Abrasion by Aeolian Particles: Earth and Mars

j NASA Contractor Report 3788

R. Greeley, 'S. Williams, B. R. White, LOAN COPY: RfTURN TO AFWL TECHNICAL LIBRARY J. Pollack, J. Marshall, and D. Krinsley KIRTUND AFB, N.M. 87117

GRANTS NSG-2284 and NCC2-60 MARCH 1984 TECH LIERARY KAFB, NY

00b2345 NASA Contractor Report 3788

Abrasion by Aeolian Particles: Earth and Mars

R. Greeley J. Pollack Arizona State University NASA Ames Research Center Tempe, Arizona Moflett Field, California

S. Williams J. Marshall and D. Krinsley University of Santa Clara Arizona State University Santa Clara, California Tempe, Arizona

B. R. White University of California at Davis Dauis, California

Prepared for NASA Office of Space Science and Applications under Grants NSG2284 and NCC2-60

National Aeronautics and Space Administration Scientific and Technical Information Office

1984

TABLE OF CONTENTS

Section Page

1 . Introduction ...... 1 1.1 The Martian Dilemma ...... 3 1.2Approach ...... 5 1.3 Organization ...... 6

2. Experimental Materials and Rationale For Their Selection ...... 7

2.1 Target Materials ...... 7 2.2 Windblown Particles ...... 7 2.3 Summary ...... 10

3. Dynamics of Windblown Particles ...... 11

3.1 Methodology ...... 11 Wind Tunnel Experiments...... 11 Field Experiment ...... 13 3.2 Particle Velocity ...... 13 3.3 Saltation Model ...... 15 3.4 Particle Flux ...... 16 Model for Calculation of Flux ...... 18 3.5 Summary ...... 23

4 . Susceptibility of Rocks to Aeolian Abrasion ...... 24

4.1 Previous Investigations ...... 24 4.2 The Mechanics of Abrasion ...... 24 Fracture Patterns Produced by RoundedParticles ...... 24 Fracture Patterns Producedby AngularParticles ...... 26 The Influence of Fracture Patterns on Abrasion Rates ...... 26 4.3 Abrasion Experiments ...... 27 Sa Versus Particle Diameter ...... 27 Sa Versus Particle Velocity ...... 28 Sa Versus Particle Impact Angle ...... 29 Sa Versus Atmospheric Density...... 29 Sa Versus Particle Composition ...... 29 4.4 Summary ...... 31

5 . WindFrequencyData ...... 34

5.1 TerrestrialEnvironment ...... 34 5.2 Martian Environment ...... 34 5.3 Summary ...... 36

iii Section Page

6 . Results and Discussion ...... 37

6.1 Aeolian Abrasion on Earth ...... 37 6.2 Rate of Wind Abrasion on Mars ...... 38 Climatic Effects...... 40 Surface History ...... 41 Material Properties ...... 42 Agents of A brasion ...... 42 Conclusions ...... 44

Acknowledgments ...... 46

References...... 47

iv 1. INTRODUCTION

“Underthe stimulus of aridity, wind scour becomes an erosive agent more constantthan the rain,more potent where Sa is thesusceptibility to abrasion, thanstreams, and more persistent than q is particleflux, and f is windfrequency. the sea”. Althoughmany previous studieshave addressed windabrasion, most analyses So wrote A.K. Lobeck(1939) in his intro- have lacked various critical elements and are ductorytextbook on geomorphology.On thereforedifficult to usein determining Earth, aeolian processes areparticularly ratesof wind abrasion.For example, engi- important in hot and cold deserts, along many neering studies of abrasion are generally well coastlines, and in glacial outwash plains. documented and quantitativein the approach, Mars experiences frequentdust storms and but do notinvolve geological materials such as lacks theinhibiting effects of liquid water rocks. On theother hand, studies ofwind- andvegetation. Thus, aeolian processes play abradedrocks - ventifacts - in thenatural an important role in the geological evolution field environment generally lacksufficient of the surfaces of bothEarth and Mars. wind frequency data and particle flux information to quantify the results. The objective ofour investigation was todetermine rates of wind abrasion fora wide range of conditions on Earth and Mars. Thestudy, however, was restricted to con- siderationsof rock abrasionand does not includedeflation or the erosion of land- forms.

In orderto determine rates of aeolian abrasion,knowledge threeof factors is required (Fig. 1):(a) windcharacteristics, (b)particle characteristics, and (c) target characteristics. Wind characteristicsinclude windstrengths, durations, and directions. Particle Characteristics includeparticle size, shape,and composition and the velocities andtrajectories of windblown grains; the Figure 1. Diagram showing the three principal parameters involved in determining rates of aeolian lattertwo are functions of freestream wind abrasion: (I) wind factors, including wind velocity, speeds and heights above the ground. Target duration and direction; (2) particle characteristics, characteristicsinclude the rock shape and a including sizes, velocities, and fluxes as functions measureof theresistance of the rock to of wind speed andheight above the ground, and abrasion,termed the susceptibility to abra- (3)a measure of the resistance to abrasion of various sion. Therate of abrasion may then be rocks, termed the susceptibility to abrasion (fiom calculated as Greeley et aL, 1982). 2

TABLE 1. Nomenclature am maximum impact contact radius final particle speed A particle cross-sectional area wind friction speed

CD coefficient of drag threshold wind friction speed CL coefficient of lift freestream wind speed D drag force wind speed DH mean crack diameter particle speed DP particle diameter relative speed betweenparticle and wind E subscript denoting Earth initial particle upward velocity f wind frequency average initialparticle upward velocity g acceleration due to gravity final particle downward velocity number of saltating grains landing in a unit velocity in x-direction GO area per unit time acceleration in x-direction H maximum height above surface of saltation trajectory velocity in y-direction KE kinetic energy acceleration in y-direction L saltation path length height above surface LF lift force surface roughness particle mass angle betweensaltation trajectory and mP horizontal surface at impact M subscript denotingMars angle subtended by crack ‘horns’ (Fig. 23) AP pressure difference average saltation path length Pa atmospheric pressure density of atmosphere 9 saltation area flux gm/cm2*sec density of particle Q saltation lane flux gm/cm*sec total surface shear stress Sa susceptibility to abrasion surface shear dueto particle impact t time surface wind shear maximum value (at U wind speed threshold) initial particle speed surface wind shear constants 3

In this study, we have attempted to gain but it remained for Sagan and Pollack (1969) sufficientinformation on the characteristics to first fully propose this idea. ofwindblown particles, wind frequencies, andmaterial properties appropriatefor esti- In 1971the Mariner 9 orbiterand the mating rates of wind abrasion on Earth and Soviet spacecraft Mars 2 and 3 arrived at Mars Mars. during a major global dust storm and verified the predictionsof aeolian activity. In addition to providing information onthe dynamics of dustclouds, the Mariner 9 cameras 1.1 THE MARTIAN DILEMMA revealed abundant surface features attributed to aeolian processes (McCauley, 1973). Surfacemarkings have been observed tele- Yardangs, dune fields, andvarious deflation scopically on Mars for several hundred years. features were identified, along with bright and One of the earliestobservations that some dark surface patterns that changed their size, markings varied seasonally was made in 1884 shape,and position during the -1 1 month bythe French astronomer E. Trouvelot mission (Sagan et al.,1972). These patterns whoattributed the variations to seasonal (Fig. 2) were correctlyinterpreted as repre- changes ofvegetation (Veverka and Sagan, senting aeolian erosion and deposition. It 1974).Speculations that the markings were was concludedfrom Mariner 9 that the sea- related to vegetation, water, or even martian sonal variations in surface markings were the civilizations persisted well into the twentieth consequence of major dust storms. century. None of these speculations has been shown to be valid. The first consideration of Ultraviolet and infrared spectrometers on the markingsbeing related to aeolian pro- board Mariner 9 also provided information on cesseswas apparentlymade by Dean the atmosphere, including chemical composi- McLaughlin (1954), who published a series of tion,temperature, anddensity. The atmo- papers on potentialmartian wind activity, sphere was found to be composedpredomi-

Figure 2. Mars is subject to frequent wind stonns which cause changes on the surface, as shown by the variable features in these Viking orbiter images. Variable features are defined (Sagan et al., 1972) as surface markings that change their size, shape andlor appearance, suchas the dark streaks associated with craters visible herein the Daedalia region on Mars (30's. 118'W). In (a) streaks record winds fiom the upper left (VO frame 56A41); (b) shows same area for another season in which streaks have two prominent orientations (VO frame 603A08) (from momas and Veverka. 1979). 4

nantly of carbon dioxide with a density very lowcompared to Earth.In this low atmosphericdensity, threshold wind speeds are more than an order of magnitude higher on Mars thanon Earth (Fig. 3). Because of these high winds, it was reasoned that windblown particles would also travel much faster on Mars thanon Earth and wouldbe very effective agents of abrasion, on a per impact basis. Thisconsideration, coupled with the frequentdust storms, led to predictionsof extremely highrates of aeolianerosion (Sagan, 1973; Henry, 1975).

