Streaks on martian slopes are dry - No Water On Mars

Slope streaks are dark albedo features on martian slopes that form spontaneously and fade over years to decades. Along with seasonally recurring slope
lineae, streak formation has been attributed to aqueous processes, implying
the presence of transient yet substantial amounts of liquid water or brines on
Mars’ surface, with important implications for present-day Mars’ habitability.
Here, we use a deep learning-enabled approach to create the first consistent,
global catalog of half a million individual slope streaks. We show that slope
streaks modify less than 0.1% of the martian surface, but transport several
global storm equivalents of dust per Mars year, potentially playing a major role
in the martian dust cycle. Our global geostatistical analysis challenges wet
streak formation models and instead supports dry streak formation, driven by
seasonal dust delivery and energetic triggers like wind and meteoritic impacts.
We further identify a qualitative, spatiotemporal relation between recurring
slope lineae formation, seasonal dust deposition, and dust devil activity. Our
findings suggest that modern Mars’ slopes do not commonly experience
transient flows of water or brines, implying that streak bearing terrain can be
explored without raising planetary protection concerns.

1. Article https://doi.org/10.1038/s41467-025-59395-w
Streaksonmartianslopesaredry
Valentin Tertius Bickel
1
& Adomas Valantinas
2
Slopestreaksaredarkalbedofeatures on martian slopes that form sponta-
neously and fade over years to decades. Along with seasonally recurring slope
lineae, streak formation has been attributed to aqueous processes, implying
the presence of transient yet substantial amounts of liquid water or brines on
Mars’surface, with important implications for present-day Mars’habitability.
Here, we use a deep learning-enabled approach to create thefirst consistent,
global catalog of half a million individual slope streaks. We show that slope
streaksmodifylessthan0.1%ofthemartian surface, but transport several
global storm equivalents of dust per Mars year, potentially playing a major role
in the martian dust cycle. Our global geostatistical analysis challenges wet
streak formation models and instead supports dry streak formation, driven by
seasonal dust delivery and energetic triggers like wind and meteoritic impacts.
We further identify a qualitative, spatiotemporal relation between recurring
slope lineae formation, seasonal dust deposition, and dust devil activity. Our
findings suggest that modern Mars’slopes do not commonly experience
transientflows of water or brines, implying that streak bearing terrain can be
explored without raising planetary protection concerns.
First observed in Viking images in 1977
1,2
, slope streak occurrence and
formation have been subject to a continuous debate ever since. Dark
slope streaks form sporadically and fade over years and decades (~3 to
~20 Mars years, MY)
3,4
, occasionally turning brighter than the sur-
rounding regolith
3,5
. Slope streak populations have been observed to
occur in dusty, low thermal inertia equatorial regions
6,7
.Instarkcon-
trast, recurring slope lineae (RSL),first discovered in 2011, are pre-
dominantly located in rocky, southern, high thermal inertia
highlands
8,9
. The formation of streaks and RSL has been explained by
dry and wet mechanisms, such as airfall- or thermal cycling-induced
dust avalanching and impact-, quake-, dust devil-, or rockfall-induced
mass/dust movements (dry)
10–18
, as well as groundwater spring-, frost-,
and brine-inducedflows (wet)
7,9,19–23
.Recently
4
,and
24
identified an
enhanced streak formation rate in northern autumn and winter (L
s
~140°–270°). Similarly, RSL have been observed to predominantly
form in southern summer (L
s~270°
8
). To date, it remains unclear
whether slope streaks and RSL are different expressions of the same
process, or fundamentally different features (e.g. refs.5,25). If con-
firmed, a wet formation mechanism for streaks or RSL would indicate
that present-day Mars’hydrological cycle is substantially more pro-
nounced than previously anticipated, with substantial implications on
our understanding of martian weather, climate, surface evolution, and
habitability. In addition, liquid water or brines on Mars’surface would
evoke serious concerns about planetary protection. Here, we fuse
several decades’worth of orbital data and use a global-scale deep
learning- and data-driven approach to create thefirst global catalog of
slope streaks, shedding new light on a fundamental, decades-old
debate.
Results
Martian streak global geostatistics
We identify a total of 13,026 bright and 484,019 dark slope streaks with
a wide range of morphologies in the pre-v01 version of the global MRO
CTX (Mars Reconnaissance Orbiter Context Imager) mosaic
26,27
(‘detected population’,p
D), including previously unknown slope
streak-bearing regions (Fig.1; Supplementary Figs. S1, S2). There is no
clear, universally applicable, quantitative way to separate‘dark’from
‘bright’streaks, especially in CTX data, as the transition between dark
and bright is continuous. Here,‘dark’streaks include recently-formed
streaks that appear dark in CTX images, and‘bright’streaks refer to
old, partially-faded streaks that appear bright in CTX images
(Fig.1A, B). The deep learning-driven detector is able to identify about
Received: 2 May 2024
Accepted: 22 April 2025
Check for updates
1
Center for Space and Habitability, University of Bern, Bern, Switzerland.
2
Department of Earth, Environmental and Planetary Sciences, Brown University,
Providence, RI, USA.e-mail:[email protected]
Nature Communications| (2025) 16:4315 1
1234567890():,;
1234567890():,;

2. 20% and 55% of all bright and dark streaks in a given CTX image, on
average; at the same time, we estimate that about 6 % of all detections
are duplicates caused by the spatial overlap of some of the CTX images
that make up the mosaic (Supplementary Fig. S3), suggesting that the
actual number of bright and dark streaks observable on the surface of
Mars are ~60,000 and ~830,000 (at the CTX-scale, ~6 m,‘statistically-
corrected population’,p
C). We note that the deep learning-predicted
detections contain an average of 2.33 individual streaks per detection
box due to streak co-location and overlap (see Fig.1A, B, Supplemen-
tary Fig. S1), indicating there might be up to ~140,000 and ~1,900,000
individual, CTX-scale bright and dark slope streaks on Mars, which is
several factors more than was previously estimated (~800,000 dark
Viking Merged ColorViking Merged Color
Viking Merged ColorViking Merged Color
70°N70°N
150°E150°E100°E100°E50°E50°E0°E0°E50°W50°W100°W100°W150°W150°W
50°N50°N
30°N30°N
10°N10°N
10°S10°S
30°S30°S
50°S50°S
70°S70°S
70°N70°N
50°N50°N
30°N30°N
10°N10°N
10°S10°S
30°S30°S
50°S50°S
70°S70°S
70°N70°N
50°N50°N
30°N30°N
10°N10°N
10°S10°S
30°S30°S
50°S50°S
70°S70°S
Acidalia Planitia
Tempe Fossae
Sinai Planum
Arabia Terra
Hellas Planitia
Terra Sabaea
Noachis Terra
MY36_014887_165_0_RGB
PSP_005787_1475_MRGB
ESP_022689_1380_MRGB
MY35_010269_009_0_RGB
MY35_009504_159_0_RGB
MY36_014975_168_0_RGB
Tharsis Montes
Ceraunius Fossae
Amazonis Planitia
Terra Sirenum
Utopia Planitia
Elysium Planitia
Terra Cimmeria
bright > dark
bright > darkbright >> darkbright >> dark
Bright streaksBright streaks
800 m
~23 %~23 %
~8 %~8 %
~68 %~68 %
RSLRSL
RSLRSL
RSLRSL
dark > brightdark > brightdark >> brightdark >> bright
5 01
Dark streaksDark streaks
RatioRatio
~79 %~79 %
A D
E
F
B
C
1600 m
1000 m
1500 m
50 m
25 m
-5115
-5115 12561256MOLA topographic elevation (m)MOLA topographic elevation (m) 15831583 22362236 38703870-1848-1848
Bright streaksBright streaks
Dark streaksDark streaks
Fig. 1 | Global distribution of slope streaks and RSL on Mars. AthroughC: CaSSIS
(Colour and Stereo Surface Imaging System
89
) and HiRISE color-infrared images of
bright and dark slope streaks, as well as RSL; downslope direction is always pointing
down- or side-wards; image IDs MY36_014887_165_0, MY36_014975_168_0,
MY35_009504_159_0, MY35_010269_009_0, ESP_022689_1380, ESP_034830_1670.
Raw image credits ESA/TGO/CaSSIS CC-BY-SA 3.0 IGO and NASA/MRO/UoA.D)
throughF): Maps of slope streak (bright, white; dark, black) and RSL (red)
distribution (e.g. refs.8,51) as well as bright-to-dark slope streak ratio per 10° by 10°
quadrangle, overlain on a Viking merged color mosaic and MOLA topography
81
.
Percentages of total population per streak hotspot indicated on the map. Our
results agree well with earlier maps of streak-bearing regions
3,5
, while identifying a
series of previously unknown streak hotspots, e.g. in Xanthe Terra and Lunae Pla-
num (Supplementary Fig. S2).
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 2

