Arguments relative to the age and origin of Kite’s unit F3 and relationship of F2 to F1 and F3.
Bold, italicized letters refer to locations on f3_relationships.jpg. “A:” denotes an argument for
the bold assertion heading.
I agree with this list of arguments, so have made notes/comments inline. I agree that the rivercontaining deposits are not necessarily fluvial deposits and so (in the ‘v2’ draft):Fluvial 1 (F1) is renamed as River-containing unit #1 (R-1);
Fluvial 2 (F2) is renamed as River-containing unit #2 (R-2); and
Fluvial 3 (F3) is renamed as River-containing unit-Southwest (R-SW).
The figure below summarizes the different interpretations of R-SW (i.e., F3) and R-1 (i.e., F1).
The color ramps in all three maps (SW Aeolis Dorsa, Grand Staircase in SW Utah, and Blue
Ridge Scarp in North Carolina) are linear from white to red with a range of 750m. In the Blue
Ridge Scarp case, the highest peaks are saturated at red.
The Grand Staircase example is both a good and a poor analogy for the hypothesis that the F1
and F3 channels are separated by a stratigraphic offset. It is a good analogy in that the paleochannel orientation rotates moving up in the stratigraphy (as the Nevadan orogeny ~150 Ma
started to divert the drainage toward the Sundance Sea). However, it is a less good analogy
because the Earth paleo-channels cannot be seen in the DTM (unlike Mars), because the most
erosionally resistant parts of the Grand Staircase are not necessarily the fluvial members (on
Mars the fluvial parts of the Aeolis Dorsa succession are the most erosionally resistant parts of
the succession), and because the current erosion surface of the Grand Staircase is a fluvial
erosion surface rather than an aeolian erosion surface as on Mars.
In SE Aeolis Dorsa, there is a similar relationship between N-S oriented R-1 materials separated
by a trough from topographically highstanding channels that appear to drain in a different
direction. Screenshot:
This is perhaps a closer analogy to the Grand Staircase in that the paleochannel orientations are
perpendicular to one another (rather than parallel/antiparallel). In my mapping, I lumped the
southern E-W oriented channels with “Southern materials” and did not discuss them.
1. Age of F3 relative to F2 and F1
1.1. F3 is younger than F1
1.1.1. A: The whole system is aggradational during the fluvial epoch, so that higher
deposits are younger (holds true in the area mapped by Matsubara et al., submitted)
Yes, I think this is locally true, although there is long-wavelength deformation. In an
aggrading fluvial deposit, if deposit A is more than 1 channel-deposit thickness
higher than deposit B then A is probably younger than B. Channel-deposit
thicknesses are typically 50x smaller than channel-deposit width – so the relevant
‘topographic offset for time ordering’ is much smaller than any of the topographic
offsets being discussed here. If there are deeply incised valleys (e.g. the Wonoka
paleovalleys; Christie-Blick et al., American Journal of Science, 1990), then
deposits can get out of time order.
1.1.2. A: The original extent of the F3 unit to the northeast appears to be larger than its
present extent due to the abrupt termination of some fairly large inverted channels.
If they extended more than ~3-4 km farther NE, then they must clearly have
originally covered N-draining F1 channels.
Yes, I think this conclusion is hard to avoid because the F3 channels have a large
width and correspondingly wide wavelengths close to the drainage divide. (If only
the widths were large, then perhaps we could be looking at an indurated hyporheic
zone that is much wider than the channels). From
PSP_002002_1735_REDmos_hijitreged_o_25cm (in
‘Transect_3_with_relative_paths’) I get wavelengths of ~1km. The simple peak
discharge estimate is Qflood = 0.011 Lm ^ 1.54 (Burr et al. JGR 2010), which gives
Qflood ~ 500 m3/s. The corresponding drainage area (extending it all the way to the
heads of the F1 channels) is ~50 km2. This gives a runoff-generation rate of 36
mm/hr, which is rather high.
