suggest that each this dust input, the albedo would b e c o m e
water ice particle arrives at the polar cap 0.85.
surrounding a dust particle so that N~ = Nd. Because of the strong sensitivity o f al-
Hence, the dust-to-H,O ice mixing ratio in bedo to dust content, a strong feedback
the north polar cap, if it has an albedo of exists between dust storm frequency, polar
0.4, is about 15%. albedo, and CO2 pressure. At high obli-
We cannot use N i / N d = I for COs, quity, the release of COs from clay reser-
however, since it is likely that much more voirs may provide enough additional pres-
COs is deposited directly onto the cap from sure to increase dust storm frequency. The
the vapor phase than is formed as dust- greater dust storm frequency in turn should
contaminated snowflakes in the atmo- cause the COs albedo at both poles to be
sphere. The difference in the behavior of low and so encourage higher pressures. On
H20 and CO2 frost occurs because, unlike the other hand, at low obliquity the lower
H,O, the vapor abundance of COs is quite pressures (Fig. 4) will decrease the dust
high above the pole, so that COs can di- storm frequency, the polar cap albedo will
rectly condense onto the surface. An addi- increase, and still lower pressures should
tional factor is that large COs flakes falling be favored. This feedback should greatly
rapidly through the atmosphere will prefer- enhance the possibility of the formation of
entially sweep up large numbers of dust permanent COs polar caps at both poles
particles in their path. A few large COs when the obliquity is lower than at present.
flakes will therefore be mixed with many Dust storms under present conditions
smaller dust particles. occur preferentially during southern hemi-
F r o m Viking observations o f the abun- sphere summer. Therefore, the albedo of
dance of dust in the Martian atmosphere CO2 frost should be higher in the southern
Pollack et al. (1979) estimated that 2 × l0 -2 hemisphere than in the northern hemi-
g cm -2 year -1 of dust is deposited poleward sphere. If the dust-to-COs mixing ratio is
of 60 ° in the north polar region of Mars. lower in the southern hemisphere than in
Hess et al. (1979) estimated from the ob- the northern hemisphere by a factor of 10,
served pressure variation on Mars that the the albedo difference could cause the solar
mass of COs deposited in the south polar energy budgets of the northern and south-
region is about 35 g / c m 2 o f CO2 spread o v e r ern poles to differ by a factor of 2. Better
45 ° latitude. The mass of COs in the sea- observations of the difference in the al-
sonal north polar cap should be less (Fig. bedos between the two seasonal caps and
7). H e n c e the volume mixing ratio of dust the variations of albedo as the caps sublime
to COs in the north polar cap is approxi- during the spring might provide insight into
mately l0 -3. Using Eq. (8) and a polar the importance of dust to the polar cap
albedo of 0.65 we find that for each COs albedos.
particle there are about 100 dust particles. The effect of embedded dust on the polar
If the average dust particle radius is 1 /~m, ice albedo is very large. Atmospheric dust
then the average COs radius is 50 ~m. could also affect the polar energy budget by
If we assume that the COs deposition is blocking solar and thermal radiation. Pol-
constant, we can calculate the polar albedo lack et al. (1979) showed that the surface
change caused by variation in the dust energy budget at low latitudes is not sub-
mixing ratio by using Eq. (8) scaled from stantially influenced by atmospheric dust
 576 TOON ET AL.

because the increased ground-level infrared manent polar cap albedo should be greater
radiation from the dust offsets the ground- than the seasonal cap albedo now. At high
level solar energy deficit. Calculations of obliquity the COs seasonal caps at both
the radiation balance in the polar regions poles should have albedos similar to, or
have been performed by James et al. perhaps slightly less than, that of the
(1979). They found that dust with a t~0 of present seasonal north polar COs cap. At
0.75 would reduce the amount of solar times of moderate 0 and high e, such as the
energy absorbed at the polar surface if the present, the two COs seasonal polar caps
polar albedo were 0.5, whereas if the polar should have significantly different albedos
albedo were 0.6 the energy absorbed would because of the seasonal asymmetry in dust
increase. If ~0 were 0.85, James et al. storm occurrence. The solar energy budget
determined that the dust decreased the of the two polar caps could easily differ by a
amount of energy absorbed at the pole for factor of 2 because of the difference in the
both polar cap albedos. James et al. ignored dust content of CO~ frost. Better observa-
the contribution of the dust to the infrared tions of the present polar cap albedoes
radiation at the ground, which Pollack et al. would be valuable.
(1979) found to be important. Suspended
dust may either increase or decrease the v. THE SEASONALCYCLE OF H~O
energy available at the polar surface de- Water vapor in the Martian atmosphere
pending upon the values of polar albedo, undergoes a strong seasonal variation as it
single scattering albedo, and the ratio of condenses on and sublimes from the perma-
infrared-to-visible dust extinction. Much nent H20 polar caps. Over long times,
further work is required on this difficult water vapor also interacts with permafrost
problem, but the sensitivity suggests that and with water bound to soil particles. We
the net effect is probably not large, espe- now consider the water seasonal cycle un-
cially when it is compared with the factor- der different orbital and obliquity condi-
of-2 change in the energy budget that could tions with most concern being given to the
result from admixing different amounts of stability of the polar caps. Then we discuss
dust and COs in the polar ice. long term changes in the amount of perma-
frost and soil water.
(c ) Summary
In agreement with previous studies (a ) 1-I20 R e s e r v o i r s - - - A t m o s p h e r e a n d
(Ward et al., 1974; Pollack et al., 1979; Polar Caps
Pollack, 1979), we find that at low obliquity Presently the Martian atmosphere con-
low pressures will prevent dust storms. At tains seasonally varying water vapor equiv-
high obliquity, slightly higher pressures will alent to 0.3 to 1.3 km 3 of ice (Farmer and
enable dust storms to occur throughout the Doms, 1979; Farmer et al., 1977). Two
Martian year independently of the preces- permanent water ice caps are located at the
sion of the equinoxes. Since dust storms Martian poles; their combined area is about
are self-limiting, the intensity of individual 10~ km z, approximately one-half the area of
storms, their lifetimes, and their frequency the Earth's Greenland ice cap or one-tenth
of occurrence should be similar to those the area of the Earth's Antarctic ice cap.
found now during southern hemisphere The thickness of the Martian polar caps is
summer. At moderate obliquities, dust not known, but Cutts et al. (1976) suggest
storms will be most prevalent at large ec- that ice is very thin (~< 1 cm) in areas of the
centricity when perhelion occurs near sol- south pole where the visible ice cover is
stice. A strong feedback occurs between variable. Dzurisin and Blasius (1975) de-
dust storm frequency and polar cap albedo. duced that the ice cover is typically a few
At low obliquity the COs seasonal or per- hundred meters or less since ice does not
 MARTIAN CLIMATE 577

c o v e r topographic areas with greater height since bare ground n e a r the pole is w a r m e r
differences. Murray and Malin (1973) than ice at the pole, which will lead to a
identified one polar region in which ice did w a r m a t m o s p h e r e o v e r the pole, and
s e e m to c o v e r the t o p o g r a p h y , implying since observations show that the atmo-
that the thickness in that region m a y exceed sphere o v e r the pole is stable (Kieffer et
a few hundred meters. al., 1976). F o r heat conduction in a stable
a t m o s p h e r e we follow Gierasch and Toon
(b) M a x i m u m S u m m e r Ice Cap
(1973 t and use
Temperatures and Sublimation Rates
Altl = f l P U ( T A - T),
We have shown b y using a simple model
(Figs. 3 and 6) that s u m m e r polar ground where
t e m p e r a t u r e s can be quite high at large
fl = 2.6 x 10 -5 °K -1
obliquities. The thermal history of an ice
cap is m o r e c o m p l e x than that of the soil U = typical wind speed = l0 a cm see -~,
and, thus, a m o r e detailed model is required P = pressure in dynes cm -2. (10)
to explore its t e m p e r a t u r e variation and the The atmospheric heating due to CO2 and
associated water v a p o r loss rate. The heat H20 are found by using the emissivities of
balance equation for an ice cap is Gierasch and G o o d y (1968) and Cess
(1 - A ) I + A H - et~T 4 (1974):
OT OT A H C O 2 = 6 × 103In(1 + 1.97 > 10-3p)
- Kite O---z- E c = p C p - ~ " (9)
e x p ( - 9 6 0 / T g ) e r g s c m -2 sec -~,
The initial t e r m is the daily averaged A H H 2 0 = 0.75 O-TA4[1 -- e x p ( - ( P n ~ o P ) It2
available solar radiation. This t e r m con- 9.1 × 10-5)]ergs cm -2 sec -t.
tains an ice albedo A taken as 0.4 in
a g r e e m e n t with present values for the In calculating the w a t e r emissivity PH2o, the
north polar cap (Kieffer et al., 1976). The water v a p o r pressure (in dynes cm -2) at the
Martian south polar cap has a higher al- surface is used. The w a t e r is taken to be
bedo (>0.55, F a r m e r and D o m s , 1979). uniformly mixed, and it is a s s u m e d that the
The term i is the daily average solar inso- column a b u n d a n c e of water does not ex-
lation. At the pole there is no variation o f ceed that which had been previously sub-
the insolation during the day. Ice is mod- limed from the w a t e r cap.
erately t r a n s p a r e n t to solar radiation, but The energy loss t e r m s in Eq. (9) include
Martian ice contains a large quantity o f ground radiation losses (we take an emis-
dust, as evidenced by the low albedo, so sivity o f 0.95) and conduction into the ice.
we have a s s u m e d that all the solar energy The conductivity o f pure ice at 273°K is
is deposited in the top layer (1.5 cm a b o u t 5 × 10 -a cal cm -~ sec -~ °K-~ and at
thick) o f our ice cap model. Depositing 150°K it is about 10-3 cal cm -~ sec -~ °K-~
the solar energy in a thicker layer would ( H o b b s 1974). Snow at 273°K has a density
result in lower temperatures. dependent K of 6.8 x 10 -a P]now. Dense
The second term in Eq. (9) represents arctic snow with p = 0.33 g cm -a has K = 7
a t m o s p h e r i c heating caused by thermal × 10 -4 cal cm -1 sec -1 °K-1 (Maykut and
conduction from the a t m o s p h e r e to the Untersteiner, 1971). Permafrost with 20%
ground, infrared radiation from COs, and soil content can have a 50% lower K than
infrared radiation from H~O. In order to does pure ice (Clark, 1966). To e n c o m p a s s
evaluate these t e r m s we a s s u m e that the the possible range of conductivities we per-
s u m m e r a t m o s p h e r e is isothermal and is form calculations with K = 8.6 × l0 -a
always w a r m e r than the polar cap b y TA -- (Davies et al., 1977) and with K = l × 10 -a
T = 10°C. This assumption is reasonable cal cm -~ sec -~ °K-1.
 578 TOON ET AL.

T h e final e n e r g y loss t e r m is e v a p o r a t i v e 300

I I I I I [ I I I
cooling. Two evaporative mechanisms are ..... 45 ° OBLIQUITY (45 cm, 20 cm)
35 ° (20 cm, 5 cm)
at w o r k . A t u r b u l e n t w a t e r v a p o r flux oc- 25 ° (1.5 cm, 0.2 cm)
c u r s as w i n d s b l o w a c r o s s t h e ice (Inger-
250 t f~,,
soll, 1974): I! I \"~' e = 0.14
" ~' X \ ".', A = 0.40
E C 1 = O.O02pw(O)UL, x',, /
L = l a t e n t h e a t o f ice
= 2.8 × 10'°ergs g - l ,
pw(0) = w a t e r v a p o r m a s s d e n s i t y . (11)

Also convective water vapor evaporation

o c c u r s b e c a u s e w a t e r is l i g h t e r t h a n CO2 I I I [ I '~'1 1-
( I n g e r s o l l , 1970) 200 240 280 320 0 40 80 120 160
SOLAR ANOMALY, deg

E C 2 = O.17pw(O)D[(Ap/p)(g /vZ)] v3 L FIG. 13. The seasonal dependence of the tempera-

ture of a water ice cap is a strong function of the
e r g s c m -2 sec - l ,
obliquity. Given in parentheses are the thicknesses of
where ice lost from each polar cap during the summer. The
greater thickness is lost from the pole reaching higher
~ P / P = 2 6 P H 2 o / ( 4 4 P - 26PH20), temperature. Substantial water loss rates occur at high
V = 3.75 × IO-7(TSr2/T + 2 6 9 ) ( R / P ) c m 2 obliquity. As discussed in the text, the loss rate is very
sec -~ = v i s c o s i t y o f COz, sensitive to ice albedo and thermal conductivity.
D = 32T312/P c m 2 s e c -1 = diffusion
coefficient o f HsO in COs, B o t h c a p s a r e a s s u m e d to h a v e an a l b e d o o f
R = gas c o n s t a n t . 0.4 a n d a K o f 10 -z cal c m -1 s e c -t °K -~. T h e
s o l a r i n s o l a t i o n i n t e g r a t e d o v e r t h e y e a r is
H e r e w e a s s u m e t h a t t h e r e is negligible i d e n t i c a l at b o t h p o l e s b e c a u s e t h e p o l e
w a t e r v a p o r e x c e p t j u s t a b o v e t h e ice. w i t h t h e l a r g e r i n s o l a t i o n at s o l s t i c e h a s t h e
T h e final t e r m in Eq. (9) is t h e t e m p e r a - shorter summer. However, sublimation of
t u r e c h a n g e o f t h e ice. W e u s e d a n ice h e a t ice at t h e t w o p o l e s is n o t t h e s a m e b e c a u s e
c a p a c i t y o f 1.4 × 107 ergs g-Z OK-~ w h i c h is it d e p e n d s n o n l i n e a r l y on t h e i n s o l a t i o n .
a p p r o p r i a t e for 200°K ( H o b b s , 1974). W e M o s t e v a p o r a t i o n o c c u r s w h e n t h e ice is
a l s o c o n s i d e r e d t h e e n e r g y in t h e v a p o r warmest.
phase molecules caused by vapor pressure A n e v a l u a t i o n o f t h e s e n s i t i v i t y o f the
c h a n g e s , b u t this effect is n e g l i g i b l y small. r e s u l t s p r e s e n t e d in Fig. 13 to t h e p r o c e s s e s
T o s o l v e Eq. (9) w e u s e d t h e s a m e ther- i n c l u d e d in E q . (9) s u g g e s t s t h a t the a l b e d o ,
m a l m o d e l as p r e v i o u s l y e m p l o y e d for the thermal conductivity, and evaporative
g r o u n d e x c e p t t h a t w e t o o k 20 l a y e r s o f 1.5- losses are the major uncertainties. Table I
c m t h i c k n e s s a n d 20 l a y e r s o f 30-cm t h i c k - presents the results from a number of calcu-
n e s s . W e b e g a n the c a l c u l a t i o n at t h e t i m e lations exploring the variation of the water
o f c o m p l e t e COs s u b l i m a t i o n as f o u n d f r o m m a s s l o s s r a t e c a u s e d b y c h a n g e s in t h e s e
o u r e a r l i e r c a l c u l a t i o n s (Figs. 3 a n d 6) a n d parameters.
set t h e b e g i n n i n g ice t e m p e r a t u r e at all H i g h e r w a t e r l o s s r a t e s c o u l d o c c u r if the
d e p t h s in t h e ice at t h e a n n u a l m e a n t e m - p o l a r a l b e d o w e r e less t h a n 0.4 but, b e -
perature from our earlier calculations. c a u s e t h e n o r t h e r n h e m i s p h e r e w i n t e r is
F i g u r e 13 i l l u s t r a t e s ice s u r f a c e t e m p e r a - n o w as d u s t y as is p o s s i b l e , w e d o n o t
t u r e s for v a r i o u s v a l u e s o f o b l i q u i t y w i t h believe that lower albedos are likely. A
the maximum possible eccentricity. Also c a l c u l a t i o n w i t h Kiee = 10 -4 cal c m -1 sec -~
n o t e d is the t o t a l t h i c k n e s s o f ice s u b l i m e d °K -1 p r o d u c e d negligible d i f f e r e n c e s f r o m
from each cap during the summer season. t h e r e s u l t s g i v e n in Fig. 13. D a v i e s et al.
 MARTIAN CLIMATE 579

