ContentslistsavailableatScienceDirect
Icarus
journalhomepage:www.elsevier.com/locate/icarus
Saltation under Martian gravity and its influence on the global dust distribution
Grzegorz Musiolika,∗,Maximilian Krussa,Tunahan Demircia, Björn Schrinskia, Jens Teisera, Frank Daerdenb, MichaelD. Smithc, Lori Nearyb,Gerhard Wurma
aFaculty of Physics, University of Duisburg-Essen, Lotharstr. 1–21, 47057, Duisburg, Germany
bRoyal Belgian Institute for Space Aeronomy (BIRA-IASB), Ringlaan 3, B-1180, Brussels, Belgium
cNASA Goddard Space Flight Center, Greenbelt, MD 20771, United States
a rt i c l e i n f o
Article history:
Received 17 October 2017 Revised 21 December 2017 Accepted 4 January 2018 Available online 9 January 2018 Keywords:
Mars Saltation
Microgravity experiments Cohesion
General circulation model
a b s t r a c t
DustandsandmotionareacommonsightonMars. Understandingtheinteractionofatmosphereand Martiansoilisfundamentaltodescribetheplanet’sweather,climateandsurfacemorphology.
WesetupawindtunneltostudytheliftofamixturebetweenveryfinesandanddustinaMars simulantsoil.TheexperimentswerecarriedoutunderMartiangravityinaparabolicflight.Thereduced gravitywasprovidedbyacentrifugeunder externalmicrogravity.Theonsetofsaltationwasmeasured forafluidthresholdshearvelocityof0.82±0.04m/s.ThisisconsiderablylowerthanfoundunderEarth gravity.
Inadditiontoareductioninweight,thislowthresholdcanbeattributedtogravitydependentcohe- siveforceswithinthesandbed,whichdropby2/3underMartiangravity.Thenewthresholdforsaltation leadstoasimulationoftheannualdustcyclewithaMarsGCMthatisinagreementwithobservations.
© 2018ElsevierInc.Allrightsreserved.
1. Introduction
Wind tunnel experiments simulating dust lifting on the Mar- tian surfacedateback intothelast century (Greeley etal., 1980).
These studies usedifferent low-densitymaterials to simulatethe reducedgravityonMarsof0.38gandprovidethefirstthresholds fortheonsetofsaltation.Comparedtotheavailable meteorologi- caldatawhichallowsanestimationoftheMartianboundarylayer winds (Hess etal., 1977;Schofield et al., 1997; Magalhãeset al., 1999;Holstein-Rathlouetal.,2010)andtopredictionsfromglobal circulationmodels(GCMs)(Forgetetal.,1999;Haberleetal.,1999, 2003), this threshold should be exceeded only rarely (Jerolmack et al., 2006; Kok et al., 2012; Wang and Zheng, 2015; Newman etal.,2017). Incontradictiontothis, themotionofdustandsand canbeobservedfrequentlyandhasalargeimpactontheMartian climate (Zurek et al., 1992; Smith, 2004; Heavens et al., 2011;
Guzewichetal.,2017).
Strongeffortshavebeenmadeinrecentyearstodetailthepic- tureofsoil-atmosphereinteraction(Whiteetal., 1987;Strausberg et al., 2005; Sullivanet al., 2005; Greeley et al., 2006; Merrison et al., 2007;Almeida etal., 2008; Merrison et al., 2008; Sullivan et al., 2008; Kok, 2010b,a; Bridges et al., 2012). Even though, it
∗ Corresponding author.
E-mail address: gregor.musiolik@uni-due.de (G. Musiolik).
stillremainsquestionableifduststormscangenerallybeinitiated by wind drag. For example, a lower shear velocity would suffice tokeep saltationactive butcannotexplain the onsetofsaltation.
Hence,alsosupportingeffectsarestudied.Forexample,insolation ofthesoilleadstothermalcreepandasub-surfaceoverpressure, capable of reducing the threshold wind velocity significantly (de Beuleet al., 2014;Küpper and Wurm, 2015). Also dust devils go along with pressure excursions which can support grain lifting (Balme and Hagermann, 2006). In any case, numerical models often use an artificially reduced threshold which is needed to initiateliftingeventstosimulatesaltationonMars(Haberleetal., 2003;Kahreetal.,2006;Daerdenetal.,2015).
However,aeoliantransportexperimentsatMartiangravityand pressure,ase.g. byWhite etal.(1987),arerare.Inthiswork, we investigatetheinfluenceofreducedgravityonsaltationandshow thatthethresholdvelocityforasandbedpreparedandsubjectto gasflowatMartiangravityandpressureisstronglyreduced.
