1
Data Repository materials
2
1. Methods: determination of layer orientations
3
1m-resolution stereo terrain models were produced from High-Resolution Imaging Science
4
Experiment (HiRISE) images, using the method of Kirk et al. (2008). To confirm that our
5
procedure is measuring layers within the mound, and is not biased by surficial weathering
6
textures nor by the present-day slope, we made measurements around a small reentrant canyon
7
incised into the SW corner of the Gale mound (a DTM illustrating this may be obtained from the
8
authors). Within this canyon, present-day slope dip direction varies through 360°, but as
9
expected the measured layer orientations dip consistently (to the W).
10
11
2. Methods: assessment of alternative mechanisms for producing outward dips
12
Few geologic processes can produce primary outward dips of (3±2)° (Figures 1, 2). Spring
13
mounds lack laterally continuous marker beds of the >10 km extent observed (Anderson & Bell,
14
2010). Preferential dissolution, landsliding/halotectonics, post-impact mantle rebound, and
15
lower-crustal flow can produce outward tilting. On Early Mars, viscoelastic isostatic-
16
compensation timescales are <<106 yr. In order for subsequent mantle rebound to produce
17
outward tilts, the mound must have accumulated at implausibly fast rates. Mars’ crust is
18
constrained to be ≲90 km thick at Gale’s location (Nimmo & Stevenson, 2001), so lower-crustal
19
flow beneath 155km-diameter Gale would have a geometry that would relax Gale Crater from
20
the outside in, incompatible with simple outward tilting. Additionally, Gale is incompletely
21
compensated (Konopliv et al., 2011) and postdates dichotomy-boundary faulting, so Gale
22
postdates the era when Mars’ lithosphere was warm and weak enough for limited crustal flow to
23
relax the dichotomy boundary and cause major deformation (Irwin & Watters, 2010; Watters et
1
24
al., 2007). Any tectonic mechanism for the outward dips would correspond to ~3-4 km of floor
25
uplift of originally horizontal layers, comparable to the depth of a fresh crater of this size and
26
inconsistent with the current depth of the S half of the crater if we make the reasonable
27
approximation that wind cannot quickly erode basalt. Tectonic doming would put the mound's
28
upper surface into extension and produce extensional faults (e.g., p.156 in Melosh, 2011), but
29
these are not observed. Preferential dissolution causes overlying rock to fail and leaves karstic
30
depressions (Hovorka, 2000), which are not observed at Gale. Landsliding/halotectonics can
31
produce deformed beds in layered sediments on both Earth and Mars (Metz et al., 2010; Jackson
32
et al., 1990; Hudec & Jackson 2011),. These sites show order-unity strain and contorted bedding,
33
but the layers near the base of Gale’s mound show no evidence for large strains at kilometer
34
scale, except for a possible late-stage landslide on the Gale mound’s north flank (Anderson &
35
Bell, 2010).