3. streaks
28
). Applying the power law relation between the number of
resolved streaks and image resolution as established by Bergonio
et al.
29
, we estimate there might be ~14 times more streaks resolvable at
the HiRISE-scale (High Resolution Imaging Science Experiment,
~0.3 m
30
), i.e., ~2,000,000 bright and ~28,000,000 dark streaks.
We identify streaks between ~40°N and ~20°S, with a mean lati-
tude of 12.7°N and 16.7°N for bright and dark streaks, respectively,
indicating that dark streaks are systematically shifted northwards
relative to bright streaks. Notably, ~89% and ~90% of all bright and dark
streaks are located north of the equator, and there are a factor ~37
more dark than bright slope streaks. Bright streak hotspots (>100
features per 5° x 5° quadrangle) are located in Arabia Terra - Terra
Sabaea, E Kasei Valles, Ius Chasma, and Noctis Labyrinthus, where the
Arabia Terra–Terra Sabaea hotspot is the most pronounced. Dark
streak hotspots (>5000 features per 5° × 5° quadrangle) are located in
Olympus Rupes - Lycus Sulci, Ceraunius Fossae, Arabia Terra–Terra
Sabaea, Tartarus Montes, and NW Medusae Fossae–Marte Vallis,
where the Olympus Rupes–Lycus Sulci hotspot is the most pro-
nounced (holding ~68% of the global dark streak population). Only 15%
of all 10 × 10° quadrangles across the surface host more bright than
dark streaks (Fig.1).Thoseregionsmighthaveexperiencedreduced
streak formation rates in recent years; we note a particularly sub-
stantial percentage of bright streaks in Kasei Valles and Nili Fossae. The
vast majority (~85%) of all bright streaks are located on Noachian ter-
ranes, only ~4% are located on Amazonian terranes. In turn, about ~50%
of all dark streaks are located on Amazonian terranes, and only 31% on
Noachian terranes (Fig.2A).
Generally, bright and dark streaks are located at a) below-average
topographic elevations, with average values of−126 m (bright) and
−1126 m (dark); b) above-average slope angles with average values of 7°
(bright) and 8.6° (dark); c) in above-average TES (Thermal Emission
Spectrometer
31
) albedo terrain with average values of 0.27 (bright) and
0.28 (dark); and d) in above-average OMEGA (Observatoire pour la
Minéralogie, l’Eau,lesGlacesetl’Activité
32
) nanophase ferric oxide
-
-4500 -3500-4000 -3000 -2500 Ma-2000 -1500 -1000 -500 0
NoachianNoachian HesperianHesperian AmazonianAmazonian
11 %11 %
85 %85 %
Dark streaksDark streaks RSLRSLBright streaksBright streaks
4 %4 %
50 %50 %
31 %31 %
9 %9 %
42 %42 %
49 %49 %
19 %19 %
S
NorthNorth
SouthSouth
89 %
90 %
89 %
N
S
N
S
N
RSL
RSL
-1.1-1.1
Bright streaksBright streaks
-3.9-3.9
Dark streaksDark streaks
-3.8-3.8
randomrandom
-8000 -6000 -4000 -2000 2000 4000 6000 60000051015253035 20
1x103
1x104
1x105
1x106
1x102
1x101
1x100
0 50 100 150 250 300 350 200
MOLA surface elevation (m) MOLA slope angle (°) MOLA slope aspect (°)
N-facingN-facing N-facing N-facingS-facing S-facingE-facing E-facing W-facing W-facing
TES surface albedo OMEGA nanophase ferric oxide Estimated feature area (m2)
N
0.26
1.9°
-119 m
0.21
0.99
8.6°
8.6°
24.8 %24.8 % 26.5 %26.5 % 26.3 %26.3 %22.4 %22.4 %22.4 %22.4 %
-1126 m-1126 m
0.280.28
1.021.02
7.1°7.1°
1.021.02
0.270.27
-126 m-126 m
9.2°9.2°
0.970.97
0.150.15
-8 m-8 m
1x103 1x104 1x105 1x106 1x1071x1021x101
A
B
C
Fig. 2 | Global streak and RSL geostatistics. APie charts of terrane age distribu-
tion of bright streaks, dark streaks, and RSL; co-located bar chart indicates their
distribution over the northern (black) and southern (gray) hemispheres
82
.
BHistograms for slope streak (bright, tan; dark, solid black), RSL (red), and back-
ground (black line) distributions over a range of thermophysical properties
31,81,83
).
Note the blue line at TES albedo 0.26, as spatially visualized in Fig.3. Average values
indicated. Note that the relatively low spatial resolution of most of the global
thermophysical datasets might blur certain relations, most prominently for the
MOLA slope angle dataset, which largely underestimates the actual slope angle at
the streak-scale.CAFDs of estimated feature area;
p
2 binning is used, power law
fits indicated by lines, power law exponents indicated by numbers. Note that the
lack of small slope streaks is caused by the spatial resolution of the used CTX
images (6 m); RSL are only resolvable at the HiRISE scale (0.3 m).
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 3

4. terrain with average values of 1.02 (bright and dark) (Figs.2B,3A).
Globally, we observe a weak power law relation between the density of
slope streaks and the mean slope angle (Supplementary Fig. S4). Dark
slope streaks do not appear to feature a preferred slope orientation,
but equator-facing slopes host slightly more dark streaks (global dif-
ference to pole-facing slopes is 4.1%, i.e., ~20,000 streaks, Fig.2B);
bright streaks feature a distinct west-facing orientation. We do not
observe any substantial differences when considering northern and
southern hemisphere populations separately.
Bright and dark streaks feature a strong correlation between
estimated size, MOLA elevation (Mars Orbiter Laser Altimeter
33
), MCD
diurnal surface temperature amplitude (Mars Climate Database
34
),
albedo, and nanophase ferric iron: larger streaks appear to feature
lower MOLA elevation and higher diurnal temperature amplitude,
albedo, and nanophase ferric iron values (Supplementary Fig. S5). On
average, bright streaks are larger than dark streaks, with an average
estimated size of ~775 vs ~609 meters. We do not observe any general,
longitudinal or latitudinal variations in streak size. Streak length and
width are weakly linearly related; assuming a rectangular streak shape,
we estimate a mean surface area for bright and dark streaks of ~0.047
and 0.034 km
2
, respectively. The area-frequency-distribution (AFD) of
estimated streak area follows a power law with a negative slope of 3.9
and 3.8, respectively. Interestingly, an AFD derived for RSL presents a
negative exponent of 1.1 (Fig.2B). There is no correlation between
streak size and the age of the host terrane (Supplementary Fig. S6).
Using the mean number of streaks per detection (2.33), we esti-
mate that one bright or dark slope streak detection modifies a surface
area of ~0.1 km
2
, on average. This indicates that the detected slope
streaks modify an area of 49,705–103,500 km
2
globally, i.e.,
0.03–0.07% of the martian surface (for p
Dand p
C, respectively).
< 0.2< 0.2 0.2 - 0.260.2 - 0.26
TES surface albedoTES surface albedo
70°N70°N
150°E150°E100°E100°E50°E50°E0°E0°E50°W50°W100°W100°W150°W150°W
50°N50°N
30°N30°N
10°N10°N
10°S10°S
30°S30°S
50°S50°S
70°S70°S
0.26 - 0.320.26 - 0.32
12 PM12 PM
Bright streaksBright streaks
Dark streaksDark streaks
randomrandom
L
s
0° 0°
L
s
90° 90°
L
s
180° 180°
L
s
270° 270°
MCD daily mean dust deposition rate on horizontal surface (kg mMCD daily mean dust deposition rate on horizontal surface (kg m
-2-2
s s
-1-1
)
1x101x10
-8-8
RSLRSL
0.08
0.06
0.01
0.04
0.090.09
0.050.05
0.010.01
0.030.03
0.040.04
0.010.01
0.060.06
0.090.09
0.030.03
0.010.01
0.070.07
0.070.07
MCD sensible surface heat ux (W/mMCD sensible surface heat ux (W/m2)
L
s
90° 90°
12 AM12 AM
3 AM3 AM
6 AM6 AM
9 AM9 AM
0.7 W/m2
0.8 W/m2
0.4 W/m2
0.2 W/m2
1.2 W/m1.2 W/m2
1.4 W/m1.4 W/m2
0.3 W/m0.3 W/m2
0.8 W/m0.8 W/m2
1.1 W/m1.1 W/m2
0.2 W/m0.2 W/m2
0.4 W/m0.4 W/m2
0.4 W/m0.4 W/m2
0.4 W/m0.4 W/m2
0.1 W/m0.1 W/m2
0.150.10.050.00
A
B
Fig. 3 | Global relation between streaks, RSL, and dust. AMap of slope streak
(bright, white; dark, black) and RSL (red) distribution, overlain on TES albedo
(<0.2 transparent, 0.2–0.26 gray, 0.26–0.32 blue
31
), highlighting several spatially
coinciding, distinct cut-offs of slope streak and high albedo terrain (>0.26), such as
in southern Arabia Terra, north-west Elysium Planitia, southern Amazonis, and
north-western Tharsis. This implies that streaks can only form in the highest albedo
(smallest particle size) terrains. Viking merged color mosaic in the background.
BJoyplots (showing kernel density estimates) of slope streak (bright, tan; dark,
solid black), RSL (red), and background (black line) MCD daily mean dust deposi-
tion rate on a horizontal surface for L
s0°, 90°, 180°, and 270° (left
34
) and sensible
surface heatflux for L
s90° (northern summer) at 12, 3, 6, and 9 AM (right
34
).
Average values indicated.
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 4

5. Considering a streak survival time of 3–20 MY (where streaks are
observable
3,4
), the global surface modification rate would be 15,568 to
2485 km
2
/MY and 34,500 to 5175 km
2
/MY, i.e., 0.01–0.002% and
0.02–0.004% of the surface per MY (for p
Dand p
C, respectively).
Considering a streak thickness of ~1 m (i.e., the maximum value derived
by
14,15
, an average streak detection would modify a volume of
~100,000 m
3
or ~0.0001 km
3
, which would result in a total modified
volume of ~50 to ~100 km
3
(for p
Dand p
C, respectively). Considering a
dust bulk density of ~2500 kg/m
3
or ~2.5 × 10
9
t/km
335
, the total mod-
ified mass would account for ~1.3 × 10
11
to ~2.5 × 10
11
t, i.e., ~4.2 × 10
10
to
~6.3 × 10
9
t/MY and ~8.3 × 10
10
to ~1.3 × 10
10
t/MY (for p
Dand p
C,
respectively). References36,37report that around 4.0 × 10
8
to 2.9 × 10
9
t/MY of dust is exchanged between the martian surface and atmo-
sphere, i.e., between ~1 to ~46% and ~0.5 to ~22% of the dust mass
modified by streaks (for p
Dand p
C, respectively). Reference38reports
a total dust mass of ~4.3 × 10
8
t lifted during a global dust storm
observed in 1977, which is ~1 to ~7% and ~0.5 to ~3% of the estimated
mass modified by slope streaks per year (for p
Dand p
C, respectively).
These estimates indicate that slope streaks move several dust-
equivalents of a global dust storm every MY. The injection of dust
into the atmosphere by slope streaks has not been observed yet, but
would make slope streaks an important contributor to the martian
dust cycle and climate, potentially accounting for a substantial fraction
of the surface-atmosphere dust exchange budget.
Endo- and exogenic drivers of streak formation
We observe that dark streaks and date-constrained impact events
39
are
partially co-located in dusty regions (Fig.4A). To suppress the obser-
vational bias related to the detection of new impacts in dusty regions
(e.g. refs.39,40) we perform a statistical analysis of streak and new
impact locations in the dust-rich Arabia Terra, Elysium Planitia, and
Tharsis regions, which host both streaks and new impacts. Wefind that
70°N70°N
150°E150°E100°E100°E50°E50°E0°E0°E50°W50°W100°W100°W150°W150°W
50°N50°N
30°N30°N
10°N10°N
10°S10°S
30°S30°S
50°S50°S
70°S70°S
Viking Merged ColorViking Merged Color
Minimum distance between streaks and events / random points (°)
ArabiaArabia
ElysiumElysium
ElysiumElysium
ElysiumElysium
ArabiaArabia
TharsisTharsis
TharsisTharsis
MarsquakesMarsquakes
MarsquakesMarsquakes
New impactsNew impacts
New impactsNew impacts
Bright streaksBright streaks
Bright streaks - random
Dark streaksDark streaks
Dark streaks - random
172 impacts
0.2°0.2° 0.8°0.8°
0.3°
0.5°
0.5°
0.6°
0.4°0.4°
0.6°0.6°
0.6°0.6°
0.6°
2.8°
2.5°2.5°
3.5°3.5°
2.6°
2.6°
2.3°2.3°
30 impacts
615 impacts
17 quakes
1 quake
1023456
0.5° 2° 5°
0.5° 2° 5°
bd n/a
A
B
C
Fig. 4 | Global relation between streaks, RSL, tectonic activity, and meteoritic
impactflux. AMap of slope streak (bright, white; dark, black) and RSL (red) dis-
tribution as well as new impacts (blue stars
39
). InSight-recorded marsquakes (LF and
BB events with BAZ, yellow diamonds
42
), and tectonic features (white lines
82
); TES
albedo and Viking merged color mosaic in the background
31
. Red dashed lines
outline the three areas used for the statistical analysis of streak/RSL and impact/
quake proximity (Arabia Terra, Tharsis, and Elysium Planitia).BJoyplots (showing
kernel density estimates) of minimum distance between streaks (bright, tan; dark,
solid black) and real/synthetic impact/seismic events for Arabia Terra, Elysium
Planitia, and Tharsis; differences between distance dark streaks–real impacts and
dark streaks–synthetic points are statistically significant for Arabia Terra, but not
for Elysium and Tharsis. Note that there are no InSight-recorded marsquakes
located in Arabia Terra and only one in Tharsis (per
41
). We note that we do not
consider the relation in Tharsis representative due to the availability of just one
single data point (InSight event S0183a). Average values indicated.CPie charts
visualize the ratio between the number of streaks (b = bright, d = dark) within a 0.5°,
2°, or 5° area around real (tan/black) vs synthetic (brown) impact events/epicenters.
RSL statistics not shown as they are not deemed statistically sound (see Supple-
mentary Table S1).
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 5