A-simple qualitative experiment was run in our wind tunnel (Greeley et al., 1981) to compare wind erosion onEarth and Mars (Fig. 4). Inthe Earth case,a small front- surface mirror was positioned downwind from a patch ofsand. The tunnel wind velocity was increased to achieve particlethreshold speed and was maintained until all of the sand was blown away. The experiment was dupli- cated with a fresh mirror in a simulated martian atmospheric environment of about 7 mb pressure. Again, the windvelocity was just above threshold speed, but because of the low atmospheric density, the threshold speed was Figure 4. Qualitative comparison of aeolian abrasion on Mars (a) and Earth (b);mirrors were placed in a wind tunnel and abraded by windblown particles of quartz; in both cases, the wind speed was just above threshold, but because of the low atmospheric den- sify on Mars, the wind speed is greater than on Earth to accomplish the sameparticle movement. The mirror in the martian case (a) is more severelyabraded than the Earthcase (b), primarily from the more energetic impacts resulting from the higher velocity. Area shown in both cases is about 500 pn by 750~. 10 100 1000 10,000 PARTICLE DIAMETER D,, pm

about 16 times higher than in the Earth case. Figure 3. Threshold curves for Earth and Mars show- Both mirrors were then examined microscopi- ing threshold piction speeds required for particles of cally to assess the relative damage. The mir- different sizes; about a factor of IO higher speeds are ror that was abraded in the martian environ- required on Mars than on Earth because of the lower density atmosphere on Mars. The particle size most ment was much more severely damaged than easi& moved (lowest friction speed) is about 80 pm the one abraded in the Earth environment, a on Earth and about 100 P on Mars (after Iversen resultof the greater kinetic energies of the et al., 19 76;and Greeley et al., 1980). windblownsand grains on Mars. Thus,at 5 least qualitatively, higher rates of wind abra- little modifiedby erosion ordeposition of sionmight be expected on Marsif aeolian any type (Arvidson et al., 1979). If the wind materialsand wind frequencies of threshold erosionrates predicted prior to the Viking strength are comparable to Earth. mission were correct, then these surface rocks and craters should have been obliterated long The Viking mission to Mars (1976-1981) ago. Thus, there is aconflict between the returnedadditional evidence ofextensive predicted high rate of aeolian abrasion on aeolian activity,including images of asand Mars and the apparent constraints posed by sea equal in size to the largest erg on Earth the Viking results. (Tsoar et al., 1979) and views of the martian surface(Fig. 5) showingdrifts of aeolian 1.2 APPROACH sediments (Mutch et al., 1976). However, the Viking results have also caused a reassessment Ideally, all of the parameters involved in wind of aeolian abrasion on Mars (Williams, 198 1 ; erosion on Earth and Mars would be observed Greeley et al., 1982).The pictures returned and measured in situ. However, even if a field from the Viking landersshow numerous experiment were designed to obtain complete surface rocks, none showing extensive aeolian wind and particle flux data, it would be very modification. Viking Orbiterpictures reveal difficult to separate the effects of the wind surfaces that have abundant small (-10 m) from other erosive agents such as water. Field impactcraters, the presence of which indi- measurementson Mars would be even more catessurface ages that are millions or even difficult and must await a future mission with hundreds of millions of years, but have been an appropriately designed experiment.

Figure 5. The surface of Mars as viewed from Viking Lander1 in the Chlyse Planitia region, showing a2 m boulder, nick-named ‘Big Joe, ’ numerous smaller rocks, and light-toned drifts of material interpreted to be aeolian deposits modified by wind erosion. 6

The approach in this investigation was to 1.3 ORGANIZATION use resultsfrom laboratory simulations, supplemented by field experiments and theo- We present our resultsin six sections.The reticalconsiderations, inorder to derive introduction is followed by asection de- empirical results and calculated values for the scribing the rocks, materials, and windblown factors required in estimating aeolian abrasion particles used in our experiments and giving the rationale for theirselection. Thenext rates. Laboratory experiments have the advan- three sections present results on the dynamics tage that conditions can be controlled and the of windblownparticles, susceptibilities to parameters, such as particle velocity, can be abrasion of the test materials, and wind fre- isolated and studied separately. Furthermore, quency data. The last section gives estimates theatmospheric environment can be simu- of aeolianabrasion rates on Earth and Mars lated to derive dataappropriate for Mars. and discusses the results. 2. EXPERIMENTALMATERIALS AND RATIONALEFOR THEIR SELECTION

In this section we discuss the rocks and other be variable, including very friableduricrust materials used inour experiments to deter- sediments observed atthe Viking landing minesusceptibilities to abrasion appropriate sitesand thick, apparently highly indurated for geological studies on Earth and Mars. We deposits observed in the walls of the ‘Grand also consider the common sourcesof wind- Canyon’ of Mars, Vallis Marineris (Cam, blownparticles and describe their character- 198 1). Possible origins of the various man- istics in terms of aeolian abrasion. tling depositsinclude windblown sediments, fluvial deposits, glacial deposits,and others. 2.1 TARGETMATERIALS Thus, there appear to be fine-grained to glassy volcanic rocks on Mars, plus fine-grained Eight materials(Table 2) were selected for pyroclastic materials and possible sedimentary abrasionexperiments to representa wide clastic deposits ranging from breccias (gener- range ofpotential targets on Earth and atedpartly from impact cratering) and con- Mars. The materials selected have asuffi- glomerates to clays. ciently wide range of texturesand compositions to representmost rock types on Earth.Although there is relatively little 2.2 WINDBLOWN PARTICLES information on martian rock types, the Sam- ples within thesuite selectedare probably Most wind-transportedparticles onEarth appropriatefor Mars. Datafrom the Viking range in size from “micrometers (‘dust’) to landerinorganic chemistry experiments several millimeters indiameter (Bagnold, (Toulmin et al., 1977) suggest mafic to 1941). Given the similar shapein the thres- ultra-maficcompositions, probably involving hold curve (Fig. 3) it is likely that windblown basaltic rocks and their weathering products particleson Mars would be of comparable such as various montmorillonitic clays. Spec- size. Furthermore,most rock abrasion on traldata from remote sensing also suggest Earth is apparently accomplished by sand-size that large parts of the martian surface involve (-60 to 2000 pm indiameter) grains trans- mafic and ultra-mafic materials (Singer et al., ported in saltation,although finer material 1979). These data on the composition, com- (Whitney andDietrich, 1973)and perhaps bined with photogeological mapping, suggest even wind alone can be responsible for that more than half of the surface of Mars is shaping ventifacts (Whitney, 1979). Sand-size of volcanic origin (Greeley and Spudis, 1981) particles result from primary sources such as andprovide the basis for selectinga large volcanism andfrom various processes in- suite of volcanic rocks for the abrasion tests. cludingweathering and fragmentation from impact cratering (Pettijohn et al., 1972). On Inaddition to volcanic materials,sub- Earth, most sand is quartz derived from the stantialparts of Mars alsoappear to be weathering of granite. Moreover, most aeolian mantledwith sediments (Scott and Cam, sand has gone through some part of the hy- 1978) which appear to range in thickness drologiccycle, such as beachsand that is from less than 1m to perhapshundreds of transportedby the wind to form coastal meters. The degree of induration appears to dunes, or windblowndesert sands derived TABLE 2. Materials usedin determining susceptibilities .~

MATERIALS SOURCE DESCRIPTION .__

Basalt Amboy lava flow,California Typical basalt inthin section. Contains euhedral plagioclase crystals that show some zoning. Crystal size is 0.5 - 1 mm. Glassy groundmass with some alteration.

Obsidian Lookout Mountain, Long In thin section the obsidian is almost totally Valley Caldera, Mono Lake, glass. Contains some verysmall crystals and California shows flowtexture.

Tuff, welded Bishop Tuff, Owens River The matrix shows both flow and compaction Gorge near Bishop, features. Contains elongated and somewhat California devitrified pumice fragments. Has glass shards in matrix. Euhedral sanidinecrystals of 1.5 - 2 mm diameter. Contains asmall amount of quartz.

Tuff Bishop Tuff, Owens River Contains somelarge pumice fragments. Matrix Gorge near Bishop, has glass shards but noevidence of flowor California compaction. Euhedral sanidinecrystals of 1.5 - 2 mm diameter.

Rhyolite SanCarlos Indian Reservation, Granular texture with all crystals smaller than Arizona Containsmicrocline1 mm.phenocrysts a with few quartz crystals and some accessory biotite and garnet.

Granite Unknown Coarse grained (upto 5 mm) granite. Primarily quartz and microcline, with someplagioclase and thin biotite crystals and accessoryzircon.

Brick Common red clay In hand specimen,the brick is primarily red clay with some quartz and feldsparcrystals, ranging in size up toapproximately 2 mm.

HydrocalCommon gypsum cement specimen,hand In hydrocalthe is uniform buildingmaterial chalky mass,with no visible crystalsinhomo- or geneities, apart from an occasionalvesicle.

