Orbital monitoring of martian surface changes

Icarus 278 (2016) 279–300

Contents lists available at ScienceDirect

Icarus

journal homepage: www.elsevier.com/locate/icarus

Orbital monitoring of martian surface changes

Paul E. Geissler a , ∗, Lori K. Fenton

b , Marie-therese Enga

c , Priyanjoli Mukherjee

d

a Center for Astrogeology, U.S. Geological Survey, 2255 N Gemini Drive, Flagstaff AZ 86001, USA b SETI Institute, 189 Bernardo Ave, Suite 100, Mountain View, CA 94043, USA c Macomb Community College, South Campus, 14500 E. 12 Mile Road, Warren, MI 48088-3896, USA d Mesa Community College, 1833 W Southern Ave., Mesa, AZ 85202, USA

a r t i c l e i n f o

Article history:

Received 23 January 2014

Revised 10 May 2016

Accepted 11 May 2016

Available online 24 May 2016

Keywords:

Mars

Eolian processes

a b s t r a c t

A history of martian surface changes is documented by a sequence of global mosaics made up of Mars

Global Surveyor Mars Orbiter Camera daily color images from 1999 to 2006, together with a single mosaic

from the Mars Reconnaissance Orbiter Mars Color Imager in 2009. These observations show that changes

in the global albedo patterns of Mars take place by a combination of dust storms and strong winds.

Many of the observed surface changes took place along the tracks of seasonally repeating winter dust

storms cataloged by Wang and Richardson (2015). These storms tend to sweep dust towards the equator,

progressively shifting albedo boundaries and continuing surface changes that began before the arrival of

MGS. The largest and most conspicuous changes took place during the global dust storm of 2001 (MY

25), which blanketed Syrtis Major, stripped dust from the Tharsis region, and injected dust into Solis

Planum. High wind speeds but low wind stresses are predicted in Syrtis, Tharsis and Solis by the NASA

Ames GCM. Frequent changes in these regions show that dust accumulations are quickly removed by

stronger winds that are not predicted by the GCM, but may result from smaller-scale influences such as

unresolved topography.

Published by Elsevier Inc.

1

m

s

r

s

t

(

o

w

o

m

M

a

l

e

p

(

d

u

a

e

(

o

s

M

i

i

f

M

b

t

a

o

r

h

0

. Introduction

Changes in the appearance of Mars’ surface have fascinated hu-

ans for centuries. Once thought to be due to vegetation, modern

pacecraft have shown that the changes are caused by relentless

edistribution of sand and dust by the martian winds. Major

urface albedo changes took place during the decades between

he Viking Orbiter mission and the arrival of Mars Global Surveyor

MGS) in 1998 ( Geissler, 2005 ). The long interval between the

bservations made these changes difficult to interpret, because it

as not known whether the surface changes took place abruptly

r gradually. Beginning with MGS, Mars has been continuously

onitored by a succession of orbiting spacecraft since 1999. The

ars Orbiter Camera (MOC) ( Malin et al., 1991 ) on MGS produced

daily photographic record of the entire planet’s surface until

ate 2006, documenting almost 4 complete martian years ( Malin

t al., 2010 ). MGS maintained a steady orbit over this period that

rovided consistent illumination and viewing conditions ideal for

∗ Corresponding author. Tel.: 928 556 7257; fax: 928 556 7014.

E-mail addresses: [email protected] (P.E. Geissler), [email protected]

(L.K. Fenton), [email protected] (M.-t. Enga), [email protected]

P. Mukherjee).

o

c

c

t

c

m

ttp://dx.doi.org/10.1016/j.icarus.2016.05.023

019-1035/Published by Elsevier Inc.

etecting surface changes. The frequent MOC observations allow

s to examine the sequence of events that led to surface changes

nd better infer how the winds act upon the martian surface.

MOC was followed by the Mars Color Imager (MARCI) ( Malin

t al., 2001; Bell et al., 2009 ) on Mars Reconnaissance Orbiter

MRO), which began daily imaging in March 2006 and continues to

perate at this writing. This report focuses on the MGS MOC ob-

ervations and describes the construction and interpretation of a

ars “movie”, a time series of global mosaics of Wide-Angle MOC

mages. A MARCI mosaic made after the 2007 global dust storm

s also included in the analysis, but the history of martian sur-

ace changes detailed here is derived from the MOC observations.

any significant changes can be seen in these data. In the sections

elow, we point out specific examples grouped for discussion by

he timescales of the changes. Progressive changes include moving

lbedo boundaries that advanced incrementally over a succession

f martian years, continuing changes that began before MGS ar-

ived. Episodic changes typically took place during dust storms that

ccurred in the perihelion season (within 65 º Ls of Ls 251 º). Quasi-

ontinuous changes were observed in Tharsis and in the Solis La-

us region south of the Valles Marineris. High resolution images of

hese targets now available from the HiRISE camera on Mars Re-

onnaissance Orbiter (MRO) offer more insight into the nature and

echanisms of the surface changes. Here we document the various

http://dx.doi.org/10.1016/j.icarus.2016.05.023

http://www.ScienceDirect.com

http://www.elsevier.com/locate/icarus

http://crossmark.crossref.org/dialog/?doi=10.1016/j.icarus.2016.05.023&domain=pdf

mailto:[email protected]





280 P.E. Geissler et al. / Icarus 278 (2016) 279–300

s

i

w

T

g

z

v

t

l

w

n

p

b

d

p

s

t

s

a

t

v

a

t

l

i

i

a

i

t

s

p

e

g

f

t

t

r

q

I

b

h

f

e

c

m

c

e

t

d

a

e

o

T

2

c

s

h

e

t

a

timescales of martian surface changes and suggest hypotheses for

the eolian processes responsible.

2. Approach

2.1. Data processing

Over 40 0 0 selected color image pairs from the MOC Wide An-

gle Camera (WAC) were processed into 43 separate mosaics for

this project. These mosaics record snapshots of Mars at intervals

of about two months (30 ° of solar longitude, Ls) from June 1999

to October 2006. Uneven coverage meant that the time steps were

not always equal, but nearly so. Each mosaic is the average of ap-

proximately 100 individual red and blue filter image pairs spanning

many days of observations (typically 10 º of Ls, more if incomplete

coverage), so that meteorological phenomena such as clouds and

dust storms are blurred but surface features are emphasized.

The data were downloaded from the Planetary Data System and

processed with the USGS ISIS 2 software using the approach de-

scribed in Geissler (2005) . The images were radiometrically cali-

brated and geometrically reprojected to a simple cylindrical map

projection at a resolution of 10 pixels per degree. A Lunar-Lambert

limb-darkening correction was applied with coefficients 0.6 for

blue images and 0.75 for red. The images were cropped to show

only regions viewed at incidence and emission angles less than

70 ° and phase angles greater than 15 °. The images were averaged

together and a procedure was applied to remove seams between

images by adding a low-pass filtered version of the mosaic to-

gether with a mosaic of high-pass filtered versions of the individ-

ual frames ( Soderblom et al., 1978 ). For display purposes, a green

channel was synthesized by interpolating between the WAC red

and blue images.

These procedures reduced but did not eliminate imperfections

in the mosaics. Artifacts including seams and striping are visible

along the orbit tracks, from top left to bottom right. However,

the mosaics are sufficient for the task of tracking surface albedo

changes; the artifacts are easily recognizable and do not interfere

with the interpretation of surface features.

MRO MARCI data were processed using USGS ISIS 3 software

on a cluster of 50 CPUs. The even and odd channels were cali-

brated separately and map projected before being mosaicked to-

gether. Values of the local incidence, emission, and phase angles

for each pixel were calculated and stored by the program phocube.

These values were used to apply a Lommel–Seeliger photometric

correction to the images, weighted by the cosine of the incidence

angle divided by the sum of the cosines of the incidence angle and

the emission angle. The images were then cropped to restrict the

longitude range to within 40 ° of the image center longitude, to re-

strict the emission angles to be less than 60 °, and to restrict the

difference in incidence and emission angles to be greater than 7.5 °to remove opposition effects and specular reflection. The images

were then averaged and seam-corrected as for the MOC images.

These procedures are still experimental, and produce mosaics with

more noticeable striping artifacts than the MOC mosaics.

MOC and MARCI imaged the surface using different filter sets.

MOC’s wide angle color filters were sensitive to wavelengths from

600 to 630 nm in red and from 420 to 450 nm in blue ( Malin

et al., 1992 ). MARCI visible images are acquired in 5 filters, two of

which approximate the MOC coverage with sensitivities from 573

to 635 nm in red and from 405 to 469 nm in blue ( Malin et al.,

2001; Bell et al., 2009 ).

2.2. Data analysis

The time series of mosaics was then assembled to produce an

animated sequence of images in which surface changes are easy to

pot. A 1/3 scale gif animation of the MOC Mars movie is presented

n the Supplementary materials as Supplementary Fig. 2 (along

ith the full resolution individual mosaics, named year-month-Ls).

he gif animation was produced with the publicly available pro-

ram gifsicle and has additional artifacts caused by color quanti-

ation, but all of the surface changes described below are easily

isible in the animation.

To quantify the changes that took place from one martian year

o the next, difference images were created by subtracting an ear-

ier mosaic from a later mosaic taken at the same season (Ls 330 °,hen the southern hemisphere was clear and the equatorial volca-

oes were free of clouds). We used the red filter mosaics for this

urpose because they show much greater surface contrast than the

lue filter mosaics. Areas that darkened during the interval appear

arker in the difference images. This approach unfortunately em-

hasized seams and artifacts but helped minimize the effects of

easonal variations in cloud cover and surface frost. The magni-

udes and areal extents of the albedo changes can be easily mea-

ured from the difference images. The average speed of advancing

lbedo boundaries was calculated from the width of the bands in

he difference images divided by the time interval between obser-

ations.

The USGS ISIS program qview was used to make measurements

utomatically corrected for latitude. Potential sources of uncer-

ainty include differences in viewing angle or conditions, meteoro-

ogical impacts, and georeferencing errors. Photometric variations

n illumination and viewing geometry were minimized by limit-

ng the incidence, emission, and phase angles considered and by

veraging together many images within these ranges. Meteorolog-

cal changes were minimized by comparing images taken during

he same season each year, at a time of year when no major dust

torms happened to take place, however they are clearly present in

laces such as the Hellas Basin, as we discuss later. Georeferencing

rrors were absorbed by the process of averaging many images to-

ether, which resulted in blurry mosaics with little misregistration

rom one mosaic to the next. The major source of uncertainty in

he measurement of the speed of advancing albedo boundaries is

he variability of the boundary itself.

Individual surface changes were measured by subtracting the

ed filter image acquired before each change from the image ac-

uired after the change in equal area cylindrical map projections.

n most cases the interval between successive images was 30 º Ls,

ut the timing uncertainty was sometimes longer during the peri-

elion season when the surface could not clearly be seen. The dif-

erence images were thresholded and then edited to remove differ-

nces caused by changes in contrast (due to changing atmospheric

onditions and lighting) and differences caused by artifacts in the

osaics. The threshold was chosen as the highest value that in-

luded all the genuine surface changes, excluding artifacts to the

xtent possible, and was different for each scene. An example of a

hresholded difference image is shown as Fig. 1 . These thresholded

ifference images were then used to mask the difference image

nd isolate the regions involved in the change. The masked differ-

nce images were used to compute the area and mean magnitude

f each change. The results of these measurements are listed in

able 1.