1.1.3. A: If E is part of north-flowing F1 channels, this deposit may extend beneath F2
deposits, assuming that deposits F and B are F2 deposits.
I am not sure whether or not “E” is a channel deposit.
1.1.4. A: If deposits K, L, M, N are F3 deposits, and the deposits O, P, Q, R, S, T
connect to them and are F3 deposits, then the original extent of F3 deposits to the
east was 40+ km greater than the present fan-like deposit Z, bolstering argument
1.1.2.
I agree with this argument. I would go further and link up the chain of mesas
extending 80+km further E than the present fan-like deposit Z (shown as the
easternmost left-right running line in the screenshot below). This chain of mesas
runs perpendicular to the trend of F1 ~300m topographically below the summits of
the mesas, but the summits of the mesas are along trend from Z and at roughly the
same elevation as Z. The poor preservation state of the mesa-chain deposits is
consistent with erosional unzipping of the strata (erosion proceeding from E to W).
These mesas have been isolated and subjected to wind erosion and mass wasting for
longer, and so are more poorly preserved.
1.2. F3 is contemporaneous with F1
1.2.1. A: The deposits have similar erosional presentation
F3 has similar meander wavelengths and channel widths to F1, but a slightly lower
channel-deposit proportion, and a much lower proportion of lateral-accretion
structures. I think F3 is also less sinuous than F1 although I have not quantified this.
1.2.2. A: Erosional windows into (below?) the F3 unit do not reveal older fluvial
deposits with a different flow direction.
The anaglyph at http://www.uahirise.org/ESP_026818_1740 shows narrow sinuous
ridges ~100m topographically below F3. This is the yellow-tinted region equidistant
from B, D and E in ‘f3_relationships.’ The channels in this low-lying region run EW. I am unable to trace these sinuous ridges into F3 channels, which surround the
low-lying region. The adjacent F3 channels are broader and better-preserved, and
locally run N-S except at “B”. Especially because the pattern of drainage in this
local low cannot be reconciled with modern topography, I interpret this area as a
wind-eroded window into an earlier (pre-F3) generation of channel deposits.
1.2.3. A: If fluvial deposits near A are part of F1, flowing northward and units near D
are part of F3, flowing southeast and then bending to the southwest, Then they head
at relatively comparable elevations. The same is true for F1 channels near C.
I agree that the channels at A and C are <100m below channels at D. The channels at A and
C are hard to trace into the rest of F1.
Channel tracing might be misleading if the network structure of river channels tends to get
preserved through multiple stratigraphic layers. In that case, channels that are aligned in
map-view might actually correspond to different stratigraphic layers.
The network structure of river channels must be preserved if the only processes acting on the
landscape are fluvial cut-and-fill and diffusive / draping airfall. In this idealized two-process
model of landscape evolution, diffusive airfall during dry periods cannot create new
topographic depressions and so the (buried) network of fluvial depressions from the previous
wet episode’s cycle of fluvial cut-and-fill must set the template for fluvial cut-and-fill during
the next wet episode.
This is an idealized model - it obviously does not apply to N. Aeolis Dorsa where there is no
evidence for channels stacking vertically - and it might not be a good approximation even in
S. Aeolis Dorsa. However, if the tendency to stack channel networks vertically above one
another is real in S. Aeolis Dorsa, then small breaks in a channel trace (if accompanied by a
big topographic break) might correspond to a shift to a different stratigraphic level.
1.2.4. A: Asymmetric divides occur in terrestrial landscapes. For example, the
asymmetric divide defining the Blue Ridge Mountains in SW Virginia and northern
North Carolina between the steeper, Atlantic-draining streams from the gentler, but
higher Ohio-draining streams. If this were the case, the divide separating lowerelevation N-flowing F1 drainage from higher, SW-flowing F3 drainage might have
been migrating to the SW through time, beheading F3 headwaters.