~~ -. -.. .~...... ~. -~". ~ . .

from alluvium. Quartz is thecommon com- on Earth, quartz was selected as one of the position of aeolian sands,apparently because primaryagents of abrasion inour experi- it is chemicallystable and mechanically resis- ments. tant and therefore is able to sustain repeated cycles of weatheringand transportation. Granite appears to beabsent or rareon Because it is dominantin aeolian processes Mars andconsequently there may be very 9 littlequartz sandpresent (Smalley and be an important process on Mars (Malin, Krinsley,1979). Furthermore,under the pre- 1974). sent martian climate, and with the apparent lack of active volcanism andtectonism (Cam, 1981) and the present-day low rate of Laboratoryexperiments simulating the impact cratering, the generation of sand-size comminution of windblown sand onMars give particles on Mars is probablysuppressed in support to the idea of the kamikazeeffect comparison to Earth. Nevertheless, the pre- (Greeley, 1979). Experiments also show that sence of sand dunes (Cutts, 1973; Cutts and Smith, 1973), some of which may be active windblownsediments generate electrical (Tsoar et al., 1979), and other aeolian fea- charges, a result confirmed by field observa- tures (Sagan, 1973; Binder et al., 1977) tions on Earth (Mills, 1977). Both the magni- suggest the presence of both sand-size par- tude of the charge and the polarity (positive ticles and winds capable of transporting them. or negative) appear to be functions of particle Remote sensing datafor areas of aeolian size, composition, impact velocity, and atmo- activity on Mars and analyses of thermal sphericdensity. Furthermore, the experi- inertiameasurements also suggest the pre- mentsshowed thatthe dust generatedfrom sence of sand-size particles(Peterfreund, the breakup of the kamikaze grains clumped together, forming aggregates. These aggregates 198 1) as do analyses of materials atthe grew to diametersgreater than several mm Viking Landers(Moore and others,1977). (Greeley,1979). Evidently, differences in electrical charges among thedust particles Most martiandunes are very dark, in- were sufficient for aggregate formation. cludingthose of the large ergin thenorth Because the electrical effects are enhanced in polarregion, and it has been suggested that the rarified martian atmospheric environment, they are composed of basalt particles derived Greeley and Leach (1 978) and Greeley (1979) fromnearby volcanic plains (Tsoar et al., suggested that aggregates maybe an impor- 1979). Thus, sand-size grains of basaltic com- tantpart of the aeolian regime on Mars. positions were used in many of our abrasion experiments to simulate Mars. Windblown aggregates also occuron Earth.Some are generatedduring volcanic eruptions (Krinsley et al., 1980) and apparently are bonded by electrostatic charges. Because of the high wind speeds needed Most aeolian aggregates, however, are asso- for sand entrainmenton Mars, it has been ciated with playa deposits. Silt and clay playa suggested that sand-size particles would be depositsare often slightly induratedand/or very short-lived; the term ‘kamikaze’ (mean- cementedwith various salts. During wind ing divine wind inJapanese) grains was storms,sedimentary accumulations that are coinedby Sagan et al. (1977)to describe fragmented by dessicationare picked upby particles that smashed against rocks and each the wind,rounded into sand-size grains and other, forming fine-grained dust.Estimated deposited in typical sand bedforms, including particle sizes in the martian dust clouds is a dunes (Huffman and Price, 1949; and others). few pm and smaller(Pollack et al., 1979b) and the kamikaze effect may be the principal Sand-size aggregates of fine grains may be mechanism to produce the dust. In addition, an importantpart of aeolian processes on salt weathering, mechanical fracturing caused Mars, and are known to occur on Earth. Thus, by the growth of salt crystals, has been shown aggregates were used in some of our abrasion to be important in theproduction of fine experimentsforcomparison with results grains on Earth (Goudie et al., 1979) and may using crystalline quartz and basalt sand grains. 10

2.3 SUMMARY commonon Earth. Experiments involving Windblownparticles rocktargets. aof basaltic rockswere emphasized for Mars,as wide range were selected for studyin our basalts appear to becommon on the Red experimentsto simulate aeolian abrasionon Planet.Three types of particles - quartz, Earth and Mars. Potential rock targets include basalt, and aggregates offine grains - were igneous andsedimentary rocks which are used as the abrading particles. 3. DYNAMICS OF WINDBLOWNPARTICLES

Of the three parameters required to determine characteristics as possible forEarth condi- the rates of wind abrasion, theone dealing tions,then to repeat theexperiments under with the dynamicalcharacteristics of wind- simulatedmartian conditions. Some of the blown particles is the most complicated and results for the terrestrial cases were compared difficult to assess. Forany given wind, the withresults from field experiments in order saltationflux and trajectories of windblown to verify the wind tunnel simulations and to particles vary with surface roughness, height providea basis forextrapolation to Mars. above the surface,and atmospheric density. Because not all conditions for Earth or Mars Thus,in order to consider wind erosion on couldbe satisfied in laboratorysimulation, Earthand Mars fortimes in the past when numericalmodels were derived based on a the climate and atmosphere might have been combination of theoryand experiments to different,these characteristics must also be allow calculations of fluxand saltation tra- known for a range of atmospheric densities. jectoriesfor a wide range of conditions.

3.1 METHODOLOGY Wind Tunnel Experiments

Our approach was to run wind tunnel experi- Experiments were performed in an open- ments to determine as many of the particle circuit wind tunnel (Fig. 6) using particle

Diffuser

Boundary. layer probe

Figure 6. Diagram of the Martian Surface Wind ntnnel (MARSWIT). MARSWITis an open-circuit, atmospheric boundaly layer tunnel 14 m long, with a 1.1 m2 test section. The tunnel is housed in an environmental chamber in which the atmospheric pressure can be varied from 1 bar to about 3 mb formartian simulations; at low pressures, carbon dioxide is used as the atmosphere, appropriate for Mars.

I 12 collectors to determineparticle flux and the surface.Each collector was a wedge- instruments to determineparticle velocity. shaped box, open at the upwind end to sam- An experimental matrix was developed com- ple a cross sectional area of 2.4 cm2 perpen- bining atmosphericpressure, particle diame- dicular tothe flowdirection. The larger, ter,freestream wind velocity, and height downwindend was covered with 40 pm above the surface (Appendix 1). Atmospheric mesh. A small baffle was placed in the collec- pressures of 1 bar (Earth case) and an average tor to prevent particles from bouncing out of of "7 mb (Mars case) were used. Three par- thefront end or damaging the screen. The ticle sizes were selected for analyses: 71 5 pm, design used here was developed to maximize representing coarse sand, 350 pm for typical the collection of the particles with minimum dunesand, and 92 pm, the size most easily interference of the air flow, although no par- moved bylowest strength winds (Fig. 3). ticle collector is 100% efficient.

For both the Earth and Mars experiments Particleflux (9) asa function of height a wide range of freestream wind velocities above the surface was determined as the mass was selected,including high speeds to simu- of particles caught per unit area of the collec- late storm conditions when substantial aeolian tor perunit time. Although the collectors abrasion probably occurs. Crushed and sieved couldbe stacked to sample the entire cross walnut shells were used inexperiments to section of the wind tunnel from the floor to simulate Mars. Theirlower density helps to the ceiling, preliminarytests showed erratic offsetthe differencesin gravity between particle distribution at heights above 65 cm, Earth and Mars,as discussed below.Quartz evidently as aresult of particlesbouncing sand wasused interrestrial experiments. fromthe ceiling. Because of thisbehavior and because very few particles occur at these Particlecollectors (Fig. 7) were used to heights,collection in most experiments was determine flux as a function of height above madefrom thesurface to 65 cm above the surface.Total particle flux, Q, was determined by summing the masses caught in the individual collectors.

Particlespeed was determinedwith a velocimeter (Schmidt, 1977). The velocimeter producesa light beam perpendicular tothe wind stream,focused ontwo light-sensitive

ENTR semiconductors that detect the shadow of any particle as it crosses two separate portions of the light beam (Fig. 8). Parallel windows limit the field of view of the light receivers; an elec- AIR FLOW AND PARTICLES tronic unit connects the two phototransistors such that a particle passing through the light beam produces a positive voltage pulse as it Figure 7. Diagram of particle collector used to deter- shadows the firstwindow, and a negative mine flux. Collector is madeof plexiglass; windtunnel pulse atthe second window. To determine smoke tests were carried out to determine optimum angle for the wedgewhich would allow maximum particle speed the time interval between posi- efficiency for particle collection withminimum inter- tive and negative pulses is measured with an ference to air flow. Baffleprevents particles from oscilloscope. Thistime interval, in conjunc- bouncing back out the frontor damaging the screen. tion with the known separation of the sensor Tabs on the sideenable collectors to bestacked. windows,enables the particlevelocity to be 13

\ ,PHOTOTRANSISTORS Waddell Creek State Beach, about100 km south of San Francisco. The beach consists of wind-transported marine sands and was readily accessible to equipmentrequired for the experiment.In November 1981, a Hycam 16 mm motion picture camera was installed to record saltating particles at framing rates ,...,. -. . YYL up to 8000 framesper second. Concurrent measurements of wind velocity were made I DOWNWIND 1m above the surface.Particle collectors !? VI were placed in the wind stream to determine particleflux. Seven successful experiments Figure 8. Schematic diagram of particle velocimeter; were conducted for wind speeds ranging from deviceconsists of alight source and two photo- about 4.5 to 7.0 m/sec. transistors; as a particle passes through the beam, the upwindphototransister registers a positive voltage, then the second (downwind)phototransister registers 3.2 PARTICLE VELOCITY anegative voltage; the differencebetween the two signals yields time (hence velocity); the amplitude is Most windblown particles near the ground are a function of grain size (from Greeley et al., I982). transported in saltation. The path that a saltating grain follows is called thesaltation trajectory (Fig. 9). The velocity of the grain determined. About two hundred data points generally increases throughout its trajectory, were obtainedfor each run. To investigate although by the time it has reached maximum ratesofrock abrasion corresponding to height above the ground, most grains will have various heights above thesurface, particle achieved about 50% of their maximum veloc- velocities were measured at various heights ity.For most of theirtrajectory, saltating particles travel at some fraction of the free- above the wind tunnel floor. stream wind speed. Because meteorological data provide only wind speeds (not particle High speed 16 mm motion pictures were velocities) it is necessary to knowparticle taken for some of the Earth-case wind tunnel experiments to determinesaltation trajec- toriesand to validate the velocities determinedwith the velocimeter(Greeley et al., WIND PROFILE 1 1982, 1983). The camera was positioned per- WIND- + pendicular to the particle flow and recorded I i? the saltation cloud at 2,000 to 10,000 frames u1 I persecond. The films were studiedwith a Vanguard Motion Analyzer using film timing marks at 0.00 1-second intervalsrecorded during the experiment.