.3. Comparison with simulated winds

The timing and direction of observed dust-removal events were

ompared to wind patterns expected in each region of interest, as

imulated by the NASA Ames Global Climate Model (GCM), which

as been used extensively to investigate Mars’ climate ( Haberle

t al . , 1999; Kahre et al., 2006 ). The model numerically represents

he atmosphere as a set of points that forms a 3-dimensional grid

round the surface of Mars, calculating and outputting atmospheric

P.E. Geissler et al. / Icarus 278 (2016) 279–300 281

Fig. 1. Individual surface change measurement. Successive red-filter images were subtracted and then thresholded to isolate the regions affected by this darkening of Tharsis.

(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

a

t

i

S

T

G

g

s

w

3

p

(

n

r

b

t

a

T

2

p

i

(

t

s

a

t

f

3

o

d

s

A

n

s

i

o

T

s

p

nd surface parameters at each grid point (e.g., wind velocity, air

emperature, surface temperature, air pressure). Surface properties

nclude topography from the Mars Orbiter Laser Altimeter (MOLA;

mith et al., 2001 ) and both albedo and thermal inertia from the

hermal Emission Spectrometer (TES; Christensen et al., 2001 ). The

CM uses a 2-stream radiative transfer scheme that accounts for

aseous absorption and scattering aerosols. The simulations in this

tudy utilized a grid spacing of 5 º of latitude by 6 º of longitude

ith a temporal resolution of 1.5 “hours” (in which an “hour” is

698.7 s long), saving output throughout a full martian year. Winds

resented in this work represent those averaged over a large area

e.g., at the equator each grid spans ∼300 ×350 km horizontally),

ot resolving smaller-scale flows such as those produced by un-

esolved topography, varying surface roughness, unresolved albedo

oundaries, or daytime convective turbulence (i.e., they may effec-

ively be regarded as the background wind at a given time of day

nd season). Some key results of these simulations are listed in

ables 2 and 3.

.4. Comparison with dust storm observations

The timing and direction of the surface changes were also com-

ared with orbital observations of dust storms, which are also vis-

ble in MOC images and are known to be active in specific regions

Cantor et al . , 2001 ). For this purpose we relied primarily on the

abulation of regional dust storms compiled by Wang and Richard-

on (2015) , which covers the entire duration of MGS observations

nd beyond to MARCI. These data include the area, duration, and

iming of dust storms which we will graphically compare to sur-

ace changes in the figures that follow.

. Overview of global changes

The MOC movie shows many seasonal changes such as clouds

ver the giant volcanoes in Tharsis and Olympus Mons and Elysium

uring northern spring and summer, winter hoods at the north and

outh poles, and winter clouds and seasonal frost in the Hellas and

rgyre basins. To minimize the effects of these atmospheric phe-

omena, it is useful to examine a subset of images taken at the

ame season on subsequent martian years. Fig. 2 shows a compar-

son of mosaics acquired during late northern winter, at Ls 330 °,ver 5 seasons (2007 is missing because of the global dust storm).

he surface changes are highlighted in the panels at right, which

how the differences after subtracting the red filter mosaic of the

revious year from the next.


Table 1

Specific surface changes during the MGS era.

This table lists the specific albedo changes that were measured in our study sites, giving the details of the global mosaics assembled from data before and after the changes

took place, along with the measured characteristics of the surface changes. Full resolution versions of these mosaics are included in the Supplementary Materials, named as

the year-month-Ls of the mosaics as listed in this table. The “mosaic phase” refers to the MGS Mission phase from which the MOC data were drawn. The “mosaic start date”

is the date of acquisition of the first image in the mosaic, typically 5 º of Ls before the “mosaic center Ls”, because the mosaics averaged many images taken over an interval

of about 10 º Ls. The “threshold” was chosen as the highest value that included all the genuine surface changes, and was different for each scene. The “estimated Ls” is the

midpoint in time between the two mosaics in which the change was detected, and is used to define the position of the changes in the bubble plots to follow. Notes: #1.

Includes Propontis brightening at Ls 240 º. #2. Area of Propontis brightening at Ls 240 º.

Location Before Before Before After After After Threshold Area Mean Estimated Mars

mosaic mosaic mosaic mosaic mosaic mosaic in km

2 change Ls year

phase start date center Ls phase start date center Ls I/F

Solis A07 1999-09-14 210 A09 1999-11-01 240 0 .01 431,793 0 .013 225 24

Solis A10 1999-12-25 270 A13 20 0 0-03-29 330 0 .015 89,784 −0 .019 300 24

Solis A15 20 0 0-05-23 0 A18 20 0 0-08-07 33 0 .015 687,036 −0 .019 16 .5 25

Solis E05 2001-06-16 184 E09 2001-10-12 255 0 .04 485,284 0 .058 219 .5 25

Solis E05 2001-06-16 184 E09 2001-10-12 255 0 .04 784,472 −0 .065 219 .5 25

Solis E10 2001-11-05 270 E12 2002-01-02 305 0 .015 21,832 −0 .019 287 .5 25

Solis E13 2002-02-13 330 E15 2002-04-11 360 0 .03 29,413 −0 .034 345 25

Solis E15 2002-04-11 0 E17 2002-06-22 32 0 .03 525,017 −0 .038 16 26

Solis E17 2002-06-22 32 E19 2002-08-20 60 0 .03 21,060 −0 .035 46 26

Solis E23 2002-12-31 120 R03 2003-03-03 150 0 .015 88,766 −0 .022 135 26

Solis R04 2003-04-21 180 R09 2003-09-23 270 0 .015 3,402,536 0 .022 225 26

Solis R04 2003-04-21 180 R09 2003-09-23 270 0 .015 681,350 −0 .023 225 26

Solis R08 2003-08-06 240 R09 2003-09-24 270 0 .005 36,714 −0 .008 255 26

Solis R09 2003-09-24 270 R11 2003-11 300 0 .011 28,887 −0 .015 285 26

Solis R11 2003-11 300 R13 2004-01-01 330 0 .015 664,397 −0 .024 315 26

Solis R23 2004-11-17 120 S02 2005-01-18 150 0 .03 280,865 −0 .037 135 27

Solis S02 2005-01-18 150 S07 2005-06-25 240 0 .01 562,433 0 .014 195 27

Solis S06 2005-05-12 210 S07 2005-06-26 240 0 .01 901,528 0 .014 225 27

Solis S06 2005-05-12 210 S07 2005-06-26 240 0 .01 70,234 −0 .014 225 27

Solis S10 2005-09-24 300 S12 2005-11-18 330 0 .03 443,376 −0 .034 315 27

Solis S18 2006-05-27 60 S20 2006-08-03 90 0 .005 57,668 −0 .008 75 28

Tharsis A13 20 0 0-03-29 330 A15 20 0 0-05-23 360 0 .015 267,212 −0 .022 345 24

Tharsis A15 20 0 0-05-23 0 A18 20 0 0-08-07 33 0 .015 74,060 0 .018 16 .5 25

Tharsis A15 20 0 0-05-23 0 A18 20 0 0-08-07 33 0 .015 413,541 −0 .022 16 .5 25

Tharsis A18 20 0 0-08-07 33 E05 2001-06-16 184 0 .015 559,239 0 .022 108 .5 25

Tharsis A18 20 0 0-08-07 33 E05 2001-06-16 184 0 .015 427,581 −0 .024 108 .5 25

Tharsis E05 2001-06-16 184 E10 2001-11-05 270 0 .03 2,995,067 −0 .049 227 25

Tharsis E13 2002-02-13 330 E15 2002-04-11 360 0 .015 6915 −0 .019 345 25

Tharsis E21 2002-10-25 90 E23 2002-12-31 120 0 .01 1,255,646 0 .014 105 26

Tharsis R11 2003-11 300 R13 2004-01-01 330 0 .015 78,482 0 .020 315 26

Tharsis R11 2003-11 300 R13 2004-01-01 330 0 .015 57,528 −0 .021 315 26

Tharsis R13 2004-01-01 330 R14 2004-02-26 360 0 .01 32,397 −0 .016 345 26

Tharsis R14 2004-02-26 0 R17 2004-04-28 30 0 .015 146,049 −0 .020 15 27

Tharsis R23 2004-11-17 120 S02 2005-01-18 150 0 .015 221,056 0 .025 135 27

Tharsis R23 2004-11-17 120 S02 2005-01-18 150 0 .015 456,854 −0 .026 135 27

Tharsis S10 2005-09-24 300 S12 2005-11-18 330 0 .015 33,344 −0 .019 315 27

Tharsis S14 2006-01-13 0 S16 2006-03-16 30 0 .015 392,973 −0 .022 15 28

Amazonis A07 1999-09-14 210 A09 1999-11-01 240 0 .005 98,454 −0 .009 225 24

Amazonis A13 20 0 0-03-29 330 A15 20 0 0-05-23 360 0 .005 61,529 0 .009 345 24

Amazonis E05 2001-06-16 184 E13 2002-02-13 330 0 .01 29,027 0 .015 257 25

Amazonis R03 2003-03-03 150 R04 2003-04-21 180 0 .01 6669 0 .013 165 26

Amazonis R04 2003-04-21 180 R13 2004-01-01 330 0 .02 212,773 0 .029 255 26 (#1)

Amazonis R09 2003-09-23 270 R11 2003-11 300 0 .015 156,157 0 .021 285 26

Amazonis R04 2003-04-21 180 R13 2004-01-01 330 0 .02 69,356 0 .029 240 26 (#2)

Amazonis R11 2003-11 300 R13 2004-01-01 330 0 .015 12,741 −0 .020 315 26

Amazonis R14 2004-02-26 0 R17 2004-04-28 30 0 .005 46,542 0 .008 15 27

Amazonis R14 2004-02-27 0 R17 2004-04-29 30 0 .015 12,039 −0 .018 15 27

Amazonis S04 2005-03-15 180 S10 2005-09-24 300 0 .015 40,470 0 .025 240 27

Amazonis S04 2005-03-16 180 S10 2005-09-25 300 0 .015 88,345 −0 .022 240 27

Amazonis S10 2005-09-24 300 S12 2005-11-18 330 0 .01 61,670 −0 .014 315 27

Amazonis S12 2005-11-18 330 S14 2006-01-13 360 0 .015 119,970 0 .021 345 27

Syrtis A13 20 0 0-03-29 330 A15 20 0 0-05-23 360 0 .01 70,234 −0 .013 345 24

Syrtis E05 2001-06-16 184 E10 2001-11-05 270 0 .03 1,132,798 0 .045 227 25

Syrtis E12 2002-01-02 305 E13 2002-02-13 330 0 .03 1,022,410 −0 .042 317 .5 25

Syrtis E19 2002-08-20 60 E21 2002-10-25 90 0 .015 60,476 −0 .019 75 26

Syrtis E21 2002-10-25 90 E23 2002-12-31 120 0 .03 158,474 −0 .034 105 26

Syrtis R04 2003-04-21 180 R13 2004-01-01 330 0 .04 781,102 0 .053 255 26

Syrtis R11 2003-11 300 R13 2004-01-01 330 0 .01 112,353 −0 .014 315 26

Syrtis R14 2004-02-26 0 R17 2004-04-28 30 0 .01 42,891 −0 .013 15 27

Syrtis R17 2004-04-28 30 R19 2004-07-04 60 0 .02 78,026 0 .024 45 27

Syrtis S10 2005-09-24 300 S12 2005-11-18 330 0 .015 15,058 −0 .018 315 27

Syrtis S10 2005-09-25 300 S16 2006-03-16 390 0 .015 34 8,74 8 0 .017 345 27-28

Syrtis S12 2005-11-18 330 S14 2006-01-13 360 0 .02 95,365 −0 .024 345 27

( continued on next page )