36
37
3. Methods: MarsWRF Gale runs
38
MarsWRF is the Mars instantiation of Planetary Weather Research and Forecasting
39
(planetWRF), developed by Ashima Research as an open-source branch of the terrestrial NCAR
40
WRF model (Richardson et al., 2007). For the Gale runs, MarsWRF was run in mesoscale model
41
nudged at the boundaries by
42
43
4. Methods: scaling sediment transport
44
Conservation of sediment mass (Anderson, 2008) in an atmospheric boundary-layer column can
45
be written as:
46
2
47
dz/dt = D – E = CWs -E
48
49
Here C is volumetric sediment concentration, Ws is settling velocity, and E is the rate of
50
sediment pick-up from the bed. In aeolian transport of dry sand and alluvial-river transport,
51
induration processes are weak or absent and so the bed has negligible intergrain cohesion. C
52
tends to E/Ws over a saturation length scale that is inversely proportional to Ws (for dz/dt > 0) or
53
E (for dz/dt <0). This scale is typically short, e.g. ~1-20m, for the case of a saltating sand on
54
Earth (Kok et al., 2012). Our simplifying assumption that D  f ( x ) and therefore C  f ( x )
55
implies that this saturation length scale is large compared to the morphodynamic feedback of
56
interest. For the case of net deposition (dz/dt > 0) this could correspond to settling-out of
57
sediment stirred up by dust storms (Vaughan et al., 2010). These events have characteristic
58
length scales >102 km, larger than the scale of Gale’s mound and justifying the approximation of
59
uniform D (Szwast et al., 2006). For the case of net erosion (dz/dt < 0), small E implies a
60
detachment-limited system where sediment has some cohesion. The necessary degree of
61
induration is not large: for example, 6-10 mg/g chloride salt increases the threshold wind stress
62
for saltation by a factor of e (Nickling, 1984). Fluid pressure alone cannot abrade the bed, and
63
the gain in entrained-particle mass from particle impact equals the abrasion susceptibility, ~2
64
×10-6 for basalt under modern Mars conditions (Bridges et al., 2012) and generally <<1 for
65
cohesive materials, preventing runaway adjustment of C to E/Ws. Detachment-limited erosion is
66
clearly appropriate for slope-wind erosion on modern Mars (because sediment mounds form
67
yardangs and shed boulders, indicating that they are cohesive/indurated), and is probably a better
68
approximation to ancient erosion processes than is transport-limited erosion (given the evidence
3
69
for ancient near-surface liquid water, shallow diagenesis, and soil crusts) (McLennan &
70
Grotzinger, 2008; Arvidson et al., 2010; Manga et al., 2012).
71
Our model makes testable predictions for MSL. For example, the key role attributed to
72
suspended sediment during mound growth predicts that particles too large to be suspended will
73
be uncommon, except as aggregates (Sullivan et al., 2008).
74
75
5. Methods: parameter choices and mound growth dynamics
76
Coriolis forces are neglected because almost all sedimentary rock mounds on Mars are equatorial
77
(Kite et al., 2012). Additional numerical diffusivity at the 10-3 level is used to stabilize the
78
solution. Analytic and experimental results show that in slope-wind dominated landscapes, the
79
strongest winds occur close to the steepest slopes (Manins & Sarford, 1987). L will vary across
80
Mars because of changing 3D topography (Trachte et al., 2010), and will vary in time because of
81
changing atmospheric density. Ye et al. (1990) find L ~ 20km for Mars slopes with negligible
82
geostrophic effects, and Equation 49 in Magalhaes & Gierasch (1982) gives L ~ 25 km for Gale-
83
relevant slopes. Simulations of gentle Mars slope winds strongly affected by planetary rotation
84
suggest L ~ 50-100 km (Saviarvi & Siili, 1993; Siili et al., 1999). Entrainment acts as a drag
85
coefficient with value 0.02-0.05 for Gale-relevant slopes (Manins & Sawford, 1987; Ellison &
86
Turner, 1959; Horst & Doran, 1986), suggesting L = 20-50 km for a 1km-thick cold boundary
87
layer. Dark strips in nighttime thermal infrared mosaics of the horizontal plateaux surrounding
88
the steep-sided Valles Marineris canyons are ~20-50 km wide, which might correspond to the
89
sediment transport correlation length scale for anabatic winds (Spiga & Forget, 2009). Therefore
90
we take L ~ 101-2 km to be reasonable, but with the expectation of significant variability in L/R,
91
which we explore in the next section.