Field Experiment

A field experiment was conducted to provide Figure 9. Diagramshowing two typicalsaltation a calibration of the wind tunnel results under trajectories; solid path correspondsto low wind speed natural aeolian conditions using the same (profile I) and has low (HI)and short path (LI)in techniquesand instruments as employedin comparison to path of particle driven by high wind the laboratory. The field site was located at speed (profile 2). 14 velocity as a function of wind speed in order to calculaterates of abrasion. Furthermore, at any given instantand at any given place within the saltation cloud there is a complex spectrum of particle paths and speeds. Thus, it is also necessary to know the velocity dis- MARS CASE tributionsof particles as functions of grain 350pm SHELLPARTICLES size, wind speeds,and heights above the umz65rn/a ground.

Figure 10 showstypical velocity distri- 15 10 5 20 25 30 35 MEANPARTICLE VELOCITY (rn/sl bution withinsaltation clouds in terrestrial I I I I 0 10 20 simulations, measured at a single height above MEAN PARTICLE VELOCITY30 40(%Urn) 50 the floor of the wind tunnel. Velocities are given in m/s and as percentages of the free- Figure 11.Particle velocities for four different stream wind velocity. In general, particles heights above the surface for 350 m walnutshell travelling at greater heights are also travelling particles subjected to a freestream wind speed of 65 at higher speeds. Evidently, this is a reflection mls (u, = 3.4 m/s) in a low atmospheric pressure (6.6 mb) to simulate the martian environment of the increasein wind speedwith height (from Greeley et al., 1983).

380pm OUAllTZ above theground, through the turbulent l0.lcm HElOHT boundary layer. In addition, because particles at greater heights tend to have longer saltation trajectories, there is a longer interval of time for them to be accelerated by the wind. Note also thatsome grains travel at velocities exceeding the wind speed, which probably results from elastic rebound of grains in saltation. Figure 11 shows similar results obtained in experiments simulatingMars.

Figure 12 comparesvelocity datafor

PARTICLEVELOCITY (mlmoc) particles on Earth with those of Mars under similar dynamic conditions for both low and I J 0 20 40 no 80 100 120 high wind speeds. ‘Low’ refers to speeds

X OF FREESTREAMWIND SPEED about 25% greaterthan threshold; ‘high’ refers to stormconditions, where winds are about twicethreshold. In all cases, particles Figure 10.Dism’bution of particlevelocities at a travel at a higher percentage of the freestream height of 16.1 cm above the surface for 350 pm wind speedon Earththan on Mars. This diameter quartz grains subjectedto a freestream wind suggests morea efficient coupling of the speed of -10.22 mls a simulation of Earth (u* in = particleswith the wind in the denserterres- 0.6 m/s). The velocity distribution is typical and can be expressed as a mean value with an associated one trialatmosphere. Because wind speedsare standard deviation range about the mean, as shown. about an order of magnitude higher on Mars Later figures will present only the mean and range for particle motion, the grains would have to (from Greeley et al., 1983). beaccelerated to amuch greater speed on 15

in the wind flow when the wind speed is below threshold (White et al., 1976). How- ! EARTH ever, when minimumthreshold speeds (u,,) arereached, the optimum-sizeparticles (Fig. 3) begin to saltate, which causes the viscous sublayer to degenerateand form a

I MARS fully turbulentboundary layer.

Ourcomputational model of saltation trajectories is based on the equations of motion of a single particle (White andSchulz, 1977; Fig. 13):

0 155 10 20 25 30 HElGHTABOVESURFACE,cm

Figure 12. Average velocity of particles 350 and 92 pm in diameter as a function of height above surface at IO3 mb pressure (Earth case) and 6.5 mbpressure (Mars case); both casesare run at freestream wind where m is the particle mass, u is the wind P. speeds just above threshold (‘low’u,) and for storm velocity, (x, 9) are the particle’s horizontal conditions (‘high’ u~). and vertical velocity, (x, y) are the particle’s horizontal and vertical acceleration, and vr is the velocity of the particle relative to the Mars in comparison to Earth in order to reach velocity of the wind and is equal to the same percentage of the freestream. Even [ (2- u)* + 9’ 3 s. Aerodynamic force on a thoughthe saltation trajectory is longeron particle is proportional to its cross-sectional Mars, they may not be in flight long enough area, A, times the difference in pressure, Ap, for an accelerationcomparable to Earth. across the particle. Bernoulli’s equation states However, the lowerpercentage could also resultpartly from experimentalerror. The

saltation path length on Marsis longerthan Y on Earth (discussed below)and in the wind tunnel the saltation cloud in martian simula- tionsmay not becomefully developed because ofthe limited wind tunnellength.

3.3 SALTATION MODEL

A computational model of particle saltation was developed in order to determine velocities and trajectories of particles for a wider range of conditions than could be simulated in the laboratory. In themodel, aone-dimensional flow regime is assumed for calculations of the Figure 13. Force diagram for a saltating particle. trajectories. The velocity inthe vertical (y) L denotes lift force;D, drag force; up is the instanta- direction is zeroand the velocity, u, in the ’ neous particle Velocity and vr is the veiocity of the flux(x) direction is a functionof height particie relative to the wind velocity (modified fiom above the surface. A viscous sublayer develops White andSchulz, I977).

I 16

producedby spin) are taken intoaccount (White andSchulz, 1977). The percentage where pa is the atmospheric density and v is values at various positions along the trajectory the air velocity. Hence are theratio of particlespeed tothe freestream wind speed. In general, the percentage L a 1/2 A pa vr2 of freestream wind speed that particlea reaches is aboutthe same for all calculated D a 1/2pa A vr2 (6) particles at similar positions intheir trajectories. Particles accelerate to nearly half of Assuming the particles are sphericaland of the freestream wind speed in the initial lifting uniform density, we can write phase of their trajectories, which corresponds to about1/4 of their path length, L. Note, L= 1/8 CL~~T Dp2 vr2 however, that particles more nearly approach freestream wind speed on Earth than they do D = 1/8 CD pa 7~ Dp2 vr2 on Mars, similar tothe resultsdetermined experimentally. where CL is the coefficient of lift, CD is the coefficient of drag,and Dp is the particle 3.4 PARTICLE FLUX diameter. We can now substitute (7) and (8) into (2)and (3) and rearrange Numerousinvestigators have analyzed mass transport, both from experimental and theo- reticalapproaches. Bagnold (1 94 1), for example,carried out experiments to determine Q and then considered the flux in terms of momentum loss of the air due to sand in saltation.Sharp (1 964) collected grains within saltation clouds at various heights over longperiods of time,then determined the total masses andparticle size distributions; The drag coefficient of a sphere is strong- however,concurrent wind data are not lydependent upon the Reynold's number. available. Thus,inorder to obtaindata The empirical formulas of Morsi and Alexan- appropriate for both Earth and Mars, experi- der (1972) are used for its determination in ments were carried out andnumericala our numerical calculations. The initial condi- model was developed. tionsare the same as those used by White et al. (1976), in which the particle is at rest In ourexperiments, particles were placed onthe surface and, as the wind speed is inthe wind tunnelupwind of an array of increased above threshold,particles begin to collectors(Fig. 15). The collectors have a move. For calculations in the martian atmo- remotely-operatedtrap door that can be sphere it is necessary to includethe effects openedand closed to sample thesaltation of slip flow caused by the low density atmo- cloud for adiscrete interval of time.The sphere,although the slip flowfactor only wind speed was increased and held steady to becomessignificant forthe smaller-size produceafully developed saltationcloud. particlesin the calculations (White, 1979). Thetrap door was thenopened to sample thesaltating particles. At the conclusion of Figure 14 displays typical computed parti- the run, particle masses were determined for cle trajectories for Earth and Mars when lift each collector to obtain values of q for each forcesdue to the Magnus effect(lift force height. 17

SALTATION PATH LENGTH, rn

EARTH MARS VELOCITY, PERCENT, VELOCITY, PERCENT, POSITION rnlsec vca rnlsec v, 1 2.6 10 5.3 5.3 2 4.7 18 12.9 12.9 3 7.2 28 22.6 22.6 4 12.6 49 37.3 37.3 5 15.4 60 53.2 53.2 6 18.8 73 61.8 61.8 7 22.1 90 65.7 65.7 8 23.1 90 67.6 67.6 9 23.5 91 10 23.6 91 11 23.7 92 12 23.7 92 13 23.5 91

Figure 14. Gzlculated trajectory for grains in saltation on Earth and Mars for wind speeds about 25% above threshold, showing much longer trajectory for martian case. Positions along the trajectory show the instantaneous velocity (shown in both m/s and as a percentage of freestream wind speed) ofthe particle. Note that vertical scale is greatly exaggerated.