Table 1 ( continued )

Location Before Before Before After After After Threshold Area Mean Estimated Mars

mosaic mosaic mosaic mosaic mosaic mosaic in km

2 change Ls year

phase start date center Ls phase start date center Ls I/F

STB E03 2001-04-15 150 E05 2001-06-16 184 0 .03 228,918 −0 .034 167 25

STB S02 2005-01-18 150 S04 2005-03-16 180 0 .03 24 8,74 9 −0 .036 165 27

Oxia R11 2003-11 300 R13 2004-01-01 330 0 .012 45,594 0 .017 315 26

Oxia S10 2005-09-24 300 S12 2005-11-18 330 0 .018 29,834 0 .023 315 27

Oxia A13 20 0 0-03-29 330 E13 2002-02-13 330 0 .01 5160 0 .012 330 24-25

Oxia E13 2002-02-13 330 R13 2004-01-01 330 0 .0 0 01 55,211 0 .004 330 25-26

Oxia R13 2004-01-01 330 S12 2005-11-18 330 0 .02 48,999 0 .025 330 26-27

Oxia A13 20 0 0-03-29 330 S12 2005-11-19 330 0 .015 81,080 0 .023 330 24-27

Oxia A13 20 0 0-03-30 330 S12 2005-11-20 330 0 .015 34,678 −0 .019 330 24-27

Nilosyrtis A01 1999-06-06 150 E03 2001-04-15 150 0 .015 41,663 −0 .018 150 24-25

Nilosyrtis E03 2001-04-15 150 R03 2003-03-03 150 0 .015 96,488 −0 .018 150 25-26

Nilosyrtis R03 2003-03-03 150 S02 2005-01-18 150 0 .035 37,205 −0 .040 150 26-27

Table 2

Predicted friction velocities in the study sites from the NASA Ames GCM.

Listed are predictions from the NASA Ames global circulation model of the local wind friction velocities at each of the study sites.

The simulations in this study utilized a grid spacing of 5 º of latitude by 6 º of longitude with a temporal resolution of 1.5 H, saving

output throughout a full martian year. The maximum binned values were drawn from 36 bins of 10 º Ls .

Location Maximum binned values All values in this grid point set Number of

Mean u ∗ (m/s) Median u ∗ (m/s) Range of u ∗ (m/s) Mean u ∗ (m/s) Median u ∗ (m/s) grid points

Nilosyrtis 0 .73 0 .81 0 .29–1.15 0 .29 0 .25 3

Alcyonius 0 .79 0 .77 0 .52–1.11 0 .37 0 .34 1

Syrtis Major 0 .96 0 .97 0 .46–1.41 0 .48 0 .46 24

Amazonis 0 .78 0 .76 0 .59–1.03 0 .35 0 .32 2

Tharsis 1 .11 1 .11 0 .62–1.69 0 .43 0 .56 63

Solis 1 .06 1 .03 0 .50–1.98 0 .51 0 .48 48

S. Trop. Bndry 0 .94 0 .86 0 .50–1.71 0 .46 0 .39 8

Oxia Bndry. 0 .85 0 .85 0 .46–1.21 0 .39 0 .36 2

Table 3

Predicted wind stress in the study sites from the NASA Ames GCM.

Listed are predictions from the NASA Ames global circulation model of the local wind stresses at each of the study sites, calculated as the

product of the atmospheric density times the square of the friction velocity.

Location Maximum binned values All values in this grid point set Number of

Mean τ max (mPa) Median τ max (mPa) Range of τ max (mPa) Mean τ (mPa) Median τ (mPa) grid points

Nilosyrtis 15 16 1–36 3 1 3

Alcyonius 13 12 5–24 3 2 1

Syrtis Major 12 11 2–28 4 3 24

Amazonis 12 11 8–18 3 2 2

Tharsis 13 12 3–28 4 3 63

Solis Planum 11 10 3–40 3 2 48

S. Trop. Bndry. 11 9 3–35 4 2 8

Oxia Bndry. 15 14 4–31 4 2 2

G

l

o

t

e

t

T

p

2

s

e

c

e

h

o

e

b

s

i

a

L

m

e

s

t

a

t

p

s

i

t

c

b

c

p

t

The most significant meteorological event witnessed by Mars

lobal Surveyor was the global dust storm of 2001, lasting from

ate June to late September. The storm started in Hellas and soon

bscured all of the planet’s surface with airborne dust except for

he poles and the peaks of the tallest volcanoes (e.g., Strausberg

t al., 20 05; Cantor, 20 07 ). A storm of similar magnitude was not

o be seen again on Mars until 2007, after MGS ceased operations.

he storm produced many abrupt surface changes including an ap-

arent brightening of Hellas itself, as seen in the “2002 minus

0 0 0 (red)” panel in Fig. 2 . Hellas darkened again during the sub-

equent years, beginning at the center and later spreading to the

dges. However, Hellas and the lower Valles Marineris are compli-

ated by the appearance of afternoon hazes, due to their low el-

vation. We therefore defer discussion of Hellas in favor of other,

aze-free targets.

Fig. 3 shows the temporal distribution of the surface changes

bserved from 1999 to 2006. This figure shows that deposition and

rosion of dust went on during all seasons throughout the year,

ut the largest and most conspicuous changes took place in the

outhern spring, mostly between Ls 210 º and 240 º.

The spatial distribution of surface changes from 20 0 0 to 20 09

s shown in the upper panel of Fig. 4 , calculated by summing the

bsolute values of difference images from one year to the next, at

s 330 ° (the right side panels of Fig. 2 ). A small bias in the MARCI

inus MOC subtraction, possibly caused by the bandpass differ-

nce discussed in Section 2.1 , was corrected by subtracting a con-

tant to make the mean pixel value equal to zero. Bright areas in

his map changed greatly in albedo over this period whereas dark

reas changed little. This map is full of artifacts (stripes) and me-

eorological phenomena (clouds and hazes) but it gives some com-

arative measure of the magnitude and primarily the frequency of

urface changes on Mars over this period. It is qualitatively sim-

lar to a map of just the frequency of the changes, calculated by

hresholding the difference images and adding up the number of

hanges.

The lower panel of Fig. 4 shows the locations of the regions to

e described in more detail below. These sites were chosen be-

ause they display a variety of temporal behaviors ranging from

rogressive changes through episodic dust deposition and erosion,

o quasi-continuous changes in the case of Solis Lacus. Progressive


Fig. 2. Global mosaics and differences at Ls 330 °. These mosaics of MOC and MARCI images show changes in the appearance of Mars from 20 0 0 to 20 09. They show the

portion of the planet from 90 ° S to 60 ° N during late northern winter. The changes are shown in the panels on the right, calculated by subtracting each red filter mosaic

from the next. Bright areas in the difference images are regions that brightened from one year to the next, whereas dark areas are regions that darkened. The difference

images accentuate the artifacts in the mosaics, such as the stripes from top to bottom along the orbit tracks. Changes in clouds and hazes are suspected in Hellas, the lower

Valles Marineris, and near the summit of Arsia Mons. Values of the differences in I/F portrayed range from −0.05 (black) to + 0.08 (white) in 20 02–20 0 0, −0.05 to + 0.10 in

20 04–20 02, −0.03 to + 0.03 in 20 05–20 04, and −0.10 to + 0.07 in 20 09–20 05. (For interpretation of the references to color in this figure legend, the reader is referred to the

web version of this article.)


Fig. 3. Seasonal distribution of surface changes 1999–2006. Points that plot above

zero on the vertical axis are new dust deposits, and points below zero represent

dust erosion. The area affected by each change is represented by the size of the

bubble. The horizontal position of the bubble is the midpoint of the interval be-

tween the successive observations that detected the change ( Table 1 ), with a timing

uncertainty of typically 30 º Ls. The duration of the events is not known and is not

represented. The largest and most conspicuous changes took place in the southern

spring, mostly between Ls 210 º and 240 º.

c

m

t

t

r

M

b

a

f

d

e

4

4

E

S

w

M

r

d

m

o

t

w

w

d

(

F

f

V

i

hanges that accumulated from one Mars year to the next were

ost obvious along moving albedo boundaries at Utopia Planitia,

he Southern Tropical Band, and western Oxia. Episodic changes

hat took place during local and global dust storms and sometimes

ig. 4. Map of surface changes and locations. The top panel is a map of surface changes f

rom one year to the next, at Ls 330 ° (the right side panels of Fig. 2 ). Bright areas in this

alues of the cumulative absolute changes in I/F portrayed range from 0.01 (black) to 0.37

n this paper.

eversed soon afterwards were seen in both dark terrain (Syrtis

ajor) and bright terrain (Amazonis), showing similar temporal

ehavior in areas of differing albedo and surface cover. The high

ltitude equatorial regions Tharsis and Solis Lacus had the most

requent and conspicuous changes on Mars during the MGS era,

isplaying quasi-continuous changes that took place in multiple

vents within a single Mars year.

. Progressive changes

.1. Southern Tropical Band

The “Southern Tropical Band” is the dark region around 180 °, 20 ° S that includes the classical albedo features Cimmeria and

irenum. It is the largest of 3 regions discussed in this section

here dust clearing took place between Viking and the arrival of

GS that continued during the period of MGS observations. This

egion was mantled by dust during the Viking era but had largely

arkened again by the time MGS observations began (see Supple-

entary Fig. 1). The region continued to darken during the tenure

f MGS, and the timing of these changes is an important clue to

he cause of the changes.

A rapidly moving albedo boundary was seen advancing north-

ards in this region. This boundary had a distinctive appearance,

ith a bright fringe of freshly deposited dust separating the

ark terrain to the south from the brighter terrain to the north

Fig. 5 , see also Supplementary Fig. 3 ). Interestingly, the bright

rom 20 0 0 to 2009, calculated by summing the absolute values of difference images

map changed greatly in albedo over this period whereas dark areas changed little.