I agree and have included this as a terrestrial analog in the new Figure 19 in the draft. However,
my (subjective) impression of Aeolis Dorsa is that channels are being exposed at many
stratigraphic levels – more than the 2 (or 3) that we are discussing. If so, then mass is being
added to the landscape so that older river deposits are preserved. If so, this is different from a
mountain belt on Earth where mass is being removed from the system and only one generation of
bedload can be preserved (even if we shut the water off instantaneously). Rivers versus uplift on
Earth, rivers versus airfall on Mars? I wonder what the topography of the Colorado Plateau
would look like if (at every point on the map) we removed all the material down to the top-most
channel deposit in the stratigraphy?
1.2.5. A: In the trough arcing from E to J to G F1 channels cannot be seen plunging
below F3 channels – This depression may have been a highlands forming the divide
between F1 and F3 channels, presumably composed of easily eroded material by
both fluvial and, later, aeolian processes.
This contradicts 1.1.2, which I think is a stronger argument. From G to J I think we are looking
at an outcrop of F2, but I have no explanation for why F2 isn’t exposed in the arc from E to G.
1.2.6. A: If the fluvial deposits at T and U represent F3 and F1 deposits, respectively,
then they head at similar elevations in a E_W trending uplands at and beyond the
southern limit of f3_relationships.jpg.
This is a key area. In my GIS, I filed this under “Illustrating_Key_Relationships”
and not “Important_And_Not_Yet_Understood” because the recognizable channels
in T are at about -1700m, and the recognizable channels in U are at about -1800m.
This offset is big enough to be consistent with the stratigraphic-offset hypothesis,
but only if the fluvial deposits in U are actually part of F1 (not F3 as sketched in my
GIS). ESP_036167_1730 shows that the fluvial deposits in U truly do disappear
beneath F2, as suggested by CTX.
1.3. Relative ages of F2 deposit relative to F1 and F3
1.3.1. F2 is younger than F1 deposit
I agree with this conclusion.
1.3.1.1.
A: There are numerous instances of F1 deposits that terminate laterally
against F2 deposits which rise to a higher elevation than the projected trend of
the F1 deposits.
1.3.1.2.
A: The F2 deposits host fluvial features with a different presentation than
the F1 deposits.
1.3.1.2.1.
In some cases these are aligned depressions or slightly inverted
channels which are narrower, and with less sinuosity than the adjacent F1
inverted channel/floodplain deposits, less meandering, and with a
dendritic pattern that suggests they headed more or less from the present
divides in the F2 unit.
The fine-scale channels in ESP_024497_1745 are a good example of this.
Antoine Lucas’ HiRISE DTM (stored in
“Transect_2_with_relative_paths.zip”) shows that the fine-scale channels do
not narrow in proportion to their distance from the ridgeline of the modern F2
lobe. Instead, they are still wide very near the ridgeline of the modern F2 lobe.
Therefore these dendritic channels did not form on modern F2 topography –
they had bigger catchments than their modern catchments (though probably
very much smaller than the catchments of the F1 rivers).
1.3.1.2.2.
Other slightly inverted channels are somewhat wider, more
sinuous, and longer-wavelength meandering than the case above, and they
follow the general trend of F1 channels/floodplains, and appear to
connect between F1 deposits on either side of the F2 deposits, but they
are a single channel with little evidence of prolonged flow with meander
migration, cutoffs, etc.
1.3.1.2.3.
In at least one case, a now-inverted channel that appears to have
been part of an F2 deposit superimposes and flows in a trend that strongly
diverges from the trend of lower, adjacent F1 deposits.
1.3.1.3.
A point to be made about F2 deposits is that the bulk of the deposits
appears to be non-fluvial. It presents itself as a smooth-surface at multidecameter scale in broad domes, which at multi-meter scale hosts a partial to
complete cover of megaripples, apparently sourced from the unit. This
contrasts with the “yardangy” erosion presentation of the materials encasing
the F1 unit.