Figure 16 shows q as a function of height determiningflux in terms of numbers of for 92 pm grains at three wind speeds in the particles, the experimentaldata for flux wind tunnel compared to results obtained in mass must be adjusted (Greeley et al., 1982). the field experiment at Waddell Creek where Thus, in Figure 18the gravity-adjusted data the average grain size was about 300 pm. areshifted upwards by a factor of 2.6(the In general, there is an increase in flux closer density difference between walnut shells and to thesurface, and at greater freestream particlesexpected on Mars, such as basalt). velocities. As in the terrestrial cases, there is a general decrease in particle flux with height above the Figure 17 showsexperimental results surface. In addition there is an abrupt break for Mars to determine q for 92 pm particles in the flux at about 30 cm above the surface. versus height for wind speeds ranging from 58 This break may represent the transition from to 123 m/sec. Inthese and other martian grains that aretransported primarily in simulationsparticles composed ofwalnut saltationnear the surface to grains moving shells approximately one third the density of mostly in suspension at levels above 30 cm. grains expected on Mars were used to com- pensate partly for the lower martian gravity Total mass flux, Q, was determinedby (Greeley et'al., 1977a).Although this tech- summingincremental flux, q, forthe total nique isvalid forthreshold tests and for column sampled. Figure 19 shows total flux, 18

Model for Calculation of Flux

An estimate of the total amount of surface material transportedby the wind maybe madeby examining the physics of the process. Thekey parameter in estimating surface-material movementrates is the charac- teristic pathlength of saltatingparticles. Several investigators have developed both empirical and theoretical methods to calculate path lengths(Bagnold, 1941 ; Kawamura, 1951 ; Tsuchiya,1969). Tsuchiya has a theoreticaldevelopment based on successive saltating leaps. However, it is necessary to

10- +

+ EARTH CASE, U, 92 pm QUARTZ 6.92 mlsec 10-2 0 Figure 15. View of stackedparticle collectors (1) .+ 0 9.0 m/sec used to determine flux of saltating grains. Samples + + 11.94 mlsec are taken in intervals of 2 cm above the floor; trap P - 300 pm BEACH SAND N X 7.1mlsec doordevice allows precise timing of collection E P .+ during filly-developed saltation cloud. Also shown is & particle velocimeter (2) and electrometer probe (3) e + 1o-~ 0. used to measure electrical current generated by wind- 2 -I blowngrains. Air flow is from right to left (from U + w Greeley et al., 1982). J 0 u XX I- + U a L ox x+. 0 Q, for freestream wind tunnel speeds ranging 1o-~ X + from -6 to 12.0m/s for 92pm quartz x. particles,simulating terrestrial conditions. 0 X Martian simulationsare shown in Figure 20 + which give total flux, Q, for freestream speeds 0 ranging from about 55 to 90 m/s for 92 pm; 10-5 bothexperimental and gravity adjusteddata 0 10 20 30 40 50 are shown. Few particles in suspension were HEIGHT ABOVE SURFACE, cm counted. Ingeneral, there is an increasein fluxwith increasing wind velocity.Thus, Figure 16. Particle flux, q. of 92 pn quartz grains QS flux is higher on Mars thanon Earth for a finction of height for threefreestream winds dynamically similar conditions (same u*/u*t), (Earth case), showing general increase in flux with because ofthe higher wind speedsrequired wind speed, and flux for 300 pm particles as deter- on Mars for particle entrainment. mined from field experiment. 19

10-ly MARS CASE, 92 pm SHELL To = rs + rw

Bagnold (1941) and Kawamura state that the stress on the surface caused by the impact of saltating grains is equal to the momentumlost by thegrains:

\ 54-81-C 58 mls where Go is the number of grains landing per unit area per unit time, uf is the final grain speed at surface impact, and uiis the initial grain speed at particle lift off.

It can be assumed (Kawamura, 1951) that there is a linearrelationship between the 10-51 I I I I I I I 0 10 20 30 40 50 60 70 particle momentum lost occumng in the HEIGHT ABOVE SURFACE. cm horizontal and vertical directions

Figure 17. Particle flux. q, for 92 pn particles as a function of height for 5 wind speeds in the martian wind tunnel at -6 mb atmospheric pressure. MARS CASE, 92 prn SHELL GRAVITY ADJUSTED know the initial or maximum vertical component of velocity at lift-off for the particle in order to calculate its path length. This com- 10 ponent is generally a function of the flow conditions.Formulation of this type, with 0 thepath length a functionof the vertical I N velocity, does not lend itself well to calcula- E e tions of surface material movement. For our 10 application it is useful to express thepath d 2 length in relation to thethreshold shear 3 stress exerted on the surface by the wind, as U presented by Kawamura (1 95 1). The surface \ shear stress ro consists of two elements. The 5-4-81 -c first is the ordinary wind shear, rw, caused by 58 mls the motion of thewind flow over the surface. The second stress, rs, comes from the impact of particles colliding withthe surface. lo-\ 10-441 0 10 20 30 40 50 60 70 Kawamura found that rw increases with HEIGHT ABOVE SURFACE, cm increasing wind frictionspeed, u*, until wt is reached, then it remains approximately Figure 18. Same data as in Figure I7 corrected for constant ata value of rt, hence, martian gravity. 20

.-e Substituting (1 5) into (14) into (13) yields

~3=2[GoWi (1 6)

EARTH, 92 pm PARTICLES OEXPERIMENTAL Substituting into(1 2) .4 -THEORETICAL .FIELD EXPERIMENT

P i P E, .3 0 x 3 U MARS, 92 pm PARTICLES VI VIa -THEORETICAL 5 .2 0 EXPERIMENTAL a EXPERIMENTAL k 0 GRAVITY-ADJUSTED k

0 I I I I I 7 8 9 10 11 12 FREESTREAM WIND SPEED, m/sec

35 45 40 50 55 60 WIND FRICTION SPEED, cm/sec

Figure 19. Particle flux, Q, as afunction of freestream wind speed, u, for 92 wn quartz particles in 0 Earthwind tunnel simulations comparedwith flux predicted by Equation 33.

where g is a constant,and u and w are the horizontal and vertical components of the

~ particle velocity, v, respectively. 5!? 60 70 80 90 FREESTREAM WIND SPEED, m/sec I I I 1 I Afterthe collision withthe surface a 4.5 4.0 2.5 3.5 3.0 bouncingparticle traveling vertically with WINDFRICTION SPEED, mlsec speed wi will again return to the surface with the same speed wf = -wi onthe average. Figure 20. Particle flux, Q,for 92 ~.unparticles in Hence experiments, as adjusted for Mars gravity, and compared withflux predicted by Equation 33 (from Greeley et al., 1982). 21

From the definition of surface friction speed Because iii 0: Wi (from 14)

If2 U* 3 [~o/~al (1 8) where pa is theatmospheric density, Equa- tion (1 6) can be rewrittenas Substituting(22) into(26) andcollecting 2 Go Wi = pa(u*, - u,~') (19) constants

From field studies of saltation,Kawamura (1951) found that

Go = K1 Pa (u* - U*t) (20) We now have the necessary information to calculate a total saltation lane flux, Q. If we which also agreed withnumerical solutions consider a zone with length h, downwind of a of particle motion (White, 1979). Substi- unitwidth lane, no particlecan possibly tuting (20) into (19)yields landin thatzone without first passing over that lane width; also, no particle can pass over the lanewidth and fail to landin the zone. Hence

Thus, the amount of material passing over the lane width is equal to the amount of material landing on the surface per unit area times the length of the zone. We can substitute (20) and (27) into (28) The average saltation path length, h, can be calculatedfrom this result in the following manner.From elementary physics we know that the time aloft for a particle with initial upward velocity, wi, is The final form of the theoretical total flux is t = 2 Wi/g (23) where t is time aloft and g is the acceleration due to gravity. During timet, the particle will be moving downstreamwith an average On Mars, thereappear to be different velocity Ui. Applying basic physics equations interparticleforces as well as effectsfrom of distance traveled, the saltation path length, changes inReynolds number. These changes L, is will alter the saltationcharacteristics by affecting the thresholdfriction speed. The flow field around individual particles (Knud- sen and Reynolds numbers effects), shearing Substituting (23) into (24): rate within theboundary layer, and the existence of a comparatively large (to Earth) viscous sublayer fornonsaltating flow will cause change in threshold values. 22

Using Equation 30, a comparison between single line in support of theidea of auniversal Earth and Mars can be made function describing the movement process. Hence, the following equation can be used in estimating mass flux on either Earth or Mars once the friction speed and threshold friction speed are known:

P Q = 2.61 -5 (u, - u,~)(u* + u*~)?(33) where the C's are constants of proportional- g ity. Assuming that the ratio of friction speed Theflux of material on Marsis consis- to thresholdfriction speed, u*/u*~,is the tently higher than on Earth, if the frequen- same for both Earth and Mars (most areas on cies of winds abovethreshold are the same, Earth are windier than Mars): in spite of the lowermartian atmospheric density,due to: a) the higher u*~required on Mars, b) the lowermartian gravity and c) the typically longer saltation path lengths on Mars (discussed above). FromEquation 32 the difference in Q between the Earth and White (1979) has shownexperimentally Mars may be expressed as thatunder martianconditions substantially less stress is needed to move materialthan on Earth. This result supports the numerical results that show similar increases in particle path lengths on Mars and that the path length is almostdirectly related to the mass flux. For example,the relationshipbetween the Thus, it would be useful if Earth and Mars materialmovement on Earth andon Mars, cases could beadequately described bya for equal ratios of u*/u*~of 1.47, would be single universal equation. According to the QM/QE = 6, or 6 times as great on Mars for analyticalderivation, the onlydifferences 200pm particles.A direct estimate of flux between the relations describing the two cases are the empirical constants CE and CM. 1.2 Irn If theconstants of proportionality CE and X CM are the same for both cases, this would indicatethat the physics between thetwo testsremained unchanged, and would imply a universality tothe mass flux process. AS one would expect, the process of movement of surface material on Mars should be nearly the same as the process on Earth. Hence, the constants of proportionality between the two 0 .1 .2 .3 .4 .5 FRICTION SPEED PARAMETER, p(u, - uSf) test cases should be nearly the same.From (u, + u,J2/g. grn/(crn X SI the wind tunnel data of both cases, a unitless empirical constantwith a value of 2.61 was Figure 21. Particle flux, Q, as a function of friction determined (White, 1979). speed parameter; one bar (Earth) and low pressure (Mars) datacolrclpse to a singlecurve which is the Figure 21 displays the mass flux versus straightline on the plot. Note the origin for both a functiona of frictionspeeds. Here, both cases are data points, as they are used in determining high- and low-pressure data collapse to a thethreshold friction speeds (from White,1979). 23 rates on Mars can be made by employing the them to achieve higher velocities. This,in results shown in Figure 21. To obtain a nu- turn, results in higher rates of mass transport. merical value, the thresholdfriction speed Because of the higher wind speedsrequired mustbe known. The value will vary with for threshold on Mars, both particle velocities size of particles but, once known, adirect andfluxes are higher thanon Earth, if the estimate of the mass flux can be made. frequency of wind above threshold is the same. However, particle velocities seldom are 3.5 SUMMARY greater than about 50% of freestream velocity In general, increases in freestream wind speed on Mars, in contrast to Earth where particle result in increases in the trajectory height and velocities commonlymeet and sometime pathlength for grains insaltation, causing exceed freestream velocity. 4. SUSCEPTIBILITY OF ROCKS TO AEOLIANABRASION

Rocks and minerals of different compositions (1981) have conductedexperiments of and texture should have different resistances abrasion in a geological context. to abrasionby windblown particles. In this section we review previousstudies of wind 4.2 THE MECHANICS OF ABRASION abrasion, discuss the mechanicsof abrasion, and present the results of laboratory experi- Themechanisms by which aeolian particles ments that were conducted under controlled disruptand remove materialin the abrasion conditions in order to derive values that can process is not well understood. Discussion be applied to estimates of abrasion on Earth here is limited to aeolian abrasion of brittle and Mars. materials (as opposed to ductile), appropriate for rocks and minerals, and is based primarily 4.1 PREVIOUSINVESTIGATIONS oninterpretations ofSEM photographs of materialsabraded under laboratory condi- Exceptfor the work of Kuenen (19601, tions.Not all particleimpacts lead directly very few studies have been made to quantify to erosion; some may damage the surface but aeolian erosion of natural materials, although not necessarily remove material. Under most ventifacts have been long recognized and conditionsmaterial removal occurs where described. Abrasion of metals, glasses, and propagating fractures intersect. ceramics is important inmanufacturing and industrial processes. Much research has been FracturePatterns Produced By Rounded doneon the topic for engineering applica- Particles tions. Theeffects of targetcomposition, impactor size and composition, and impact Well-rounded particlessuch as sand grains angle have beeninvestigated experimentally impacting the flat surface of an elastic, brittle (Neilson and Gilchrist, 1968; Sheldon and surfacebecome flattened at the contact and Finnie, 1966b).Experimental and theore- the targetsurface is bent downwards.The tical studies of the mechanisms of material area of contact increases with particle radius removal for both brittle and ductile materials andapplied load. If the load exceeds a were conductedby Adler (1976a, 1976b), critical limit,circulara crack - called a Smeltzer et al. (1970), Sheldon and Finnie Hertzianfracture - developsaround the (1966a),Sheldon (1970) and Sheldon and indentationsite, generally at a radius 10 to Kanhere (1972) anda numerical prediction 30% greater than the contact radius (Johnson of abrasion on both ductile and brittle mate- et al., 1973). In its ideal form,the crack rials was made by Finnie (1 960). The mech- extends a short distance vertically intothe anism ofabrasion of brittle fused silica was specimen before turning outward and propa- made by Adler (1 974).The behavior of gating into aconical frustrum,the depth of impacting particles during abrasion was stud- which increases with increasing load. ied by Tilly and Sage (1970). The initial and ‘steady-state’ erosioncharacteristics of a Andrews (193 l), Knight et al. (1977), target in response to particleimpact was andothers have shown that the surface dia- noted by Tilley (1969) and discussed by meter of a Hertzian fracture (DH) is constant Marshall (1979).Suzuki andTakahashi andindependent of the impacting grain 25 velocity, a result in agreementwith theory forperfectly spherical impactors. However, for well-rounded, but irregularly-shaped particles, the meandiameter of a population of cracks increases withvelocity (Marshall, IMPACT PATH OF 1979; Fig. 22a). Because the impacting grains were not perfectspheres, the targets were subject toindentation with a wide range of radii andat low impact velocities only the smaller radii (which concentrate the impact stresses) were capable ofproducing Hertzianfractures. At high impact velocities (vp) both large and small radii were effective and mean crack diameter (DH) increased with velocity. The relationship DH a vpo .47 was derived by Finnie (1960), which in comparison tothe relationshipbetween maximum Figure 23. Effect of particle trajectov on opening angle, 8, defined in inset (a, top diagram), diameter contact radius (am)and velocity (am a (b, middle diagram) and Orientation (c, bottom dia- vpo m4 ), implies thatthe cracksformed at gram)of Hertzian fractures. Each point represents the mean of 100 measurements (afterMarshall,1979).

about the time ofpressure release. Johnson et al. (1973) observed cracks smaller than maximum contact size formed during unloading and attributed the behavior to an elastic mismatchbetween indentor andspecimen which affects the stress field via frictional E 20 401 forces at the contact interface. ~ 0 5 10 15 20 25 30 VELOCITY, ms-’ Figure 22bshows theeffect ofparticle size (Dp) on the surface diameter of Hertzian fractures forimpact of irregularly-shaped 200 1 - particles. Perfectlyspherical indenters give rise to anexponential relationship of the 5160 - LT w- form DH a Dpm when Dp < 7.0 cm and a + - linear relationship when Dp 7.0 cm (Tillett, g 120 > 5- 1956). 0 80- a- [r At impact angles less than 90” the tangen- 0 40 - tial component of force introduced into the - stress system inhibits cracking in front of the 1lllllllllJ ~~ 0 400 800 1200 1600 2000 grain and the result is an arc-shaped or appar- PARTICLE DIAMETER,prn ently incomplete Hertzian fracture (Fig. 23). As the angle decreases theopening angle 0 Figure 22. Effect of particle velocity (a, top diagram) increases (0 = the angle subtendedby the and diameter (b,bottom diagram) on Hertzian - fracture diameter; each point represents the mean of ‘horns’ of the crack at an imaginary center of 100 measurements of quartz grains impacting on curvature of the arc), the diameter decreases, quartz plates (after Marshall, 19 79). and the preferred orientation of the individual 26

crack - increases (orientation is defined as thedirection pointed to by aline bisecting 8 1.

FracturePatterns Produced By Angular Particles

Angular particles do not abrade in the same manner as roundedparticles. A great diver- sity of erosion patterns can arise fromonly minor variations inparticle shape and from the manner in which the particle strikes the surface. Lawn and Wilshaw (1975) consider the fracture propagation shown in Figure 24 to be a basic sequence during normal loading of surfacea with sharpa point. Initial contact induces a zone of inelastic deforma- Figure 25. Surface pattern produced by intersection tion(crushing and/or plasticdeformation) of lateral and median vents. and at some threshold stress initiates avertical crack (or medianvent) which propagates influence of neighbors; several intersecting downwardwith increasing load.Upon the surfaceproduce star-shaped patterns unloading the medianvent closes (but does (Fig. 25). Inanisotropic systems the vents not heal)and lateral vents develop which tend to follow cleavage directions. continue to propagate after removal of the impacting grain. Chipping can result if the lateral vents reach the surface of the indented TheInfluence Of FracturePatterns On specimen(Fig. 25). Oftenthe medianvent Abrasion Rates will break through to the surface of the specimen and give rise to a radial fracturetrace. Abrasion rates of ventifacts reflect the stage The length of the surface trace is an indica- of surface-textureevolution. For rounded tion of the depth of the vent. For isotropic particlesimpacting an initially smooth sur- materials thenumber of medianvents is face, texturaldevelopment andrelated limitedonly by themutual stress-relieving abrasion have the following evolution. During theembryonic phase of fracturenetwork development, abrasion is inhibited (Fig. 26). Atsome critical spatial density of conical frustra, material becomes easily removed, but abrasion diminishes thereafter because the isolatedfrustra aredifficult to remove. Impact on the top of a cone apparently only deepens thestructure, whereas cone spacing prevents contact of roundedparticles with surfaces between the cones. Abrasion on the final surfaceshould besomewhat less rapid Figure 24. Crack propagation below a pointed parti- thanduring cone removal because surface cle; ZID is zone of inelastic defomation; MV is roughnessprohibits cone formation - such median vent;L Vis lateral vent. fracturesprobably have deeperpenetration