(white). The lower panel shows the locations of albedo changes discussed in detail


Fig. 5. Advancing albedo boundary in the Southern Tropics. A fast moving albedo boundary was seen advancing northwards in this region (arrows), driven by southern

winter frontal storms. A bright fringe separating the dark terrain to the south from the brighter terrain to the north advanced ahead of the dark terrain. Another, fainter

albedo boundary is seen roughly parallel to the Southern Tropical Band but hundreds of kilometers to the north, where bright dust drapes the edge of the southern highlands

and the start of the northern plains. Grid spacing is 15 ° (889 km at the equator). Values of the differences in I/F portrayed range from −0.05 to + 0.08 in 20 02–20 0 0, −0.05

to + 0.10 in 20 04–20 02, and −0.03 to + 0.03 in 20 05–20 04.


f

i

i

a

2

o

1

8

m

p

o

M

T

t

d

t

o

s

B

t

p

w

i

b

l

w

d

o

e

d

s

r

m

2

T

2

s

t

t

s

o

s

a

a

t

3

i

t

s

w

4

e

b

a

t

t

d

F

t

r

d

Fig. 6. Surface changes and dust storms in the Southern Tropical Band. The tim-

ing of surface changes in the Cimmeria/Sirenum region corresponds exactly to the

frontal dust storms spotted by Wang and Richardson (2015) . Bubble size represents

the areas affected by the surface changes and the dust storms, respectively. Bubbles

are labeled with the Mars Year of the event.

o

b

e

s

s

a

e

(

e

t

c

3

N

s

h

e

w

w

n

i

t

d

o

s

p

f

w

ringe advanced ahead of the dark terrain. The changes took place

n two episodes on non-subsequent years, so the Southern Trop-

cal Band appears to be an example of recurring changes which

ccumulate over the years. The first darkening took place in MY

5 between April and June 2001 (Ls 150 ° to 180 °), just before the

nset of the global dust storm. The dark terrain advanced up to

20 km northwards during this period across a region more than

00 km in extent. The following year saw a brightening of the

argin of the newly darkened terrain as a bright fringe reformed,

roducing a higher albedo band typically 30 km in width ahead

f the shifted albedo boundary. The next advance took place on

Y 27 between January and March 2005 (again, Ls 150 ° to 180 °).he albedo boundary shifted farther northwards a distance of up

o 150 km, and a new bright band formed to the north of the

arkened terrain (bottom right panel of Fig. 5 ). The advance of

he albedo boundary continued through 2009 (bottom right panel

f Fig. 2 ), with further enlargement of the dark region occurring

ometime in the interval between 2005 and 2009.

There are 8 GCM grid points along the Southern Tropical

oundary, ranging from 171–207 º E, −32.5 to −22.5 ºN. Wind fric-

ion velocities here are relatively high, however wind stresses are

redicted to be relatively low. There is a distinct seasonality to the

ind regime here, with weak southeasterly winds blowing dur-

ng the southern winter and much stronger westerly to NW winds

lowing during southern summer (these are midlatitude wester-

ies). The period from Ls = 150–180 º is transitional from a southern

inter SE-dominating wind pattern to a southern spring westerly-

ominating wind pattern.

The GCM alone does not give any indication that this time

f year would produce erosional winds, suggesting that a more

pisodic weather pattern may be responsible for the removal of

ust in this region. Wang and Richardson (2015) identified a dust

torm track in this area, which they called the Cimmeria/Sirenum

oute. They cataloged five regional dust storms that formed and

oved northward along this path from 2001 through 2010 (MYs

5 through 30), each of which occurring between Ls = 152–173 º.he first observed storm took place from May 26 through June

, 2001 (in MY 25), coinciding with the first observed northward

hift of the Southern Tropical Band (see Fig. 6 ). The second and

hird storms occurred from February 5 through 13 and March 2

hrough 10 of 2005 (in MY 27), respectively, coinciding with the

econd observed northward shift of the albedo boundary. The lack

f advancement during MY 26 is consistent with the lack of ob-

erved dust storms along this pathway over the same time period,

nd the observed continuing advancement of the albedo bound-

ry through 2009 may have been caused by another storm that

ook place in 2008 (MY 29). A fifth storm occurred in 2010 (MY

0), suggesting that this is a common corridor for dust storms. It

s significant that the Cimmeria/Sirenum path is taken by most of

he southern hemisphere storm sequences with frontal structure,

imilar to erosive dust storms in the northern hemisphere that we

ill examine next.

.2. Oxia

Another rapidly moving albedo boundary was found in west-

rn Oxia Palus, near longitude 340 ° E, latitude 10 º to 30 º N. The

oundary advanced significantly between the Viking era and the

rrival of MGS (Supplementary Fig. 1; Geissler, 2005 ). MOC moni-

oring observations here show that the boundary shifted towards

he east and south each year as the western dark terrain (Aci-

alia) expanded southeastwards ( Fig. 7 , see also Supplementary

ig. 4). The advance of the boundary continued after the end of

he MGS mission, at a rate of up to 13 km/martian year in more

ecent MRO MARCI observations ( Mukherjee and Geissler, 2010 ). A

istinctive feature of the Oxia Palus albedo boundary is the fringe

f locally high albedo ahead of the boundary, which moved and

roadened as the boundary encroached on the bright terrain to the

ast. We suspect that the fringe is an apron of fresh dust that was

wept off of the advancing dark terrain and deposited downwind,

imilar to that observed in the Southern Tropical Band discussed

bove.

Fig. 7 shows that changes accumulated along this boundary

ach year, however the changes that took place from 20 0 0 to 2002

MY 24-25) were too small to be measured in successive differ-

nce images. Only two conspicuous changes were seen such that

he area of new dust deposits could be measured ( Fig. 8 ). These

hanges took place between November 20 03 and January 20 04 (Ls

00 to Ls 330 MY 26) and again between September 2005 and

ovember 2005 (Ls 300 to Ls 330 MY 27).

There are two GCM grid points along the Oxia boundary. Wind

peeds are moderate here, however the wind stresses are relatively

igh because of the low elevation ( Table 3 ). Typical of the north-

rn plains, there is a distinct seasonal pattern in which northern

inter winds are stronger than northern summer winds. The peak

inter winds rotate clockwise diurnally, blowing from the NNW to

orth in the afternoon and swinging around to blow from the NE

n the evening. Both of these wind directions match the orienta-

ions of wind streaks (e.g., Edgett, 2002 ) as well as the migration

irections of ripples ( Silvestro et al., 2011 ) in Arabia Terra, just east

f the Oxia boundary. The winter northwesterlies could be respon-

ible for the migration of the Oxia albedo boundary.

Another possible source of surface dust erosion could be the

resence of a well-populated storm corridor that produces many

rontal storms during northern fall and winter, which travel south-

ard from Acidalia to Chryse Planitia and often cross the equator


Fig. 7. Advancing albedo boundary in Oxia Palus. A bright fringe advanced ahead of the southeastwards moving albedo boundary between Acidalia and Oxia Palus (arrows).

Minor changes are seen in MY 25 but more significant changes took place in MY 26 and MY 27. Values of the differences in I/F portrayed range from −0.05 to + 0.08 in

20 02–20 0 0, −0.05 to + 0.10 in 20 04–20 02, and −0.03 to + 0.03 in 20 05–20 04.

w

p

a

a

( Wang et al . , 2005; Wang and Richardson, 2015 ). These storms may

be driven by the same northwesterlies just mentioned. The move-

ment of these storms is consistent with the southeastward migra-

tion of the Oxia albedo boundary and also with the southwards

expansion of Acidalia between longitudes 300 º E and 330 º E that

as observed between the Viking era and the arrival of MGS (Sup-

lementary Fig. 1).

Like other northern hemisphere frontal storms, those observed

long the Acidalia storm track by Wang et al. (2005) and Wang

nd Richardson (2015) occurred over two seasonal periods, with


Fig. 8. Surface changes and dust storms in Oxia. The surface changes in Oxia during

MY 26 and MY 27 could have been caused by frontal dust storms spotted along the

Acidalia storm track by Wang and Richardson (2015) .

r

3

b

t

b

e

e

d

b

r

w

t

4

f

G

T

t

t

d

s

t

a

M

a

a

l

a

d

l

F

d

s

t

g

a

w

c

M

d

i

h

c

N

n

r

t

w

p

m

i

c

a

1

m

s

w

f

d

r

s

s

b

d

e

d

i

g

(

L

d

t

t

w

r

m

a

N

a

fl

m

b

5

5

m

t

w

egional storms occurring from Ls = 183–245 º and again from Ls =05–20 º. The timing of the observed migration of the Oxia albedo

oundary is constrained to Ls = 300–330 º in MYs 26 and 27, al-

hough incremental changes took place in other years that cannot

e well constrained in time, but are visible in year-to-year differ-

nces ( Table 1 ). These surface changes took place during the north-

rn post-winter-solstice season (Ls 270 º to Ls 360 º) in which many

ust storms were sighted ( Fig. 8 ). These regional dust storms could

e responsible for dust lifting in these periods. In any case, the

egularity of storms in this area during periods of seasonal north-

esterly winds is consistent with the observed long-term shift of

he albedo boundary.

.3. Nilosyrtis/Utopia Planitia

The albedo patterns in western Utopia Planitia changed pro-

oundly over the twenty Earth years between Viking and Mars

lobal Surveyor (Supplementary Figs. 1 and 5A; Geissler, 2005 ).

he dark terrain in the north of the region is known as Nilosyr-

is in the west and Utopia in the east. The isolated dark patch

o the south is called Alcyonius or sometimes Thoth ( Fig. 4 ). The

ark terrain of Nilosyrtis/Utopia enlarged by more than a million

quare kilometers over the interval between the Viking observa-

ions (1976–1980) and the first MOC images from MGS (1999). The

lbedo of the affected area dropped by up to 50%. Images from

ariner 9 show that surface changes in this region began as early

s 1972. The albedo boundaries in this region are strikingly sharp

nd irregular in comparison to the sweeping, diffuse transitions

eft by wind streaks elsewhere on Mars. High resolution MOC im-

ges taken during local summer revealed that the dark terrain is

ominated by the tracks of dust-devils, particularly in the higher

atitudes from 45 ° to 65 ° north ( Geissler, 2005 ).

Monitoring by MGS shows ( Fig. 9 , see also Supplementary

ig. 5 ) that the northern albedo boundary (between the northern

ark regions of Nilosyrtis and Utopia and the bright plains to the

outh) continued to shift southwards over the 4 martian years be-

ween 1999 and 2005. Part of this advance was reversed by the

lobal dust storm of 2001, which reset the western portion of the

lbedo boundary (see black arrow in Fig. 9 ). The rate of south-

ards movement (7–19 km/Mars yr) is consistent with the total

hange seen from Viking to the arrival of MGS ( ∼150 km over 10

ars years), suggesting that the earlier albedo changes were pro-

uced by a similar process. In contrast, the albedo boundary divid-

ng the bright plains from the southern dark terrain in Alcyonius

as been stationary over the past few martian years. Why the Al-

yonius albedo boundary remains fixed in location while nearby

ilosyrtis advances is a mystery.

The timing of the surface changes is also mysterious, because

o discrete changes are seen during the summer months when the

egion is not obscured by polar haze. The prevalence of dust devil

racks led Geissler (2005) to suggest that erosion by dust devils

as responsible for the darkening. While such gradual processes

ossibly contribute to the movement of the albedo boundary, a

ore likely explanation is that the surface changes take place dur-

ng local winter, under the cover of the polar hood.

Three GCM grid points cover the albedo border in Nilosyrtis,

overing the area spanning 93–111 º E, 42.5–47.5 º N. The strongest

nd most persistent winds blow from the north and NW (towards

35–180 º) throughout most of the year, excluding northern sum-

er. These winds blow for most of the day, typically reaching peak

trength in the afternoon. During local summer, winds are much

eaker and typically blow from the west to WSW. Although the

riction velocities are relatively low, the wind stresses here are pre-

icted to be among the highest of the study sites. The persistent,

elatively strong winds from the N/NW are consistent with the ob-

erved southward progression of the albedo boundary in Nilosyrtis,

uggesting that prevailing winds could be responsible for removing

right dust cover (possibly with the aid of dust devils).