4
92
Early in mound evolution (Phase I; Figure 3b), mound topography can be a positive
93
feedback on mound growth because the mound’s adverse slope decelerates erosive winds
94
flowing down from the canyon walls. As the mound grows, winds flowing down the mound
95
flanks become progressively more destructive, and a kinematic wave of net erosion propagates
96
inward from the mound toe (Phase II). During the all-erosive Phase III, decreasing the mound
97
height reduces the maximum potential downslope wind. However erosion also decreases mound
98
width, which helps to maintain steep slopes and correspondingly strong winds. Erosion decreases
99
everywhere at late stage, and the model mound can either (i) enter a quasi-steady state where
100
slow continued slope-wind erosion is balanced by diffusive geologic processes such as
101
landsliding, or (ii) reduction in windspeed in the widening moat can lead to cycles of satellite-
102
mound nucleation, autocatalytic growth, inward migration and self-destruction. There is strong
103
evidence for secular climate change on the real Mars, which would break the assumption of
104
constant external forcing (Main Text). Uo is set to zero in Figure 3, and Uo sensitivity tests show
105
that for a given D', varying Uo has little effect on the pattern of erosion because spatial variations
106
are still controlled by slope winds. Equation (3) implies the approximation E ~ max(U)α ~ ∑Uα,
107
which is true as α  ∞. To check that this approximation does not affect conclusions for α = 3-4
108
(Kok et al., 2012), we ran a parameter sweep with E ~ (U+α + U-α). For nominal parameters
109
(Figure 3), this leads to only minor changes in mound structure and stratigraphy (e.g., 6%
110
reduction in mound height and <1% in mound width at late time). For the parameter sweep as a
111
whole, the change leads to a slight widening of the regions where the mound does not nucleate or
112
overspills the crater (changing the outcome of 7 out of the 117 cases shown in Figure DR2). The
113
approximation would be further supported if (as is likely; 24) there is a threshold U below which
114
erosion does not occur. If MSL shows that persistent snow or ice was needed as a water source
5
115
for layer cementation (Niles & Michalski, 2009; Kite et al., 2012), then additional terms will be
116
required to track humidity and the drying effect of föhn winds (Conway et al., 2012; Madeleine
117
et al., 2012).
118
119
6. Controls on mound growth and form: sensitivity to parameter variation
120
To confirm that our results were not the result of idiosyncratic parameter choices, we carried out
121
a parameter sweep in α, D’, and R/L (Figure DR2). Weak slope dependence (α = 0.05) is
122
sufficient to produce strata that dip toward the foot of the crater/canyon slope (like a sombrero
123
hat). Similarly weak negative slope dependence (α = -0.05) is sufficient to produce concave-up
124
fill.At low R/L (i.e., small craters) or at low α, D' controls overall mound shape and slope winds
125
are unimportant. When D' is high, layers fill the crater; when D’ is low, layers do not
126
accumulate. When either α or R/L or both are ≳1, slope-wind enhanced erosion and transport
127
dominates the behavior. Thin layered crater floor deposits form at low D', and large mounds at
128
high D'.If L is approximated as being constant across the planet, then R/L is proportional to
129
crater/canyon size. There is net deposition everywhere for small R/L, although there a small moat
130
can form as a result of lower net-aggradation rates near the crater wall. For larger R/L, moats
131
form, and for the largest craters/canyons, multiple mounds can form eventually because slope
132
winds break up the deposits. This is consistent with data which suggest a maximum length scale
133
for mounds (Figure DR3). Small exhumed craters in Meridiani show concentric layering
134
consistent with concave-up dips. Larger craters across Meridiani, together with the north polar
135
ice mounds, show a simple single mound. Gale and Nicholson Craters, together with the smaller
136
Valles Marineris chasmata, show a single mound with an undulating top. The largest canyon
137
system on Mars (Ophir-Candor-Melas) shows multiple mounds per canyon. Gale-like mounds
6
138
(with erosion both at the toe and the summit) are most likely for high R/L, high α, and
139
intermediate D' (high enough for some accumulation, but not so high as to fill the crater) (Figure
140
DR2). These sensitivity tests suggest that mounds are a generic outcome of steady uniform
141
deposition modified by slope-wind enhanced erosion and transport for estimated Early Mars
142
parameter values.