TABLE I

POLAR CAP WATER ICE LOSS RATES ( g c m -2 year -~) FOR ORBIT OF MAXIMUM ECCENTRICITY

Fixed parameter Variable Obliquity Loss rate Loss rate

parameter from from
hot pole cool pole

25 1.5 0.2

K = 1 x 10-3 cal cm -~ see -1 °K -~

t A= 0.4
35
45
20
45
5
20

P = 5 mbar A =0.5 25 0.5 0.05

U = 10 m see -1 35 10 2
45 30 10

A = 0.6 25 0.1 0.01

35 5 0.3
45 15 2

K = 8.6 × 10-a cal cm -~ sec -1 °K-t

P = 5 mbar
A = 0.4 25
35
45
0.1
7
25
1
0.02

8
U = 10 m sec -~
A = 0.6 25 0.005 0.001
35 0.1 0.05
P = 5 mbar 45 43
K = 1 × 10-a cal cm -~ see - ' °K P = 25 mbar 35 40
U = 10 m sec -~ 45 78/0.3 melt
A = 0.4
P = 50 mbar 35 52/77 melt
45 69/282 melt
A = 0 . 4 ; P = 5 mbar 25 0.7
U = 10 m see -1
A = 0.4; P = 50 mbar 35 22
K = 8.6 × 10-a cal cm -~ sec -t OK
A = 0.6; P = 5 or 50 mbar 35 0.1
U = 10 m sec -1
A =0.4;P= 5 mbar 45 53
K = 1 x 10-4 cal cm -~ sec -t °K

(1977) argue that such low values of K fixed water vapor amount causes a smaller
would lead to inconsistencies with pres- f r a c t i o n a l d e n s i t y c h a n g e t h a n o c c u r s at
ently observed water vapor abundances smaller COs pressures. Without wind, but
over the north polar ice cap. Suppressing w i t h a n a t m o s p h e r i c p r e s s u r e o f 25 m b a r ,
evaporation by shutting off the wind has a w e f i n d t h a t f o r e = 0 . 1 4 a n d 0 = 45 ° i c e
small effect because convective evapora- temperatures at the warmer pole reach
tion limits the temperature. 2 7 3 ° K , 78 c m o f i c e s u b l i m e , a n d 0 . 3 - c m
I n g e r s o l l (1970) p o i n t e d o u t t h a t t h e c o n - melt. If the atmospheric pressure reaches
vective evaporative flux declines as the 50 m b a r , t h e n t h e m e l t i n g p o i n t o f i c e is
a t m o s p h e r i c p r e s s u r e i n c r e a s e s . T h e f l u x is r e a c h e d a n d m a i n t a i n e d f o r a b o u t 60 M a r -
inversely dependent upon atmospheric t i a n d a y s , 70 c m o f i c e s u b l i m e s , a n d 280
p r e s s u r e b e c a u s e it is d r i v e n b y d e n s i t y cm of ice melts. Ice melting also occurs
differences between the atmosphere just with P = 50 m b a r w h e n 0 = 35 °.
above the ice and the atmosphere at higher Now let us consider the factors that may
altitudes. With larger pressures, adding a reduce the water vapor loss rates. At low
580 TOON ET AL.

obliquity, suppressing evaporation reduces decline radically as 0 decreases because the

the water loss rate, as can be seen in Table I polar ice albedo and conductivity will rise
by comparing the 0 = 25 ° cases with and at low obliquity as dust storm f r e q u e n c y
without wind. At high obliquity, evapora- drops because of declining CO2 pressure.
tion is the t e m p e r a t u r e limiting process. In More importantly, if p e r m a n e n t CO2 polar
order to balance the energy, a p p r o x i m a t e l y caps form, the w a t e r loss rate will drop to
the same physical mass must be lost e v e n if an e x t r e m e l y small value. For e x a m p l e , at a
the e v a p o r a t i o n rate is reduced. This oc- t e m p e r a t u r e of 150°K the w a t e r v a p o r loss
curs by raising the temperature, which in rate is about 10-6 g cm -2 s u m m e r -I with a
turn increases the incoming energy from wind speed of 10 m / s e c at 5 m b a r of
a t m o s p h e r i c infrared radiation and hence pressure. Ingersoll (1974) and Leighton and
causes greater mass loss. Murray (1966) previously pointed out the
The most effective way to limit the mass great importance of p e r m a n e n t CO2 de-
loss is to increase the thermal conductivity posits as traps for water vapor.
or the albedo. Table I shows that, at large
obliquity, large w a t e r loss rates (several (c) Water Ice Deposition R a t e s at the
c e n t i m e r s / y e a r ) do not occur if the albedo Poles and Exchange between Polar and
and the conductivity are both high. Ice Nonpolar Ice Reservoirs
melting will not o c c u r at 0 = 35 °, even at 50- Water ice not only sublimes from the
m b a r pressure, if the conductivity is high. s u m m e r poles, but it also is transported
At a moderate 0, such as 25 °, the albedo toward the winter pole. Some ice m a y
and thermal conductivity also strongly con- reach the winter p e r m a n e n t pole to replace
trol the w a t e r loss rate. The present south ice lost during the previous s u m m e r , other
polar cap o f Mars m a y have an albedo of ice m a y be deposited in the annual CO2
>0.5 ( F a r m e r et al., 1979), even though the polar cap, and yet more ice m a y exchange
northern cap has an albedo of 0.4. Table I with mineral surfaces and permafrost.
shows that the southern cap with an albedo The available evidence does not allow us
of 0.6 would lose less water than the north- to estimate the net gain or loss rates for the
ern cap with an albedo of 0.4, as is ob- current p e r m a n e n t polar caps. The only
served. In addition, Table I shows that numerical data on water return to the poles
changing the thermal conductivity in the are those of F a r m e r et al. (1979) who ob-
range from 1 x 10- a t o 8 × 10-a alters the served that the globally integrated atmo-
water loss by a very large factor. spheric water v a p o r declined during the
Comparing the hot- and cold-pole loss northern hemisphere winter by an amount
rates in Table I with the loss rates for equivalent to 4 × 10-3 g cm -~ spread pole-
changes in various p a r a m e t e r s shows that ward at 60°N. The latitude was chosen to
the w a t e r loss rate d e p e n d e n c e on eccen- agree with the b o u n d a r y of the seasonal
tricity is no stronger than its d e p e n d e n c e north polar cap. Pollack et al. (1977, 1979)
upon obliquity, thermal conductivity, or noted that dust particles probably serve as
albedo. H e n c e a s y m m e t r i e s between the the carriers of water into the polar regions
p e r m a n e n t polar caps could be caused and that the mass of ice per dust particle is
equally well by differences in albedo, ther- about the same as the mass o f dust, sug-
mal conductivity, or position in the preces- gesting a water v a p o r deposition rate of
sional cycle, and the differences may well about 2 × I0 -z g cm -z y e a r -1 o v e r the area
be frozen in from an earlier phase of the of the north annual polar cap. The n u m b e r
obliquity cycle. is approximately equal to the deposition
Figure 13 and Table I do not show water each year of all the w a t e r v a p o r in a hemi-
loss rates f r o m the polar regions for obliqui- sphere containing 10 precipitable microme-
ties less than 25 ° . The loss rates are likely to ters o f water. H e n c e this value is close to
 MARTIAN CLIMATE 581

an upper limit for the possible current ice Clark (1978) shows that montmoriUonite
deposition rate. clays strongly adsorb H20 and that sul-
These water vapor return rates are less fates, such as MgSO4, undergo rapid hydra-
by 1 to 2 orders of magnitude than the water tion at temperatures near 210°K with
loss rates shown by Fig. 13 for O = 25°. The present Martian water vapor abundances.
implication is that both Martian polar caps A soil dominated by clays, but also contain-
are currently being destroyed or (Table I) ing about 10% magnesium sulfate, would
that the ice conductivity is close to pure ice contain 30, 15, or 5% of its dry weight as
for both caps and additionally that the water for temperatures of 195, 215, and
south cap has an albedo near 0.6. The 235°K, respectively, for a column water
highest temperature reached in the south content of 10 pr /zm of uniformly mixed
cap calculation for A = 0.6 and K = 8.6 × water. At 215 ° about half this H20 would be
10-3 cal-i g-Z OK-i was 193°K, which is attached to the sulfates. By contrast, a
much higher than the CO~ frost point on basalt regolith at 215°K along with a water
present-day Mars. pressure o f l0 pr /xm would only adsorb
There is no reason to believe that polar about 0.3% of its weight as HzO (Fanale
caps on Mars are permanent features. At and Cannon, 1974).
other times in Martian history the caps may The storage of H~O in clays and sulfates
have been larger, smaller, or even nonexist- is likely to be a strong function of latitude,
ent. A similar physical situation exists on depth, and season. Because o f the strong
Earth, and the evidence shows that the temperature d e p e n d e n c e of the adsorption
Earth's permanent polar caps were much and because of hysteresis effects, the diur-
larger 2 × l05 years ago and were absent 4 nal thermal wave probably permanently
x 107 years ago. Large ice sheets have dessicates the equatorial and midlatitude
probably existed on Earth for a total of less soil within a few centimeters of the surface,
than 10% o f the Earth's geologic history while the upper meter of soil in the polar
(Pollack, 1979). regions could exchange water because of
Since Mars has no oceans, its polar caps the seasonal thermal wave (see Clark,
can undergo long-term alterations only by 1978).
exchanging H20 with the regolith. The Figure 14 illustrates the adsorbed water
Martian permanent polar caps o c c u p y 1% on smectite clays for various temperatures
o f the planet. If the regolith were 10% H20 and pressures (Anderson et al., 1978). For
by volume, then only 10 to 100 m of glob- an average water vapor amount o f 10 pr
ally distributed regolith is needed to accom- /xm, soil colder than 195°K will be in vapor
modate all the ice in polar caps 100 to 1000 equilibrium with permafrost rather than
m thick. Carr and Schaber (1977) suggest with adsorbed water. On the other hand,
that the ice content of some Martian soils soil only a few degrees warmer than 195°
may be as high as 50%. If permafrost with will be strongly desorbed. In polar regions,
50% ice content spreads from 45 to 40° where substantial adsorption occurs, the
latitude, then the ice in a 100-m ice cap major exchange will be between permafrost
could be stored in only 30 m of soil. The and atmospheric water vapor as long as
available data (Farmer et al., 1977, Farmer permafrost is present. Of course, the con-
and Doms, 1979) do not show any obvious tact between the permafrost and atmo-
indication o f a seasonal regolith source or sphere will be indirect because even in the
sink of water, but it is likely that water is polar regions the annual thermal wave will
currently stored reversibly in the regolith as seasonally remove the permafrost in the
permafrost, adsorbed water, and water o f upper soil layers. At slightly lower lati-
hydration. We shall consider each o f these tudes, just b e y o n d the permafrost margin,
reservoirs in turn. substantial clay regolith adsorption will oc-
582 TOON ET AL.

H 2 0 C O L U M N , or/am
.1 1 10 100
.5

>- .4
..d 180 ° 190 ° 200 ° 210 °
o

%.a
-r

~.2
o
a
o .1 o
N

0 I I A I i
.01 .1 1 10
W A T E R PRESSURE, #bar

FIG. 14. The mass of water adsorbed per gram of clay is a very strong function of the water pressure
and of the temperature. Adsorption will precede permafrost formation, but when enough water is
available to fill completely the adsorption sites so as to make permafrost, then permafrost formation
will be the dominant physical process occurring. Because of the large quantity of water probably
available, permafrost formation is likely to be more important for Martian climate than adsorption. The
precipitable water column abundance was found by assuming that the water was uniformly mixed.