It is also notable that this boundary is often crossed by winter

ust storms that sweep southward through Utopia Planitia ( Wang

t al., 2005; Wang and Richardson, 2015 ). Like Acidalia and Arca-

ia, Utopia is a nexus for frontal storms that tend to concentrate

n low topographic regions ( Wang et al., 2005 ). Most of the re-

ional dust storms in this area spotted by Wang and Richardson

2015) took place between Ls 195 º and Ls 236 º, and again between

s 310 º and Ls 335 º ( Fig. 10 ). It is possible that seasonally repeating

ust storms are responsible for the particle entrainment and long

erm surface erosion observed here.

The pattern of predicted winds at Alcyonius is broadly similar

o that along the Nilosyrtis albedo border: northerly afternoon

inds are the strongest during most times of the year, with

elatively weaker westerly winds prevailing during northern sum-

er. Friction velocities from the GCM are comparable to, and

ctually slightly stronger than the predicted winds along the

ilosyrtis albedo boundary, however the predicted wind stresses

re somewhat lower. Neither northerly winds nor southwards

ushing dust storms are expected to enlarge Alcyonius’ northern

argin upwind, consistent with the observations that this albedo

oundary remained fixed in place.

. Episodic changes

.1. Amazonis

An area where dust was repeatedly deposited and then re-

oved is in northern Amazonis Planitia, on a smooth plain in

he northern lowlands at 190–210 ° E, 25–35 ° N. This is a region

here dust is stripped from the surface by giant dark streaks that


Fig. 9. Advancing albedo boundary at Nilosyrtis/Utopia Planitia and Stationary Boundary of Alcyonius. Seasonally repeating frontal dust storms advanced the dark terrain of

Nilosyrtis and Utopia Planitia in the north of the scene (white arrows), southwards towards the equator at a rate consistent with cumulative change from Viking to MGS

( Geissler, 2005 ). A portion of this boundary was reset during the global dust storm of 2001 (black arrow). The dark terrain at the center of the scene, Alcyonius, did not

spread to the north during this time period. Values of the differences in I/F portrayed range from −0.06 to + 0.08 in 2001–1999, −0.05 to + 0.07 in 20 03–20 01, and –0.04 to

+ 0.04 in 20 05–20 03.


Fig. 10. Dust storms and wind stresses in Nilosyrtis/Utopia. Northerly winds push winter frontal dust storms south across the albedo boundary while the region is shrouded

by the polar hood. The upper frame shows the timing of dust storms spotted by Wang and Richardson (2015) . The lower frame shows the maximum wind stress (binned to

10 º of Ls) and its corresponding wind direction in each of 3 GCM grid points.

t

t

f

F

“

l

“

t

b

s

m

w

F

i

b

i

t

i

f

m

T

3

w

t

E

a

o

t

d

2

(

(

b

y

w

t

i

l

f

e

2

z

s

b

s

p

s

fi

i

e

aper upwind like spire streaks in Tharsis (see Section 6.1 ), but ex-

end hundreds of kilometers in length in parallel sets that change

rom year to year ( Thomas et al., 2003 ; see also Supplementary

ig. 6B). Thomas et al. suggested that the even spacing of these

mesoscale linear streaks” was caused by longitudinal boundary

ayer rolls, spiral motions of air that blows downwind and cause

cloud streets” on Earth. The significance of the streaks is that they

ell us that freshly deposited dust is mobile and frequently eroded

y the winds, even in flat places and places where there are no

altating sand particles to kick the dust off of the surface.

The wind streaks are too small to be seen in the MOC global

osaics, but many changes in the distribution of dust in the region

ere observed from 1999 to 2006 ( Fig. 11 , see also Supplementary

ig. 6A). Three major brightening events took place, each time dur-

ng northern winter/early northern spring ( Fig. 12 ). The first event,

etween Ls 330 ° and Ls 0 ° in 20 0 0 (MY 24), spread bright dust

n northern Amazonis and partially coated the low albedo terrain

o the north, classically called Euxinius Lacus. A minor brighten-

ng of the area east of Euxinius was completed by Ls 330 ° of the

ollowing martian year (2002) but the preceding global dust storm

akes it difficult to discern exactly when the changes took place.

he second major brightening took place between Ls 270 ° and Ls

0 0 ° of 20 03 (MY 26), when dust was deposited to the south-

est of Euxinius. The last major dust deposition was observed be-

ween Ls 330 ° and Ls 0 ° in late 2005 and early 2006 (MY 27).

ach of these events was followed by rapid darkening of the south

nd west margins of the dust deposits, possibly by the formation

f mesoscale linear streaks. Smaller changes took place at other

imes, primarily concentrated in the season from Ls 210 ° to Ls 30 °.Also seen in Fig. 11 is a brightening in western Propontis, the

ark patch west of Euxine. This brightening took place in 2003 (MY

6) between Ls 180 ° and Ls 240 °. Propontis disappeared in 2009

Fig. 2 ) during a period of local dust storm activity at Ls 322–327 ° Lee et al., 2013 ) and has not reappeared since.

Two GCM grid points in this area show that peak winds here

low during the afternoon from the north throughout most of the

ear, strengthening in the winter. As in Nilosyrtis and Alcyonius,

inds are weaker in the summer, with the strongest blowing in

he morning from the west. Friction velocities are similar to those

n Nilosyrtis and Alcyonius, however wind stresses are somewhat

ower.

There is a storm track through Arcadia Planitia that sends

rontal storms southward through Amazonis Planitia during north-

rn fall and winter ( Wang et al., 2005; Wang and Richardson,

015 ). Comparison between the timing of surface changes in Ama-

onis and sightings of regional dust storms by Wang and Richard-

on in Arcadia does not show perfect correspondence ( Fig. 12 )

ut it does show a broad agreement between the seasons of both

urface changes and dust storms, with a peak in activity before

erihelion and a second peak in activity after the northern winter

olstice. Many smaller frontal storms form along this track, identi-

ed by Wang et al. (2005) , which may be responsible for deposit-

ng dust and brightening the surface, or removing dust and dark-

ning the surface. These seasonally repeating dust storms are also


Fig. 11. Fresh dust deposits in Amazonis. An area where dust was deposited onto an already dust-covered surface is in Amazonis Planitia, a smooth plain in the northern

lowlands where dust is deposited and then stripped from the surface each year by giant dark streaks that taper upwind. Propontis is the dark spot in the upper left of the

scene, and the dark band in the top right is called Euxinius Lacus. Fresh dust is blown into the region by frontal dust storms during northern winter, and removed again soon

afterwards. Dust devils quietly vacuum the surface during northern summer and fall. Values of the differences in I/F portrayed range from −0.06 to + 0.08 in 2001–1999,

−0.05 to + 0.07 in 20 03–20 01, and −0.04 to + 0.04 in 20 05–20 03.


Fig. 12. Surface changes and dust storms in Amazonis. Most of the surface changes

in Amazonis took place in northern winter, a season when several frontal dust

storms were spotted in the same region. Dust storm data from Wang and Richard-

son (2015) .

l

c

p

l

c

(

m

a

b

p

5

2

f

1

z

u

n

N

a

t

b

(

b

b

(

r

f

c

a

b

t

b

o

a

r

u

w

e

w

f

b

t

o

i

2

S

2

S

o

P

s

t

t

s

e

t

3

n

t

S

t

T

s

m

r

t

a

s

s

i

d

m

l

2

t

t

a

g

d

6

6

t

s

H

ikely to be the cause of the advance of the albedo boundary in Ar-

adia Planitia between the Viking era and the arrival of MGS (Sup-

lementary Fig. 1, north of the Southern Tropical Band between

atitudes 30 ° N and 60 ° N).

At other times of the year, the surface of Amazonis is quietly

leaned by dust devils, which are plentiful from Ls 35 ° to Ls 180 ° Cantor et al., 20 02, 20 06; Fisher et al., 20 05 ) and continue to be

onitored today by the HiRISE and CTX cameras on MRO ( Fenton

nd Lorenz, 2015 ). The episodic activity in Amazonis is punctuated

y local summer seasons when winds are light and a more gradual

rocess of erosion takes over.

.2. Syrtis Major

Syrtis Major is a broad low shield volcano ( Hiesinger and Head,

0 04 ), over 10 0 0 km across, at the center of a classical low albedo

eature known for historical changes in appearance (e.g., Capen,

976 ). Located adjacent to the bright Isidis impact basin, the dark

one is decorated by bright dust streaks along its western margin,

nchanged since Viking, that suggest that winds from the east and

ortheast have removed dust from the surface in the recent past.

umerous sand dunes have collected in local topographic lows

round the region ( Hayward et al., 2007 ), including those con-

ained within Nili Patera at the center of the volcano, confirmed

y MRO HiRISE to be among the most active of martian dunes

Bridges et. al, 2012 ). These dunes indicate that modern day winds

low sands from the east and northeast.

Syrtis was a typical example of episodic deposition followed

y episodic erosion during the period of MGS MOC observations

Figs. 13 and 14 ; see also Supplementary Fig. 7 ). In 1999, the ter-

ain north and west of Syrtis was dark and mostly clear of dust as

ar as Antoniadi crater. Only minor changes took place in 20 0 0, in-

luding subtle brightening in the east half of Syrtis Major Planum

nd darkening in Isidis Planitia and in Antoniadi and Baldet craters

etween Ls 330 ° and 0 ° of MY 24. The region continued to main-

ain its appearance until late the next year, when it was blanketed

y the 2001 global dust storm (largely prior to the top right frame

f Fig. 13 ). The storm left a bright dust coating that mantled an

rea of more than a million square kilometers. Most of the dust

emained on the surface until it was abruptly blown away in Jan-

ary 2002 (MY 25 Ls 300–320 °) during a time span of only a

eek ( Cantor and Edgett, 2002 ). Not all of Sytis was cleaned in the

vent; large areas of the west and north of Syrtis remained coated

ith dust, including Antoniadi crater. Two minor cleaning events

ollowed in late 2002, MY 26 between Ls 60 ° and 90 ° and again

etween Ls 90 ° and 120 °. Only minor changes were seen in subsequent MOC observa-

ions. In 2003, minor dust deposition took place in the eastern half

f Syrtis and along the western margin, while Isidis darkened dur-

ng the same interval between Ls 180 ° and 300 ° MY 26. In late

005 (MY 27 after Ls 300 °), dust was deposited to the north of

yrtis in the Baldet crater region that was removed again by early

006 (Ls 0 °). MARCI observations show that the western margin of

yrtis had darkened by 2009 (MY 29), after the global dust storm

f 2007.

Ayoub et al. (2014) measured sand fluxes of barchans in Nili

atera, finding that the highest values occurred during southern

ummer from Ls = 207–355 º, but that winds blew ripples towards

he WSW throughout the year. Some of their results suggest that

he wind varies in strength on timescales of weeks or less, con-

istent with the idea that dust erosion in Syrtis Major may be

pisodic. Michaels (2011) similarly found higher wind stresses from

he northeast during southern spring and summer (Ls = 210 º and

00 º, respectively) in mesoscale model simulations.