... 27

allows impact velocity, impact angle, impact- ingparticle size and type, sample type, and atmosphericdensity and composition to be controlled(Appendix 2). Marshall (1 979) found that anomalous weight losses occur in samples that were not pre-abraded and that have not attaineda steady state in aeolian abrasion.Thus, all targets were pre-abraded prior to experiments to establish fracture patternsin their surfaces. Each target was weighed to kO.1 mg before and after each run. Dust clinging to thetarget was removed by compressed air. Thefraction of thetotal mass of particlesimpacting any target was Figure 26. Mass lost in aeolian erosion as a finction determined from the total mass expended in of time; profire in lower part of diagram is deduced from SEM obsewations of Marshall (I 9 79), compared theexperiment by calculating the angle withlaboratory experiments of Adler (1976b). subtendedby the target from the geometry of theapparatus and dividing by 360".A check onthe calculatedimpact mass was than cleavage-controlled chipping.Figure 27 made by placing aparticle catcher in the shows scanning electron microscope photos of place of one of the sample stations. fourevolutionary phases inthis sequence; a,b, c and dcorrespond to the four stages Quartzsands were used as the abrading signalled in Figure 26. Although this sequence particles in most of the experiments because of events is dependent upon the initial surface they are readily available, theirbehavior is of the material, it may be typical for venti- fairly well understood, and nearly all previous facts if periodsbetween abrasion permit abrasion experiments have involved quartz. chemical smoothing of the roughened abraded However, basalt sands and sand-size aggre- surface. For a ventifact impacted with angular gates of fine grains .were also utilized in some particles, the textural development may pro- experiments (Appendix 2). ceed in themanner depicted in Figure .28. Susceptibilities to abrasion were assessed in terms of five parameters:particle size, 4.3 ABRASIONEXPERIMENTS particlevelocity, angle of particle impact, atmosphericdensity, and impacting particle The susceptibility to abrasion (Sa) for differ- composition. ent materials is expressed as the ratio of the material mass eroded to eitherthe mass of Sa Versus Particle Diameter impacting particles or the number of impacting particles. Sa varies withparameters such as impacting particle velocity (v,) and angle Figure 30 shows Sa as a function of diameter of impact. Angle of impact is defined as the of impactingparticles (DP = 75,105, 140, angle between the surface and the impacting and 160 pm) for basalt, obsidian, and hydro- particleand is thecomplement of the inci- cal. Particles impacted at an angle of 90" at dence angle. a velocity of 20 m/s in a CO, atmosphere of 5 mbpressure, conditions comparable to a In order to determine Sa values for vari- martianenvironment. The effect of particle ous rocks under a wide range of conditions, size in the range analyzed is fairly well dean apparatus(Fig. 29) was fabricatedthat fined bya slope of "3 on a log-log plot 28

Figure 27. Stages in erosion of quartz plate, impactedby quartz particles 710-840p in diameter at speedsof 20 to 2.5 mIs, a) embryonic stage, showing tiangular, crescentic, and polygonal depressions, b)youthful stage, showing surface representedby cone tops, c)advanced stage, showing curved escarpments, conchoidal fractures. irregular and cleavage-controlled fractures, andd) mature stage, showingblocky, cleavage-controlled structure. Frame widths are: a)420 pm, b)420 pm,c) 455 p,and d)150 pm.

givingrelationship theimpacting normal totarget the in a CO, atmosphere at 5 mbpressure. Although the Sa a Dp3 (35) effect of particlevelocity (vp) is poorly defined for volcanictuff (both welded and nonwelded) and rhyolite, the graphs for most Sa VersusParticle Velocity materials showslope a of 2 on a log-log plot, giving the relationship Figure 3 1 shows the effect of impact velocity on Saa for wide range ofmaterials; rocksand Sa a vp2 (36) particlevelocities ranged from 5 to 40 m/s. Particles were 125 to 180 pm quartz grains, Combiningrelationships 23 and24 gives 29

Sa 0: Dp3 vp2 (3 7) Sa Versus Particle Impact Angle

which shows that mass lostper impact is Figure 32 shows the effect of particle impact directlyproportional to thekinetic energy angle on Sa for the test materials; Figure 33 (KE) of the impacting particle. shows relative erosion as a function of impact angle of ideally ductile and brittle materials KE = 95 mp vp2 , mp a Dp3 (a.fter Ohet al., 1972)for comparison. Most of the rocks and materials tested appear The very slight concave-upwards curvature to mimica combination of typicalbrittle of the sizeeffect plots and many of the velo- andductile behavior, except obsidian which cityeffectplots isan indication that some behaves as a brittle material. factorother than kinetic energy maybe involved. This may be the size of the contact The high susceptibilityat low impact areawhich increases with both velocityand angles forsome materials is interpreted as particle size andincreases in the number of chipping and gouging of microblocks and the preexisting cracks encountered by the stress efficiencywith which theyare removed field. The behaviorof brick apparently is would increase as the angle declines. Fractures due to the development of a sintered crust at propagate to shallower depths as the vertical low and medium velocities which chipped off component of force diminishes. However, the in large pieces at high velocities. high susceptibility at high angles suggests an apparent high ratiobetween indenter and target hardness which could lead to crushing and powdering of the surface at high impact angles. Thismay provide some degree of protectionfor the surface at normal (90") impact and could account for the reduction in susceptibility. The downturn could also be due to impactingparticles rebounding from thetarget and interfering with incoming particles, butthis possibilityhas not been assessed fully.

C. Sa Versus Atmospheric Density

Tests to determine the effect of atmospheric pressure on abrasion (Fig. 34) were inconclu- D. sive. Generally at 1 barthere was slightly more erosion forthe three materials tested thanat martian pressure (5 mb). However, this may be related in some way to air turbu- lenceinthe confined abrasionchamber.

Sa Versus Particle Composition

Figure 35 shows Sa for basalttargets as a function of impact velocity forquartz, Figure 28. Possible fiacturing and chipping sequence basalt, and basalticash particles 125 to 175 of a surface impacted with angular particles. pm in diameter, impacting at 90". Although

I 30

thedata for basalt and ash particles are Experiments were also conducted using limited, there appearsto be little difference in aggregates as the agents of abrasion. A abrasion by quartzand basalt, whereas ash different apparatus than MRED was used, so particles appear to be less effective as agents theresults are not entirely compatible for of abrasion. comparisons;however, the critical result

Figare 29. View of the apparatus used to conduct experiments to determine susceptibilities to erosion (Sa). Szmples to be abraded are placed in holders (a) positioned within cylindrical chamber(b); cover for chamber (c) swings into position such that abrading particles (quartz sand, etc.) in hopper (d) are fed into rotating arm (e)where they are slung against the samples; speed of arm controls the impact velocity; sample holders (a) can be adjusted to control impact angle; 12 samples can be abraded per experiment; one sample station (f)is modified as an open port to col- lect the same amount of material as stnkes each sample to enable an accurate determination of particle flux. Control panel is shown in foreground; chamber can be evacuated to low atmospheric densities and carbon dioxide can be introduced for martian simulations(from Greeley et al., 1982). 31

I 0-3 0 0 GRANITE 0 BASALT A BASALT 0 + 0 # + OBSIDIAN A 0 HYDROCAL 8 m 0 1 o-~ 1 cn .cn z & >- A k -I g 10-5 k UI 2 10-3 RHYOLITE OBSIDIAN 3 v) 0 E 0 W i s 10-4 i 0 0 O. 0 Y + 0. I I 10 100

TUFF, NONWELDED TUFF, WELDED

"I I I I 0 25 50 - 100 250 LOGl0 PARTICLE DIAMETER, prn 0. 0 0 0 Figure 30. Susceptibilities to abrasion of basalt, obsidian, and hydrocal as a finction of particle 0 0 0. 0 diameter; susceptibility (Sa) is expressed as the ratio 0 0 of target mass eroded (g) tothe number of impacting z particles (i) (from Creeley et al.,1982). 0 0 0

was thatfor impact velocities below about I I 20 m/s, the aggregates were plastered against the target, causing a net gain in mass of the targetand forming a protective coating. At higher impact velocities, the coating sloughed off,but there was no measurable loss in 0 target mass. t 0 0 4.4 SUMMARY 0 0 Susceptibility to abrasion is highly dependent 0. onthe texture of therock, composition, velocity, and size of the impacting particles, 0 10 100 0 10 100 LOGlO PARTICLE IMPACT VELOCITY, rnls and the angle of impact.On a perimpact basis, glassy materials will erode very quickly for surfaces perpendicular to thewind; Figure 31. Shsceptibilities to abrasion of test mate- crystallinematerials such as granite and Mls as functions of particle impact vebcities (from basalt erode more quickly when surfaces are Greeley et al.. 1982). 32

PARTICLE IMPACT ANGLE, deg IMPACT ANGLE, deg

Figure 32. Susceptibilities to abrasion of test materials as fitnctions of particle impact angle (from Greeley et al., 1982). 33 sub-parallel tothe wind. Quartzand basalt grains appear to be of equal efficiency as 0 OBSIDIAN agents of abrasion; basaltic ash isless effi- A BASALT HYDROCAL cient. Sand-size aggregates of fine grains do 0 not abrade targets at low velocities, but form a veneer onthe surface which shields the target from abrasion.