cur. At still lower latitudes only a few liming from equatorial regions o v e r the age
percent of the regolith could be adsorbed of Mars.
H=O. The low-latitude (high-temperature) The edge of the equilibrium p e r m a f r o s t
adsorption is a weak function o f t e m p e r a - b o u n d a r y depends upon t e m p e r a t u r e and
ture but a m o d e s t function o f pressure. F o r a t m o s p h e r i c water content. Presently, ice
example, if the H~O pressure fell from 10 to is in equilibrium at a t e m p e r a t u r e near
1 p r / z m , soil at 205 ° would lose about 2% o f 200°K. A factor-of-10 change in water va-
its weight in water. In this w a y a 102-m por a m o u n t would change the equilibrium
thick regolith covering half the planet might t e m p e r a t u r e b o u n d a r y b y about 10 to 15°K,
lose about 4 × 10~ g c m -2 of H20, which is which is equivalent to a latitude shift o f
sufficient to add about 200 m o f ice to the about 100 for lower w a t e r amounts and 20 °
poles. for higher w a t e r amounts. Likewise, for
P e r m a f r o s t has been considered by sev- fixed H 2 0 abundance the b o u n d a r y m a y
eral authors. Carr and Schaber (1977) dis- shift if the m e a n t e m p e r a t u r e shifts. At 35 °
cussed geologic evidence for p e r m a f r o s t obliquity the permafrost should extend 5°
and suggested that p e r m a f r o s t was wide- further e q u a t o r w a r d than it does now. At
spread with H=O c o m p o s i n g 50% of the soil 15° obliquity the p e r m a f r o s t should extend
volume in some regions. Fanale (1976) and about 3° further poleward than at present.
F a r m e r and D o m s (1979) have shown "that In the a b o v e discussion we have sug-
p e r m a f r o s t could be in equilibrium with the gested that substantial amounts of perma-
present atmospheric w a t e r content at lati- frost and a d s o r b e d w a t e r could m o v e into
tudes a b o v e 40 °. H o w e v e r , Smoluchowski and out o f the soil as Pw and T vary. We
(1968) has pointed out that only a few tens have already indicated that the t e m p e r a t u r e
of meters of fine-grained regolith are w a v e can penetrate 100 m in only 104 years.
sufficient to prevent p e r m a f r o s t f r o m sub- We m u s t also consider the diffusion of H 2 0
 MARTIAN CLIMATE 583

into and out of the soil. This problem has all 104 years, for temperatures [which control
the complexity of the COo diffusion prob- the water vapor pressure pw(Z)] from 200
lem discussed earlier. to 200°K, for pw(0) = 0, and for diffusion
The H00 diffusion problem differs from coefficients D between 0.01 and 10 cm 2
that of COo in several ways. First, relative sec -1. Diffusion coefficients D cannot be
to the atmospheric abundance much more much larger than 10 cm 2 sec -1 because that
H~O is adsorbed on soil than is CO2. In limit is set by the present COo pressure. For
solving the COo problem we showed that, pore sizes of a few microns or less D in
even if all the CO2 in one pressure wave- centimeters squared per second is approxi-
length were desorbed, it would contribute mately numerically equal to the pore size in
only a small amount of CO2 to the atmo- microns.
sphere and, therefore, we did not actually It is obvious from Table II that only a
solve the diffusion problem. For H~O, how- thin layer of permafrost can be lost or
ever, we must indeed solve the diffusion gained during a 5 × 104-year period. There
equation because of the large water abun- are two physical limitations expressed by
dance. A single molecule of HoO can diffuse these numbers. Most important, the mass
to a great depth based upon diffusion in the vapor phase over the ice is very low.
lengths such as those shown in Fig. l l. For example, at 210°K uniform mixing
However, the gradient of the vapor density would allow only about 50 pr/ zm of vapor
caused by diffusion greatly limits the actual phase H00 above exposed ice. Diffusion
mass flux. Second, because the H00 in the greatly limits the amount of water in the
atmosphere is a very tiny fraction of that column to a lesser value at any given time
adsorbed or present as permafrost, damp- and also prevents rapid replacement of any
ing of pore diffusion by adsorption is very vapor removed from the column. Suppose,
important and surface diffusion along the for example, that exposed ice at 210°K is
adsorbing surfaces could also be sig- allowed to replenish by diffusion the
nificant. Finally, the diffusion mean free column of air above it with H00 10 times
path for HoO depends on both the pore during a year. Then during 5 × l& years
radius and the COs pressure. For small pore only 25 m of ice could be lost.
sizes (r < 1.0 tzm) HoO molecules collide As we pointed out earlier, the ice-loss
mainly with pore walls in attempting to rates from exposed polar ice can be quite
escape, while for larger pore sizes HoO large at high obliquity but quite small at low
molecules collide with CO~ molecules. obliquity. At low obliquity the atmospheric
Hence, H~O diffusion may depend upon H00 pressure will drop and temperatures at
CO2 pressure changes. latitudes equatorward of 60° latitude will
As a guide to the diffusion problem we increase. Both factors will push both the
can ignore the adsorption processes and edge of the equilibrium permafrost bound-
calculate at time t the distance Z to the top
of an evaporating permafrost layer by as-
T A B L E II
suming that the diffusion occurs essentially
in steady state. Then DEPTH OF PERMAFROST (cm) AFTER 5 X 104 YEARS

Z = [2Dt((pw(Z) - pw(O))/piee] 1'2 (12) Diffusion coefficient T e m p e r a t u r e (°K)

(cm2 sec-1)
while the mass lost during t is EpieeZ, where 200 210 220
E is the porosity of the soil. This calcula-
tion, and the results, are essentially identi- 0.01 10 20 40
cal to those of Smoluchowski (1968) for the 0.1 35 65 130
1 110 200 410
same temperature and diffusion coefficient. 10 340 650 1300
Table II gives the values of Z for t = 5 ×
584 TOON ET AL.

ary and the edge of the regolith adsorption obliquity, then the permfrost boundary
boundary poleward at low obliquity. could move to about 25° latitude. The vapor
H e n c e , the regolith will attempt to supply pressure above ice at about 210°K is 60 pr
ice to the pole. The actual amount of ice /zm, and about 20% of the planet's area is
supplied is very difficult to determine. between 25 and 40° latitude. Table II sug-
Because of the great temperature depen- gests that under these circumstances the
dence for the vapor pressure of ice, most regolith might consume 20 m of ice from the
of the supply o f vapor will come from the caps. The amount of ice consumed could
present permafrost region whose annual increase if higher water vapor abundances
mean temperature is about 200°K. Table occur, or if the regolith has a very large
II suggests that this region could lose pore size in its upper ten meters and the
about 0.5 m of ice over 5 × 104 years. If CO2 pressure does not increase dramati-
this permafrost region of Mars occupies cally at high obliquity. Wind-blown dust
10% o f the planet's area, and the perma- might also speed the processes by hydrat-
nent polar cap occupies 1%, then perhaps ing in suspension and then becoming bur-
5 m of ice would be added to the ice cap. ied, or by simply covering exposed sur-
It is conceivable that ten times as much face ice. High ice cap loss rates may
as this could be added if the upper few force increased permafrost. If the polar
meters of regolith had a very large pore cap loss rates exceed the gain rate, and if
size. In that case the low COz pressure at the permafrost cannot take up the differ-
low obliquity could greatly increase the ence, then the high atmospheric vapor
HzO diffusion rate. Another possibility is pressures will occur. These high pres-
that geologic processes, such as slumping sures will drive ice into the regolith and
on slopes, could be important and might they will force the polar cap gain rate to
continually keep fresh permafrost ex- increase until balance is achieved.
posed. It is unlikely, however, that winds The reader may note that several differ-
would expose fresh permafrost because of ent processes control the polar cap budget.
the inability of winds at low atmospheric If the two polar caps are out of balance with
pressure to move dust. The seasonal ther- each other the feedbacks reinforce the im-
mal wave may also play a significant role balance by lowering the albedo o f the cap
in allowing permafrost to escape when losing net water. If both caps are out of
the pressure is low. The region poleward balance with the atmosphere-regolith sys-
of 60° latitude has an annual temperature tem, the feedbacks depend on the sign of
of 175°K. At such a low temperatures the imbalance. At low obliquity, permanent
only a few centimeters of permafrost CO2 caps will have net H20 gain rates, and
could be desorbed. H o w e v e r , summer- the atmospheric water will be reduced until
time temperatures would increase the the ice cap gain balances the loss by the
temperature of the upper meter of soil so permafrost. At high obliquity, when large
that diffusive losses would increase. If polar-ice-loss rates occur, the loss will con-
the upper meter of permafrost over one- tinue if permafrost can adsorb the loss. If
third of the planet is lost, then 30 m of not, atmospheric water will build up, in-
ice would be added to the caps. creasing the polar cap gain rate. A high
At high obliquity, the poles will lose water gain will increase polar cap albedos
more water during the summer season than and decrease the ice loss rate. Hence at
now, and temperatures equatorward of 55 ° high obliquity the polar caps may balance
latitude will drop. These processes Will their gain and loss rates so that the differ-
move the equilibrium permafrost boundary ence is equal to the regolith removal rate.
to lower latitudes. For example, if the mean The m o v e m e n t of water vapor through
1-120 amount increased to 60 pr tzm at 35° adsorbing soil is difficult to calculate theo-
 MARTIAN CLIMATE 585

retically. Experimental studies of this ice and sometimes not, so that the ice-to-
process, using terrestrial analogs, could be dust mixing ratio varies with depth. Ice as
quite valuable for improving our estimates permafrost is stable at the latitudes of the
of permafrost transfer rates. laminae, but the laminae are too thick to
have formed by vapor diffusion of ice into
(d) Formation o f Ice Sheets on the Polar soil. A dusty ice sheet covered with a meter
Laminae
of dust would be stable at the latitudes of
The polar laminae appear to be dusty ice the laminae because the dust would prevent
deposits extending slightly beyond the high summertime surface temperatures
north polar cap to 80°N latitude and well from reaching the ice and causing it to
beyond the south polar cap to 70°S (Cutts sublime. Buried ice would be at the annual
1973). Pollack (1979) has proposed that the mean temperature and would be cold
permanent polar cap, the polar laminae, enough to be in equilibrium with atmo-
and the polar debris mantle form a contin- sphere water vapor amounts or more likely
uous sequence of geologic formations dif- to grow by picking up atmospheric water.
fering, according to their latitude, in their Therefore the laminae ice would not diffuse
ability to retain water ice. Specifically, the out through a thin overlying dust layer.
debris mantels originated from atmospheric Here we attempt to identify the conditions
dust that was deposited in the annual CO~ that might have led to ice being perma-
caps but was unable to retain any attached nently deposited in the area of the polar
ice during the summer. This process was laminae before a dust layer covered it or
observed by Viking (Jones et al., 1979). before a thick, high-albedo ice sheet was
The polar caps retain water ice for very present to stabilize the ice.
long periods. The laminae are the interme- Table III presents water loss rates at 80
diate case of dust sometimes able to retain and 70° latitudes for an albedo of 0.25,

TABLE III

LAMINAE WATER LOSS RATE (g c m -2 y e a r -~)

Fixed parameter Variable parameter Obliquity Loss rate Loss rate

from from
h o t pole cool pole

P = 5 mbar 15 0.5 0.05

lat 80 ° 20 5 0.7
25 15 5
A = 0.25
U = 10 m s e c -~ P = 1 mbar 15 0.5 0.03
K = 1 x 10 - 4 c a i c m -1 s e c - 1 ° K -~ lat 80 ° 20 5 0.5

P = 1 mbar 15 4 0.2
lat 70 ° 20 l0 0.7

U = 10 m s e e -1
K = 1 × 10 -4 c a l c m -I s e c -1 ° K - t 80 ° lat 20 2 0. l
P = 5 mbar 70 ° lat 20 5 0.2
A = 0.4

U = 10 m s e c -1
K = 1 x 10 -a c a l c m - t s e c -~ ° K 80 ° lat 20 0.2 0.005
P = 5 mbar 70 ° lat 20 1 0.01
A = 0.4
586 TOON ET AL.

typical soil conductivity, and for obliquities In the dust-free conditions at low obliquity,
of 15, 20, and 25 ° . In performing these an e v e n higher albedo m a y be anticipated.
calculations we used the daily averaged At 70 ° latitude the calculated loss rates o f
solar insolation, even though there is a ice are larger than those at 80° , but the
diurnal cycle at both 70 and 80 ° . It m a y be differences are not always striking. Several
noted that the lower CO2 pressures at 0 = additional factors not included in our calcu-
15° play only a small role in the w a t e r lations may contribute to the laminae being
budget b y reducing the CO~ greenhouse confined to high latitudes. Permanent
warming slightly. CO2 deposits are unlikely to have e v e r
These calculations suggest that, at very extended m u c h b e y o n d 80° , so large
low obliquity, ice can begin to accumulate H~O ice sheets at low latitudes could not
if the rate of supply is on the order o f 0.1 to have been established. In addition, the cal-
0.5 g cm -z year -1 . This is a very high rate of culations we p e r f o r m e d ignore the diurnal
supply, as we pointed out previously, and it cycle. We h a v e already shown (Figs. 3 and
suggests that the ice c o m p o n e n t o f the 6) that the daily average s u m m e r t e m p e r a -
laminae does not form simply as an ex- ture is not a strong function o f latitude at
tended thin dusty ice sheet. low obliquity, which is why the loss rates
It is m o s t likely that the laminae originate differ only slightly b e t w e e n 70 and 80° .
at low obliquity when a large p e r m a n e n t Water loss rates are likely to be controlled
polar cap o f CO2 forms f r o m the present even at 70 ° b y diurnal t e m p e r a t u r e cycles.
a t m o s p h e r e and from CO2 d e s o r b e d out of A m u c h more complex model, including an
the regolith. Such a p e r m a n e n t cap of CO~ atmospheric b o u n d a r y layer, diurnal con-
would trap w a t e r released from permafrost densation and e v a p o r a t i o n cycles, and diur-
reservoirs v e r y effectively. The laminae nal transport out of the b o u n d a r y layer, is
m a y also form at m o d e r a t e obliquity by needed to examine fully the water loss rates
growth of p e r m a n e n t ice sheets if one pole under these conditions. Finally, at low lati-
has m u c h lower ice loss rates than the tudes, e x p o s e d ice would not be stable at
other. Then the preexisting p e r m a n e n t ice high obliquities, and the ice c o m p o n e n t o f
sheet at the low-loss-rate pole m a y grow at the low-latitude laminae might therefore
the e x p e n s e of the other polar cap. Of have been eroded, whereas the CO2 annual
course, alternatives exist. Perhaps wind caps partly protect the laminae at higher
deposits H 2 0 ice into small piles, or per- latitudes.
haps shadowing by surface relief allows The reader m a y note that there is no clear
s o m e ice to remain each summer. These distinction in o b s e r v e d thickness, struc-
patches of ice might then be able to grow ture, or geographical extent between the
into a more extensive, stable ice sheet. present north polar cap and a layer of
Large-scale slopes could also play a role by northern laminae with a thin cap of dust.
preventing solar energy from directly Ice surrounding dust grains n e v e r seems to
reaching the ice. Once enough ice (proba- be stable on bare polar ground according to
bly a few centimeters) is present to raise the our calculations of ice loss rates. Rather, it
albedo of the ice well a b o v e that of the seems more likely that the laminae are
ground, the loss rate of w a t e r ice without remnant, buried, p e r m a n e n t H~O ice caps
the p e r m a n e n t CO~ might be low enough to that formed at both poles during periods of
allow frost to accumulate. low obliquity or at the colder pole during
Table III shows s o m e typical loss rates periods of m o d e r a t e obliquity and high ec-
from established polar ice sheets at 20° centricity.
obliquity. With K = 10a cal c m -1 sec -1 °K -1
and high albedo (0.4), the water ice loss (e) Summary
rates at 80 ° for the cooler pole are very low. At low obliquity, CO2 p e r m a n e n t caps
 MARTIAN CLIMATE 587