1.1. Experimentalsetup
TheMartianenvironment issimulatedinalow pressurewind tunnel designedsimultaneously asa centrifuge to simulateMar- tian conditions (Fig. 1). In detail, the experiment consists of a vacuumchamber whichisevacuatedtoapressureof6mbarand a gasmixture of 95% CO2 and5% air.It has a radius of100mm https://doi.org/10.1016/j.icarus.2018.01.007
0019-1035/© 2018 Elsevier Inc. All rights reserved.
Fig. 1. A schematics of the experiment. The outer vacuum chamber has a diameter of 200 mm and is designed as a centrifuge. The wind tunnel is placed inside this centrifuge and has a cross section of 100 mm × 100 mm.
andcanberotatedatmorethan2Hz.Thewindtunnelislocated inthecenterof theexperimentchamber andhasa crosssection of100mm × 100mm.Thewindflowiscreatedbyafanrotating withupto 11.000rpmatan air flowrateofup to 570m3/h. The gas flows through the wind tunnel over the sand bedand back againontheouter sideofthewindtunnel.The totalmassofthe experimentis161kg. TheReynoldsnumber forthisconfiguration insidethewindtunnelisontheorderofRe≈800.
The set up is used in parabolic flights on the ZERO-G Airbus operatedbyNOVESPACE inBordeaux(Pletseretal.,2016).Aflight consistsof31parabolaswithadurationof22sperparabolaanda residualaccelerationonthescaleof ±0.05g(Pletser etal.,2016).
Thecentrifugalforceonthesurfaceofthedustbedissetto0.38g, whileadditional experimentson groundwere carriedout at 1g.
Theparticle samplewas ∼ 50g ofa mixture betweenvery fine sandanddustconsistingofthe JSC1AMartianregolithsimulant, which was tempered at 600 K before to remove volatiles and organics.This simulantis made out of alteredvolcanic ash from a Hawaiian cinder cone and is a representative species for the reflectance spectrum, mineralogy, chemical composition, density, porosityandmagneticpropertiesoftheMartiansoil (Allenetal., 1997).ThesizedistributionoftheusedsampleisshowninFig.2. Beforeeachparabola,thesampleisclosedbyashuttermecha- nismtoprotectthesampleagainstuncontrolledaccelerations.The experimentrunsautomatically.Withtheonsetofthemicrogravity phase,the chamberstartsto rotate. The shutterisremovedonce thesetrotationfrequencyisestablished.Duetothemomentumof theshutter,thesandsampleisfirstliftedandthensettlesbackto theground. Thisway,thesurfaceofthesand sampleis prepared atMartiangravitylevel beforeeach measurement. Theerosion is observed optically with a camera installed perpendicular to the wind flow at 457 frames per second and an exposure time of 200μ s,usingbacklightillumination(s.Fig.3).Thisprovidesares- olutionsufficient totrace thefractionofthelargerparticles from Fig.2,butnotsufficienttoresolvethefractionofsmallerparticles.
Fig. 2. Grain size distribution of the used sample. The fraction of larger grains dom- inates the mass distribution and therefore the mechanical properties of the sample.
2. Results 2.1. Dataanalysis
WeuseaMartiansimulantJSCMars1Aassoilwithaparticle densityof1.9g/cm3 (orthebulk densityof0.87g/cm3 including 54% porosity) (Allenet al., 1997) anda particle size distribution asdepicted in Fig. 2.The shownsize distribution represents the volume densityof theparticle sizes. We cannot exclude that the smallerdust mighthavean impacton thecohesionpropertiesof thesample.Nonetheless,whilethesmallerparticlesgetsustained in the atmosphere more easily, saltation is probably dominated by the fraction of the larger particles. The larger particles might haveeitheragrain-likeoraggregatestructure.Ingeneral, theyfit in size to particles in Martian dunes, which are given to 87μm (ClaudinandAndreotti,2006;Koketal.,2012).Thoughevenlarger
Fig. 3. Snapshot of particles lifted at 0.38 g close to threshold wind velocity. The wind is flowing from left to right. The shown surface roughness is typical.
Fig. 4. Sample trajectories in wind direction for 1 g (left) and 0.38 g (right). The motion of the particles was fitted according to Eq. (1) . The fits are overplotted in black. The particles are accelerated by the gas motion until they finally couple to it.
particles,e.g. 40−400 μm(HighDune Samples)or50−400μm
(Namib Dune Sample) are discussed in the literature as well (Ehlmann et al., 2017; Tirsch et al., 2012; Sullivan et al., 2008;
EdgettandChristensen,1991)thesampleallowsanestimationfor theminimumshearvelocityneededtoliftparticles.