Agree. The draft now says up front (§2.2):
“In this study, we describe sedimentary units containing numerous river deposits as “river-containing
deposits.” At some stratigraphic levels, especially within R-2, river-containing deposits have a small
channel-deposit proportion. Where the channel-deposit proportion is small, the river-containing deposits
could represent atmospherically transported material with limited fluvial reworking (Haberlah et al. 2010,
Schiller et al. 2014, Ewing et al. 2006); fluvial deposits with a high proportion of overbank (floodplain)
materials (Bridge 2003); or some intermediate (e.g. Grotzinger et al. 2006). At other stratigraphic levels,
especially within R-1, river-containing deposits have a channel-deposit proportion approaching 100%.
Where the channel-deposit proportion is high, the distinction between river-containing deposits and
fluvial deposits is less important – because whether or not the material was delivered to Aeolis Dorsa by
the wind, most or all of it has been fluvially reworked.”
1.3.2. The F1 unit was aeolian-deflated, creating inverted channels prior to
deposition of the F2 unit
I have added this as an unconformity to the stratigraphic column, and added text on
“Is R-2 concordant on R-1?”. The F1/F2 paleosurface (F1/F2 contact) isn’t as
obviously eroded as the sub-Yardang paleosurface (or the modern topography or the
sub-F1 paleosurface), but it is not completely flat and there is some evidence for
short-wavelength ~10m amplitude erosion.
A: I feel that in several places where F1 channel systems disappear beneath the F2
unit that the F2 unit wraps around the F1 inverted channels, as in my diagram.
Clear examples need to be found and documented.
1.3.3. The F2 unit is older than the F3 unit. (this interfingers with the issue of the
relative ages of the F1 and F3 units).
1.3.3.1.
A: In the vicinity of the G and H F3 channels F2 deposits are
topographically lower, and wrap around the F3 channels, with the simplest
interpretation that F2 underlies F3 and is being exposed by erosion of F3. This
would imply that the F3 deposits are more readily eroded than the F2 deposits
and with a different erosional style (F2 characterized by deflation over broad
surfaces rather than scarp retreat for F3).
Agree. A stratigraphy that explains the observations is
---Highest: F3 (saw lots of water; cliff-forming / caprock)
--Intermediate: Foo-foo dust (rarely seen in outcrop; mantled by talus from F3;
represents most of the height of the scarp that elevates F3 above F2)
-Lowest: F2 (saw some water; not as erosionally resistant as F3 caprock, but
more resistant than intermediate foo-foo dust).
---1.3.3.2.
A: F2 deposits do not superimpose the F3 deposits in the main F3 unit
exposure at Z. However, if the knolls marked J and some similar hills west of
H are part of the F2 deposit, then they do superimpose F3. (alternatively the J
deposits are a wholly different, later deposit, or they are through-poking
basement).
I don’t see anything diagnostic in CTX for the knolls marked “J”.
1.3.3.3.
A: If I, P, Q, R, S and T materials are F3 deposits, it could be argued that
they were deposited on top of F2 materials
Agree.
1.3.4. The F2 unit is younger than the F3 unit. In this case, the F2 deposits either
originally covered the F3 unit and has been more rapidly eroded/stripped than F3
deposits, or the F2 deposits were deposited primarily in eroded depressions
surrounding already-inverted F3 deposits.
1.3.4.1.
A: The probable F3 deposits connecting between O and M appear to be
covered with F2 deposits (and a further capping of J deposits) near Y. Similar
apparent mantling appears to occur west of R.
Agree. If F3 postdates F2 then these deposits are “F4.” Would need lots of
measurements of river-deposit dimensions to confirm that Y this has to be
different stuff than O and M, but I agree that that the intra-ocular impression is
that Y is different stuff.