will effectively trap water vapor. Perma- wetter than now, may have played a role in
nent H20 ice caps will form partly from their formation. Pollack (1979), Sagan and
atmospheric water vapor, but mainly Mullen (1972), and Sagan (1977) pointed
through water supplied from permafrost. out that reduced atmospheres composed of
The thickness of such caps could be tens of CI-I4 and NH3 could maintain high surface
meters, but slow diffusion of water vapor temperatures through a greenhouse effect
through the upper several meters of regolith with small (100 mbar) pressures of CH4.
limits the cap thickness. At intermediate Such atmospheres are consistent with
obliquities, water loss rates from the polar many models of outgassing for Mars and
caps will be very large unless the conduc- Earth (Fanale, 1971), with theories of the
tivity of the polar ice is close to that of pure origin of life on Earth, with ideas on the
ice. At high eccentricity one polar cap will nature of the early terrestrial atmosphere,
have a significantly larger water ice loss and with concepts about the maintenance
rate than the other if both caps have the of the early Earth's high surface tempera-
same albedo. However, the cap with the tures despite low solar luminosity. The
higher insolation will probably have a great disadvantage of reduced atmospheres
higher albedo than the other cap because is that they are unstable with respect to
dust storms will preferentially occur during hydrogen escape following photodissocia-
the winter season of the low insolation cap. tion and may therefore have very short
The albedo difference equalizes the water lifetimes.
loss rate, or perhaps causes the high insola- Another possibility is that high surface
tion pole to lose less water than the low temperatures were maintained by a green-
insolation pole. This situation probably oc- house effect in a CO2-H20-cloud atmo-
curs on present-day Mars. At high obliquity sphere (Pollack, 1979; Cess et al., 1980).
both poles experience large ice loss rates, Such an atmosphere has the advantages of
moving the permafrost boundary equator- having a relatively long lifetime and of
ward. At high obliquity it is possible that a being chemically identical to the presently
small amount of ice could melt in wind- observed atmosphere. Many investigators
sheltered regions, if the atmospheric pres- (e.g., Walker, 1976) believe that the Earth's
sure increased substantially. Ice is stable in early atmosphere were dominated by CO2
the region of the polar laminae only when rather than CH4. Owen et al. (1979) show
thick, high-albedo ice sheets are present. that the temperature of the early Earth
No obvious mechanism exists to incorpo- could have been maintained above freezing
rate ice into forming dusty laminae. It is despite lower solar luminosity by a plausible
most likely that the laminae are buried, CO2-H~O atmosphere. Fanale (1971)
remnant, permanent polar caps. Hence pointed out that the outgassed atmosphere
over long time periods the latitude bound- of Mars would be dominated by CO2 and H2
ary of the polar debris mantle marks the assuming the H20 to be cold trapped. The
furthest equatorward boundary of the an- formation of CH4 occurred only if the es-
nual CO~ caps and the boundary of the cape of H~ was not rapid compared with the
laminae marks the furthest equatorward outgassing rate. He adopted a catastrophic,
boundary of the permanent water ice caps. accretional degassing model so that Hz had
a relatively long enough lifetime to convert
VI. T H E R E L A T I O N B E T W E E N O R B I T A L CO2 to CH4. In 1971 it was believed that
VARIATIONS AND H20-CO2-CLOUD Mars was not geologically active, but now
GREENHOUSES
the observed prevalence of large volcanoes
Numerous dessicated channels cover the suggests continual geologic activity. More-
surface of Mars. Their origin is unknown, over, most evidence (Anders and Owen,
but an early Martian climate, warmer and 1977; Pollack and Black, 1979) suggests
588 TOON ET AL.

that Mars is not fully degassed, which may CO2 pressures and a significant CO2-H20
imply a more gradual degassing epoch. The greenhouse, we must return to a period
question of degassing history and the before the formation of an extensive clay-
CO2/CH4 ratio needs to be reanalyzed carbonate regolith.
based upon the data obtained since 1971. The processes forming the Martian rego-
The disadvantage of a COs atmosphere is lith are uncertain. Fanale (1976) argued that
that relatively large amounts of COs are meteoroid bombardment produced a 2-km
required to maintain high temperatures. thick basaltic megaregolith. As pointed out
Early estimates by Anders and Owen (1977) earlier, basalt is not strongly adsorbing at
suggested that less than 500 mbar of COs high surface temperatures so, if the CO2
had been outgassed on Mars. However, atmosphere appeared before the formation
more recent estimates (Pollack and Black, of a basalt regolith, a high-surface-tempera-
1979) that consider revised theories of noble ture climate could be stable. However, if
gas abundances on the terrestrial planets the CO2 outgassed slowly as the regolith
suggest that 1 to 3 bars of COs may have formed, it would be incorporated into the
been outgassed. These outgassing esti- regolith. Likewise, if at low obliquity the
mates make an early CO~-H20 greenhouse CO2 pressure were reduced and tempera-
much more plausible. tures fell, then CO2 would be permanently
Below we explore in greater detail the lost to the regolith. Higher obliquity could
nature of CO2-H20-cloud greenhouses. not return the CO2 to the cold atmosphere.
First, we discuss the maintenance of sub- Channel features are found only on heavily
stantial quantities of atmospheric gases. cratered terrain, so a scenario of early CO2
Then we elucidate some difficulties in de- outgassing, gradual regolith production by
termining precisely the pressures needed meteoroid impacts, carbonate formation,
for keeping high surface temperatures. It and loss of an atmosphere is consistent with
should be noted that the climate considered the timing of the end of channel formation.
represents a substantial departure from that Another plausible scenario that might
of present-day Mars. A complete analysis lead to high surface pressures is that mete-
of such climates is beyond the scope of the oroid bombardment produced a basaltic
present work. regolith only 100 m thick which could not
adsorb all the available CO2 as it was out-
(a) Maintenance of COs and H~O in the gassed. However, geochemical processes
Atmosphere operating at high surface temperatures may
The present atmospheric pressure on have slowly produced more adsorbing clays
Mars is far too low to produce an effective from the basalts. Some CO~ may have been
greenhouse. High COs pressures are re- converted into carbonates, and clays as
quired to produce appreciable greenhouse well as basalt may have been transported to
warming by the 15-/.tm CO2 bands and other polar regions where low temperatures in-
weaker COs bands, and also to pressure creased adsorption. The outgassing rate,
broaden H~O bands. Since atmospheric the rate of formation of clays and carbon-
H~O is restricted by the atmospheric tem- ates, and the rate of transport of regolith to
perature, an increased CO2 greenhouse ef- the poles would have determined the lon-
fect is also needed to raise temperatures to gevity of the high-temperature state. These
allow more H20 into the atmosphere. As geochemical processes may cause any high-
we have shown, changes in the orbital temperature climate to be self-destructive
elements of Mars are unable to drive more since the rate of carbonate and clay forma-
than a few tens of millibars of COs out of tion is greatly enhanced by high tempera-
the regolith and into the atmosphere-polar tures and by the availability of liquid water.
cap system. Hence to achieve significant Future studies of the rate of loss of atmo-
 MARTIAN CLIMATE 589

spheric pressure due to carbonate and clay by such an atmosphere, in a manner analo-
formation would be quite interesting. F. gous to that on the Earth, could maintain
Fanale (private communication, 1980) sug- polar temperatures above 230°K. Hence
gests that a high-CO~-pressure, high-tem- CO2 once established at 0.5- to 2-bars pres-
perature climate might not last more than sure would not tend to form permanent
106 years. On Earth, volcanism associated polar caps since CO2 would be far below its
with continental drift recycles clays and vapor pressure at polar temperatures.
carbonates into CO2 and other volatiles in The process stabilizing the atmosphere in
the atmosphere, but continental drift has the above example is atmospheric heat
not occurred on Mars. transport to the poles. Gierasch and Toon
During the period before the loss of CO2 (1973) examined the importance of atmo-
to geologic reservoirs, it is necessary to spheric heat transport to the stability of the
maintain the CO= in the atmosphere, and atmospheric pressure on Mars against
this requires high polar temperatures. As changes in the polar heat budget. Their
we shall discuss in the next section, a 0.5- conclusions are illustrated in Fig. 15 where
to 2-bar CO= atmospheric pressure can the solid lines represent solutions of a
maintain global average temperatures simplified polar heat balance equation for
above 273°K. It is likely that heat transport various levels of annual average available

I I I , I I I I I I I I ] I
ip. ,111,
CLATHRATE
e=O.1 /
190
/~ = 0.013 /
/
/A
o

~170
I--
,¢
CC
LU
a_

LLI = 2.6 x 104 erg cm -2 s-1

I--

150 2.2

1.8

1.4 "~ I

~ i i i ] i i i i I ! I I I J I i I 1
10 -1 100 101 102 103 104
ATMOSPHERIC PRESSURE, mbar
FIG. 15. Because of the strong dependence of atmospheric heat transport on atmospheric pressure,
the polar temperature increases rapidly with atmospheric pressure. The various solid lines illustrate
this dependence for different amounts of solar energy at the pole. The dashed curves are the vapor
pressure relations for CO= and for CO=-H=O clathrate. Single intersections of solid and dashed curves
represent stable climates. When multiple intersections occur, both the higher- and lower-pressure
intersections are stable. As the solar energy available at the pole increases, more CO= is added to the
atmosphere. Because of feedback, the pressure is very sensitive to solar insolation. ~ and fl are
variable parameters entering the heat balance equation. (See Gierasch and Toon, 1973.)
 590 TOON ET AL.

solar insolation, S. The heat balance equa- readily occur. Figure 15 indicates that CO2
tion determines the t e m p e r a t u r e of the po- clathrate greatly limits the Martian atmo-
lar cap consistent with the a t m o s p h e r i c spheric pressure, and, for the range of
heat transport, which in turn is controlled values of S shown, a t m o s p h e r i c pressures
by the pressure of the a t m o s p h e r e . An greater than 20 m b a r are difficult to obtain.
additional constraint when p e r m a n e n t CO2 It is possible that increased greenhouse
polar caps are present is that the atmo- warming would increase the slope of the
spheric pressure is controlled by the polar thermal balance solution (the solid lines in
cap t e m p e r a t u r e through the v a p o r pres- Fig. 15) and thereby allow m u c h higher
sure relation. This relation is illustrated in pressures. The value of S m a y also be m u c h
Fig. 15 by dashed lines for COs and C O s - larger than the values indicated in Fig. 15.
H 2 0 clathrate. Stable climatic states are Figure 15 shows that a high-pressure
represented by intersections of the dashed climate must either be initiated by, or main-
and solid lines. F o r example, the S c u r v e tained by, an increase in available solar
labeled 1.8 × 104 ergs cm -2 sec -~ intersects insolation. Figure 2 indicates that obliquity
the COs v a p o r pressure c u r v e at a polar variations are capable of considerably mod-
t e m p e r a t u r e of about 144°K and an atmo- ulating the polar heat budget. The solar
spheric pressure of 6 mbar. If Mars had insolation is also very sensitive to the polar
p e r m a n e n t CO2 caps with albedos of 0.77, cap albedo. Normalizing to a CO2 albedo of
then S would n o w have this value and the 0.6 and on obliquity of 25 °, we find
model would represent present-day Mars.
If the obliquity increased to 35 °, then S S(A,0)/S(0.6,25°) = 5.91(1 - A) sin 0. (13)
would increase to about 2.4 (for fixed al-
bedo) and the only stable climate would This ratio is 0.6, 1.35, and 1.67 for obliqui-
have a pressure a b o v e 1 bar. ties of 15, 35, and 45 °, respectively, w i t h A
The climate illustrated by Fig. 15 has fixed at 0.6. For an obliquity o f 25 ° it takes
multiple solutions. As a result of the multi- values of 1.25 and 0.75 for polar cap al-
ple solutions, if there are p e r m a n e n t CO, bedos of 0.5 and 0.7, respectively. With A
polar caps, then, once initiated, a high- = 0.6 a n d 0 = 25 ° , S = 3.1.
pressure climate is stable until the solar Hence, if the CO2 albedo is now 0.6, then
energy input to the pole drops well below Fig. 15 suggests that if COz were available,
the solar input needed to initiate the high- a runaway condition could occur at
t e m p e r a t u r e state. Gierasch and Toon present. If, earlier in Martian history, the
(1973) did not include the greenhouse solar luminosity were 0.7 of its present
warming caused by increased COz, so the value, S at 0 = 25 ° would be 2.0, e v e n for an
high-pressure climate is even more stable albedo of 0.6.
than Fig. 15 suggests. O f course, the pro- The polar energy would be greatly in-
duction of clays and carbonates would creased by lowering the albedo or increas-
eventually destroy the high-pressure cli- ing the obliquity. Ward e t al. (1979) sug-
mate. gested that during its early history Mars
Another very likely possibility is that m a y have had a mean obliquity near 32 °
C O , - H z O clathrate might dominate the po- with excursions to 45 and 25 ° . Such an
lar cap (Dobrovolskis and Ingersoll, 1975; obliquity would make a high-pressure cli-
Miller and Smythe, 1970). Clathrate is ther- mate easier to achieve and maintain. For
m o d y n a m i c a l l y favored o v e r CO, at Mar- example, with 0 = 45 °, A = 0.5, and current
tian polar t e m p e r a t u r e s and, since H~O ice luminosity, S could reach a value o f 6 × l04
is brought into the caps as finely divided ergs cm -z sec -1. H e n c e we can conclude
grains in good contact with the a t m o s p h e r e , that reasonable modifications of the polar
it is likely that clathrate formation would albedo and obliquity are probably
 MARTIAN CLIMATE 591

sufficient to initiate a high-pressure cli- ature of 273°K are quite sensitive to the
mate for either COs or its clathrate. details of the radiative model. The major
differences between the models of Pollack
(b ) C02-H~O-Cloud Greenhouse (1979) and Cess et al. (1980) are that the
Pollack (1979) computed surface temper- latter authors employ a greater pressure-
atures for various CO~ pressures, relative broadening coefficient than Pollack and that
humidities, and tropospheric lapse rates, the details of the treatment of the CO2
and he found that global average tempera- bands are slightly different. Here we shall
tures above 273°K occurred when the CO2 point out that clouds and a possible low
surface pressure was about 2 bars. Cess et surface albedo might have been important
al. (1980) performed similar calculations to climate early in Martian history. Very
but they concluded that a pressure of only 1 low surface pressures are found to produce
bar would produce a temperature above high surface temperatures.
273°K. Both pressure estimates greatly ex- Table IV presents radiative-convective
ceed early predictions of the Martian CO2 calculations of surface temperatures for
inventory (Anders and Owen, 1977). How- various pressures of CO2, a planetary al-
ever, new predictions of outgassed CO2 bedo of 0.25, and for various values of an
exceed the pressures required to obtain a insolation factor,f, which is the ratio of the
surface temperature greater than 273°K by available solar energy to that presently
either o f these radiative calculations (Pol- available with globally averaged condi-
lack and Black, 1979). The dependence of tions.
surface temperature upon surface pressure Most of the cloud-free values given in
increases rapidly and becomes highly non- Table IV were taken from the calculations
linear as the surface temperature ap- of Pollack (1979), who assumed a fixed-
proaches 273°K. Hence predictions of the temperature profile with a lapse rate, F, of
pressure needed to obtain a surface temper- 3.5°K km -1 and a relative humidity Rh of

T A B L E IV

RESULTS OF GREENHOUSE CALCULATIONS

Insolation Surface pressure (mbar)

factor,
f 6 100 300 600 1000 1500

No cloud
Global average 1.0 217 228 233 (248; F = 3.5) 252 (272)
(218) (250; F = 5) (259; F = 3.5)
(263; F = 5.0)
(254;Rh = 0.5)
Global average 1.25 232 244 255 (284; F = 5)
w i t h A = 0.1 or
equator with
A = 0.25
Equator with 1.5 242 258 273
A=0.1
2.0 265 295
Cloud
Global average 1.0 285
1.25 260 298
E q u a t o r with 1.5 280
A=0.1
 592 TOON ET AL.