143
144
Data Repository References
145
Anderson, R.S., 2008, The Little Book of Geomorphology: Exercising the Principle of
146
147
Conservation, http://instaar.colorado.edu/~andersrs/The_little_book_010708_web.pdf
Arvidson, R.E. et al., 2010, Spirit Mars Rover Mission: Overview and selected results from the
148
northern Home Plate Winter Haven to the side of Scamander crater, J. Geophys. Res. 115,
149
E00F03.
150
Buczkowski, D.L. & Cooke, M.L., 2004, Formation of double-ring circular grabens due to
151
volumetric compaction over buried impact craters: Implications for thickness and nature of
152
cover material in Utopia Planitia, Mars, J. Geophys. Res. 109(E2), E02006.
153
154
155
156
157
Ellison, T.H., & Turner, J.S., 1959, Turbidity entrainment in stratified flows. J. Fluid Mech. 6, p.
423-48.
Horst, T. W., & Doran, J. C., 1986, Nocturnal drainage flow on simple slopes. Boundary-Layer
Meteorol. 34, p. 263-286.
Hovorka, S. D., 2000, Understanding the processes of salt dissolution and subsidence in
158
sinkholes and unusual subsidence over solution mined caverns and salt and potash mines,
159
Technical Session: Solution Mining Research Institute Fall Meeting, p. 12–23.
7
160
161
162
163
164
165
Hudec, M.R., & Jackson, M.P.A., 2011 The salt mine : a digital atlas of salt tectonics. Austin,
Tex: Jackson School of Geosciences, University of Texas at Austin.
Irwin, R. P., III, and T. R. Watters, 2010, Geology of the Martian crustal dichotomy boundary, J.
Geophys. Res. 115, E11006, doi:10.1029/2010JE003658.
Jackson, M.P.A., et al., 1990, Salt diapirs of the Great Kavir, central Iran: Geological Society of
America, Geol. Soc. Am. Memoir 177, 139 p.
166
Kirk, R.L., et al., 2008, Ultrahigh resolution topographic mapping of Mars with MRO HiRISE
167
stereo images: Meter-scale slopes of candidate Phoenix landing sites. J. Geophys. Res.
168
113(E12), CiteID E00A24.
169
170
171
Konopliv, A.S. et al., 2011, Mars high resolution gravity fields from MRO, Mars seasonal
gravity, and other dynamical parameters, Icarus 211, p. 401-428, 2011.
Madeleine, J.-B., Head, J. W., Spiga, A., Dickson, J. L., & Forget, F., 2012, A study of ice
172
accumulation and stability in Martian craters under past orbital conditions using the LMD
173
mesoscale model, Lunar and Planet. Sci. Conf. 43, abstract no. 1664
174
175
176
177
Magalhaes, J., & Gierasch, P., 1982, A model of Martian slope winds: Implications for eolian
transport, J. Geophys. Res., 87, p. 9975-9984.
Manga, M., Patel, A.,Dufek, J., and Kite, E.S., 2012, Wet surface and dense atmosphere on early
Mars inferred from the bomb sag at Home Plate, Mars, Geophys. Res. Lett. 39, L01202.
178
Manins, P. C., & Sawford, B. L.,1987, A model of katabatic winds, J. Atmos. Sci. 36, p. 619-630
179
Metz, J., Grotzinger, J., Okubo, C., & Milliken, R., 2010, Thin-skinned deformation of
180
181
sedimentary rocks in Valles Marineris, Mars, J. Geophys. Res. 115, E11004.
Melosh, H.J., 2011, Planetary Surface Processes, Cambridge University Press.
8
182
183
184
185
Nimmo, F., & Stevenson, D.J. 2001, Estimates of Martian crustal thickness from viscous
relaxation of topography, J. Geophys. Res. 106, 5085-5098, doi:10.1029/2000JE001331.
Parish, T.R., & Bromwich, D.H., 1987, The surface windfield over the Antarctic ice sheets.
Nature 328, p. 51-54.