Anexamplefortheobservationofliftedsandparticlesat0.38g isshowninFig.3.Theroughnessofthesurfaceisconsistentwith the roughness map of Hébrard et al.(2012) derived from MOLA data,in whichthe meansurfaceroughness onMarsis 4.435mm and the median surfaceroughness is 11.05mm, with 36% of the Martian surfacehavinga roughnessvalue higherthan 5mm.The gas flow is just set high enough for lifting events to occur and thefluidthresholdshearvelocityu∗isdetermined.Saltationtakes placeaswellassuspension. Onceinitiated, alower windvelocity attheimpactthresholdisneededtosustaintheparticleflow, but thisisnotfurtherinvestigatedinthiswork.
ForMartiangravityof0.38g,51trajectoriesofliftedsandpar- ticlesareanalyzed,while53trajectoriesareanalyzedfor1g.From these trajectories,the horizontalgas velocity andits dependency on the height above the sand are calculated. The eroded sand particles coupleto the motionof thegas inside thewind tunnel andare used totrace the gasvelocity closeto thesand bed. For a given height z, the trajectory of the sand aggregatesalong the (horizontal)x-axiscanbedescribedby(Wurmetal.,2001) x(t,z)=(vg(z)−v0)tCexp
−t tC
+vg(z)t+c. (1)
This equation is valid for spherical particles with a constant coupling time but can be used as an approximation for bumpy particles asshowninFig. 4.Thefollowing fitparameters are ob- tainedfromfittingthetrajectoriesofthesandparticles according to Eq.(1): The initial velocity v0 ofthe grain ata certain height z,thegas-aggregatecouplingtime tC,aconstantc andfinallythe gas velocity vg(z) for a given height z above the dust sample.
Furthermore, t is the time after the lifting event. Note, that the
Fig. 5. Gas velocity profile over height at the threshold of particle lifting for 1 g (green) and 0.38 g (blue). The data are binned, including 53 individual values for 1 g and 51 values for 0.38 g. The slopes d vg(z)/d zresulting from the linear fits are 680 s −1for 1 g and 453 s −1for 0.38 g. The threshold u ∗was calculated using Eq. (2) . (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
Coriolis force is negligible for the lifting process ofthe particles (astheyare atrest) aswell asforthegrain motionata constant heightzatwhichtheparticlesaretracked.
Forthe0.38gtrajectories,thevaluesforthegasvelocitiesare binnedin 0.5mm steps. For each bin,the values forthe median gasvelocity arecalculated with7–8individual valuesforthe gas velocity from the fitted trajectories according to Eq. (1). For the 1g trajectories,the binsizeis setto 0.6mm. The binneddatais giveninFig.5.Bothprofilesindicate alinearcorrelationbetween thehorizontalgasvelocityandtheheight abovethesandsurface.
A linear profile close to the ground is also in agreement with former experimentsin wind tunnels (Merrison et al., 2008). The
ω
fromtheaccelerationdata.Iftheamplitudeofthe vibrationswas A≈100μm,whichequals thegrainsize andhenceisa maximum estimation,theerrorinaccelerationwouldbeontheorderofgv= Aω2≈0.01 m/s2. Comparedto the Martian gravitationof 0.38g, thisisarelativeerrorof2.7%,whichweconsiderasnegligible.
2.2.ThresholdshearvelocityandcohesionreductionforMars
Interactionofaturbulentwindflowwithasurfacecanbechar- acterizedby the shear velocity u∗=√τ/ρ with the shear stress τ and the fluid density ρ (Schlichting and Gersten, 2016). This quantity can be interpreted as the wind velocity acting directly atthesoil. Theshearstress canbe expressedbyNewton’slawof viscositytoτ=ηdvg(z)/dz withthe dynamicviscosity η andthe
flowheight profiledvg(z)/dz dependingonthe gasvelocity vg(z) andtheheightz.Thus,u∗canalsobeexpressedas
u∗=
η
ρ
dvg(z)
dz . (2)
The gas velocity vg(z) is logarithmic in z within a turbulent sublayer and linear in z within a viscous sublayer close to the groundasmeasured inthis work. Considering η≈15 μPa·s and
ρ≈0.01 kg/m3 (CO2 at6 mbar and 300 K) as well as dvg(z)/dz from Fig. 5, the threshold shear velocity can be derived di- rectly from Eq. (2) and yields 0.82±0.04 m/s for 0.38 g and 1.01±0.04m/sfor1g.
The threshold shear velocity at 0.38 g is lower than values determinedinprior experimentson ground(Greeley etal.,1980;
Merrisonet al., 2008) which are generallysomewhat larger with
∼1.5−2 m/s. However, u∗ wasmeasured in a different gravita- tional environment in this work and depends also on the grain species.Thus,itcannotbecompareddirectlytotheseotherworks.