unity. He obtained the surface temperature planetary albedo was near 0.1 in early Mar-
by balancing the outgoing thermal flux tian history. The insolation factor for this
against the available solar energy. Pollack albedo is 1.2, and COz pressures near 600
showed that his results were not very sensi- mbar could then yield global temperatures
tive to the lapse rate, nor the relative hu- above freezing!
midity. Global average temperatures o f 273°K
Also presented in parentheses in Table may not be the relevant criteria for fluvial
IV are new results for cloud-free condi- features. For example, if one only desires
tions. These calculations were performed equatorial average temperatures of 273°K,
with a radiative-convective scheme that then for an obliquity of 25°, f = 1.22 on a
converges to a radiative equilibrium ther- yearly mean and P = 600 mbar is needed. If
mal profile in regions where the profile is the albedo is 0.1 and only high equatorial
subadiabatic. The solar energy is assumed temperatures are needed, then a n f o f about
to be deposited only at the surface. The 1.5 and 300 mbar of CO2 will suffice. At
calculations also include a pressure broad- high surface pressures, atmospheric heat
ening coefficient appropriate for CO2 that transport will lower equatorial tempera-
Pollack (1979) inadvertently overlooked. tures and raise polar ones. Hence precise
However, the difference in the surface tem- estimates o f the COz pressures needed to
peratures calculated with the two methods achieve high equatorial temperatures are
is only a few degrees except at high pres- difficult to make.
sures. The new calculations show that only The calculations tabulated in Table IV
1.5 bars of CO~ are needed to obtain a consider only the greenhouse warming due
273°K mean temperature. to CO2 and H20. Perhaps other trace gases
The basic conclusion to be drawn from that are effective at producing a greenhouse
the numbers in Table IV is that CO2 pres- might have been present in the early Mar-
sures near 1.5 bars are required under tian atmosphere. For example, Oyama et
cloud-free conditions in this model to al. (1979) suggest that SO2 may be impor-
achieve global average surface tempera- tant for the Venus greenhouse. We plan to
tures above the freezing point of water. The consider further the importance of trace
temperatures are not very sensitive to rela- gases in future work.
tive humidity nor to lapse rate, as is appar- Another possibility for enhanced green-
ent from the i - b a r values in Table IV. house effects is that high clouds may form.
Table IV also suggests several means by The advantage of clouds is that pressure
which lower pressures might yield high broadening is not needed to make them
temperatures. For example, the present al- efficient infrared radiators. The disadvan-
bedo of Mars is quite high, A - 0.25, tage o f clouds as greenhouse constituents is
because of the presence of abundant iron that they reflect solar insolation. Whether
oxides. As we have already pointed out clouds warm or cool depends upon their
large atmospheric pressures could only optical depth in the visible and infrared,
be achieved before extensive regolith their area, their altitude, and their composi-
modification by geochemical processes. At tion (Schneider, 1972).
this stage in Martian history, iron oxides Several cloud materials might contribute
may not have been formed. Kieffer et al. to a greenhouse. Dust now dominates the
(1977) report that Mars presently has a thermal balance on Mars. However, it is
bimodal surface albedo distribution, sug- unlikely that dust could form more exten-
gesting that the Martian surface is com- sive clouds than at present because solar
posed of one material with an albedo near energy deposition in the dust stabilizes the
0. I and another, much brighter material. atmosphere and stops the dust lifting pro-
Hence, it is reasonable to assume that the cesses. CO2 ice has very narrow infrared
 MARTIAN CLIMATE 593

bands and would not be an efficient green- substantial fall velocities, so they could
house material. Our r a d i a t i v e - c o n v e c t i v e remain in the atmosphere for long periods
calculations confirm the speculation o f o f time. Particles of such size have large
Gierasch and T o o n (1973) that a high pres- visible a s y m m e t r y factors, so the clouds
sure COz thermal profile has its upper atmo- have relatively low visible albedos, but the
sphere at temperatures below the CO2 frost clouds' drops are large enough so that the
point. This situation occurs in calculations clouds effectively block infrared radiation.
for present-day Mars, which is corrected by A large visible optical depth insures that the
appealing to efficient large-scale vertical clouds will have infrared optical depths that
heat transport through dynamics (Pollack et are large (1 to 3). Finally, optically thick
al., 1979). A similar process, and perhaps clouds radiate effectively at the ambient
solar energy deposition in the gas as well, is temperature. The higher the ambient tem-
needed to prevent the formation of CO2 perature, the more energy they radiate to
clouds over most of Mars at high pressures. space and the less effective they are at
Below we consider only water ice clouds creating a greenhouse. Therefore, high-alti-
because they have greater infrared emis- tude clouds are the best greenhouse candi-
sion than CO~ clouds and, unlike dust, dates, but once they reach an altitude
water cloud particles do not have to be where the temperature varies slowly with
lifted from the surface. Table IV presents a height, further increases in altitude do not
series of cloud greenhouse calculations per- substantially improve the ability o f the
formed with a r a d i a t i v e - c o n v e c t i v e 1-D clouds to create a greenhouse. We chose a
model. As we have noted, the effect of cloud top pressure altitude of 10% o f the
clouds on the radiation budget depends surface pressure since it was the lowest
upon various cloud properties. Rather than altitude at which the thermal profile became
attempt to explore the entire range of possi- nearly constant. Typical terrestrial radia-
bilities, we have studied only those clouds t i v e - c o n v e c t i v e models place clouds at 0.4
that should be among the most effective at to 0.5 of the surface pressure.
increasing the surface temperature. This We assumed in our cloud calculations
deliberate bias is intended to determine the that all the available solar energy was de-
maximum additional warming that clouds posited at the ground. The solar energy was
might plausibly contribute, and is not reduced o v e r 75% of the planet by
meant to represent the most likely case. reflection at the cloud top surface cal-
Hence we chose a water cloud fractional culated with a two-stream approximation
coverage of 75%, a particle size distribution for a conservatively scattering cloud (It-
with a mean size of 4 / x m , a cloud visible vine, 1975), including multiple ground
optical depth of about 5, and a cloud top reflections.
pressure o f 10% of the surface pressure. For particles of the size o f interest the
We also used a relative humidity of unity a s y m m e t r y factor, g, is about 0.85. With
and an adiabatic lapse rate of 5°K km -x. ground albedo A = 0.25 and cloud optical
If the clouds are effective as greenhouse depth of 5, the planetary albedo is about
warmers, then using as large an area of 0.43. This case represents f = 1. If the
coverage as possible enhances their impact. surface albedo were 0.1, the planetary al-
H o w e v e r , it did not seem reasonable to bedo would be 0.35 a n d f = 1.15.
expect 100% cloud c o v e r since the water Table IV shows that favorably placed
vapor sources would be localized to the clouds may raise surface temperatures over
equatorial and summer polar regions and a cloud-free case by about 20 to 40°K.
restricted to a small fraction o f the planet's Global average temperatures above 273°K
surface since no oceans are anticipated. then require less than 600 mbar of pressure.
Particles o f sizes near 4 /xm do not have If the surface albedo is 0.1, a global surface
594 TOON ET AL.

temperature of 273°K occurs at pressures variations in the CO2, dust, and HsO cycles
near 300 mbar. A high equatorial tempera- caused by changes in the orbit and obliquity
ture occurs near 300 mbar for an albedo o f on Mars. We also discussed the implica-
0.25 and, if the surface albedo is low, then tions of enhanced COs pressures for green-
equatorial temperatures can be 273°K for house warming. Here we would like to
pressures near 100 mbar. unify the ideas developed by describing
how the climate might differ between the
(c) Summary present orbital configuration and various
New estimates of the COs outgassed over alternatives. Then we will mention the ob-
Martian geologic history suggest that more servational evidence for climate change and
than enough CO2 is available to produce a finally propose models for the formation of
significant greenhouse effect (Pollack and the polar laminae and of the fluvial features.
Black, 1979). A high-obliquity state might At present, the Martian obliquity is about
be required to initiate a high-pressure cli- 25 ° , but periodically in the recent past the
mate, but once initiated, greenhouse warm- obliquity has reached values less than 15°
ing and atmospheric heat transport might (Ward, 1979). As the obliquity decreases,
prevent polar cap formation even at low the poles receive less solar energy and, if,
obliquity. The atmosphere would be de- the polar cap albedo is not too low, as
stroyed as carbonates and adsorbing clays seems likely, permanent CO2 polar caps
form and as the regolith is transported to will form. The formation of permanent po-
polar regions. If COs had been slowly out- lar caps will reduce the atmospheric pres-
gassed after the formation of a thick, cold, sure to values of 1 mbar or less. The precise
adsorbing regolith, the atmospheric pres- value of the pressure depends on the al-
sure could never have built up. New green- bedo; higher albedos will cause lower pres-
house calculations by Cess et al. (1980) sures. The reduced atmospheric COs pres-
suggest that only 1 bar of COs is needed to sure will strongly desorb COs from the
achieve global average temperatures above regolith augmenting the permanent CO2
273°K with the current Mars albedo and caps with the limit being set by the ability of
solar luminosity. We studied a less sensi- CO2 to diffuse from the regolith and also by
tive radiation model that predicts that 1.5 the amount of COs presently adsorbed on
bars are needed to achieve 273°K. We the regolith. The regolith could contribute a
found with this less sensitive model that if several-hundred-meter thickness of COs
the surface albedo of Mars were lower in its over the area of the present caps, while the
early history, or if equatorial conditions are present atmospheric pressure corresponds
more relevant than global averages ones, or to only about 10 m.
if extensive high clouds form, then only 500 Permanent COs caps will be strong water
mbar of COs are needed to obtain tempera- vapor sinks and thus atmospheric water
tures of 273°K. A great many details of a vapor will decline at low obliquity. In addi-
high-pressure COs climate need to be con- tion, equatorial and midlatitude annual
sidered further. In particular, the upper mean temperatures will be several degrees
atmosphere needs to be warmed to prevent higher than current values. Lessening
COs frost formation. Atmospheric heat amount of atmospheric water vapor and
transport to polar regions needs to be taken increased midlatitude temperatures will
into account, for it may significantly de- force the equilibrium permafrost boundary
press equatorial temperatures and raise po- to a higher latitude than now. Water will be
lar ones. supplied to the permanent polar caps, but
the water supply to the caps will be strongly
VII. DISCUSSION limited by diffusion so that during a half
In the preceding sections we considered obliquity cycle of 5 × 104 years a few
 MARTIAN CLIMATE 595

meters of H~O might be added o v e r the area The orbital eccentricity and longitude of
of the present permanent polar caps. Other perihelion will probably play no significant
processes assisting diffusion, such as role in determining the basic climate at low
slumping that exposes fresh ice, or the obliquity.
presence of large pore sizes or cracks near Ward (1979) shows that periodically the
the surface, could substantially augment Martian obliquity may reach values as high
the H20 supplied to the poles. The seasonal as 35°. Ward et al. (1979) suggest that the
thermal wave might drive high latitude per- obliquity could have had a mean value of
mafrost to the poles, thus adding a few 32 ° and reached an extreme near 45 ° prior
tens of meters of ice to the caps. If per- to the formation of the Tharsis ridge. As the
manent CO2 polar caps c o v e r a larger geo- obliquity increases the poles receive more
graphic area than the present permanent solar energy so that temperatures of the
H20 caps, or if local slopes and other ground and ice will increase.
thermal balance factors are favorable, ex- Polar ground temperatures during sum-
tensive water ice sheets may develop mer can exceed 260°K at 0 = 35° when e =
over the area now covered by the polar 0.14. H o w e v e r , at 0 = 35 °, annual average
laminae. If these sheets reach great polar temperatures will exceed present
enough depths to have high albedos and temperatures by only about 10°K whereas
high thermal conductivities they will be equatorward of 60 ° latitude annual average
stable at higher obliquities than those at temperatures will be several degrees cooler
which they formed. than at present.
As the CO2 pressure drops, dust-storm The small changes in annual average tem-
activity will cease, because at low pressure perature will slightly decrease adsorption in
only unreasonable wind speeds, in excess the polar regolith and increase adsorption
of the speed of sound, would be able to lift in the equatorial regolith. A globally distrib-
dust. Since no dust will be available, both uted basalt regolith would desorb enough
the CO2 and HzO caps would have much CO2 to provide an increase of only a few
higher albedos than the present 1-120 and millibars of pressure. Larger increases
CO2 caps. Hence the decline of dust storms might occur if the regolith were concen-
strongly augments the trend o f the CO2 and trated in polar regions or if the regolith
HzO cycles. were composed of clays. A clay regolith
The atmospheric circulation at low obli- could potentially release several tens of
quity will be quite different than at present. millibars of COz to the atmosphere. Such
Since there will be no dust storms, the small pressure increases are significant for
vertical thermal structure will be under dust storm activity, but not for greenhouse
greater dynamic control. Feedback be- warming.
tween tidal winds and local dust storms will High temperatures for water ice can oc-
not occur. Large temperature gradients, cur at large obliquity. In wind-sheltered
perhaps similar to those at the present regions, at fairly high atmospheric pres-
southern summer pole, will drive violent sures (50 mbar), low-albedo water ice could
winds at the edge of the summer permanent melt at 0 = 35°. H o w e v e r , it is more likely
polar caps. Larger meridional winds will be that thermal conduction and evaporative
caused by the necessity of supplying the heat losses will prevent water ice from
mass of the annual CO2 polar caps with a reaching the melting point. The water vapor
less massive atmosphere. The radiative loss rate from the polar cap is a strong
time constant will be less than at present function of temperature and therefore of
and the ground temperature will be under albedo and thermal conductivity; neverthe-
even greater radiative control than it is less, it is likely that large water loss rates
now. will occur. Dusty water ice would be highly
 596 TOON ET AL.