186
Richardson, M.I., Toigo, A.D., & Newman, C.E., 2007, PlanetWRF: A general purpose, local to
187
global numerical model for planetary atmospheric and climate dynamics, J. Geophys. Res.
188
112, E09001.
189
190
191
192
Szwast, M., Richardson, M. and Vasavada, A., 2006, Surface dust redistribution on Mars as
observed by the Mars Global Surveyor and Viking orbiters. J. Geophys. Res. 111, E11008.
Savijarvi, H., & Siili, T., 1993, The Martian slope winds and the nocturnal PBL jet. J. Atmos.
Sci. 50, p. 77-88.
193
Siili, T., Haberle, R.M., Murphy, J.R., & Savijarvi, H., 1999, Modelling of the combined late-
194
winter ice cap edge and slope winds in Mars’ Hellas and Argyre regions. Planet. & Space
195
Sci. 47, p. 951-970.
196
Trachte, K., Nauss, T., & Bendix, J., 2010, The impact of different terrain configurations on the
197
formation and dynamics of katabatic flows: idealized case studies. Boundary-Layer
198
Meteorol. 134, p. 307-325.
199
200
201
202
203
204
Vaughan, A.F., et al., 2010. Pancam and Microscopic Imager observations of dust on the Spirit
Rover: Cleaning events, spectral properties, and aggregates, Mars 5, p. 129-145.
Watters, T.R., McGovern, P.J. & Irwin, R.P., 2007, Hemispheres apart: The crustal dichotomy
on Mars, Ann. Rev. of Earth and Planet. Sci. 35, p. 621-652
Ye, Z.J., Segal, M., & Pielke, R.A., 1990, A comparative study of daytime thermally induced
upslope flow on Mars and Earth. J. Atmos. Sci. 47, p. 612-628.
9
205
206
Data Repository Captions
207
208
Figure DR1. Results from MarsWRF
209
The winds are extrapolated to 1.5m above the surface using boundary layer similarity
210
theory (the lowest model layer is at ~9m above the surface).
211
212
MarsWRF (Toigo et al., 2012) is the Mars version of planetWRF (Richardson et al., 2007),
213
a planetary extension of NCAR’s widely-used Weather Research and Forecasting model.
214
For this application, MarsWRF was run as a global Mars model at 2deg resolution, with
215
three increasingly high-resolution domains 'nested' over Gale Crater to increase the
216
resolution there to ~4km in the innermost nest. Each nested domain is both driven by and
217
fed information back to its parent domain, but also responds to surface variations (e.g.
218
topography, albedo) at the resolution of the nest.
219
220
221
Figure DR2. Overall growth and form of sedimentary mounds – results from a model parameter
222
sweep varying R/L and D', with fixed α = 3. Black square corresponds to the results shown in
223
more detail in Figure 3. Symbols correspond to the overall results:– no net accumulation of
224
sediment anywhere (blue open circles); sediment overtops crater/canyon (red filled circles);
225
mound forms and remains within crater (green symbols). Green filled circles correspond to
226
outcomes where layers are exposed at both the toe and the summit of mound, similar to Gale.
227
10
228
Figure DR3. Width of largest mound does not keep pace with increasing crater/canyon width,
229
suggesting a length threshold beyond which slope winds break up mounds. Blue dots correspond
230
to nonpolar crater data, red squares correspond to canyon data, and green dots correspond to
231
polar ice mound data. Gray vertical lines show range of uncertainty in largest-mound width for
232
Valles Marineris canyons. Blue dot adjacent to “G” corresponds to Gale Crater. Craters smaller
233
than 10km were measured using Context Camera (CTX) or HiRISE images. All other craters,
234
canyons and mounds were measured using the Thermal Emission Imaging System (THEMIS)
235
global day infrared mosaic on a Mars Orbiter Laser Altimeter (MOLA) base. Width is defined as
236
polygon area divided by the longest straight-line length that can be contained within that
237
polygon.
11