6. dark streaks tend to be located slightly closer to real impact sites than
to randomly placed (synthetic) impact sites in Arabia Terra (statisti-
cally significant,P= 0.038) and Elysium Planitia (not significant,
P= 0.231), on average, but not in Tharsis. In Arabia and Elysium, in
circles with radii of 0.5, 2, and 5°, there are a factor ~1.5, ~1.3, ~1.1, as well
as ~1.9, ~1.6, and ~1.3 more streaks around new impacts vs synthetic
impacts, on average (Fig.4B, see Supplementary Table S1), decreasing
with increasing distance from a given impact. Beyond Arabia Terra and
Elysium, we do not see a global relation between the spatial density of
streaks and new impacts (Supplementary Fig. S7).
Some of the global streak hotspots are spatially co-located with
InSight-recorded, tecto-genetic broadband (BB) and low-frequency
(LF) events with back azimuth (BAZ)
41–44
, but a) not all streak hotspots
feature a co-located epicenter and b) not each epicenter features a
streak hotspot (Fig.4A). We perform a statistical analysis of the two
global streak hotspots with associated seismic BB or LF event(s) (Ely-
sium Planitia and Tharsis). Dark streaks do not tend to be located
closer to real quake epicenters than to randomly placed (synthetic)
epicenters (no statistically significant difference,P= 0.167). In Elysium,
in circles with radii of 0.5, 2, and 5°, there are a factor ~1.4, ~1,4, and ~1.0
less streaks around real epicenters vs synthetic epicenters, on average
(Fig.4B, see Supplementary Table S1). We note that a substantial
fraction of the CTX images used by the global mosaic were acquired
prior to the InSight mission, rendering our analysis not ultimately
conclusive. We do not observe a general, global relation between the
distribution of streaks and the geomorphic expressions of past tec-
tonic activity, such as graben and fossae, despite those features’steep
slopes, but identify noteworthy exceptions such as Ceraunius, Tractus,
Noctis, Ulysses, Elysium, and Labeatis Fossae (Fig.4A). Similarly, we do
not observe a general, global relation between rockfall locations
45
and
slope streak distribution other than a small number of co-located
features in the Cerberus Fossae and Ius Chasma areas (Fig.5).
Slope streaks tend to be located in regions with above-average
MCD wind speed, particularly in the morning (9 AM, statistically sig-
nificant,P<2.2×10
−233
,Fig.6). We observe that slope streaks and active
dust devils
46
are strongly inversely correlated, i.e., regions with dust
devil activity are nearly entirely devoid of slope streaks. Notably,
regions of active dust devil occurrence appear to coincide with several
of the globally distinct boundaries of slope streak distribution, such as
located in NW Lycus Sulci, Syria Planum, N Phlegra Dorsa, E Terra
Sabaea, and S Arabia Terra (Fig.6A). Slope streaks are strongly asso-
ciated with above-average MCD diurnal temperature amplitude regions,
particularly pronounced in northern spring, summer, and autumn
(Fig.7). Slope streaks tend to be located in regions with above-average
MCD atmosphere-to-surface heatflux during the night (12 and 3 AM),
wherethecorrelationdisappearsinthemorning(6AM)(Fig.3B).
We observe that slope streaks appear to be located in regions with
slightly above-average WEH (water equivalent hydrogen, statistically
significant,P=~0), H
2O, and H, particularly bright slope streaks.
Notably, not all high-H
2O/H/WEH regions feature slope streaks and
many streak hotspots are located in low H
2O/H/WEH regions (Fig.8;
e.g., in Elysium, 10°N 170°E, and E Tharsis, 20°N 110°W). Slope streak
locations feature slightly enhanced water vapor column values in
northern summer and autumn (Fig.8B). The vast majority/parts of the
slope streak population pass through the ~273 to ~283 K temperature
band that theoretically allows for transient liquid surface water, in
northern summer and southern summer, respectively (northern
summer is particularly pronounced, Fig.7B). Streak locations do not
show any substantial overlap with the valley networks mapped by
Hynek et al.
47
, expect for a few locations in Naktong, Deva, Minio,
Padus, and Sabis Vallis (Fig.5A). We note that dark streaks are located
in regions that see above-average dust deposition rates in southern
summer (statistically significant,P= ~0), but below-average rates for
the rest of the year (Fig.3B).
Discussion
Broader implications for streak formation and Mars’present-
day habitability
Our observations discard three previously proposed dry streak for-
mation mechanisms: dust devils (e.g. refs.11,48), rockfalls (e.g. ref.15),
and thermal cycling (e.g. ref.25) do not appear to play a globally
important role in triggering slope streaks (Figs.5–7). Dust devil,
rockfall, and slope streak occurrence are predominantly inversely
correlated, and slope streak populations experience higher-than-
average diurnal temperature amplitudes around the year, not just
during streak peak formation season in southern summer (L
s~270°); in
fact, the southern summer peak formation season experiences the
least pronounced temperature amplitude difference (Fig.7B).
Viking Merged ColorViking Merged Color
70°N70°N
150°E150°E100°E100°E50°E50°E0°E0°E50°W50°W100°W100°W150°W150°W
50°N50°N
30°N30°N
10°N10°N
10°S10°S
30°S30°S
50°S50°S
70°S70°S
Fig. 5 | Global map of slope streak, RSL, rockfall, and valley network dis-
tribution.Map of slope streak (bright, white; dark, black) and RSL (red) distribu-
tion, overlain on a Viking merged color mosaic. Blue lines indicate valley networks
as mapped by Hynek et al.
47
; yellow shapes represent confirmed rockfall locations
mapped by Bickel et al.
45
.
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 6

7. Our observations question many of the hypothesized wet for-
mation mechanisms: streak populations a) do not express any pre-
ferred slope orientation (Fig.2B), challenging a potential CO
2frost
related trigger
21
; b) only experience conditions suitable for liquid water
in northern summer (L
s~90°)–despite being located in regions that
experience above-average diurnal thermal amplitudes and heatflux–,
which conflicts with observations of peak streak formation occurring
in northern autumn and winter (L
s~140–270°
4,24
); in fact, streak
populations experience unfavorable conditions at L
s~ 270° (statisti-
cally significant,P=5.8×10
−158
)(Figs.3Band7B); and c) streak popu-
lations experience peak water vapor column values in northern
summer and autumn (L
s~90–180°), which also conflicts with obser-
vations of enhanced streak formation in northern autumn and winter
(L
s~140–270°
4,24
)(Fig.8B). We observe that streak populations feature
slightly enhanced WEH, H
2O, and H values at the global-scale (Fig.8B),
but note that recent, local-scale investigations using CRISM (Compact
Reconnaissance Imaging Spectrometer for Mars) have not found any
signs of hydration at slope streak locations
49
.
We identify three global-scale, statistically significant relations
that support dry formation hypotheses for slope streaks: streak
populations a) tend to be slightly closer located to new impacts than
to the same number of randomly placed locations in some regions
(Fig.4B); b) consistently experience above-average (near-surface)
wind velocities (Fig.6B); and c) experience above-average dust
deposition rates in northern winter (L
s~270°) (Fig.3B), coinciding
with the seasonal formation peak
4,24
). Notably, we identify a distinct
00 10.410.4
MCD surface wind speed Ls 270° 12 PM (m/s)MCD surface wind speed Ls 270° 12 PM (m/s)
MCD surface wind speed (m/s)MCD surface wind speed (m/s)
70°N70°N
150°E150°E100°E100°E50°E50°E0°E0°E50°W50°W100°W100°W150°W150°W
50°N50°N
30°N30°N
10°N10°N
10°S10°S
30°S30°S
50°S50°S
70°S70°S
15.615.6 20202.62.6 5.25.2 7.87.8 1313 1818
A
9 AM9 AM
L
s
90° 90°
12 PM12 PM
3 PM3 PM
RSLRSL Bright streaksBright streaks
Dark streaksDark streaks
3.8 m/s
3.7 m/s
3.4 m/s 4.4 m/s4.4 m/s
4.9 m/s4.9 m/s
4.8 m/s4.8 m/s
5 m/s5 m/s
4.9 m/s4.9 m/s
3.7 m/s3.7 m/s
1.9 m/s1.9 m/s
2.4 m/s2.4 m/s
2.6 m/s2.6 m/s
B
L
s
270° 270°
randomrandom
4.4 m/s
4.5 m/s
5.1 m/s
5.3 m/s5.3 m/s
3.6 m/s3.6 m/s
4.9 m/s4.9 m/s
5.5 m/s5.5 m/s
3.4 m/s3.4 m/s
4.6 m/s4.6 m/s3 m/s3 m/s
2.9 m/s2.9 m/s
2.5 m/s2.5 m/s
L
s
180° 180°
3.6 m/s
3.7 m/s
4.2 m/s
5.8 m/s5.8 m/s
4.9 m/s4.9 m/s
5.2 m/s5.2 m/s
5.8 m/s5.8 m/s
3.9 m/s3.9 m/s
3.8 m/s3.8 m/s
1.9 m/s1.9 m/s
3.0 m/s3.0 m/s
3.3 m/s3.3 m/s
Fig. 6 | Global relation between streaks, RSL, and atmospheric processes.
AMap of slope streak (bright, white; dark, black) and RSL (red) distribution as well
as CTX-observed dust devils (brown triangles
46
, covering MY28 to MY36); MCD
surface wind speed in the background
34
.BJoyplot (showing kernel density
estimates) of slope streak (bright, tan; dark, solid black), RSL (red), and background
(black line) MCD surface wind speed for 9 AM, 12 PM, and 3 PM LST for L
s90° (left),
180° (center), and 270° (right). Differences between dark streaks and background
are statistically significant for all L
sand LST. Average values indicated.
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 7

