A N

ew G

lobal Database of M

ars River Dim

ensions

Acknowledgem

ents: Particular thanks to Jonathan Sneed and Sam H

olo for contributing to fieldwork.

Thanks to Marjorie Chan for logistic and intellectual support for fieldw

ork. Thank you to Sharon Wilson,

Caleb Fassett, Joel Davis, and M

arisa Palucis, who each shared unpublished w

ork. We acknow

ledge useful discussions w

ith Becky William

s, Robert Jacobsen, Ben Cardenas, Alan How

ard, Devon Burr, Tim

Goudge,

Alastair Hayden, Caleb Fassett, N

icolas Mangold, D

avid Mohrig, Ben Cardenas, Rom

an diBiase, and especiallyRoss Irw

in. Grants: N

ASA (NN

X15AM49G

, NN

X16AG55G

).

We have ruled out the possibilities that m

y observations result from lim

ited image resolution, postfluvial

modification, flash-snow

melt due to reentry heating from

distal impact ejecta, dam

-overtopping, karst-like m

odification of paleochannels, or misinterpretation of

debris-flow deposits. W

e might be m

easuring strath terraces or channel belts at som

e sites, but not most. Published w

ork on Earth perm

afrost-river hydraulic geometry does not support a large

permafrost correction. Tw

o possibilities remain:

Fine grain sizes, or extreme runoff production on Early M

ars?

Rivers on Mars w

ere wider than rivers on Earth for the sam

e drainage area

The main surprise so far: Channels are frequently too big (relative to their drainage area) to be

easily reconciled with a seasonal-snow

melt clim

ate. As expected from Earth data, scatter is

high. At Earth sites, the width of flat-topped ridges can record channel-belt w

idth (not channel width).

On M

ars, width-inferred and w

avelength-inferred paleodischarges agree, consistent with the

interpretation that well-preserved flat-topped sinuous ridges record channel w

idth.

Evidence for regional variations

Repeated ArcGIS traces are m

erged and analysed in MATLAB.

172 drainage areas are represented. ~83%

of sites have HiRISE coverage

(~60% of those w

ith HiRISE anaglyphs or H

iRISE DTM

s) an advance over the pioneering THEM

IS study of Irwin et al., Geology, 2005

Two m

ain biases in the database: (1) O

nly well-preserved channels and drainage areas are included.

(2) We m

easured the largest visible channels in a catchment.

Discharge Q

inferred from w

idth w:

(using the Ref. 24 scaling; the Ref. 23 scaling gives very sim

ilar results).

Discharge Q

inferred from w

avelength λ:

R:R:

Edwin S. Kite

1,* (kite@uchicago.edu), D

avid P. Mayer 1,2, Casey J. D

uncan3, D

eirdre Edward

1

Rivers on Earth adjust their depths to just mobilize sedim

ent. Therefore, if grainsize on M

ars was sm

all, then modest

river depths could transport sediment, so discharge

could be small (consistent w

ith snowm

elt) for a given width.

Extremely high runoff production (rainfall?) on Early M

ars.

12

1. University of Chicago. 2. N

ow at U

SGS Astrogeology Center. 3. U

niversity of Utah. * poster author.

Motivations &

Findings:

Method:

The scale of ancient Mars rivers – their paleochannel w

idths and m

eander wavelengths – is a proxy for past river discharge.

Discharge, divided by drainage area, yields runoff production, w

hich is a key param

eter for selecting between clim

ate models (e.g rain

versus snowm

elt).

Gale region (1)

+ supraglacial channels (2,8)

SW M

elasChasm

a (9)

Colors correspond togeographic clusters

k-means clustering of geographic distribution of m

easured rivers

High-quality data onlyO

pen circles: widths

Closed circles: wavelengths

Do the flat tops of sinuous ridges m

easured from orbit corre-

spond to channel-belt widths, or channel w

idths? Sinuous ridges near G

reen River in SE Utah preserve channel deposits

and point-bar deposits. At this site, channel deposits preserved w

ithin sinuous ridges can be less than the width of the flat top

of the ridge, and in some interpretations the w

idth of the flat top of the ridge corresponds to channel-belt w

idth, not a channel w

idth. These Earth-analog data are a warning against

over-interpreting HiRISE D

TMs, because the decisive grain-size

and bedset-thickness measurem

ents are very difficult from

orbit. For m

any Mars river deposits, the sim

plest interpretation of sinuous-ridge w

idth is nevertheless that ridge-top width

corresponds to inverted channel width. This is because

0

3D view

of drone-derived Digital Terrain M

odel (DTM

) of SE Utah inverted

λ-only global trend(λ)

(w)

w-only global trend

yellow trend-line: all-data fit, high quality only

cyan trend-line: all-data fit, all qualities

Exploiting preservation of river channel properties in river deposits

Author contributions. E.S.K. analysed M

ars data and supervised fieldwork.

D.P.M. built the M

ars GIS, and the M

ars D

TMs, for the project. C.J.D

. collected the drone data and built the Earth D

TMs. D.E. analysed drone data.

(n = 577; 106 channels)

Automatic extraction from

hand-picked traces (Kite et al., EPSL, 2015)

centerlines picked in triplicate

R: runoff production (mm

/hr)

channel (Oct 2017). Flat top is ~20 m

wide. N

o vertical exaggeration.

(1) sandblasting-away of floodplain deposits allow

s diagnostic bedform

s, e.g. lateral-accretion deposits, to be recognized from

orbit; (2) agreement betw

een meander-w

avelengths and paleochannel w

idths allows cross-checking; (3) rivers can

transition from negative-relief valley floors to positive-relief

channel floors, which decreases the likelihood that either is an

artifact.

Earth data from Bieger

et al., JAWRA 2015