unstable in the polar laminae area. Atmo- increase proportionately so that tidal and
spheric water vapor amounts will increase, topographic winds will be damped.
and since equatorial temperatures will de- The orbital eccentricity and the longitude
cline at high obliquity, the equilibrium per- of perihelion may be less important at high
mafrost boundary will move to lower lati- obliquity than it is now. The dust-storm
tudes. The polar water ice will attempt to cycle will be unaffected by the eccentricity
transfer to the regolith, but regolith diffu- (if atmospheric pressures increase). How-
sion will limit the amount of new perma- ever, the water balance between the poles
frost formed. Only a few tens of meters of may be controlled by the eccentric orbit. If
polar ice might find its way into the regolith permanent water ice polar caps are found at
unless atmospheric water vapor builds up high obliquity, they will be much more
to high levels or soil pore sizes are very likely to transfer between the poles with the
large near the Martian surface. The polar precession of the equinoxes than they are
caps will disappear at high obliquity if per- now. Then the pole with summer contain-
mafrost formation is rapid enough. ing perihelion will have the smaller polar
The dust storm cycle will be strongly cap.
modified at high obliquity if a few tens of Ward (1974, 1979) and Murray et al.
millibars of CO2 are desorbed from regolith (1973) show that the eccentricity of Mars
reservoirs. In that case wind speeds needed varies from 0.0 to 0.14 and that the longi-
to lift dust will decline significantly. Dust tude of perihelion slowly varies. The ec-
storms will occur during the summers in centricity barely affects the annual aver-
both hemispheres. H o w e v e r , it is unlikely age temperatures. Therefore, permafrost
that dust storms will be more prevalent or boundaries (now located near ___40° lati-
intense than they are today during southern tude) and regolith adsorption processes
hemisphere summer. Atmospheric heating cannot be affected substantially by changes
due to solar energy deposition in the dust in e or to. Eccentricity and to do not
will stabilize the atmosphere and will pre- significantly affect the climate at very low
vent substantially more dust storm activity 0, but they might make significant changes
than at present. Since dust storms will in the water cycle at very large 0. Ward
occur during both summers, the rate of (1979) shows that during a modest fraction
supply of dust to the two polar regions will of Martian history, 0 oscillations about the
be more nearly comparable than is the case mean are fairly small. During these periods,
now. Because more water will be mobilized variations in e and 0 play important roles by
in summer and then deposited with the dust creating hemispherical asymmetries.
in winter than now, it is likely that both When the Martian orbit is circular, the
permanent polar caps, if they exist, will hemispheres have comparable energy
have a slightly higher albedo than the budgets. It i s likely that global scale dust
present northern H20 polar cap but a lower storms do not occur during such periods at
albedo than the present southern H20 polar modest 0, but our knowledge of Martian
cap. dust-storm generation is not sufficient to
The atmospheric circulation at high obli- determine whether dust storms occur at e =
quity will be characterized by strong sum- 0. If dust storms do not occur, then the
mertime easterlies because summer polar albedos of both polar caps should be large
temperatures are higher than equatorial during such periods, while if dust storms do
ones. Unlike the present Martian summer, occur, both poles would have albedos like
which is relatively free of baroclinic distur- that of the present northern water ice cap.
bances, a high-obliquity summer may have Detailed stratigraphy o f the polar caps may
frontal activity. If the atmospheric pressure some day answer this question.
increases, the radiative time constant will At large eccentricities, hemispherical
 MARTIAN CLIMATE 597

asymmetry is accentuated, especially when not be easy to separate symmetric and

perihelion occurs near a solstice, as at asymmetric variations without precise dat-
present. Then global dust storms will occur ing.
preferentially during the summer in which The most obvious evidence that the orbit
perihelion occurs, the albedo of that sum- affects the climate is that global dust storms
mer cap will be higher than the albedo of now occur only during southern hemi-
the other cap, and the water ice cap in that sphere summer. This asymmetry most
hemisphere will be smaller than the other likely is due to higher winds occurring
because of larger water loss rates. This when solar insolation is maximized near
simple picture can be upset by feedbacks perihelion. As perihelion precesses, the
between the dust cycle and the water and time of year when dust storms occur will
CO2 cycles. For example, the present change.
smaller southern water ice cap, found at the A second suggestion of the importance of
pole receiving greater summer insolation as eccentricity and perihelion passage, long
expected, lost less water vapor than the used as evidence for strong orbital insola-
larger northern cap during the year of Vi- tion control over the water budget, is that
king observations. Presumably this contra- the area of the southern polar cap is much
diction is caused by a recent change in the less than that of the northern polar cap.
heat balance caused by dust storms creat- Unfortunately, there are several unre-
ing a lower ice albedo at the northern pole. solved difficulties. The observed water
In order to verify or improve upon these budget of Farmer et al. (1979) showed less
ideas concerning climate changes due to water loss from the south pole than the
orbital element forcing, we need the follow- northern one, and James et al. (1979) sug-
ing basic knowledge: the permanent polar gest that the south pole might have a resid-
cap albedo in the southern hemisphere; the ual CO2 cap. These observations imply that
thermal conductivity of the permanent po- the orbit does not contro| the polar heat
lar caps and polar laminae; the geographical budget directly through the insolation.
distribution, depth, pore size, and composi- Probably the dust-storm asymmetry has
tion of the regolith; the amount of CO2 created different albedos at the north and
and H20 adsorbed on the regolith; the south poles so that orbital insolation varia-
amount of CO2 in the southern permanent tions control the heat budget indirectly
polar cap; and the thickness of the water through the dust storms. Another possibil-
ice in the polar caps. Experimental ity is that the difference in the water loss
studies of the diffusion of CO2 and H20 rates between the caps is so small at
through porous, adsorbing soils would be present obliquities that only a small fraction
quite useful. We also need better theoret- of the polar cap mass can be transferred
ical models of the mechanisms generating between the caps during a precessional
dust storms, of regolith desorption, of the cycle. In that case the present areal cover-
transport for 1-120 and dust, and of the age of the two caps is a fossil of a higher
polar heat budget. This information would obliquity state. A final possibility is that
improve our understanding of the pro- geological factors such as regional slope
cesses causing climatic change. differences may control the polar heat
Evidence for climatic change is also re- budget. The importance of regional topog-
quired. It is important to note that varia- raphy is strongly suggested by the fact that
tions in eccentricity and longitude of peri- neither polar cap is symmetric about the
helion will be recorded as hemispherical rotational axis. Detailed models of the polar
asymmetries, while changes in the obliquity cap heat balance are necessary to deter-
cycle will be recorded symmetrically. Un- mine the importance of topography to the
fortunately, since both cycles occur, it will heat balance.
598 TOON ET AL.

Another asymmetric geological feature, could form saltating grains from dust. They
possibly related to climate, is the vast (105 suggested that the greater dust supply to
km2), low-albedo, dune field partially encir- the north pole favored dunes there. While
cling the north pole in the region poleward we agree that the dust supply is significant,
of 80° latitude which does not contain lami- we feel that the greater longevity of the ice
nae or permanent ice. Cutts et al. (1976) at northern latitudes is also o f great impor-
find that the dunes have a complex relation tance.
to the polar ice, sometimes overlying the Although the southern polar laminae are
ice and sometimes underlying it, suggesting larger in area than the northern ones, the
that the dunes and ice are interactive. northern laminae are thicker than the
It is possible that the north polar dunes southern ones. By mass the laminae are
are the remnant of some easily erodable symmetric about the equator. The structure
geologic structure found only in the north of the laminae has been described in detail
polar region and therefore have no climatic by Cutts et al. (1973, 1976, 1979). They are
significance. However, the mechanism o f layers of dust and possibly ice exposed in
supply of dust and water to the polar cap places as fine terraces of approximately
region can naturally explain them. Individ- equal thickness (-~30 m) that extend uni-
ual dust grains entering polar latitudes be- formly over hundreds of kilometers (Fig.
come coated with H20 ice and eventually 1). The scarps along which the terraces are
fall out onto the annual caps when COz exposed were originally thought to be edges
condenses on the particles. During the fol- of a series of stacked plates, each com-
lowing summer, at most latitudes, the dust posed of many terraces. More recent obser-
grains are freed when the COs and ice vations (Cutts et al., 1979) indicate that the
sublime away, and a general polar debris scarps, which spiral about the poles are
mantle, rather than dunes, forms. How- primarily the edges of troughs cut into the
ever, the dust which falls on the permanent laminae, although isolated escarpments
polar caps does not completely lose its coat also exist. Cutts et al. (1979) have also
of water ice, and so is trapped on the polar identified an undulating character to the fiat
cap. The dunes are found within the same regions separating the troughs. The nearly
latitude belt as the permanent cap, but complete absence of craters in the laminae
because of their lower albedo and presum- suggests that they are fairly youthful or that
ably lower thermal conductivity, the dunes they are continually being resurfaced.
are less able to retain water ice than the We require mechanisms for creating
polar caps, although they are better able to three properties of the laminae. First, it is
keep water ice than the lower latitude dust necessary to explain why dust preferen-
deposits. The small amount of ice left tially collects in the polar areas. Second,
around the dust grains at least until sum- we must elucidate the formation of the
mer serves to cement many of them to- terraces. Finally, it is necessary to account
gether into large saltating particles. The for the troughs and large-scale escarp-
single-scattering albedo of a large particle is ments. All three of these observations can
lower than that of a small particle com- be explained through the use of a model
posed of the same material. Therefore, having a variable dust-to-ice mixing ratio.
dunes composed of clumps of fine dust The variation is driven by the orbital ele-
particles could have lower spherical al- ment changes that alter the polar heat
bedo. Under this scenario the dunes would budget and the dust storm cycle.
not occur in the southern hemisphere be- Both poles are probably permanent net
cause ice is less stable, due to the higher dust sinks because dust storms do not pref-
summertime solar insolation. Haberle et al. erentially remove dust from the polar re-
(1979) also pointed out that an ice cement gions, and because particles from dust
 MARTIAN CLIMATE 599

storms are preferentially deposited in the of grains, variations in particle size,

polar regions. Global dust storms now orig- changes in particle composition, or some
inate in the southern hemisphere and other property is not known. A reasonable
spread northward. Hence the dust is not hypothesis is that the terracing is caused by
preferentially obtained from the north polar variations in the dust-to-ice mixing ratio
region. We do not know the source regions through the laminae. The more resistive
of dust in the southern hemisphere, but fiats are primarily compacted dust, while
there is no reason to believe that the pole is the more easily eroded segments are largely
a preferred source region. Dust and ice are water ice with a minor component o f un-
codeposited as small particles in the north consolidated dust. When the terrace edges
annual cap and to a lesser extent in the are exposed to the atmosphere the water
south annual cap because CO2 snow forma- ice sublimes and the surrounding ground
tion is a strong atmospheric clearing mech- loses its strength and therefore slumps.
anism (Pollack e t al., 1979). Strong evidence for this mechanism is that
Cutts et al. (1973, 1979) suggested that debris mantles are not terraced at lower
dust collects at the poles because erosion is latitudes, where surface ice is probably
prevented in the polar regions by water ice never stable. The cause of the varying dust-
cementing the grains. Such a process is to-ice ratio may partly be changes in the
probably significant in the permanent cap relative deposition rate of dust and ice as
region. H o w e v e r , extensive debris mantles suggested by Pollack (1979). In addition,
surrounding both poles to fairly low lati- and perhaps more importantly, water ice
tudes (35°), but not covering equatorial can be preferentially removed as water
regions (Soderblom, e t al., 1973), suggest vapor sublimes from a shrinking polar cap.
that ice formation is not the dominant As water vapor sublimes, dust imbedded in
process keeping dust in the polar regions. the ice will become concentrated until
We are inclined to believe that the poles are eventually a thin layer of nearly pure dust
a net dust sink because of enhanced deposi- caps the ice so as to prevent further ice
tion there by CO2 snow, since global dust loss.
storms do not originate there, and because There are two types of changes that will
the polar dust grains cannot be easily lifted create variations in the dust-to-ice mixing
by the wind. Water ice formation about ratio: changes in the longitude of perihe-
dust particles increases their sizes by only lion, and changes in the obliquity. For an
small amounts. Winds are probably able to eccentric orbit one pole is favored to re-
suspend such ice-dust particles nearly as ceive both dust and polar ice. For example,
easily as single dust particles. However, the north polar cap of Mars now covers
dust particles of such size are not directly most of the laminae in the northern hemi-
lifted by the wind but are injected by larger sphere, and dust storms lead to high dust
saltating particles. Such large particles may deposition rates there. As perihelion pre-
be generally absent in the polar regions cesses, the north cap will begin to lose
since they are not carried there by winds water to the south cap. As the north cap
and this missing ingredient suppresses wind shrinks, the dust at the top of the cap will
erosion. b e c o m e concentrated into a layer. Eventu-
No observational evidence is available to ally, the dust layer will b e c o m e thick
strongly limit theories concerning the rea- enough to protect the underlying ice from
sons that the polar ice and dust form a the seasonal temperature cycle. If the
terraced structure in the laminae. Evidently present dust-to-ice mixing ratio is 0.15 (as
erosion-resistive soil occurs at regular implied from the present polar albedo),
depths, but whether the erosion-resistive then to obtain a 1.5-m-thick layer of dust,
segment is caused by chemical cementing 8.5 m o f ice must be lost. This is about all the
 600 TOON ET AL.

ice that could be lost from the pole unless water. This m e c h a n i s m does not require
the u p p e r dust layer is heavily eroded. large polar water loss rates at the present
H e n c e one layer will be formed during half obliquity. It does, h o w e v e r , imply that a
a precessional cycle at one pole, and one p e r m a n e n t polar-cap thickness on the order
layer will be f o r m e d during the other half of 40 m is needed to form a 30-m layer of
cycle at the other pole. The strongest evi- laminae, since at high obliquity only a sin-
dence that such a m e c h a n i s m occurs is that gle 10-m layer of ice needs to be lost to
the north polar cap now covers m o s t of the create a dust cap from dust e m b e d d e d in
northern laminae. the ice.
One p r o b l e m with this model is that large In reality, both eccentricity and obliquity
amounts of ice are being lost from the variations m a y be forming laminated ter-
p e r m a n e n t polar cap system and transfer- rain. H e n c e a complex record of formation
red to the laminae. In order to account for m a y eventually be found in the laminae
the present H 2 0 cap, the continued formation structure.
of laminae, and their nearly uniform thick- A problem with forming the terraces in
ness, one must assume that the rate of either of these ways is that one 30-m layer
supply of w a t e r to the larger polar cap from is created approximately e v e r y 105 years.
nonpolar cap reservoirs is v e r y constant in The supply of water and dust to the north
time and is equal to the rate of loss o f water cap as estimated by Pollack et al. (1979) is
caused by burial in the laminae. As we shall sufficient to build a 30-m layer during 10~
discuss, the formation of scarps in the years. H o w e v e r , the total thickness of the
laminae m a y supply the needed H20. An laminae is only a few kilometers (Dzurisin
added problem is that the w a t e r loss rate and Blasius, 1975). H e n c e only 10r years is
must be substantial for the present obli- represented in the laminae if each layer is
quity so that water can be driven between c o n s e r v e d and represents 105 years. Al-
the poles. The p e r m a n e n t polar cap thick- though the lack of craters suggets a possible
ness must be m o d e r a t e l y close to the lami- youthful age for the laminae, it is difficult to
nae thickness plus 10 m, or about 40 m. imagine why they should have begun to be
The second terracing m e c h a n i s m is obli- formed only I0 r years ago.
quity oscillations. At low obliquities dust- One possibility is that the laminae are not
storm activity will cease, and H20 ice conserved. Cutts et al. (1976) note that the
sheets will form within p e r m a n e n t COs caps laminae crater age can be explained by a
o v e r a large polar area from water supplied crater filling rate of I km/106 years. Such a
by permafrost. Once formed, these sheets time is consistent with the time for orbital
will be stable as the obliquity increases. At motions to create 1 km o f laminae. The
moderate obliquities dust storms will occur most plausible m e c h a n i s m for destroying 1
so that both ice and dust will be deposited. km of laminae e v e r y 106 years is horizontal
At larger obliquities any e x p o s e d ice will be propagation of the scarps that incise the
driven off the p e r m a n e n t caps and back into laminae. As we shall discuss, and as H o w -
regolith reservoirs. A layer of dust will cap ard (1978) has pointed out, laminae com-
the underlying ice and stablize it against the posed mainly of water ice could propagate.
seasonal t e m p e r a t u r e wave. Permafrost The troughs are spaced at intervals of about
reservoirs must be able to supply the differ- 50 km and do have depths on the order of 1
ence b e t w e e n the p e r m a n e n t cap w a t e r and km. The laminae will be obliterated as they
that lost to the laminae during each obli- form if the troughs migrate at the modest
quity cycle. This loss of w a t e r to the lami- rate of 5 cm year -1 and if the dust released
nae need not imply a huge p e r m a f r o s t is r e m o v e d gradually by winds. This
source of water. The formation of scarps in process m a y be the major one regulating
the laminae m a y also recycle the needed the thickness of the individual laminae.
 MARTIAN CLIMATE 601