2. Mode of origin and original slope of the F3 deposits, particularly near Z.
In the stratigraphy paper I would like to punt on the question of the original slope of the F3
deposits. By punt I mean state possibilities, but not take a firm stand. That is because I think
the decisive measurement would be to map out channel migration directions from HiRISE (in
combination with a strong argument that the meanders are not migrating upstream!) and I
have not yet done this carefully enough to fully trust the results. My first-cut database does
support Howard’s interpretation – these are the shapefiles grouped under ‘Bend_Translation_
and_Expansion’, specifically ‘scroll_bars_CTX,’ ‘enclosing_curves_CTX’,
‘scroll_bars_HiRISE_Batch_2’, ‘enclosing_curves_HiRISE_Batch_2’,
‘scroll_bars_HiRISE’, and ‘enclosing_curves_HiRISE’ ). I think a more careful analysis of
the meander-migration data would sit more easily in a ‘paleohydrology’ paper.
2.1. The F3 deposits are a fan or fan/delta sourced from the SW.
2.1.1. A: The overall shape of the deposits is generally delta-form (particularly in the
vicinity of Z).
2.1.2. A: The inverted channels near Z have an anabranching/anastomoshing pattern
similar to distributaries on terrestrial deltas.
2.1.3. A: This interpretation appears to require a topographic reversal of the topography
on the fan/delta, which now decreases in elevation to the SE. A model of lower
crustal flow (Nimmo – Lefort) suggests that topographic inversions of this
magnitude would be possible, maybe even likely. But this critically depends on the
timing of the topographic inversion.
The Nimmo-Lefort idea also explains the anomalously steep present-day plunge of
the meander belts to the N in the area of our ‘ESP_038_882_ngate_1m.cub’ DTM
(PSP_006683_1740). This is because areas to the N of the original step in crustal
thickness get tilted N, while areas to the S of the original step in crustal thickness
get tilted S. In this view the “hinge” between N-tilting to the N, and S-tilting to the
S, sits roughly at the latitude B-E-J-G in ‘f3_relationships.” It is an attractive feature
of the crustal flow hypothesis that it can explain both trends (and produces tilts of
the correct ~1:100 - ~1:200 order of magnitude). Given a crustal thickness contrast,
crustal flow must occur and must occur at all times: the only question is whether the
integral of crustal flow between ~3.5 Gyr and now is enough to explain the required
tilts.
2.2. The F3 deposits are a normal dendritic fluvial system that drained to the SW
It is uncommon on Earth for contributory channel deposits to get preserved in the
sedimentary record, because contributory channel deposits tend to be associated with
erosion. This was a major reason why I was unable to persuade the Mars fluvial
geomorphology reading group at Caltech to view the diBiase et al. delta candidate as a
contributory (W-draining) system. However, there are places on Earth (e.g., the Ganges
basin) where contributory channel deposits do get preserved in the sedimentary record.
2.2.1. A: This explanation does not require topographic inversion and is generally
compatible with inferred flow directions of the modern topography of tributary
system, including the distributaries to the NW (B,F,D) and E (G through T)
2.2.2. A: If this system were a fan/delta it requires a large containing basin to the North,
East and Northeast of the present extent of the delta. This would presumably have
to be deposits emplaced after F1 (and possibly after F2) that are now eroded away.
There is plenty of evidence for differential compaction of F1 and F2 river deposits
now at the surface. Because of the under-abundance of small craters, I think
everything we see in Aeolis Dorsa has been exhumed from “considerable”
overburden.
My hunch based on slope winds work at Gale crater is that the big E-W oriented
trough towards which the F3 channels converge in the SW is a wind-eroded trough
cut by katabatic winds off the dichotomy boundary, that crosscuts stratigraphy. It
could be post-fluvial. (There is a similar “hole” near 122E, 2N, along the dichotomy
boundary W of Gale Crater).
2.2.3. A: No obvious deltaic clinoforms have been identified in beds below the edge of
the putative fan/delta. (but may be covered by aeolian crap and mass wasting).
Agreed
2.2.4. By this interpretation, the anastomosing/avulsions near Z would be the result of a
fan-like aggrading into an existing basin.