There are 24 grid points in the GCM on Syrtis Major, span-

ing 51–75 º E, −2.5–27.5 º N. The Ames GCM predicts that fric-

ion velocities are high in this region but wind stresses are low.

outherly winds blow during the morning throughout the year, but

hat evening winds during southern summer blow from the east.

hese easterly winds are responsible for creating the bright wind

treaks that are visible over much of Syrtis Major, as well as the

igration of ripples observed on the Nili Patera dunes. The occur-

ence of easterly winds during southern summer is consistent with

he seasonal peak ripple migration and modeled winds mentioned

bove. It’s also consistent with the observed timing of dust ero-

ion from Ls = 300–320 º after the end of the 2001 GDS. Given the

trong winds inferred from high ripple migration rates in this area,

t is likely that loose sand on the surface erodes away any dust

eposits that have accumulated, particularly during southern sum-

er.

Dust was deposited in this region primarily during the perihe-

ion seasons of MY 25 (the global dust storm of 2001) and MY

6. There are two possible corridors for dust transportation into

his region. The first is along the northern winter frontal storm

racks from Utopia (blown by easterly winds), and the second is

long storm tracks from Hellas to the south, the source of the 2001

lobal dust storm and a region that is rife with dust storm activity

uring the perihelion season ( Fig. 14 ).

. Quasi-continuous changes

.1. Tharsis

A less conspicuous but more frequently changing region than

he above-mentioned Syrtis Major region is the Tharsis Rise. Thar-

is is a vast plateau bounded by three giant volcanoes to the west.

ere, bright dust is deposited onto and removed from a region that


Fig. 13. Syrtis Major. Episodic deposition followed by episodic clearing was seen at Syrtis, which was blanketed by the 2001 global dust storm which left a bright dust

coating that mantled an area of more than a million square kilometers. The dust was abruptly blown away in January 2002 (Ls 300–320 °). Not all of Sytis was cleaned in

the event; large areas of the west and north of Syrtis remain coated with dust as of this writing.

(

j

s

(

o

b

i

i

t

H

f

t

c

t

g

E

t

o

e

t

p

t

r

s

o

n

r

t

b

r

is already covered with dust, making the changes harder to spot.

The fresh coatings of mobile dust are only slightly brighter than

the settled dust deposits. The observations are also complicated by

seasonal water ice clouds that appear at the peaks of the giant vol-

canoes, obscuring our view of the surface.

The extreme topography of Tharsis generates strong upslope

(anabatic) winds during the day, leading to the formation of af-

ternoon condensate clouds at the summits of the giant volcanoes,

particularly during southern winter when the atmosphere is cool

and the relative humidity is high. The same topography produces

strong downslope (katabatic) winds at night. Dust deposited on the

giant volcanoes is quickly swept down the slopes by these kata-

batic winds, a process first seen in Mariner 9 images ( Sagan et al.,

1974 ). Large, dark “spire” streaks that taper upwind are formed as

the loose dust is eroded by nighttime katabatic winds ( Thomas

et al., 1981; Lee et al., 1982; Toyota et al., 2011 ). Many locations

at the bases of the giant volcanoes show downslope-trailing dust

deposits collected behind crater rims, boulders, and other positive

topographic obstacles that may be fallout from the erosion above.

The most dramatic change documented by MOC in the albedo

of Tharsis took place during the global dust storm of 2001

( Fig. 15 , see also Fig. 2 ). Surprisingly, Tharsis appeared darker af-

ter the dust storm, suggesting that dust was stripped away from

the region except for conspicuous localized patches to the east of

the giant volcanoes, particularly on Arsia Mons and Pavonis Mons.

Tharsis brightened again during subsequent years, but the process

was far more complicated than shown by the year-to-year snap-

shots in Fig. 15 . It was punctuated by many abrupt albedo changes

( Fig. 16 ), including frequent changes in the central Tharsis region

east of Pavonis Mons, northern Tharsis east of Ascraeus Mons, and

western Tharsis north and west of Arsia Mons. These regions dark-

ened abruptly during northern hemisphere spring equinox seasons

Ls 330 ° to 30 °) in 20 0 0 (MY 24), 20 04 (27) and 2006 (28). Ma-

or brightenings of Tharsis took place during southern winter and

pring, once in early 2001 (MY 25 Ls 120 °) and again in late 2002

MY 26 Ls 120 °). In 2009, two years after the global dust storm

f 2007, MARCI images at MY 29 Ls 330 showed central Tharsis

righter than its appearance at a similar season in 2005.

High resolution images of places in Tharsis where dust is be-

ng removed show dark erosional streaks that taper upwind and

gnore small-scale topography, similar to but much smaller than

he spire streaks on the flanks of the giant volcanoes. For example,

iRISE images ESP_014010_1800 and ESP_031917_1800 show the

ormation of dark streaks that taper generally upslope, suggesting

hat the dust was blown away by nighttime katabatic winds. These

hanges took place during periods without global dust storms be-

ween 2009 and 2013, indicating that the process of dust erosion

oes on even during “quiet” years.

There are 63 GCM grid points over Tharsis, spanning 231–273 º, −12.5–+ 32.5 º N. Tables 2 and 3 show that Tharsis is a rela-

ively windy place in comparison to the rest of Mars, but because

f its high elevation, wind stresses here are only moderate. How-

ver, there is substantial local topography in the Tharsis region

hat is not fully resolved by the GCM. The local topography could

roduce strong topographically-driven slope winds that may add

o the mean winds predicted by the GCM to produce the spo-

adic dust-lifting events. The mean wind pattern is relatively con-

istent throughout the year, with winds rotating around clockwise

nce a day, peaking in either the morning or afternoon. During

orthern summer, peak winds blow with a westerly component,

otating from southerly to northwesterly as morning progresses

o afternoon. In northern winter the strongest winds most often

low from the west to northwest. These westerly winds are likely

esponsible for the surface darkening after the 2001 global dust


Fig. 14. Surface changes and dust storms in Syrtis. The dust injected into the Syrtis Major region during the global dust storm of MY 25 arrived from the Hellas region

along the track of seasonally repeating dust storms. Dust can also reach this region during northern winter from the Utopia storm track. Dust storm data from Wang and

Richardson (2015) .

s

n

o

b

6

t

t

b

s

c

P

h

e

C

S

b

1

t

f

d

t

s

(

o

b

c

g

s

t

l

o

e

c

o

0

L

t

a

S

i

a

(

t

T

d

s

c

e

torm, which left bright dust patches in the lee of Arsia and Pavo-

is Montes. The modeled pattern of winds is consistent with the

bserved pattern of dust removal: regardless of the season, dust

rought into Tharsis would soon be removed by its winds.

.2. Solis Lacus

The classical low albedo feature known as Solis Lacus occupies

he center of a vast high plain south of the Valles Marineris ex-

ending more than 30 ° in longitude and 20 ° in latitude. Ringed

y mountains to the south and canyons to the north, the plateau

its at elevations from 2500 m to 4300 m above datum and in-

ludes Syria Planum, Sinai Planum, Thaumasia Planum and Solis

lanum. Made up of Hesperian aged volcanic plains, the region has

igher than average thermal inertia (240 to 340 MKS units; Mellon

t al., 2002 ) and inferred rock abundance (13% to 33%; Nowicki and

hristensen, 2007 ). Dozens of small shield volcanoes are found in

yria Planum, in the region’s northwest ( Baptista et al., 2008 ).

Surface albedo changes in this region have been documented

y ground-based observers since the 1920 s ( Slipher, 1962; Dollfus,

965 ), and the global dust storms of 1973 and 1988 were observed

o start in Solis Lacus ( Martin, 1976; JALPO, 2003 ). In addition to

requent dust cloud activity, Solis is the location of numerous dust-

evil plume sightings ( Biener et al. 2002; Fisher et al. 2005; Can-

or et al. 2006; Stanzel et al., 2008; Reiss et al., 2014 ). Significant

urface changes were observed by the Viking spacecraft in 1977

Lee, 1986 ) and during the interval between Viking and the arrival

f Mars Global Surveyor in 1999 ( Geissler, 2005 ). Continuous or-

ital monitoring since then, at first by MGS and later by Mars Re-

onnaissance Orbiter, has shown that the Solis Lacus region under-

oes nearly continual change, with drastic alterations during dust-

torm seasons punctuating more subtle changes at other times of

he year ( Fig. 17 , see also Supplementary Fig. 8 ).

The MOC observations show that the chaotic changes in So-

is have a simple pattern to their timing ( Fig. 18 ). Fresh coatings

f dust appeared in Solis every southern spring season. Cleaning

vents during the rest of the martian year caused the dramatic

hanges from Ls 270 ° to 150 ° in each year from 1999 to 2006. Six

f the 14 cleaning events recorded took place between Ls 270 ° and

°. Solis is not active continuously, remaining relatively quiet from

s 150 ° to 210 °, but it is the most conspicuously changing place on

he planet.

High resolution HiRISE and CTX images of the region show

generally rocky surface with thin, transient coatings of dust.

urfaces scoured clean of dust, showing rocks and boulders and

solated transverse aeolian ridges (TARs, deflated sediments that

re not presently active) are seen to the south in Solis Planum

e.g. PSP_006797_1545, ESP_017056_1475, ESP_018058_1555), to

he north in Sinai Planum (PSP_008788_1655), and to the east in

haumasia Planum (PSP_007693_1555). Surfaces with coatings of

ust that are too thin to show topographic expression at HiRISE

cales are suspected to be responsible for the short-term albedo

hanges, as dust is deposited or removed by the winds. In sev-

ral dust-coated areas, the dust is being actively vacuumed by dust


Fig. 15. Changes on the Tharsis Rise. The giant volcano Ascraeus Mons is near the center of these views of high elevation Tharsis. The region drastically darkened during

the global dust storm of 2001, and brightened and darkened again frequently during subsequent years. Values of the differences in I/F portrayed range from −0.05 to + 0.08

in 20 02–20 0 0, −0.05 to + 0.10 in 20 04–20 02, and −0.03 to + 0.03 in 20 05–20 04.


Fig. 16. Surface changes and wind stresses in Tharsis. Although the Tharsis region has the strongest wind friction velocities of any of the locations of surface changes that we

studied, the lower atmospheric density at this high elevation means that wind stresses here are only moderate. The lower frame shows the maximum wind stress (binned

to 10 º of Ls) and its corresponding wind direction in each of 63 GCM grid points.

d

l

d

S

i

s

H

−

t

w

t

i

w

T

a

t

f

fl

t

R

c

h

t

g

d

d

7

t

b

d

d

a

a

j

s

a

r

c

b

t

e

n

h

evils that leave behind dark tracks, such as those seen in So-

is Planum (PSP_006125_1525) and Syria (PSP_007022_1655). Fresh

ust deposits in the form of bright wind streaks are found in

yria Planum (PSP_008749_1660, CTX B01_010028_1634), suggest-

ng that dust is supplied to the region from the northwest. No

and dunes are seen in the region, consistent with the findings of

ayward et al. (2007, 2014 ).