Thismight also be an indication that prior experiments perhaps overestimatedthisvaluefortheMartiansoil.
UsingthemodelsfromShaoandLu(2000)andMerrisonetal.
(2008)withthethresholdshearvelocitiesfor0.38gand1g, the particledensityof1.9g/cm3 anda meanparticlediameterofap- proximately85μmwegetasurfaceenergyofγSL≈1.1·10−7J/m2. Thisisanunreasonably lowvalue astheused JSCspeciesmostly consistsofSiO2,Al2O3,Fe2O3 andCaO whichall exceedvaluesof 10−2J/m2forthesurfaceenergy(Heimetal.,1999;Miller,2011).
In consequence ofthe low value for γSL we consider a lower cohesive force at lower gravity influencing the ratio of the de- termined threshold shear velocities. The cohesion force at the thresholdcanbeestimatedfromtheforcebalance
CLπ
2ρr2u∗2=
j
FC,j+Mg. (3)
Theliftingforceisgivenontheleftside(KüpperandWurm,2015).
CL is the liftingcoefficient whichdepends onthe boundary con- ditionsofthewind tunnelandtheshapeoftheparticles,risthe averageradiusoftheparticles andρisthefluiddensity.Counter-
actingarethe gravitationalforce M·gwiththe particle’smass M andthegravitationalaccelerationgandthesumoverallcohesive contactsFC,j ofagrain. Grainsof ∼100 μmare usually easiestto move as cohesive forces and gravity are similar (Greeley et al., 1980). Thus, none of the addends can be neglected. We assume thatallindividual contactsaresharingthesamecontactarea and
with the particle density ρp and diameter d. Except the depen-
dencyinN,thisexpressionissimilartotheequation providedby ShaoandLu(2000).IfthecontactnumberNdependsonthegrav- itationalacceleration,u∗mightbelowerforreducedgravity.With twovaluesfortheu∗,thisdependencyN(g)canbeestimated.
The ratio betweenboth threshold shear velocities atdifferent gravitationalenvironmentscanbewrittenas
u∗12
u∗22 =N1FJ+Mg1
N2FJ+Mg2 ≡ FN+Mg1
χFN+Mg2, (5)
with the sum of all contact forces FN≡N1FJ and the contact numberratioχ≡N2/N1.Wecanderiveχ fromEq.(5)to
χ=uu∗2∗22 1
1+Mg1
FN
−Mg2
FN . (6)
Applying the values forthe fluid threshold shear velocity inthis workwiththeaveragenumberofcontactsN1 in0.38gandN2in 1ggives
χ≈3
2 ∀ FN10−8N. (7)
This resultshows, that the average numberof contactsand thus alsothe total contactforcesare only2/3 aslarge in0.38g asin 1g,ifFNexceeds10−8Nbyanorderofmagnitude.Ifweconsider N=1,γ≈0.01 J/m2 which is a typical value forsilicate spheres (Heim etal., 1999) andr=10−5 m (asminimumestimation) the additionalconditioniseasilyfulfilledwithF1= 32πγr≈5·10−7N.
Experimental work on contact forces confirms this likewise (Heimetal.,1999).Thisisthefirsttime thatitisconsideredthat cohesionisnotconstantinsoilsofdifferentplanetsasgravitydoes compressthesoildifferently.Areductionincontactnumberinthe low gravity environment of Mars can explain a reduction in the thresholdwindvelocitynecessarytoliftparticles.Absolutevalues ofthefluidthresholdshearvelocity derived fromourexperiment under 0.38 g indicate that saltation and suspension are possible undertheconditionsgivenonMarsandinagreementtoparticles beingobservedinmotion.
3. SimulationwiththeGlobalCirculationModel(GCM) 3.1. MarsGCM
AGeneralCirculationModel(GCM)fortheatmosphereofMars isappliedtocalculatesurfaceshearvelocities(Daerdenetal.,2015;
Neary andDaerden, 2018). It is operated on a grid witha hori- zontalresolution of4°×4°andwith103verticallevels reaching from the surfaceto ∼150km.The model calculates heatingand coolingofatmosphericCO2anddustandiceparticlesbysolarand IRradiationandsolvestheprimitiveequationsofatmosphericdy- namics. The geophysical boundary conditionsare taken fromob- servations andinclude a detailed surfaceroughness length map.
PhysicalparameterizationsinthemodelincludeaninteractiveCO2 condensation and surface pressure cycle, a thermal soil model, turbulent transport in theatmospheric surfacelayer andconvec- tive transport inside the planetaryboundary layer. The effects of the extremeMartiantopography are considered witha low level blockingscheme.Theshearvelocityisderivedfromthecomputed wind field in the second lowest vertical model level (at height
∼15m), following the expressions derived fromsimilarity theory