8. cascade of slope streaks on two specific slopes in the direct vicinity
of a 140-m diameter crater that formed on September 18
th
2022
(L
s305°, during peak dust deposition, Supplementary Fig. S8,
refs.90,91) as well as new slope streaks close to other new impact
events (ref.92, Supplementary Figs. S9–S12), but alsofind slopes
that do not host new streaks following close-by impact events
(Supplementary Figs. S12–S14), which illustrates how impacts can
trigger streaks, but only in sloped, dust-bearing terrain. We do not
observe any relation between streak occurrence and tectonic
marsquakes, noting that this particular comparison suffers from a
substantial observational (spatiotemporal) bias due to the limited
seismic data available (~2 MY). However, we identify a number of
Fossae with an increased number of slope streaks (Fig.4A), which
might point to recent tectonic activity in those regions
12
. The lack of
streaks in the majority of Fossae might also be rooted in a general
lack of (fine) dust in those regions.
In contrast, RSL populations are preferably located on equator-
facing slopes (Fig.2B
8,50
) and experience conditions suitable for liquid
water in southern summer (L
s~270°), i.e., during their peak occurrence
(Fig.7B); however, RSL populations simultaneously feature below-
average diurnal thermal amplitude, heatflux, and WEH, H
2O, and H
values, as well as average water vapor column values, rendering a wet
formation mechanism less likely (Figs.3B,7Band8B). We note that RSL
locations experience peak dust deposition rates in northern autumn
(L
s~180°), i.e., at the onset of RSL peak formation, and dust storms
have been shown to substantially enhance RSL formation rates
51
.We
do not observe any global-scale relation between RSL locations, new
impacts, and wind velocity (Figs.4and6B), but identify an indistinct,
00
MCD thermal amplitude Ls 90° (K)MCD thermal amplitude Ls 90° (K)
L
s
0° 0°
L
s
90° 90°
L
s
180° 180°
L
s
270° 270°
70°N70°N
150°E150°E100°E100°E50°E50°E0°E0°E50°W50°W100°W100°W150°W150°W
50°N50°N
30°N30°N
10°N10°N
10°S10°S
30°S30°S
50°S50°S
70°S70°S
6565 1301303333 9898
A
B
MCD diurnal temperature amplitude (K)MCD diurnal temperature amplitude (K)
12 AM - 12 PM12 AM - 12 PM
RSLRSL
Bright streaksBright streaks
59 K
80.6 K
63.6 K
90 K
77.5 K77.5 K
108.6 K108.6 K
82.4 K82.4 K
101.5 K101.5 K
79.2 K79.2 K
106.7 K106.7 K
83.5 K83.5 K
107.2 K107.2 K
47.5 K47.5 K
48 K48 K
50.1 K50.1 K
88.9 K88.9 K
273 K
283 K
liquid water
Dark streaks
Dark streaks
MCD diurnal noon temperature (K)MCD diurnal noon temperature (K)
243 K
253 K
250 K
276 K
241 K241 K
268 K268 K
249 K249 K
264 K264 K
252 K252 K
275 K275 K
260 K260 K
279 K279 K
234 K234 K
209 K209 K
237 K237 K
290 K290 K
12 PM12 PM
300250200150
random random
Fig. 7 | Global relation between streaks, RSL, and the thermal environment.
AMap of slope streak (bright, white; dark, black) and RSL (red) distribution; MCD
thermal amplitude (noon-midnight) at L
s90° in the background
34
.BJoyplot
(showing kernel density estimates) of slope streak (bright, tan; dark, solid black),
RSL (red), and background (black line) MCD diurnal temperature amplitude (left)
and diurnal noon temperature (right) for L
s0°, 90°, 180°, and 270°. The tem-
perature range between 273 and 283 K that (theoretically) provides conditions for
liquid water on the surface of Mars (in some regions) is indicated in blue. Average
values indicated.
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 8

9. spatial relation with dust devil occurrence, and–potentially–rockfall
occurrence (Figs.5and6A), as previously locally observed by, e.g.
refs.51,52. Here, dust devils and rockfalls might (jointly) trigger the
displacement of material through turbulence and the transfer of
kinetic energy (e.g. refs.11,53). Recently, seasonal peak dust devil
activity (number and size) has been observed in the southern high-
lands in southern summer (Fig.6A
46
), coinciding with peak RSL for-
mation. We note that one of the main limitations of this study is the
lack of information on the timing of global-scale streak and RSL for-
mation. Globally-distributed information about slope streak and RSL
formation would allow for a more robust statistical comparison of
streak and RSL occurrence and potential drivers.
Overall, our observations suggest that slope streak and RSL for-
mation may be predominantly controlled by two independent, dry
drivers, 1) the seasonal delivery of dust onto topographic inclines,
and 2) the spontaneous activation of accumulated dust by energetic
triggers–wind and impacts for slope streaks, as well as dust devils and
rockfalls for RSL. These observations showcase how the seasonal dust
delivery and the seasonal formation of streaks and RSL are inherently
linked, while alleviating the likelihood of wet formation as proposed by
preceding studies (e.g. refs.16,17,50–52). The important role offine
dust in streak formation is further highlighted by the abundance of
streaks in the highest-TES albedo terrain (Fig.3A), the positive corre-
lation between slope streak size and dust abundance (Figs.2B,3,
Supplementary Fig. S5), and the lack of a relation between streak size
and host terrane geologic age (Supplementary Fig. S6). Notably, ref.54
report how variations in the particle size could explain albedo differ-
ences between high-albedo (e.g., Arabia) and intermediate-albedo
terrain (e.g., Utopia). Per
55
, the dust in high-albedo terrain features a
particle size less than 40µm; recent work by Valantinas et al.
56
suggest
even smaller particle sizes, i.e., as small as 1µm, which agrees with
earlier observations of atmospheric dust
57
. Such small-sized particles
could facilitate the formation of some of the observed complex,flow-
like features of slope streaks, such asfingering and anastomosis (e.g.
refs.25,58–60). In addition, slope streak formation may be influenced
by slow moving dense granularflows that are supported by dispersive
22 3
FREND water equivalent hydrogen (wt%)FREND water equivalent hydrogen (wt%)
150°E150°E100°E100°E50°E50°E0°E0°E50°W50°W100°W100°W150°W150°W
50°N50°N
70°N70°N
30°N30°N
10°N10°N
10°S10°S
30°S30°S
50°S50°S
70°S70°S
7654 >10>1098
Equivalent (wt%)Equivalent (wt%) MCD water vapor column (kg/mMCD water vapor column (kg/m2)
RSLRSL
Bright streaksBright streaks
Dark streaksDark streaks
randomrandom
L
s
0° 0°
L
s
90° 90°
GRS HGRS H
2
O
GRS HGRS H
FREND WEHFREND WEH
L
s
180° 180°
L
s
270° 270°
12 PM12 PM
0.005 kg/m2
5.1 wt%
3.9 wt%
3.9 wt%
0.009 kg/m2
0.014 kg/m2
0.008 kg/m2
0.006 kg/m0.006 kg/m2
4.7 wt%4.7 wt%
6.4 wt%6.4 wt%
4.7 wt%4.7 wt%
0.014 kg/m0.014 kg/m2
0.016 kg/m0.016 kg/m2
0.008 kg/m0.008 kg/m2
0.005 kg/m0.005 kg/m2
5.5 wt%5.5 wt%
5.5 wt%5.5 wt%
7.8 wt%7.8 wt%
0.012 kg/m0.012 kg/m2
0.016 kg/m0.016 kg/m2
0.008 kg/m0.008 kg/m2
0.005 kg/m0.005 kg/m2
3.3 wt%3.3 wt%
3.3 wt%3.3 wt%
3.8 wt%3.8 wt%
0.003 kg/m0.003 kg/m2
0.008 kg/m0.008 kg/m2
0.01 kg/m0.01 kg/m2
A
B
Fig. 8 | Global relation between streaks, RSL, and volatiles. AMap of slope streak
(bright, white; dark, black) and RSL (red) distribution; FREND (Fine-Resolution
Epithermal Neutron Detector
84
) water equivalent hydrogen (WEH) map and Viking
merged color mosaic in the background.BJoyplots (showing kernel density esti-
mates) of slope streak (bright, tan; dark, solid black), RSL (red), and background
(black line) FREND WEH, GRS (Gamma Ray Spectrometer
85
)H
2O, and GRS H values
(left) and MCD water vapor column for L
s0°, 90°, 180°, and 270° at 12 PM LST
34
(right). Average values indicated. Note that the locations of bright and dark streaks
tend to feature slightly higher FREND WEH values than GRS H values; note that GRS
HandH
2O are nearly (but not entirely) identical.
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 9