There are 48 GCM gridpoints spanning the area 255–303 º E,

42.5–−12.5 º N. Tables 2 and 3 show that the friction veloci-

ies here are high, comparable in magnitude to those in Tharsis,

hereas wind stresses are relatively low at this elevation. Similar

o Tharsis, northern summer peak winds blow during the morn-

ng and afternoon from the west to WSW, but northern winter

inds are much more variable in direction and local time. As with

harsis, high speed winds throughout the year and winds gener-

ted by local topography likely sweep away the thin dust deposits

hat accumulate in the area. The relatively quiet period observed

rom Ls = 150–210 º may have more to do with changes in dust in-

ux than it does with wind strength, as the wind does not appear

o weaken at that (or any) time of the year.

The few regional dust storms spotted in Solis by Wang and

ichardson (2015) do not appear to correlate well with surface

hanges, except perhaps in MY 27 ( Fig. 18 ). However, Solis has a

istory of local dust storms during southern spring and summer

hat originated near Claritas Fossae in Syria Planum and grew into

lobal dust storms in 1973 and 1977 (e.g. Zurek, 1982 ). Such local

ust storms could have played a role in bringing dust into Solis

uring perihelion seasons over the years of the MGS observations.

. Discussion

Mars continues to maintain the same overall global albedo pat-

ern it has had for centuries in spite of significant surface changes

ecause dust is repeatedly deposited and removed over annual and

ecadal timescales. Fig. 3 shows that dust erosion roughly balances

ust deposition over the period of MGS observations. The average

lbedo pattern is produced by a continuous battle between dust

nd wind. Albedo boundaries can be abruptly reset during ma-

or global dust storms, and then seasonally repeating processes re-

ume restoring the familiar appearance of the planet. Dust tends to

ccumulate in the low latitudes of Mars, between many low albedo

egions such as Syrtis Major, because of seasonally repeating pro-

esses that sweep dust towards the equator. Most of the albedo

oundaries on Mars appear stationary over the short duration of

he orbital monitoring, but the albedo boundaries in the south-

rn hemisphere that appear to be persistently changing are moving

orthwards, while persistently moving boundaries in the northern

emisphere are creeping towards the south.


Fig. 17. Solis Lacus observed by MGS MOC over 4 consecutive Mars years. This high plateau was the site of frequent and conspicuous albedo changes, as dust was repeatedly

deposited during the perihelion season and abruptly blown away again by strong winds during much of the rest of the year.

a

t

e

d

a

t

(

n

t

o

s

8

g

c

M

i

b

s

p

a

Mars is still cleansing itself from the effects of a major global

dust storm that took place in the 1970 s or possibly earlier, be-

fore orbital monitoring began. Dust that was deposited at high

latitudes is being relentlessly pushed towards the equator along

the Southern Tropical Band, the boundary between Nilosyrtis and

Utopia, and in Arcadia and Acidalia. All of these locations corre-

spond to the paths of seasonally repeating dust storms identified

by Wang and Richardson (2015) . These storms are generated fre-

quently by wintertime circumpolar cyclones and tend to be con-

centrated at these locations by topography ( Wang et al., 2005 ). The

strong winds associated with dust storms are most likely to be ca-

pable of lifting dust which will later be deposited wherever the

storms dissipate. The timing of the observed surface changes cor-

responds exactly to the sightings of dust storms in the Southern

Tropical Band, and it is seasonally consistent in Oxia, Nilosyrtis,

and Amazonis. We infer that these seasonally repeating, episodic

dust storms probably contribute most to the progressive changes

that are seen on Mars from the 1980 s to today. Erosion of dust by

dust devils during local summer and during the perihelion season

may also contribute to the concentration of dust at the equator, but

to a much lesser extent, a conclusion previously reached by Szwast

et al. (2006) .

The dust storm tracks of Wang and Richardson (2015) show that

the equatorial region of Mars is a net sink of dust transportation,

the end of the line for dissipating dust storms from both the north

nd the south. The episodic changes that took place in the equa-

orial regions Syrtis, Tharsis, and Solis might reflect the ability of

ach area to shed newly deposited dust. These 3 regions are pre-

icted by the GCM to experience the highest friction velocities of

ny of the sites of surface changes that we studied ( Table 2 ), al-

hough wind stresses are reduced because of their high elevations

Table 3 ). It may be that local topography in these places that is

ot resolved by the GCM results in winds that are even stronger

han predicted by the global model. In any case, we infer from the

bservations that dust accumulations in these locations are rapidly

hed by winds.

. Conclusions

The continuous monitoring observations of MOC and MARCI

ive us a unique perspective into the timing of martian surface

hanges, while the high resolution views from MOC NAC and later

RO CTX and HiRISE observations provide insight into the phys-

cal mechanisms of albedo change. The findings of this study can

e summarized as follows:

1. Martian surface changes take place on a variety of time

cales.

High elevation equatorial regions such as Tharsis and Solis ex-

erience almost continual surface changes, as thin coatings of dust

re deposited and quickly removed. These albedo changes took


Fig. 18. Surface changes and dust storms in Solis. The frequent surface changes bore

little correspondence to the dust storms sighted in Solis and Argyre (southeast of

Solis), except perhaps in MY 27. Dust was deposited in Solis each southern spring

season and removed again each year by Ls 150 º. Dust storm data from Wang and

Richardson (2015) .

p

v

e

V

p

p

a

p

c

o

T

c

i

2

s

a

i

f

a

w

s

(

s

b

S

o

w

r

c

b

a

s

t

R

e

s

h

a

d

t

c

m

n

t

M

a

i

t

e

t

d

d

A

d

t

p

t

A

N

d

P

f

r

S

f

R

A

B

B

B

B

C

C

C

lace multiple times within each martian year of the MGS obser-

ations. At the opposite extreme, it takes decades to reverse the

ffects of a major global dust storm such as occurred before the

iking observations. Many of the largest surface changes that took

lace between Viking and the arrival of MGS continued during the

eriod of MGS and MARCI monitoring, in incremental changes that

ccumulated year by year.

2. Albedo changes take place on both dark and bright terrain.

Conspicuous albedo changes take place where bright dust is de-

osited on or removed from dark rocky surfaces such as Solis La-

us and Syrtis. More subtle changes occur where dust is deposited

n already bright, dust-covered surfaces such as Amazonis and

harsis. The consistency of the MOC data set makes these subtle

hanges easy to spot, whereas they were suspected but not proven

n the changes that took place between Viking and MGS ( Geissler,

005 ).

Freshly deposited dust is mobile even on already dust-covered

urfaces such as Amazonis, where it appears as bright patches that

re easily moved again by both seasonal dust storms and dust dev-

ls. Sand grains are not required to eject dust from the martian sur-

ace; freshly deposited dust is quickly removed from surfaces such

s Amazonis where sand grains are absent.

3. Surface changes are produced by both dust storms and strong

inds.

Many of the largest and most conspicuous surface changes ob-

erved by MGS took place during the global dust storm of 2001

MY 25). A large part of Syrtis Major was buried by dust, the Thar-

is region was stripped of dust, a portion of the Nilosyrtis albedo

oundary was reset, and dust was injected into Solis Planum.

trong winds in Syrtis, Tharsis and Solis appear to be capable

f removing dust from these regions but all of the changes that

e studied elsewhere on Mars appear to be caused by local and

egional dust storms. Progressive changes in the Southern Tropi-

al Band and in Oxia correspond closely to dust storms cataloged

y Wang and Richardson (2015) . Progressive changes in Nilosyrtis

nd episodic changes in Amazonis both appear along the tracks of

easonally repeating winter dust storms, and the surface changes

ook place during the dust storm seasons identified by Wang and

ichardson (2015) .

4. Seasonally repeating dust storms sweep dust towards the

quator.

Global dust storms deliver dust everywhere on Mars, but sea-

onally repeating frontal dust storms appear to do most of the

ousekeeping required for Mars to maintain the consistent appear-

nce it has had for centuries. Significant changes in the martian

ust distribution took place during the 20 year interval between

he Viking era and the arrival of MGS, and several of these changes

ontinued during the era of MGS monitoring. All of the persistently

oving albedo boundaries in the southern hemisphere shifted

orthwards, whereas the persistently moving albedo boundaries in

he northern hemisphere shifted southwards towards the equator.

uch of the dust erosion took place along the tracks of season-

lly repeating winter dust storms that are generated at high lat-

tudes by circumpolar cyclones and are generally blown towards

he equator. The dark area of Acidalia enlarged along its south-

rn margin and in the east, where it borders Oxia Palus, along the

rack of dust storms that originated in northern Acidalia. Dust was

riven southwards from Nilosyrtis along the track of the Utopia

ust storms. Dust was driven southwards from Arcadia (north of

mazonis) along the track of the Arcadia dust storms. Significant

ust was removed from the Southern Tropical band, along the

rack of the Cimmeria/Sirenum dust storms. These seasonally re-

eating, episodic dust storms continue to clean the surface of Mars

oday, perhaps assisted by summertime dust devils.

cknowledgments

This research was supported by grant NNH09AM39I from the

ASA Mars Data Analysis Program. Hannah Brower assisted with

ata processing with support from the NASA Arizona Space Grant

rogram. We are grateful to Melinda Kahre for providing output

rom the NASA Ames GCM. We thank Claire Newman and another

eviewer for many constructive comments.

upplementary materials

Supplementary material associated with this article can be

ound, in the online version, at doi:10.1016/j.icarus.2016.05.023 .

eferences

youb, F. , Avouac, J.P. , Newman, C.E. , et al. , 2014. Threshold for sand mobility on

mars calibrated from seasonal variations of sand flux. In: Proceedings of theEighth International Conference on Mars .

aptista, A.R., Mangold, N., Ansan, V., et al., 2008. A swarm of small shield volca-noes on Syria Planum, Mars. J. Geophys. Res. (Planets) 113, 9010. doi: 10.1029/

20 07JE0 02945 .

ell III, J.F., Wolff, M.J., Malin, M.C., et al., 2009. Mars Reconnaissance Orbiter MarsColor Imager (MARCI): Instrument description, calibration, and performance. J.

Geophys. Res. 114, E08S92. doi: 10.1029/20 08JE0 03315 . iener, K.K. , Geissler, P.E. , McEwen, A.S. , et al. , 2002. Observations of martian dust

devils in MOC wide angle camera images. In: Proceeding of the Thirty-thirdAnnual Lunar and Planetary Institute Science Conference, 33 #2004 .

ridges, N.T., Ayoub, F., Avouac, J.-P., et al., 2012. Earth-like sand fluxes on Mars.Nature 485, 339–342. doi: 10.1038/nature11022 .

antor, B.A., 2007. MOC observations of the 2001 Mars planet-encircling dust storm.

Icarus 186 (1), 60–96. doi: 10.1016/j.icarus.2006.08.019 . antor, B. , James, P.B. , Caplinger, M. , et al. , 2001. Martian dust storms: Mars orbiter

camera observations. J. Geophys. Res. 106, 23653–23687 . antor, B.A. , Edgett, K.S. , 2002. Martian dust devils: 2 Mars years of MGS MOC ob-

servations. Eos Trans. AGU 83 (47) Fall Meet. Suppl., Abstract P51A-0331 .

http://dx.doi.org/10.13039/100000104


http://refhub.elsevier.com/S0019-1035(16)30200-7/sbref0001





http://dx.doi.org/10.1029/2007JE002945

http://dx.doi.org/10.1029/2008JE003315






http://dx.doi.org/10.1038/nature11022











M

M

M

M

M

N

R

S

S

S

S

S

S

S

T

W

W

Z

Cantor, B. , Malin, M.C. , Edgett, K.S. , 2002. Multi-year MOC observations of repeatedmartian weather phenomena during the northern summer season. J. Geophys.