Another mechanism that could increase those in the south, where there are not
the estimated age of the laminae is plastic albedo differences between flats and
thinning due to outward flow caused by the scarps, are not forming at present. Why the
overburden load, followed by sublimation slopes are terraced is not clear in Howard's
at the edges as suggested for low-latitude model. An advantage of Howard's hypoth-
permafrost features by Squyres (1978). If esis is that the laminae continually recycle
the laminae are 90% ice, simply removing material and do not require water to be
the ice at the bottom but retaining the dust supplied from the regolith. Howard thought
extends the age of the laminae to 108 years. that the laminae are composed mainly of
A reasonable hiatus in dust storm and ice water ice.
deposition activity does not suffice to ex- Cutts et al. (1979) agreed with Howard
tend greatly the age of the laminae. For that insolation control is important in form-
example, large eccentricities only occur ing the slopes, but they believed that the
during about half of Martian history; if dust laminae are mostly dust and that the ice
storms only occur at large eccentricity, serves mainly as a dust trap. Hence, Cutts
then the age of the laminae could be in- et al. have no mechanism to extend the age
creased by a mere factor of 2. of the laminae by recycling material. Ac-
The final property of the laminae to ex- cording to their model the equatorward-
plain, and the one that has attracted most facing slopes of the troughs do not retain
attention, is the deep troughs that dissect ice because they receive higher insolation
the laminae. Earlier these troughs were than flats. Since the slopes do not retain
thought to be the edges of sets of young ice, they do not retain dust either. Cutts et
laminae stacked uniformly on top of older al. pointed out the existence of an undulat-
laminae. Such a structure suggested a dou- ing character for the fiats separating the
bly periodic constructional mechanism to troughs. They thought that these undula-
Murray et al. (1973). The present concept tions are caused by a gradually advancing
of the structure of the laminae does not or receding permanent polar cap and a
suggest a series of steps of progressively cyclic dust deposition cycle that causes the
younger age. dust retention region to gradually shift in a
More recent work has revolved around wavelike manner. Such a mechanism im-
the idea that both erosional processes and plies different time scales or phases for dust
constructional processes play an important cycles and water cycles.
role (Howard, 1978; Cutts et al., 1979). We propose a model of the laminae that
Howard assumed (incorrectly) that the is closely related to Howard's (1978)
laminae were part of a stepped topography. model. As we pointed out, we believe that
He thought that the steps were all of the approximately 40-m-thick layers composed
same age and believed that the higher inso- of ice and dust are deposited during obli-
lation on the equatorward-facing slopes quity and/or eccentricity cycles as the rem-
preferentially removed ice from the slopes nants of the permanent polar caps of these
and led to its deposition on the high-albedo eras. As the obliquity and eccentricity
flat areas separating the slopes. Winds car- change, the ice is forced off the polar caps.
ried away dust from the slopes leading to a The dust at the upper surface is concen-
poleward migration and formation of steep trated until eventually a thin layer of insu-
faces. Howard favored a model in which lating dust covers the dusty ice and pre-
the layers forming the terraces of the lami- vents further water loss through the upper
nae are being formed from the material surface. Hence, we believe that orbitally
composing the slopes as they erode. Hence forced insolation variations are responsible
he believed that the laminae in the north are for the terraced structure. However, we
forming under present conditions, while believe, like Howard, that regional slope
602 TOON ET AL.

differences create insolation differences un- Howard (1978). Propagation releases ice at
connected with orbital variations. These the laminae edges and recycles the polar
regional slopes cause water ice to be unsta- laminae, both filling and destroying craters
ble on equatorward-facing slopes. Hence and reducing the necessity for continual
these are defrosted zones. Since the lami- resupply of the laminae with permafrost
nae are largely water ice, the terraced water ice. This recycling allows the laminae
structure is exposed and eroded along these to be much older than 107 years, despite the
slopes. The slopes propagate poleward and fact that one 30-m layer forms e v e r y 10~
the propagation eventually leads to the ob- years.
served defrosted bands of rather uniform The final features thought to suggest
horizontal structure. Water ice is unstable Martian climate change are the channels.
on these slopes at all eccentricities and There are many different types of these
most obliquities, so they cut through nu- features, implying different origins. (See
merous terraces. Sharp and Malin, 1975). Some channels,
If equatorward-facing slopes are unsta- mostly found in chaotic terrain, seem to
ble, it follows that poleward-facing slopes have been formed during a single episode
are more stable than fiat terrain. Hence in following the catastrophic collapse of a
addition to equatorward-facing slopes volatile-rich terrain. Although the morphol-
propagating poleward by erosion, one ogy o f other features, such as channels on
also expects poleward-facing slopes mov- crater rims and channel networks covering
ing poleward by deposition. We propose mountain systems, suggests rainfall, it does
that such propagating slopes form the al- not require rainfall. Many geologists be-
ternate sides of the troughs There is no lieve that the channels were fed primarily
reason for the deposition and erosion pro- from ground water sources. A much better
cesses to proceed at identical rates. knowledge of the requirements for the for-
Hence one may see isolated escarpments mation of the channels, such as duration o f
in various regions. The undulating terrain flow, number of repeated episodes o f flow,
may represent such propagating, isolated water source, and whether or not the flow
features. Conceivably, the interaction of was capped by ice, is needed before any
these structures may lead to growth or reliable climatological inferences can be
decay of their amplitude. drawn.
This model has several advantages. The The geographic distribution of the chan-
terraces are formed by orbital variations. In nels is also not well known. Sagan et al.
Howard's (1978) simplest model the ter- (1973) presented data showing that the
races were formed concurrently with the channels are concentrated in the equatorial
troughs. Therefore, the south pole pre- regions, but the distribution may represent
sented a problem because it now has no observational bias rather than the true dis-
high-albedo flats that could grow. The es- tribution. Sagan et al. (1973) pointed out
carpments and troughs in our model are that, if channels are more abundant in the
formed through insolation controlled by equatorial regions, a climate-controlled wa-
slopes. We believe that both poleward and ter origin would be favored. Pieri (1976)
equatorward slopes will propagate respec- noted that small Martian channels (which
tively by deposition and erosion. Hence are the most numerous channels and are
troughs can form, as well as undulating typically less than 150 km long and 1 to 10
terrain. Unlike Cutts et al. (1979), we feel km wide) are restricted to low-albedo,
that the laminae are composed mainly o f highly cratered terrains. The former restric-
water ice so that the scarps can move. tion is possibly observational since in high-
Propagation allows the development of long albedo terrains the lack of contrast makes it
uniform scarps from original small nonuni- more difficult to see the channels, and since
formities such as craters, as pointed out by high-albedo terrains tend to be covered
 MARTIAN CLIMATE 603

with a debris mantle that may hide the Pollack (1979) has pointed out that only
stream beds. modest total pressures in CH4, NHa atmo-
The restriction of the small channels to spheres are needed to raise global average
highly cratered terrain may be a significant temperatures above freezing. Unfortu-
clue to their age. Most geologists believe nately, the photochemical stability of such
that the small channels, which are the most atmospheres is unknown. If one assumes
suggestive of a climatic origin, are older that during the climatic epoch in question
than the larger channels that appear to have the temperature of the top 100 m of previ-
been formed catastrophically. This belief is ously frozen soil must increase substan-
based on crater counts (Pieri, 1976; Sharp tially, then the atmosphere must maintain
and Malin, 1975). The age of both channel high temperatures for 104 years, the thermal
types, according to these workers, seems to penetration time.
be billions of years. Channels are absent in A massive CO2 atmosphere has several
younger geologic unit of the planet, such as advantages over a reduced one. CO2 is
most volcanic areas. An opposite point of presently observed in great abundance on
view was expressed by Masursky et al. Mars, Venus, and Earth. Some theories of
(1977), who asserted that cratering counts the evolution of the atmospheres of the
show that channels have formed during all terrestrial planets (Walker, 1976) suggest
periods from 3 × 109 until 5 × l0 s years that early CO2-dominated atmospheres,
ago. containing very little hydrogen, are more
Many Martian channels probably form as likely than highly reduced, hydrogen-rich
the result of geological processes, such as atmospheres. A CO2 atmosphere has a long
volcanic activity or other geothermal lifetime, being eventually destroyed by geo-
events which melt subsurface ice. The time chemical processes that form carbonates
history of such channels depends on the and a clay regolith. About 1 bar of CO2 is
geological thermal history of Mars. Most needed to yield global average Martian sur-
Martian channels do not have any obvious face temperatures above freezing with
geothermal region at their source. Since present solar luminosity and Martian al-
these channels are geographically wide- bedo (Cess et al., 1980). However, we have
spread, a global heat source is suggested to shown that much lower CO2 pressure--as
supply liquid H20. A plausible heat source low as 100 mbar in the extreme, but 500
is increased surface temperatures caused mbar being more reasonable--could yield
by climatic warming. sizable greenhouse effects. In particular, if
One possible cause of climatic warming the Martian surface albedo were lower
is that the solar luminosity may have been prior to the formation of abundant iron
higher at some past time. Although the oxides, then only about 500 mbar of CO~
possibility cannot be excluded, the current would be needed to yield a global average
theory of solar evolution does not suggest temperature at the freezing point of water.
that the solar luminosity was ever substan- Abundant high HsO clouds could also con-
tially higher in the past. In fact, the oppo- tribute to a greenhouse, and, furthermore,
site is most likely. In addition, a solar- it is not clear the global average tempera-
luminosity increase large enough to raise tures are the relevant ones for creating
Martian temperatures substantially would channels. Equatorial temperatures will be
also lead to very much higher terrestrial substantially higher than global average
temperatures, possibly high enough to ones. Another likely possibility is that, other
cause a runaway greenhouse on Earth (Pol- trace gases, such as SOs, may augment the
lack, 1979). COs greenhouse.
The most likely cause of warmer climate The present polar temperatures on Mars
is an early, more massive atmosphere capa- are so cold that if the abundance of COs
ble of producing a substantial greenhouse. were increased it would freeze out at the
 604 TOON ET AL.

poles or e n t e r the regolith. A m m o n i a w o u l d REFERENCES

also freeze o u t at p r e s e n t - d a y t e m p e r a -
tures. V a r i a t i o n s in the M a r t i a n o r b i t a n d , ANDERS, E., AND OWEN, T. (1977). Mars and Earth:
in p a r t i c u l a r , higher o b l i q u i t i e s at a n early Origin and abundance of volatiles. Science 198, 453-
p e r i o d (before the f o r m a t i o n o f c a r b o n a t e s 465.
or a n a d s o r b i n g regolith) m a y h a v e b e e n a n ANDERSON,D. M., SCHWARZ, M. J., AND RICE, A. R.
(1978). Water vapor adsorption by sodium mont-
i m p o r t a n t trigger in m o v i n g f r o z e n volatiles
morillonite at -5°C. Icarus 34, 638-644.
into the a t m o s p h e r e so that a s u b s t a n t i a l BAggER, R. M. (1967). Surface and volume flow in
g r e e n h o u s e c o u l d o c c u r . It is n o t c e r t a i n porous media. In The Solid-Gas Interface 2 (E. A.
w h e t h e r o r n o t a g r e e n h o u s e , o n c e initi- Flood, Ed.), Vol. 2, pp. 557-609. Dekker, New
ated, w o u l d be stable at l o w o b l i q u i t y . Cer- York.
BEGGER,A. L. (1976). Obliquity and precession for the
t a i n l y large c l i m a t e c h a n g e s , e v e n d u r i n g a
last 5,000,000 years. Astron. Astrophys. 51, 127-
greenhouse epoch, would occur because of 135.
o b l i q u i t y a n d orbital v a r i a t i o n s . T h e sug- BIEMANN, K., ORO, J., TOULMIN, P., III, ORGEL, L.
g e s t i o n b y W a r d e t al. (1979) that the e a r l y E., NIER, A. O., ANDERSON,D. M., SIMMONDS, P.
M a r t i a n o b l i q u i t y m a y h a v e a l w a y s b e e n as G., FLORY, D., DIAZ, A. V., RUSHNECK, D. R.,
BILLER, J. l . , AND LAFLEUR, A. G. (1977). The
large, a n d u s u a l l y larger t h a n , the p r e s e n t
search for organic substances and inorganic volatile
o n e m a y be q u i t e i m p o r t a n t . H o w e v e r , o u r compounds in the surface of Mars. J. Geophys. Res.
k n o w l e d g e o f the d y n a m i c s of a c l i m a t e so 82, 4641-4658.
s u b s t a n t i a l l y different from the p r e s e n t BRIGGS, G. A. (1974). The nature of the residual
M a r t i a n c l i m a t e is too s c a n t y to d e t e r m i n e Martian polar caps. Icarus 23, 167-191.
BRUNAUER, S., COPELAND, L. E., AND KANTRO, D.
clearly the significance o f c h a n g e s in orbit
L. (1967). The Langmuir and BET Theories. In The
and obliquity. Solid-Gas Interface (E. A. Flood, Ed.), Vol. 1, pp.
O u r u n d e r s t a n d i n g o f the p o s s i b l e c a u s e s 77-105. Dekker, New York.
o f c l i m a t e c h a n g e o n Mars is r a p i d l y in- CARR, M. H., AND SCHABER, G. G. (1977). Martian
c r e a s i n g as we l e a r n m o r e a b o u t the dy- permafrost features. J. Geophys. Res. 82, 4039-
4054.
n a m i c s o f the p r e s e n t M a r t i a n climate.
CESS, R. D. (1974). Radiative transfer due to atmo-
H o w e v e r , a great m a n y u n c e r t a i n t i e s per- spheric water vapor. Global considerations of the
sist as to the n a t u r e of the c l i m a t e c h a n g e s Earth's energy balance. J. Quant. Spectrosc. Ra-
that h a v e a c t u a l l y o c c u r r e d . O u r c u r r e n t dirt. Transfer 14, 861-871.
s t r a t i g r a p h y is v e r y c r u d e , so the t i m i n g CESS, R. O., RAMANATHAN, V., AND OWEN, T.
(1980). The Martian paleoclimate and enhanced
a n d d u r a t i o n o f m o s t e v e n t s are p o o r l y
atmospheric carbon dioxide. Icarus 41, 159-165.
k n o w n . I n o r d e r to u n d e r s t a n d the forma- CLARK, B. C., III (1978). Implication of abundant
tion m e c h a n i s m s o f the c h a n n e l s a n d o f the hygroscopic minerals in the Martian regolith. Icarus
p o l a r l a m i n a e we m u s t k n o w m o r e a b o u t 34, 645-665.
t h e s e s t r u c t u r e s , such as the w a t e r - t o - i c e CLARK, B. C.., III, BAIRD, A. K., ROSE, H. J., JR.,
TOULMIN, P., III, CHRISTIAN, R., KELLIHER, W.
ratio in the l a m i n a e , the n a t u r e o f the
C., CASTRO, A. J., ROWE, C. D., KEIL, K., AND
p h y s i c a l c h a n g e a c r o s s t e r r a c e s in the lami- HuSS, G. R. (1977). The Viking X-ray fluorescence
n a e , a n d the r e l a t i o n o f c h a n n e l s to t h e i r experiment, analytical methods and early results. J.
fluid r e s e r v o i r s . W e m u s t rely o n the f u t u r e Geophys. Res. 82, 4577-4594.
e x p l o r a t i o n of M a r s to r e s o l v e m a n y o f CLARK, S. P., JR. (1966). Handbook o f Physical Con-
stants, Geological Society of America Memoirs,
these uncertainties.
Vol. 97, p. 478. Geol. Soc. Amer., New York.
CUTTS, J. A. (1973). Nature and origin of layered
ACKNOWLEDGMENTS deposits of the Martian polar regions. J. Geophys.
We greatly appreciate discussions with C. Leovy, Res. 78, 4231-4249.
A. Ingersoll, A. Seiff, J. Cuzzi, and F. Fanale, and CUTTS, J. A., BLASIUS, K. R., CARR, M. H.,
preprints from J. Cutts, F. Fanale, and B. Farmer. The GREELEY, R., AND MASURSKY, H. (1976). North
work by Toon was partly supported by a NASA joint polar regions of Mars: Imaging results from Viking
research interchange with Cornell University (NCA2- 2. Science 194, 1329-1337.
OR 175-702). CUTTS, J. A., BLASIUS, K. R., AND ROBERTS, W. J.
 MARTIAN CLIMATE 605