10. pressure
60
, potentially explaining why slope streaks do not produce
observable dust clouds during their formation. It remains unclear what
amount of the dust displaced by slope streaks ultimately ends up being
injected into the atmosphere.
Our results underline the fundamental differences between slope
streaks and RSL, despite their visual resemblance. Streak and RSL
populations occur on opposite hemispheres (north vs south), at dif-
ferent topographic elevations (mostly lowlands vs mostly highlands),
in opposite thermal inertia terrain (low vs high), in different wind
speed regimes (above-average vs below-average), in dissimilar diurnal
thermal amplitude and heatflux terrain (above-average vs average), in
different WEH, H
2O, H, and water vapor column terrain (average vs
below-average), and in terrain that provides suitable (theoretical)
conditions for liquid water at different seasons (L
s~90° vs L
s~270°)
(Figs.1–8). Importantly, the exponents of the AFDs (streaks: ~3.8; RSL:
~1.1) bracket the exponents commonly experienced for landslides on
Earth (~1.5 to ~3
61–63
); we note that AFD exponents cannot be directly
used to infer triggers or causes of mass wasting events, but the
86,546 CTX images
CNN predictions
Dark streak
Bright streak
~30 km
~80 km
TPsTPs
TPsTPs
TPsTPs
TPsTPs
TPsTPs
TPsTPs
FNFN
FNFN
FNFN
FPFP
FPFP
FPFP
0.80.8
bright
dark
0.8
0.8
.4
.4
.6
.6
.8
.8
.2
.4 1.2
.4 1.2
.2
1
.4
.6
.8
.2
Recall Precision Precision
Condence
Condence
Recall
patch 1 patch 3
patch 5 patch 6
patch 1patch 4
patch 2patch 5
patch 3patch 6
B22_018396_1890_XI_09N320W
NASA MRO
NASA MRO
1000 pixel
patch 4
patch 2
.6
.8
.4
.6
.6
1
.2
1
1
AP
AP
0.980.98
AP AP
0.500.50
.8
.8
A
B
C
Fig. 9 | Visualization of the deep learning-driven mapping workflow. AMRO
CTX images are cropped into small patches and fed into the CNN, which (B)outputs
predictions of slope streak location, size, and confidence score (true positives, TP,
white boxes). The detector can miss slope streaks (false negatives, FN, yellow) or
can make incorrect predictions (false positives, FP, red); note how larger streaks
feature larger bounding boxes.CTestset performance of the detector is visualized
on the left, indicating recall, precision, and AP, as well as the score cut-off of 0.8
used for this study. Image ID B22_018396_1890_XI_09N320W. Raw image credits to
NASA/MSSS/ASU.
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 10

11. substantially different exponents add further weight to the conclusion
that streaks and RSL are fundamentally different features and might
point to different material properties (e.g., dusty, cohesion- vs sandy,
friction-controlled material, e.g. refs.18,63(Fig.2B).
Ourfindings suggest that martian slopes currently do not
experience seasonal, transientflows of liquid water or brines, under-
scoring the dry, desert-like nature of Mars. This implies that slope
streak and RSL locations are not likely to be habitable, alleviating strict
planetary protection measures for future landed missions to those
regions.
Methods
Few-shot learning-driven slope streak mapping
We select 40 globally-distributed MRO (Mars Reconnaissance Orbiter)
CTX (Context Imager) images with a pixel scale of ~6 m
64
and crop
them into ~1000 × 1000 pixel patches. With those patches we create a
training dataset by labeling a total of 1007 and 147 dark and bright
slope streaks, following previously established workflows
65–68
. Here, a
label represents a rectangular box drawn by a human expert around a
slope streak. We label an additional 35 dark and 6 bright slope streaks
for model testing and verification (~4% of the training set size; remains
unseen during training). During labeling, we carefully separate bright
and dark slope streaks following the definition provided in the main
text and showcased in Fig.1. We purposely minimize the size of the
testing dataset, to maximize the data available for model training.
Here, the usage of the term‘few-shot learning’accounts for the fact
that the number of slope streak labels available for training and testing
is lower than in traditional deep learning which commonly uses tens of
thousands of labels. Importantly, past studies have demonstrated that
a few-shot learning approach can successfully map geomorphic and
Map distributionMap distribution
Example - Arabia TerraExample - Arabia Terra
real impact
Criterion 1) Minimum distance
synthetic impact
synthetic event
n = 0
n = 5
real event
lat
lon
slope streak
20°E
10°N
30°E
Criterion 2) Number of features
A B C
D
Fig. 10 | Schematic of statistical analysis of streak and seismic/impact event co-
location. ASchematic map of streaks (black dots,n= 22), impacts (blue,n= 4), and
synthetic impacts (orange,n=4);BAssessment of minimum distance between
streaks and (real/synthetic) events;CAssessment of number of streaks in n° bins
around (real/synthetic) events. A relation between an impact/seismic event and
streaks would result in smaller minimum distances and increased numbers of
streaks closer to the events.DReal example from Arabia Terra, including bright
(white dots) and dark streak detections (black dots); diameter of circles does not
properly reflect the 0.5, 2, and 5° radii used throughout this study. Note how the
individual streak detections cluster along topographic inclines, mostly along crater
walls. Viking merged color mosaic in the background.
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 11

12. atmospheric features across various planetary bodies and
datasets
45,46,69–73
. A list of all used CTX images in shown in Supple-
mentary Table S2 and their illumination and imaging geometries are
listed in Supplementary Table S3.
We use the training dataset tofine-tune a COCO-pretrained
(‘Common Objects in COntext’), off-the-shelf PyTorch 1.7 imple-
mentation of the YOLOv5x convolutional neural network (CNN) pro-
vided by ultralytics via GitHub (https://github.com/ultralytics/yolov5).
YOLOv5 is a well-established, verified, cutting-edge object detection
framework that has been utilized for a wide range of scientific appli-
cations, such as food science, robotics, medical image processing, and
planetary science (e.g. refs.71,73–76). YOLOv5 represents the best
compromise between state-of-the-art performance and reliability, and
has been shown to substantially outperform (performance and speed)
earlier machine learning algorithms such as based on AdaBoost, His-
togram of Oriented Gradients, and Region-based CNNs (e.g.
refs.77–79) as were used by earlier studies that reported on the
automated detection of slope streaks
80
.
We apply substantial label augmentation during training to pre-
vent model overfitting and minimize any negative impact of the rela-
tively small number of labels, as is common practice in computer
vision and few shot learning, utilizing label rotation, up-downflipping,
shearing, up- and down-scaling, and brightness and contrast mod-
ifications. We do not apply left-rightflipping to preserve a proper
representation of the illumination conditions governed by MRO’ssun-
synchronous orbit in the training data. The training of the CNN
required ~15 min on one single NVIDIA RTX 3090 GPU (Graphical
Processing Unit).
We evaluate the performance of the trained detector using the
testing dataset and the three most common metrics used for object
detector evaluation: recall (i.e., the number of detected streaks, e.g.
8/10), precision (i.e., the number of correctly detected streaks, e.g. 9/
10), and average precision (AP, relation of recall and precision over a
wide range of model confidence scores, i.e., the area under the
precision-recall curve, which summarizes the performance of a
detector in one single number, Fig.9C). The CNN assigns a con-
fidence score to each detection (0, low to 1, high), representing its
confidence in the assigned detection and classification; higher con-
fidence score populations usually feature lower recall but higher
precision. At confidence 0.8, the trained slope streak detector
achieves full precision (100%), a recall of 20 % and 55 %, as well as
0.98 and 0.50 AP (0.74 mAP, mean average precision for all classes)
for bright and dark streaks in the testset, respectively (Fig.9C). This
indicates that the detector can be expected to make none to very few
mistakes, while being able to (consistently) identify about 20 and 55%
of all bright and dark slope streaks observed by CTX, at a confidence
score of 0.8 and higher.
At a score of 0.6 our detector outperforms (for dark streaks only,
recall ~85%) the performance as reported by earlier studies that used a
combination of feature selection and machine learning (binary) clas-
sification techniques and HiRISE images (ref.80,recall~79%),ata
continuously high level of precision (~94%). However, given the sub-
stantial number of expected slope streak candidates we deliberately
choose a confidence score of 0.8 for this study, to maximize the pre-
cision of the detector and minimize the amount of time required for
the subsequent review efforts, as described below.
We deploy the trained detector in a customized image download
and processing pipeline originally established by Bickel et al.
69
and
further refined by Bickel et al.
70,72
, and Conway et al.
46
. The pipeline
successively downloads all of the 86,546 CTX images contained in the
post-beta
27
, pre-v01 version of the global CTX mosaic (pers. comm. J.
Dickson), which was later officially published in ref.26,athttps://image.
mars.asu.edu/stream/, and tiles them into 1000 × 1000 pixel patches
(Fig.9A). At the time of data processing, thefinal version of the
26
mosaic was not yet available, which required us to process the
individual CTX images that made up the pre-v01 version of the mosaic
instead. All patches are scanned by the streak detector, the geographic
location of all detections are reconstructed using the image’s
metadata, and all detections as well as their associated metadata are
stored (including a size estimate, i.e., the diameter of the bounding
box), including so-called‘thumbnails’,i.e.,crop-outsofallCNN-
predicted bounding boxes that are used for the subsequent expert
review (see Fig.9B, Supplementary Fig. S1). We use YOLOv5’s default
Non-Maximum-Suppression (NMS) algorithm to remove duplicate
predictions in each patch, following computer vision best practice.
Overall, the processing of the entire dataset requires ~2 months of
processing time using one single NVIDIA RTX 3090 located in a regular
desktop computer.
Due to the potential for geo-location and co-registration inac-
curacies as well as partial overlap of the CTX images that make up the
pre-v01 version of the global CTX mosaic (pers. comm. J. Dickson)
there is a small chance that slope streaks located along the seams of
individual images are detected multiple times (‘duplicates’). However,
a case study in the Tharsis region suggests that the number of potential
duplicate detections in the overall dataset is expected to be small,
around 6 % of the total streak count (Supplementary Fig. S3). The slight
location inaccuracies paired with the large number of slope streak
detections impedes an automated and manual detection and removal
of duplicates, which is why they are not specifically identified and
removed. We note that duplicate detections are evenly distributed
across the mosaic and globe and, thus, do not systematically affect the
geostatistical science analysis.
After processing, the detector identified 661,144 slope streak can-
didates (40,855 bright, 620,289 dark) at a confidence score of 0.8 and
higher. At confidence 0.7, 0.6, and 0.5 the detector identified 1,494,416,
2,015,709, and 2,463,542 slope streak candidates, respectively.
Candidate review
We note a number of false detections among the bright and dark slope
streak detection candidates, which closely resemble the visual
appearance of streaks, such as wind streaks, linear shadowed topo-
graphy, dunes, and mountain ridges (Supplementary Fig. S15). Past
experience with object detectors being applied to very large and het-
erogeneous orbital image datasets has shown that the testset perfor-
mance is not always fully representative of the ‘real world’
performance of the detector in the full dataset, due to the inherently
limited representability of the strictly size-constrained training and
testing datasets as well as unknown/unexpected anomalies such as
related to imaging artefacts, weather-related phenomena (clouds,
etc.), and a-priori unknown look-alikes (see, e.g. refs.45,70–72). The
increased number of false detections as achieved by the bright streak
detector (Fig.9C) can likely be attributed to a) the smaller training
dataset, which in turn is caused by the relative scarcity of bright streaks
(as noted by other authors, such as, e.g. refs.3,5), and b) the more
subtle radiometric contrast between bright streaks and the sur-
rounding slope (compared to dark streaks). The presence of false
positives in the output of both detectors underlines the importance of
the manual review of all candidates and the removal of all false
positives.
In order to remove all obvious false detections and ensure the
integrity of the scientific analysis, we manually review all CNN-
identified candidates. The review consists of a human expert looking
at every single slope streak candidate thumbnail, one-by-one. Due to
the overwhelming amount of candidates at lower scores (e.g., >1 mil-
lion candidates are score 0.7) we decide to only review all candidates
with score 0.8 or higher, i.e., all very high confidence detections
(n= 661,144). Overall, the manual review of all candidates required
~two weeks, achieving a reviewing rate of ~183 detections/minute or ~3
detections/second, assuming an 8-hour workday and 5 workdays per
week.Weidentifyatotalof13,026brightand484,019darkslope
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 12