Res. 107 (E3), 3-1–3-8 . Cantor, B.A., Kanak, K.M., Edgett, K.S., 2006. Mars orbiter camera observations of

Martian dust devils and their tracks (September 1997 to January 2006) andevaluation of theoretical vortex models. J. Geophys. Res. 111, E12002. doi: 10.

1029/20 06JE0 0270 0 . Capen Jr., C.F., 1976. Martian albedo feature variations with season – Data of 1971

and 1973. Icarus 28, 213–230. doi: 10.1016/0019- 1035(76)90034- 8 .

Christensen, P.R. , Bandfield, J.L. , Hamilton, V.E. , et al. , 2001. Mars global surveyorthermal emission spectrometer experiment: Investigation description and sur-

face science results. J. Geophys. Res. 106 (E10), 23823–23871 . Dollfus, A. , 1965. Étude de la planète Mars de 1954 à 1958. Ann. Astrophys. 28, 722 .

Edgett, K.S., 2002. Low-albedo surfaces and eolian sediment : Mars orbiter cameraviews of western arabia terra craters and wind streaks. J. Geophys. Res. 107 (E6),

5038. doi: 10.1029/20 01JE0 01587 .

Fenton, L.K., Lorenz, R., 2015. Dust devil height and spacing with relation to themartian planetary boundary layer thickness. Icarus 260, 246–262. doi: 10.1016/j.

icarus.2015.07.028 . Fisher, J.A., Richardson, M.I., Newman, C.E., et al., 2005. A survey of martian dust

devil activity using Mars global surveyor Mars orbiter camera images. J. Geo-phys. Res. (Planets) 110, 3004. doi: 10.1029/20 03JE0 02165 .

Geissler, P.E., 2005. Three decades of martian surface changes. J. Geophys. Res. (Plan-

ets) 110, 2001. doi: 10.1029/20 04JE0 02345 . Haberle, R.M., Joshi, M.M., Murphy, J.R., et al., 1999. General circulation model simu-

lations of the Mars Pathfinder atmospheric structure investigation/meteorologydata. J. Geophys. Res. 104, 8957–8974. doi: 10.1029/1998JE90 0 040 .

Hayward, R.K., Mullins, K.F., Fenton, L.K., et al., 2007. Mars Global Digital DuneDatabase and initial science results. J. Geophys. Res. (Planets) 112, 11007. doi: 10.

1029/20 07JE0 02943 .

Hayward, R.K., Fenton, L.K., Titus, T.N., 2014. Mars Global Digital Dune Database(MGD3): Global dune distribution and wind pattern observations. Icarus 230,

38–46. doi: 10.1016/j.icarus.2013.04.011 . Hiesinger, H., Head, J.W., 2004. The Syrtis Major volcanic province, Mars: Synthesis

from Mars Global Surveyor data. J. Geophys. Res. 109 (E1), E01004. doi: 10.1029/20 03JE0 02143 .

JALPO, 2003. Mars Section Report. J. Assoc. Lunar Planet. Obs. 45, 38 .

Kahre, M.A., Murphy, J.R., Haberle, R.M., 2006. Modeling the Martian dust cycle andsurface dust reservoirs with the NASA Ames General Circulation Model. J. Geo-

phys. Res. 111 (E6), E06008. doi: 10.1029/2005JE002588 . Lee, S.W. , 1986. Regional sources and sinks of dust on Mars: Viking observations of

Cerberus, Solis Planum, and Syrtis Major. In: Proceedings of the LPI Contribu-tions 1986 Symposium on Mars: Evolution of its Climate and Atmosphere, 599,

p. 57 .

Lee, S.W., Thomas, P.C., Veverka, J., 1982. Wind streaks in Tharsis and Elysium –Implications for sediment transport by slope winds. J. Geophys. Res. 87, 10025–

10041. doi: 10.1029/JB087iB12p10025 . Lee, S.W. , Thomas, P.C. , Cantor, B.A. , 2013. Disappearance of the Propontis regional

dark albedo feature on Mars. Am. Astron. Soc Division for Planetary SciencesMeeting Abstracts, 45, 400.01 .

Malin, M.C., Danielson, G.E., Ravine, M.A., et al., 1991. Design and development ofthe Mars Observer Camera. Int. J. Imaging Syst. Technol. 3, 76–91. doi: 10.1002/

ima.1850030205 .

Malin, M.C., Danielson, G.E., Ingersoll, A.P., et al., 1992. Mars Observer camera. J.Geophys. Res. 97, 7699–7718. doi: 10.1029/92JE00340 .

Malin, M.C., Calvin, W., Clancy, R.T., et al., 2001. The Mars Color Imager (MARCI)on the Mars climate orbiter. J. Geophys. Res. 106, 17651–17672. doi: 10.1029/

1999JE001145 .

alin, M.C., Edgett, K.S., Cantor, B.A., et al., 2010. An overview of the 1985–2006Mars Orbiter Camera science investigation. Mars J. 5, 1–60. doi: 10.1555/mars.

2010.0 0 01 . artin, L.J., 1976. 1973 dust storm on Mars – Maps from hourly photographs. Icarus

29, 363–380. doi: 10.1016/0019-1035(76)90139-1 . ellon, M.T. , Kretke, K.A. , Smith, M.D. , et al. , 2002. A global map of thermal inertia

from mars global surveyor mapping-mission data. In: Proceedings of the 2002Lunar and Planetary Science Conference, 33 Abstract 1416 .

ichaels, T. 2011. Modeling aeolian surface interaction phenomena at Nili and

Meroe Paterae. In: Proceedings of the EPSC-DPS Joint Meeting 2011, pp. 1617. ukherjee, P. and Geissler, P.E. 2010. Investigating the cause of moving albedo

boundaries in the Oxia Palus region of Mars. AGU Fall Meeting Abstracts #1427.owicki, S.A., Christensen, P.R., 2007. Rock abundance on Mars from the Thermal

Emission Spectrometer. J. Geophys. Res. 112, E05007. doi: 10.1029/2006JE002798 .eiss, D., Spiga, A., Erkeling, G., 2014. The horizontal motion of dust devils on Mars

derived from CRISM and CTX/HiRISE observations. Icarus 227, 8–20. doi: 10.1016/

j.icarus.2013.08.028 . agan, C., Veverka, J., Steinbacher, R., et al., 1974. Variable Features on Mars.

IV. Pavonis Mons (Paper presented at The International Colloquium on Mars,28 Nov.–1 Dec. 1973, Pasadena, California). Icarus 22, pp. 24–47. doi: 10.1016/

0019-1035(74)90163-8 . ilvestro, S., Vaz, D.A., Fenton, L.K., et al., 2011. Active aeolian processes on Mars:

A regional study in Arabia and Meridiani Terrae. Geophys. Res. Lett. 38, L20201.

doi: 10.1029/2011GL048955 . lipher, E.C. , 1962. A Photographic History of Mars. Lowell Obs, pp. 1905–1961 .

Smith, D.E., Zuber, M.T., Frey, H.V., et al., 2001. Mars Orbiter Laser Altimeter: Exper-iment summary after the first year of global mapping of Mars. J. Geophys. Res.

106 (E10), 23689–23722. doi: 10.1029/20 0 0JE0 01364 . oderblom, L.A., Edwards, K., Eliason, E.M., et al., 1978. Global color variations on

the Martian surface. Icarus 34, 446–464. doi: 10.1016/0019- 1035(78)90037- 4 .

tanzel, C., Patzold, M., Williams, D., et al., 2008. Dust devil speeds, directions ofmotion and general characteristics observed by the Mars Express High Resolu-

tion Stereo Camera. Icarus 197 (1), 39–51. doi: 10.1016/j.icarus.2008.04.017 . trausberg, M.J., Wang, H., Richardson, M.I., et al., 2005. Observations of the initi-

ation and evolution of the 2001 Mars global dust storm. J. Geophys. Res. 110(E02006). doi: 10.1029/2004JE002361 .

zwast, M.A., Richardson, M.I., Vasavada, A.R., 2006. Surface dust redistribution on

Mars as observed by the Mars Global Surveyor and Viking orbiters. J. Geophys.Res. (Planets) 111, E11008. doi: 10.1029/20 05JE0 02485 .

Thomas, P., Veverka, J., Lee, S., et al., 1981. Classification of wind streaks on Mars.Icarus 45, 124–153. doi: 10.1016/0019-1035(81)90010-5 .

Thomas, P.C., Gierasch, P., Sullivan, R., et al., 2003. Mesoscale linear streaks onMars: Environments of dust entrainment. Icarus 162, 242–258. doi: 10.1016/

S0 019-1035(03)0 0 028-9 .

oyota, T., Kurita, K., Spiga, A., 2011. Distribution and time-variation of spire streaksat Pavonis Mons on Mars. Planet. Space Sci. 59, 672–682. doi: 10.1016/j.pss.2011.

01.015 . ang, H., Zurek, R.W., Richardson, M.I., 2005. Relationship between frontal dust

storms and transient eddy activity in the northern hemisphere of Mars asobserved by Mars Global Surveyor. J. Geophys. Res. 110, E07005. doi: 10.1029/

20 05JE0 02423 . ang, H., Richardson, M.I., 2015. The origin, evolution, and trajectory of large dust

storms on Mars during Mars years 24–30 (1999–2011). Icarus 251, 112–127.

doi: 10.1016/j.icarus.2013.10.033 . urek, R.W., 1982. Martian great dust storms – An update. Icarus 50, 288–310.

doi: 10.1016/0019-1035(82)90127-0 .





http://dx.doi.org/10.1029/2006JE002700

http://dx.doi.org/10.1016/0019-1035(76)90034-8








http://dx.doi.org/10.1029/2001JE001587


http://dx.doi.org/10.1029/2003JE002165

http://dx.doi.org/10.1029/2004JE002345

http://dx.doi.org/10.1029/1998JE900040

http://dx.doi.org/10.1029/2007JE002943


http://dx.doi.org/10.1029/2003JE002143


http://dx.doi.org/10.1029/2005JE002588



http://dx.doi.org/10.1029/JB087iB12p10025





http://dx.doi.org/10.1002/ima.1850030205

http://dx.doi.org/10.1029/92JE00340

http://dx.doi.org/10.1029/1999JE001145

http://dx.doi.org/10.1555/mars.2010.0001

http://dx.doi.org/10.1016/0019-1035(76)90139-1






http://dx.doi.org/10.1029/2006JE002798


http://dx.doi.org/10.1016/0019-1035(74)90163-8

http://dx.doi.org/10.1029/2011GL048955



http://dx.doi.org/10.1029/2000JE001364

http://dx.doi.org/10.1016/0019-1035(78)90037-4


http://dx.doi.org/10.1029/2004JE002361

http://dx.doi.org/10.1029/2005JE002485

http://dx.doi.org/10.1016/0019-1035(81)90010-5

http://dx.doi.org/10.1016/S0019-1035(03)00028-9

http://dx.doi.org/10.1016/j.pss.2011.01.015

http://dx.doi.org/10.1029/2005JE002423


http://dx.doi.org/10.1016/0019-1035(82)90127-0

Documents

Orbital monitoring of martian surface changes