(1979). Evolution of Martian polar landscapes: In- spheric pressure variation and the climate of Mars.
terplay of long-term variations in perennial ice cover J. Atmos. Sci. 30, 1502-1508.
and dust storm intensity. J. Geophys. Res. 84, 2975- GOODING, J. L. (1978). Chemical weathering on Mars.
2994. Icarus 33, 483-513.
Cuzzl, J. N., AND MUHLEMAN, D. O. (1972). The HABERLE, R. M., LEOVY, C. B., AND POLLACK, J. B.
microwave spectrum and nature of the subsurface of (1979). A numerical model of the Martian polar cap
Mars. Icarus 17, 548-560. winds. Icarus 39, 151-183.
DAVIES, D. W., FARMER, C. B., AND LAPORTE, D. D. HAYS, J. D., IMBRIE, J., AND SHACKLETON, N. J.
(1977). Behavior of volatiles in Mars' polar areas: A (1976). Variations in the Earth's orbit: Pacemaker of
model incorporating new experimental data. J. the ice ages. Science 194, 1121-1132.
Geophys. Res. 82, 3815-3822. HESS, S. L., HENRY, R. M., LEOVY, C. B., RYAN, J.
DAvis, B. W. (1969). Some speculations on adsorption A., AND TILLMAN, J. E. (1977). Meteorological
and desorption of CO2 in Martian bright areas. results from the surface of Mars: Viking l and 2. J.
Icarus 11, 155-158. Geophys. Res. 82, 4559-4574.
DOBROVOLSKIS, A., AND INGERSOLL, A. P. (1975). HESS, S. L., HENRY, R. M., AND TILLMAN, J. E.
Carbon dioxide-water clathrate as a reservoir of (1979). The seasonal variation of atmospheric pres-
COs on Mars. Icarus 26, 353-357. sure on Mars as affected by the south polar cap. J.
DZURISIN, D., AND BLASIUS, K. R. (1975). Topogra- Geophys. Res. 84, 2923-2927.
phy of the polar layered deposits of Mars. J. HoRns, P. V. (1974). Ice Physics, p. 359. Oxford Univ.
Geophys. Res. 80, 3286-3306. Press (Clarendon), London/New York.
DZURISIN, D., AND INGERSOLL, A. P. (1975). Seasonal HOWARD, A. D. (1978). Origin of the stepped topogra-
buffering of atmospheric pressure on Mars. Icarus phy of the Martian poles. Icarus 34, 581-599.
26, 437-440. HUMMEL, J. R., AND RECK, R. (1979). A global
FANALE, F. P. (1971). History of Martian volatiles: surface albedo model. J. Appl. Meteor. 18, 239-253.
Implications for organic synthesis. Icarus 15, 279- HUNT, G. E. (1973). Radiative properties of terrestrial
303. clouds at visible and infra-red thermal window
FANALE, F. P. (1976). Martian volatiles: Their degas- wavelengths. Quart. J. Roy. Meteor. Soc. 99, 346-
sing history and geochemical fate. Icarus 28, 179- 369.
202. INGERSOLL, A. P. (1970). Occurrence of liquid water.
FANALE, F. P., AND CANNON, W. A. (1971). Adsorp- Science 168, 972-973.
tion on the Martian regolith. Nature 230, 502- INGERSOLL, A. P. (1974). Mars: The case against
504. permanent COz frost caps. J. Geophys. Res. 79,
FANALE, F. P., AND CANNON, W. A. (1974). Ex- 3403-3410.
change of adsorbed H20 and CO2 between the IRVINE, W. M. (1975). Multiple scattering in planetary
regolith and atmosphere of Mars caused by changes atmospheres. Icarus 25, 175-204.
in the surface insolation. J. Geophys. Res. 79, 3397- JAMES, P. B., BRIGGS, G., BARNES, J., AND SPRUCK,
3402. A. (1979). Seasonal recession of Mars' South Polar
FANALE, F. P., AND CANr~ON, W. A. (1978). Mars: cap as seen by Viking. J. Geophys. Res. 84, 2889-
The role of the regolith in determining atmospheric 2922.
pressure and response to insolation changes. J. JONES, K. L., ARVIDSON, R. E., GUINNESS, I. A.,
Geophys. Res. 83, 2321-2325. BRAGG, S. L., WALL. S. D., CARLSTON, C. E., AND
FANALE, F. P., AND CANNON, W. A. (1979). Mars: PICK, D. G. (1979). One Mars year: Viking Lander
CO2 Adsorption and capillary condensation on imaging observations of sediment transport and
clays--significance for volatile storage and at- H~O-condensates. Science 204, 799-806.
mospheric history. J. Geophys. Res. 84, 8404- KIEFFER, H. H., CHASE, S. C., JR., MARTIN, T. Z.,
8414. MINER, E. D., AND PALLUCONI, F. D. (1976).
FARMER, C. B., DAVIES, D. W., HOLLAND, A. L., Martian north pole summer temperatures: Dirty
LAPORTE, D. D., AND DOMS, P. E. (1977). Mars: water ice. Science 194, 1341-1344.
Water vapor observations from the Viking orbiters. KIEFFER, H. H., MARTIN, T. Z., PETERFREUND, A.
J. Geophys. Res. 82, 4225-4248. R., JAKOSKY, B. M., MINER, E. D., AND PALLU-
FARMER, C. B., AND DOMS, P. E. (1979). Global CONI, F. D. (1977). Thermal and albedo mapping of
seasonal variation of water vapor on Mars and the Mars during the Viking primary mission. J.
implications for permafrost. J. Geophys. Res. 84, Geophys. Res. 82, 4249-4291.
2881-2888. LEIGHTON, R. R., AND MURRAY, B. C. (1966). Behav-
GIERASCH, P. J., AND GOODY, R. M. (1968). A study ior of carbon dioxide and other volatiles on Mars.
of the thermal and dynamical structure of the Mar- Science 153, 136-144.
tian lower atmosphere. Planetary Space Sci. 16, MASON, B. J. (1976). Towards the undertanding and
615-646. prediction of climate variations. Quart. J. Roy.
GIERASCH, P. J., AND TOON, O. B. (1973). Atmo- Meteor. Soc. 102, 473-498.
 606 TOON ET AL.

MASURSKY, H., BOYCE, J. M., DIAL, A. L. SCHABER, atmosphere, as inferred from Viking lander imaging
G. G., AND STROBELL, M. E. (1977). Classification data. J. Geophys. Res. 82, 4479-4496.
and time of formation of Martian channels based on POLLACK, J. B., HABERLE, R., GREELEY, R., AND
Viking data. J. Geophys. Res. 82, 4016-4038. IVERSEN, J. (1976). Estimates of wind speeds re-
MAYKUT, G. A., AND UNTERSTEINER, N. (1971). quired for particle motion on Mars. Icarus 29, 395-
Some results from a time dependent thermodynamic 417.
model of sea ice. J. Geophys. Res. 76, 1550-1575. POELARD, D. (1978). An investigation of the astronom-
MCELROY, M. B., KONG, T. Y., AND YUNG, Y. L. ical theory of the ice ages using a simple climate-ice
(1977). Photochemistry and evolution of Mars' at- sheet model. Nature 272, 233-235.
mosphere: A Viking perspective. J. Geophys. Res. SAGAN, C. (1977). Reducing greenhouses and the
82, 4379-4388. temperature history of Earth and Mars. Nature 269,
MILANKOVITCH, M. (1930). Mathematische Klemale- 224-226.
hie und astronomische theorie der klemas- SAGAN, C., AND MULLEN G. (1972). Earth and Mars:
chqankungen. Handbuch der Klimatologie (W. Evolution of atmospheres and surface temperatures.
Koppen and R. Geiger, Eds.), Vol. I, Teil A. Science 177, 52-56.
Berlin. SAGAN, C., TOON, O. B., AND GIERASCH, P. J. (1973).
MILLER, S. L., AND SMYTHE, W. D. (1970). Carbon Climatic change on Mars. Science lgl, 1045-1049.
dioxide clathrate in the Martian ice cap. Science SCHNEIDER, S. H. (1972). Cloudiness as a global
170, 531-533. chmatic feedback mechanism: The effects on the
MURRAY, B. C., AND MALIN, M. C. (1973a). Polar radiation balance and surface temperature of varia-
volatiles on Mars---Theory versus observations. tions in cloudiness. J. Atmos. Sci. 29, 1413-1422.
Science 182, 437-442. SHARP, R. P., AND MALIN, M. C. (1975). Channels on
MURRAY, B. C., WARD, W. R., AND YEUNG, S. C. Mars. Geol. Soc. Amer. Bull. 86, 593-609.
(1973b). Periodic insolation variations on Mars. SHORTHILL, R. W., MOORE, H. J., HULTON, R. B.,
Science 180, 638-640. SCOTT, R. F., AND SPITZER, C. R. (1976). The
OWEN, T., BIEMANN, g . , RUSHNECK, D. R., BILLER, environs of the Viking 2 lander. Science 194, 1309-
J. B., HOWARTH, D. W., AND LAFLEUR, A. L. 1318.
(1977). The composition of the atmosphere at SMOLUCHOWSKI, R. (1968). Mars: Retention of ice.
the surface at Mars. J. Geophys. Res. 82, 4635- Science 159, 1348-1350.
4639. SODERBLOM, L. S., KREIDLER, T. J., AND MASURSKY
OWEN, T., CESS, R., AND RAMANATHAN, V. (1979). H. (1973). Latitudinal distribution of a debris mantle
Infrared CO2 greenhouse to compensate for reduced on the Martian surface. J. Geophys. Res. 78, 4117-
solar luminosity on early Earth. Nature 277, 640- 4122.
642. SQUYRES, S. W. (1978). Mars: Flow of erosional
OYAMA, V. I., CARLE, G. C., WOELLER, F., AND debris. Icarus 34, 600-613.
POLLACK, J. B. (1979). Venus lower atmosphere THOMPSON, S. B., ARNOLD, F., AND SANDERSON, R.
composition: Analysis by gas chromatography. Sci- (1973). Optical effects of cryodeposits on low scatter
ence 203, 802-805. mirrors. Astronaut. Aeronaut. 35, 229-235.
PETERFREUND, A. R., AND KIEFFER, H. H. (1979). TOON, O. B., POLLACK,J. B., AND SAGAN, C. (1977).
Thermal infrared properties of the Martian atmo- Physical properties of the particles composing the
sphere. 3. Local dust clouds. J. Geophys. Res. 84, Martian dust storm of 1971-1972. Icarus 30, 663-
2853-2863. 696.
PIERI, O. (t976). Martian channels. Distribution of TOULMIN, P., III, BAIRD, A. K., CLARK, B. C., KEIL,
small channels on the Martian surface. Icarus 27, K., RosE, H. J., JR., CHRISTIAN, R. P., EVANS, P.
25-50. H., AND KELLIHER, W. C. (1977). Geochemical
POLLACK, J. B. (1979). Climatic change on the terres- and mineralogical interpretation of the Viking inor-
trial planets. Icarus 37, 479-553. ganic chemical results. J. Geophys. Res. 82, 4625-
POLLACK, J. B., AND BLACK, D. C. (1979). Implica- 4634.
tions of the gas compositional measurements of WALKER, J. C. G. (1976). Implications for atmospheric
Pioneer Venus for the origin of planetary atmo- evolution of the inhomogeneous accretion model of
spheres. Science 205, 56--59. the origin of the Earth. In The Early History o f Earth
POLLACK, J. B., COLBURN, D. S., FLASAR, F. M., (B. G. Windley, Ed.), pp. 537-546. Wiley, New
KAHN, R., CARLSTON, C. E., AND PIDEK, O. (1979). York.
Properties and effects of dust particles suspended in WARD, W. R. (1974). Climatic variations on Mars, 1,
the Martian atmosphere. J. Geophys. Res. 84, Astronomical theory of insolation. J. Geophys. Res.
2929-2945. 79, 3375-3386.
POLLACK, J. B., COLBURN, D., KAHN, R., HUNTER, WARD, W. R. (1973). Large-scale variations in the
J., VAN CAMP, W., CARLSTON, C. E., AND WOLF, obliquity of Mars. Science 181, 260-262.
M. R. (1977). Properties of aerosols in the Martian WARD, W. R. (1979). Present obliquity oscillations of
 MARTIAN CLIMATE 607

Mars: Fourth order accuracy in orbital e and I. J. (1974). Climatic variations on Mars. 2. Evolution of
Geophys. Res. 84, 237-241. carbon dioxide atmosphere and polar caps. J.
WARD, W. R., BURNS, J. A., AND TOON, O. B. (1979). Geophys. Res. 79, 3387-3395.
2, Past obliquity oscillations of Mars: Role of the WECHSLER, A. E., AND GLASER, P. E. (1965). Pres-
Tharsis uplift. J. Geophys. Res. 84, 243-259. sure effects on postulated lunar materials. Icarus 4,
WARD, W. R., MURRAy, B. C., AND MALIN, M. C. 335-352.