13. streaks, indicating a detector precision of ~32% (bright) and ~78%
(dark) (Fig.9C). The testset precision at score 0.8 is 100% for both
types, indicating there is a drop of performance by 68% for bright and
by 22 % for dark streaks, as was expected based on previous experi-
ence. After expert review, we conclude the validation of the dataset by
examining all remaining slope streak detections as plotted on the
26,27
global CTX mosaic, considering all detections in their geomorphic
context and removing any remaining outliers. All analyses and results
(as well as the associated dataset) utilize the fully reviewed streak
catalog.
Verification of the reviewed mapping results
The quality and reliability of the produced global slope streak dataset
rely on the a) combined quality of the used CTX images, b) perfor-
mance of the slope streak detector, and c) quality of the manual expert
review. We do notfind any direct or indirect evidence for any negative
effect of the scanned CTX images on the mapping outputs, such as
potentially caused by cloud cover, shadowing, haze, and data artifacts.
We point out that
26,27
carefully selected the images that make up the
global mosaic to minimize any such negative impacts. The detector
was carefully trained and its performance thoroughly tested and
assessed using state-of-the-art approaches and heterogeneous,
geographically-distributed training data. We do not expect any
observational bias for the dark slope streak class; the very wide range
of slope streak morphologies identified by the detector (Supplemen-
tary Fig. S1) provides evidence for a robust and capable detector.
Notably, the preferably more equatorward distribution and
W-facing orientation of bright slope streaks is likely to be attributed to
a bright slope streak detector that is more successful in retrieving
bright streaks on well-illuminated, Sun-facing slopes, given the very
faintappearanceofbrightslopestreaks(Fig.1A, C) and MRO’ssun-
synchronous orbit (imaging takes place in the early local afternoon).
Potentially, the higher mean length as reported for bright streaks could
be attributed to the lower performance of the bright streak detector
(compared to the performance of the dark streak detector) resulting in
systematically less precisely placed bounding boxes, leading to larger
size estimates. We point out that the evenly-distributed orientation of
dark streaks (Fig.2B) implies that there is no general, CTX imaging-
related bias regarding slope shadowing, as further verified by quali-
tative testing–and as expected based on the careful selection of
mosaic images performed by Dickson et al.
26,27
.
We expect the manual review to capture and remove the over-
whelming majority of all false detections contained in the original
candidate dataset. In confirmation of that expectation, our derived
streak distribution maps agree exceedingly well with previously pub-
lished maps of slope-bearing terrain (e.g., refs.3,5, Supplementary
Fig. S2), which further increases our confidence in the derived maps
and relations. Notably, we identify a number of previously unknown
streak hotspots, such as located in Xanthe Terra, Lunae Planum, E Kasei
Valles, and Syrtis Major (Fig.1, Supplementary Fig. S2). We note that
our mapping approach is relying on a stitched mosaic consisting of
CTX images taken several years apart (2006–2020
26,27
), and at different
times/seasons; as a result, our slope streak catalog does not represent
one snapshot in time, but a wide time-range. Much of the remaining
ambiguity could be addressed with globally-distributed information
about slope streak and RSL formation times.
Streak number and size estimate correction
The CNN reports a size estimate for every slope streak detection,
representing the diameter of the CNN-predicted bounding box, which
closely encompasses the streaks, i.e., represents a proxy for streak
length (see Supplementary Fig. S1). Using the spatial resolution of the
associated CTX image, we convert the diameter from pixel to meter.
Following a previously established methodology
65,66
and in order to
improve the accuracy of the relation between bounding box diameter
and actual streak length, we manually measure the actual length of a
randomly drawn subsample of 1000 slope streaks and relate it to the
respective bounding box diameter (linear relation), deriving the fol-
lowing relation:
length
actual=0:8774*length
CNN61:503 ð1Þ
We use this relation and correct all ~500,000 streak length esti-
mates; all scientific analyses use the corrected length estimate. As the
size of each bounding box is dominated by the largest contained
streak, this relation might not be entirely representative for bounding
boxes that include multiple streaks with distinct streak length differ-
ences. We further establish a linear relation between the length of
streaks and their width, measured at their morphological center, using
thesamerandomlydrawnsubsampleof1000slopestreaks:
width=0:0461*length
actual+19:038 ð2Þ
In addition, we use the subsample to quantify the average number
of individual slope streaks contained in one single detection (see
Supplementary Fig. S1), calculating an average of 2.33 streaks/detec-
tion. Multiple, spatially co-located slope streaks are not always easily
discernible, such as visualized by Fig.1A, B, Supplementary Fig. S8. We
note that the absolute counts of individual slope streaks as derived in
this work might under- or over-estimate the actual number of slope
streaks in a given region.
Area estimate
We use the corrected streak length and the relation between length
and width to produce estimates of streak area for the entire dataset,
assuming a simplified rectangular streak shape:
area=length
actual
*width ð3Þ
For RSL, we manually measure a subsample of 100 semi-randomly
selected, geographically-distributed features (taken from, e.g.
refs.8,51) and estimate their surface area using Eq. (3).
Geostatistical analysis & auxiliary dataset limitations
We compare our slope streak catalog to a large number of auxiliary
spatial raster and vector datasets using Quantum GIS (QGIS). Specifi-
cally, we use TES surface albedo
31
, MOLA topography
81
, RSL locations
mapped in MRO HiRISE (such as by McEwen et al.
8,51
), modeled envir-
onmental outputs produced by the Mars Climate Database (MCD
34
,
assuming a climatology average solar scenario), date-constrained
(new) impacts
39
, InSight tecto-genetic broadband (BB) and low-
frequency (LF) events with back azimuth (BAZ
41,42
), the most recent
version of the martian geologic map
82
, river valleys
47
, rockfall
locations
45
, OMEGA nanophase ferric iron
83
, TGO FREND water-
equivalent-hydrogen (WEH
84
), GRS H2OandH
85
,anddustdevilloca-
tions (ref.46, covering MY28 to MY36). We use the MOLA topography
data to create two derived topography products, slope angle and slope
aspect, using built-in QGIS algorithms. We quantitatively define the
significance of differences between different populations and datasets
using a two-sample T-test with a default P-criterion of 0.05.
Importantly, many of the used global auxiliary datasets feature
spatial resolutions several factors below the size of CTX-detected slope
streaks and/or depend on various assumptions and models, such as
the MCD-derived outputs. For example, the global, km-scale, MCD-
derived wind speed maps do not fully capture the complex influence of
small-scale topography on wind speed and direction. We underline the
importance of local, site-specific studies to verify the global-scale
observations and interpretations made by this work.
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 13

14. Seismic & impact event distance analysis
Using a methodology developed by Bickel et al.
70,86
, we quantify the
statistical relation between streak and date-constrained impact/mars-
quake epicenter locations (Fig.10). First (‘Criterion 1’), we calculate the
shortest distance between all streaks and a) all real impacts/quakes and
b) the same number of randomly placed (synthetic) points (in the same
area). Second (‘Criterion 2’), we count the number of streaks in 0.5, 2,
and 5° radii around a) all real impacts/quakes and b) the same number
of randomly placed (synthetic) points (in the same area). If there was a
physical relation between impacts/quakes and streak occurrence, the
average distance to streak locations should be smaller, and the number
of streak should increase in the proximity of real impacts/
quakes (Fig.10).
We note that this method is strongly affected by the quality and
spatiotemporal availability of the impact and marsquake data; ref.39
acknowledge that their impact catalog is likely incomplete, particularly
in the southern highlands; in addition, a CTX image used in the mosaic
might have been taken prior to an impact even at the same location.
However, there is a substantial number of date constrained impacts
available for a statistical analysis (n= 1203) which have been mapped
over the same time period as the images for the global CTX mosaic
have been acquired, indicating there is a robust statistical foundation
for our analyses.
In turn, a substantial fraction of the CTX images in the global
mosaic were acquired prior to the InSight mission (late 2018 to late
2022), making a meaningful comparison with the seismic data chal-
lenging. Tecto-genetic quakes are likely to periodically re-occur in the
same regions, but their periods are unknown and might exceed the
survival time of slope streaks. In addition, InSight’s records are occa-
sionally noisy, i.e., might not include all tectonic events in the mon-
itoring period, do not cover all of Mars’surface (core shadow), and
suffer from substantial localization uncertainties (e.g. ref.87). Overall,
the geostatistical comparison of streak and marsquake locations is not
expected to be entirely conclusive. In order to compensate the data
limitations, we also compare streak locations with the distribution of
permanent geomorphic expressions of geologically recent tectonic
activity, such as graben and fossae.
Data availability
This work is exclusively based on open data. The bright and dark slope
streak catalog generated in this study has been deposited in the Uni-
versity of Bern’s BORIS database under accession codehttps://doi.org/
10.48620/75966
88
. InSight seismic data is available here:https://www.
insight.ethz.ch/seismicity/catalog/v14. CaSSIS image data are openly
available here (use the search bar tofind specificimageIDsorzoom
and pan on the map):https://observations.cassis.unibe.ch/HiRISE data
areopenlyavailablehere:https://www.uahirise.org/. CTX image data
are openly available here:https://viewer.mars.asu.edu/viewer/ctx#T=0.
The used USGS map products (MOLA topography, TES albedo) are
openly available here, respectively: 1)https://astrogeology.usgs.gov/
search/details/Mars/GlobalSurveyor/MOLA/Mars_MGS_MOLA_DEM_
mosaic_global_463m/cub,2)https://astrogeology.usgs.gov/search/
map/Mars/GlobalSurveyor/TES/Mars_MGS_TES_Albedo_mosaic_global_
7410m. MCD data can be accessed here:https://www-mars.lmd.jussieu.
fr/mcd_python/.
Code availability
This work is exclusively based on open software. The used YOLOv5
architecture is openly available here:https://github.com/ultralytics/
yolov5.
References
1. Morris, E. Auerole deposits of the martian volcano Olympus
Mons.J. Geophys. Res. Planets87,https://doi.org/10.1029/
JB087iB02p01164(1982).
2. Ferguson, H. & Lucchitta, B. Dark streaks on talus slopes, Mars.
Planetary Geology and Geophysics Program Report, pp. 188–190,
https://ntrs.nasa.gov/citations/19840015431(1984).
3. Schorghofer, N., Aharonson, O., Gerstell, M. & Tatsumi, L. Three
decades of slope streak activity on Mars.Icarus191, 132–140
(2007).
4. Stillman,D.E.,Hoover,R.H.,Kaplan,H.H.,Michaels,T.I.&Fenton,
L. K. Comprehensive observations and geostatistics of slope streaks
within the Olympus Mons Aureole.Icarus415, 116061 (2024).
5. Valantinas, A. et al. CaSSIS colorand multi-angular observations of
Martian slope streaks.Planet. Space Sci.209, 105373 (2021).
6. Sullivan, R., Thomas, P., Veverka, J., Malin, M. & Edgett, K. Mass
movement slope streaks imaged by the Mars Orbiter Camera.J.
Geophys. Res. Planets106,https://doi.org/10.1029/
2000JE001296(2001).
7. Schorghofer, N., Aharonson, O. & Khatiwala, S., Slope streaks on
Mars: Correlations with surface properties and the potential role of
water.Geophys. Res. Lett.29,https://doi.org/10.1029/
2002GL015889(2002).
8. McEwen, A. et al. Seasonal Flows on Warm Martian Slopes.Science
333,740–743 (2011).
9. Bhardwaj, A., Sam, L., Martín-Torres, F. J., Zorzano, M.-P. & Fonseca,
R. M. Martian slope streaks as plausible indicators of transient water
activity.Nat. Sci. Rep.7,7074(2017).
10. Baratoux, D. et al. The role of the wind-transported dust in slope
streaks activity: Evidence from the HRSC data.Icarus183,
30–45 (2006).
11. Heyer, T., Raack, J., Hiesinger, H. & Jaumann, R. Dust devil triggering
of slope streaks on Mars.Icarus351, 113951 (2020).
12. Lucas, A. et al. Possibly seismically triggered avalanches after the
S1222a Marsquake and S1000a impact event.Icarus411,https://
doi.org/10.1016/j.icarus.2023.115942(2023).
13. Daubar, I. et al. Changes in blast zone albedo patterns around new
martian impact craters.Icarus267,86–105 (2016).
14. Phillips,C.B.,Burr,D.M.,Beyer,R.A.Massmovementwithinaslope
streak on Mars.Geophys. Res. Lett.34,https://doi.org/10.1029/
2007GL031577(2007).
15. Chuang, F. C., Beyer, R. A., McEwen, A. S., Thomson, B. J. HiRISE
observations of slope streaks on Mars.Geophys. Res. Lett.34
,
https://doi.org/10.1029/2007GL031111(2007).
16. Dundas, C. Geomorphological evidence for a dry dust avalanche
origin of slope streaks on Mars.Nat. Geosci.13,473–476 (2020b).
17. Dundas, C. An aeolian grainflow model for Martian Recurring Slope
Lineae.Icarus343, 113681 (2020a).
18. Dundas, C. et al. Granularflows at recurring slope lineae on Mars
indicate a limited role for liquid water.Nat. Geosci.10,903–907
(2017).
19. Ferris,J.C.,Dohm,J.M.,Baker,V.R.,Maddock,T.Darkslopestreaks
on Mars: Are aqueous processes involved?Geophys. Res. Lett.29,
https://doi.org/10.1029/2002GL014936(2002).
20. Abotalib, A. & Heggy, E. A deep groundwater origin for recurring
slopelineaeonMars.Nat. Geosci.12,235–241 (2019).
21. Piqueux, S. et al. Discovery of a widespread low-latitude diurnal
CO2 frost cycle on Mars.J. Geophys. Res. Planets,121,https://doi.
org/10.1002/2016JE005034(2016).
22. Kreslavsky, M. & Head, J. Slope streaks on Mars: A new“wet”
mechanism.Icarus201,517–527 (2009).
23. Chojnacki,M.etal.Geologiccontext of recurring slope lineae in
Melas and Coprates Chasmata, Mars.J. Geophys. Res. Planets121,
1204–1231 (2016).
24. Heyer, T. et al. Seasonal formation rates of martian slope streaks.
Icarus323,76–86 (2019).
25. Bhardwaj, A., Sam, L., Martin-Torres, F. & Zorzano, M.-P. Are Slope
Streaks Indicative of Global-Scale Aqueous Processes on Con-
temporary Mars?Rev. Geophys.57,48–77 (2018).
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 14

15. 26. Dickson,J.L.,Ehlmann,B.L.,Kerber,L.,Fassett,C.I.TheGlobal
Context Camera (CTX) Mosaic of Mars: A Product of Information-
Preserving Image Data Processing.Earth Space Sci.11,https://doi.
org/10.1029/2024EA003555(2024).
27. Dickson,J.L.,Kerber,L.A.,Fassett,C.I.&Ehlmann,B.L.Aglobal,
blended CTX mosaic of Mars with vectorized seam mapping: A new
mosaicking pipeline using principles of non-destructive image
editing, ID 2480. In:Presented at the 49th annual lunar and planetary
science conference,https://www.hou.usra.edu/meetings/
lpsc2018/pdf/2480.pdf(2018).
28. Aharonson, O., Schorghofer, N. & Gerstell, M. Slope streak forma-
tion and dust deposition rates on Mars.J. Geophys. Res. Planets108,
https://doi.org/10.1029/2003JE002123(2003).
29. Bergonio, J., Rottas, K. & Schorghofer, N. Properties of martian slope
streak populations.Icarus225,194–199 (2013).
30. McEwen, A. et al. Mars Reconnaissance Orbiter’s High Resolution
Imaging Science Experiment (HiRISE).JGR Planets112,https://doi.
org/10.1029/2005JE002605(2007).
31. Christensen, P. et al. Mars Global Surveyor Thermal Emission
Spectrometer experiment: Investigation description and surface
science results.J. Geophys. Res.106,23823–23871 (2001).
32. Bibring, J.-P. et al. OMEGA: Observatoire pour la Minéralogie, l’Eau,
les Glaces et l’Activité. Mars Express: the scientific payload, ESA SP-
1240, ISBN 92-9092-556-6 (2004).
33. Smith, D. et al. Mars Orbiter Laser Altimeter: Experiment summary
after thefirst year of global mapping of Mars.J. Geophys. Res.
Planets106,https://doi.org/10.1029/2000JE001364(2001).
34. Forget, F. et al. Improved general circulation models of the Martian
atmosphere from the surface to above 80 km.J. Geophys. Res.
Planets104,https://doi.org/10.1029/1999JE001025(1999).
35. Forget, F., Montabone, L. Atmospheric Dust on Mars: A Review. In:
47th International Conference on Environmental Systems,http://
hdl.handle.net/2346/72982(2017).
36. Pollack, J. et al. Properties of aerosols in the Martian atmosphere, as
inferred from Viking Lander imaging data.J. Geophys. Res.82,
4479–4496 (1977).
37. López-Cayuela, M. Á., Zorzano, M.-P., Guerrero-Rascado, J. L. &
Córdoba-Jabonero, C. Quantitative analysis of the Martian atmo-
spheric dust cycle: Transportedmass, surface dust lifting and
sedimentation rates.Icarus409, 115854 (2024).
38. Martin, T. Mass of dust in the Martian atmosphere.J. Geophys. Res.
Planets100,
https://doi.org/10.1029/95JE00414(1995).
39. Daubar, I. J. et al. New Craters on Mars: An Updated Catalog.JGR
Planets127,https://doi.org/10.1029/2021JE007145(2022).
40. Wagstaff, K. L. et al. Using machine learning to reduce observational
biases when detecting new impacts on Mars.Icarus386, 115146
(2022).
41. Marsquake Service Catalog v14 InSight Mars SEIS Data Service
2019 SEIS raw data v14, InSight Mission IPGP, JPL, CNES, ETHZ,
ICL, MPS, ISAE-Supaero, LPG, MFSChttps://doi.org/10.12686/
a21(2023).
42. Clinton, J. F. et al. The Marsquake catalogue from InSight, sols
0–478.Phys. Earth Planet. Inter.310,106595(2021).
43. Giardini, D. et al. The Seismicity of Mars.Nat. Geosci.13,205–212
(2020).
44. Stähler, S. et al. Tectonics of Cerberus Fossae unveiled by mars-
quakes.Nat. Astron.6,1376–1386 (2022).
45. Bickel, V. T. et al. The First Global Catalog of Rockfall Locations on
Mars.Geophys. Res. Lett.51,https://doi.org/10.1029/
2024GL110674(2024a).
46. Conway, S. J. et al. A global survey for dust devil vortices on mars
using MRO context camera images enabled by neural networks.
Planet. Space Sci.259, 106072 (2025).
47. Hynek, B., Beach, M., Hoke, M. Updated global map of Martian
valley networks and implications for climate and hydrologic
processes.J. Geophys. Res. Planets115,https://doi.org/10.1029/
2009JE003548(2010).
48. Heyer, T., Raack, J., Iqbal, W., Hiesinger, H. & Oetting, A. Albedo
analysis of dust devil-induced slope streaks and tracks on Mars.
Icarus423, 116270 (2024).
49. Kaplan, H. H. et al. Visible and Near-infrared Spectral Properties of
Martian Slope Streaks.Planet. Sci. J.4, 232 (2023).
50. Vincendon, M., Pilorget, C., Carter, J. & Stcherbinine, A. Observa-
tional evidence for a dry dust-wind origin of Mars seasonal dark
flows.Icarus325, 115–127 (2019).
51. McEwen, A. et al. Mars: Abundant Recurring Slope Lineae (RSL)
Following the Planet-Encircling Dust Event (PEDE) of 2018.J. Geo-
phys. Res. Planets126,https://doi.org/10.1029/2020JE006575(2021)
52. Schaefer, E., McEwen, A. & Sutton, S. A case study of recurring
slope lineae (RSL) at Tivat crater: Implications for RSL origins.Icarus
317,621–648 (2019).
53. Leovy,C.WeatherandclimateonMars.Nature412
,245–249 (2001).
54. Ruff, S., Christensen, P. Bright and dark regions on Mars: Particle
size and mineralogical characteristics based on Thermal Emission
Spectrometer data.J. Geophys. Res. Planets107,https://doi.org/10.
1029/2001JE001580(2002).
55. Palluconi,F.&Kieffer,H.Thermal inertia mapping of Mars from
60°S to 60°N.Icarus45,415–426 (1981).
56. Valantinas, A. et al. Detection of ferrihydrite in Martian red dust
records ancient cold and wet conditions on Mars.Nat. Commun.16,
1712 (2025).
57. Guzewich,S.D.,Smith,M.D.&Wolff,M.J.Theverticaldistribution
of Martian aerosol particle size.J. Geophys. Res.: Planets119,
2694–2708 (2014).
58. Pouliquen, O., Delour, J. & Savage, S. B. Fingering in granularflows.
Nature386,816–817 (1997).
59. Woodhouse, M. J., Thornton, A. R., Johnson, C. G., Kokelaar, B. P. &
Gray,J.M.N.T.Segregation-inducedfingering instabilities in
granular free-surfaceflows.J. Fluid Mech.709,543–580 (2012).
60. Miyamoto, H., Dohm, J., Beyer, R. & Baker, V. Fluid dynamical
implications of anastomosing slope streaks on Mars.J. Geophys.
Res. Planets109,https://doi.org/10.1029/2003JE002234(2004).
61. Van den Eeckhaut, M., Poesen, J., Govers, G., Verstraeten, G. &
Demoulin, A. Characteristics of the size distribution of recent and
historical landslides ina populated hilly region.Earth Planet. Sci.
Lett.256,588–603 (2007).
62. Tanyas, H. et al. Presentation and Analysis of a Worldwide Database
of Earthquake-Induced Landslide Inventories.J. Geophys. Res. Earth
Surface122,https://doi.org/10.1002/2017JF004236(2017).
63. Guzzetti, F., Malamud, B., Turcotte, D. & Reichenbach, P. Power-law
correlations of landslide areas in central Italy.Earth Planet. Sci. Lett.
195,169–183 (2002).
64. Malin, M. et al. Context Camera Investigation on board the Mars
Reconnaissance Orbiter.JGR Planets112,https://doi.org/10.1029/
2006JE002808(2007).
65. Bickel, V. T., Lanaras, C., Manconi, A., Loew, S. & Mall, U. Automated
detection of lunar rockfall using a convolutional neural network.
IEEE Trans. Geosci. Remote Sens.57,3501–3511 (2019).
66. Bickel, V. T. et al. Deep learning-driven detection and mapping of
rockfalls on Mars.IEEE J. Sel. Top. Appl. Earth Observ. Remote Sens.
13,2831–2841 (2020).
67. Bickel, V. T., Mandrake, L. & Doran, G. A labeled image dataset for
deep learning-driven rockfall detection on the moon and Mars.
Front. Remote Sens.2, 4 (2021).
68. Bickel, V. T., Mandrake, L. & Doran, G. Analyzing multi–domain
learning for enhanced rockfall mapping in known and unknown
planetary domains.ISPRS J. Photogramm. Remote Sens.182,
1–13 (2021).
69. Bickel,V.T.,Aaron,J.,Manconi,A.,Loew,S.&Mall,U.Impactsdrive
lunar rockfalls over billions of years.Nat. Commun.11, 2862 (2020).
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 15

16. 70. Bickel, V. T. Loew, S., Aaron, J. & Goedhart, N. A Global Perspective
on Lunar Granular Flows.Geophys. Res. Lett.49,https://doi.org/10.
1029/2022GL098812(2022).
71. Bickel, V. T. et al. A Global Dataset of Potential Chloride Deposits on
Mars as Identified by TGO CaSSIS. Nat. Sci. Data11,https://doi.org/
10.1038/s41597-024-03685-3(2024).
72. Rüsch, O. & Bickel, V. T. Global Mapping of Fragmented Rocks on
the Moon with a Neural Network: Implications for the Failure Mode
of Rocks on Airless Surfaces.Planet. Sci. J.4, 126 (2023).
73. Mills,M.M.,Bickel,V.T.,McEwen,A.S.&Valantinas,A.Aglobal
dataset of pitted cones on Mars.Icarus418, 116145 (2024).
74. Hossain,A.,Islam,M.T.&Almutairi,A.F.Adeeplearningmodelto
classify and detect brain abnormalities in portable microwave
based imaging system.Sci. Rep.12, 6319 (2022).
75. Jubayer, F. et al. Detection of mold on the food surface using
YOLOv5.Curr. Res. Food Sci.4,724–728 (2021).
76. Song,Q.etal.ObjectDetection Method for Grasping Robot Based
on Improved YOLOv5.Micromachines12, 1273 (2021).
77. Diwan, T., Anirudh, G. & Tembhurne, J. Object detection using
YOLO: challenges, architectural successors, datasets and applica-
tions.Multimed. Tools Appl.82,9243–9275 (2022).
78. Reddy, M., Premkumar, S. Detection of Objects Using YOLO Algo-
rithm and Comparison of Accuracy with AdaBoost Algorithm.Int. J.
Early Childhood Special Educ.14,https://www.int-jecse.net/media/
article_pdfs/5812-5818.pdf(2022).
79. Kaur, R. & Singh, S. A comprehensive review of object detection
with deep learning.Digital Signal Process.132, 103812 (2023).
80. Wang, Y., Di, K., Xin, X. & Wan, W. Automatic detection of Martian
dark slope streaks by machine learning using HiRISE images.ISPRS
J. Photogramm. Remote Sens.129,12–20 (2017).
81. USGS. Mars MGS MOLA Global Color Shaded Relief 463m v1. URL:
https://astrogeology.usgs.gov/search/map/Mars/GlobalSurveyor/
MOLA/Mars_MGS_MOLA_ClrShade_merge_global_463m (2020).
82. Tanaka, K. L. et al.Geologic map of Mars. United States Geological
Survey Science. Invest. (Map, scale 1: 20,000,000)https://doi.org/
10.3133/sim3292(2014).
83. Ody, A. et al. Global maps of anhydrous minerals at the surface of
Mars from OMEGA/Mex.J. Geophys. Res. Planets117,https://doi.
org/10.1029/2012JE004117(2012).
84. Malakhov, A. et al. High Resolution Map of Water in the Martian
Regolith Observed by FREND Neutron Telescope Onboard ExoMars
TGO.J. Geophys. Res. Planets127,
https://doi.org/10.1029/
2022JE007258(2022).
85. Rani, A. et al. Consolidated Chemical Provinces on Mars: Implica-
tions for Geologic Interpretations.Geophys. Res. Lett.49,https://
doi.org/10.1029/2022GL099235(2022).
86. Bickel,V.T.,Aaron,J.,Manconi,A.,Loew,S.GlobalDriversand
Transport Mechanisms of Lunar Rockfalls.J. Geophys. Res. Planets
126,https://doi.org/10.1029/2021JE006824(2021).
87. Ceylan, S. et al. The marsquake catalogue from InSight, sols 0-1011.
Phys. Earth Planet. Inter.333, 106943 (2022).
88. Bickel, V. T. Dataset to: Streaks on Martian Slopes are Dry [Dataset].
https://doi.org/10.48620/75966(2025).
89. Thomas, N. et al. The Colour and Stereo Surface Imaging System
(CaSSIS) for the ExoMars Trace Gas Orbiter.Space Sci. Rev.212,
1897–1944 (2017).
90. Posiolova, L. et al. Largest recent impact craters on Mars: Orbital
imaging and surface seismic co-investigation.Science378,
412–417 (2022).
91. Daubar, I. J. et al. Two Seismic Events from InSight Confirmed as
New Impacts on Mars.Planet. Sci. J.4,175(2023).
92. Bickel, V. T. et al. New Impacts on Mars: Systematic Identification
and Association With InSight Seismic Events.Geophys. Res. Lett.
52,https://doi.org/10.1029/2024GL109133(2025).
Acknowledgements
VTB acknowledges the conversations with J. Dickson about the pre-v01
global CTX mosaic in late 2022, as well as the fruitful discussions about
size- and area frequency distribution properties of terrestrial landslides
with J. Aaron. Both authors acknowledge the review and feedback pro-
vided by the three anonymous reviewers.
Author contributions
V.T. Bickel: Conceptualization, Methodology, Software, Validation, For-
mal analysis, Investigation, Resources, Data Curation, Writing - Original
Draft, Visualization A. Valantinas:Conceptualization, Investigation,
Writing–Original Draft.
Competing interests
The authors declare no competing interests.
Additional information
Supplementary informationThe online version contains
supplementary material available at
https://doi.org/10.1038/s41467-025-59395-w.
Correspondenceand requests for materials should be addressed to
Valentin Tertius Bickel.
Peer review informationNature Communicationsthanks Sarah Sutton
and the other, anonymous, reviewer(s)for their contribution to the peer
review of this work. A peer reviewfile is available.
Reprints and permissions informationis available at
http://www.nature.com/reprints
Publisher’s noteSpringer Nature remains neutral with regard to jur-
isdictional claims in published maps and institutional affiliations.
Open AccessThis article is licensed under a Creative Commons
Attribution-NonCommercial-NoDerivatives 4.0 International License,
which permits any non-commercial use, sharing, distribution and
reproduction in any medium or format, as long as you give appropriate
credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if you modified the licensed
material. You do not have permission under this licence to share adapted
material derived from this article or parts of it. The images or other third
party material in this article are included in the article’s Creative
Commons licence, unless indicatedotherwise in a credit line to the
material. If material is not included in the article’s Creative Commons
licence and your intended use is not permitted by statutory regulation or
exceeds the permitted use, you will need to obtain permission directly
from the copyright holder. To view a copy of this licence, visithttp://
creativecommons.org/licenses/by-nc-nd/4.0/.
© The Author(s) 2025
Article https://doi.org/10.1038/s41467-025-59395-w
Nature Communications| (2025) 16:4315 16