"_ 7.5 mb

I I J 0 10 20 30 VELOCITY. rnls

Figure 34. Susceptibility to abrasion for three materials as a function of impacting particles velocity in a 1 O3 mb (Earth)pressure environment and7.5 mb (Mars) pressure environment; although the dataare limited, there is no apparent difference due to atmo- 0 I I I spheric density. Sa expressed as ratio of target mass 0 30" 60' 90" loss to particle mass impacting the target. IMPACT ANGLE, deg

Figure 33. Relative erosion as a fitnction of impact 0 angle for typicalbrittle materials compared with ductile materials (after Oh et al., 1972). The downturn at 90" impact angle for brittle materials is due 0. 0 to two factors. The first is that at 90" impact, virtu- 0 ally all of the impactor KE goes into fracturing the El surfaceand there is no lateral impetus to dislodge chips. The second is that, in abrasion machines and in nature, particles hitting a surface at 90" rebound straight back and interfere with incoming particles, protecting the surface from the full impact of sub- 0 sequent impactors. El BASALT TARGET 0 QUARTZ PARTICLES Figure Susceptibility to abrasion of basalt as a 0 BASALT PARTICLES 35. El 0 ASH PARTICLES function of impactvelocity by 125 to 175 p III I I I I I I particlesof quartz, basalt and basaltic ash (from 5 10 15 20 25 30 35 40 45 Creeley et al.. 1982). IMPACT VELOCITY, mlsec 5. WIND FREQUENCY DATA

Thethird parameter required for estimating as long as fourmonths, which integrated rates of aeolian abrasion involves various wind periods of wind activitywith periods of data, including wind directions, strengths, and quiescence. Unfortunately, on-site wind data durations.Ideally, wind datashould also were not obtainedduring the experiment, include velocity profiles so that surface shear althoughSharp recognized the desirability stresses could be derived andutilized for of such data. particlethreshold and particle flux analyses. In addition, wind data should be available for The nearestmeteorological station to a sufficiently long period of time to charac- Sharp’s experimentalsite was the Palm terize the local aeolian regime. Unfortunately, Springs airport, about 9 km southeast of the even minimal wind data are seldom available site.The station has recorded winds for for sites of geologic interest and various many years,including the interval covering techniquesmust be utilized toextrapolate Sharp’s experiment. As part of a general from other areas. study of ventifacts and other aeolian features at Garnet Hill, Hunter (1979) obtained wind 5.1 TERRESTRIALENVIRONMENT data at Whitewater Wash in the fall and winter of 1977, near Sharp’s site.Hunter’s results Wind data are usually obtained from meteoro- confirmed Sharp’s earlierspeculations that logical stations, such as those associated with the Palm Springs airportmay not be in a airports.Commonly, wind velocities with favorablelocation for providing wind data direction are determined for a single height - appropriatefor the experimental plot and by convention 10 m above the ground - and demonstrates the difficulties in extrapolating reducedin theform of wind rose diagrams wind data from one area to another. which showthe frequency, velocities, and 5.2 MARTIAN ENVIRONMENT directions of winds for a given intervalof time. Profiles of the wind velocity as func- During the past decade intensive observations tions of height are seldom obtained, yet this of Mars have been made from Earth and from is the critical value for determining u*. spacecraft which providea variety of data that canbe used to assess wind frequencies. In principle, it may be possible to extra- Viking landers were equipped with meteoro- polate wind data fromnearby recording logical instruments to provide dataon the stations to sites of interest. In practice such wind velocity, wind direction, and atmospher- an extrapolation is difficult to achieve. One ictemperature and pressure (Hess et al., sitewhere this was attempted is theGarnet 1976,1977; Ryan et al., 1978) and provide Hill area in southern California,a classic directinformation for two sites on Mars. ventifact locality (Blake, 1855). It is a site of Although Viking Lander 2 ceased operation frequent strong winds and active windblown in the late 1970s, Viking Lander 1, designated sand.Sharp (1964,1980) established long- the Mutch Memorial Station,continued to term experiments to monitor the erosion of monitor the winds and to obtain images until various rocks and man-made materials and to late 1982. Theprominent winds compare determine mass transport of windblown favorablywith aeolian features observed at grains as afunction of height above the the landing site (Binder et al., 1977) and from ground. Particles were collected for intervals orbit (Greeley et al., 1978). 35

Observations from orbit provide addition- indicates that dust was not raised over all the al clues to thefrequency of threshold-strength surfaceduring the genesis of thosestorms. winds. Relevant data include measurements of More generally,sand movement probably global and local duststorms and of albedo occurred to a greater degree in the southern changes onthe surface.Global dust storms hemispherethan thenorthern hemisphere apparently begin as local duststorms which (the two Viking landers are in the northern originate atlow or mid latitudesin the hemisphere),although precession will proba- southern hemisphereduring the summer. bly cause thisenhanced sand motion to During the firstmartian year of the Viking switchfrom thesouthern to thenorthern mission, two global duststorms were ob- hemisphere and back with a cycle of - io5 served, whereas only one suchstorm took years. place during the secondmartian year. Thus, threshold wind speeds are exceededin the In addition to grain movement that occurs source regions of global duststorms with a during global dust storms, movement occurs frequency of about 3 x 10-2, averaged over a on local scales, as revealed bynumerous martian year. local duststorms (Fig. 36) andby albedo changes of the surface. During the first Strictlyspeaking, the term global dust martian year of the Viking mission, about 20 storm refers only to suspendeddust in the local dust storms were observed (Peterfreund atmosphere. However, because supersonic and Kieffer, 1979). Allowing for incomplete- winds would be required to place micron- ness in coverage, perhaps100 local dust sized dustdirectly into suspension (Pollack storms actually took place duringthis time. et al., 1976), undoubtedly saltation of -100 Atypical local duststorm covers an area pm sand grains also occurs whenever dust is between about0.01 and 0.1% of thetotal raised into the atmosphere with the saltating martiansurface and lasts fora few days. grains aiding the infectionof fine material into suspension throughimpact. Because muchlower wind speeds are required for Dark wind streakstend to form behind saltation to occur, the above estimate of the topographic obstacles such as hills and crater frequency of raising dustfrom the surface rims (Thomasand Veverka, 1979). Theyare can also be considered to apply to the fre- considered to form by wind erosion (Greeley quency of saltation. et al., 1974) and thus serve to mark localities where saltationmay occur more frequently Although ourunderstanding of the than elsewhere. Thesestreaks tend to be mechanism by which global dust storms occur concentratedbetween 25' and 40' south is still at a relatively rudimentary stage and latitude,although they may befound at all strongdifferences of opinionexist, it seems latitudes.Thus, saltation occurs more fre- likely that much of the martian surface serves quently in the lee of craters and ridges and in the upwind portion of crater interiors. as asource region fordust (Pollack et al., 1979; Leovy andZurek, 1979). This view- point is also supported by the lack of a build- Most places on Mars are therefore subject up of a superficial covering of fine dust from to saltation at some time during each martian the decay of the global storms over much of year,with thesouthern hemisphereexperi- the surface(Thomas and Veverka, 1979). encinggreater activity. Theconsiderations However, the absence of appreciablealbedo outlined here lead us to suggest that saltation changes atthe two lander sitesduring the takes place with a frequency ranging between global duststorms of the first martian year about 1 x 10- and 1 x 10" , dependingon of the Viking mission (Guiness et al., 1979) the region. 36

Figure 36. Three local martian dust storms in Arcadia Planitia. Photos were taken with a violet filter during Revs I7I-172, February 10, 1977. The largest is about I80 km in its longest dimension. Shadows along the northernmost cloud indicate a height of 5-10 km. (fiom Bnggs et al., 1979).

5.3 SUMMARY extrapolationsallow, atbest, only crude estimates of wind frequencies. Very littledirect information on wind frequencies is available for sites of geological On Mars, wind frequencydata can be interest on either Earth or Mars. Data which obtained directly for two sites from measure- are available seldom cover a sufficiently long ments on the surface from the Viking landers, period of time to be meaningful in the geolo- or indirectly through observations from orbit. gical context. For the most part, wind data However, poor knowledge of the surface on Earth must be extrapolated from meteoro- roughness on Mars hamperstranslation of logical stations to sites of interest. Because of wind frequency data into values of threshold the distances and/or varying terrain,such and particle flux. 6. RESULTS AND DISCUSSION

The previous three sectionsdescribed the wind-cutting. The inferredrate is parameters involved in calculatingrates of cm/yr based on the estimated minimum abra- aeolian abrasionand present values for each sion averaged over theentire time since the parameter derived fromexperiments, field initiation of wind-cutting the area. studies,and analytical models. These values are now used to estimate rates of abrasion on Hickox (1959)studied ventifacts in the Earth and Mars. Annapolis Valley of Nova Scotia. The ventifacts are composed of vein quartz, quartzite, 6.1 AEOLIANABRASION ON EARTH and indurated sandstone, and Hickox considered theventifact faces to have formedin Althoughnaturally occurring wind-abraded less than10 years of exposure.The ageis materials, such as ventifacts, have long been based on rocks that were quarried from a pit recognized anddescribed (Blake, 1855; of known age. King, 1936;Sharp, 1949; andothers), there have been relatively few attempts to deter- McCauley et a]. (1979) discovered sand- mine rates of abrasion. Rates of abrasion can abraded hearth stones in Egypt considered to be derived from data provided from 5 studies, be of Upper Paleolithic age (200,000 years). summarized in Table 3. However, lack of The stones are abraded to a depth of 2 to 4 wind frequencydata and information on cm, yielding a rate of abrasionof 1 to 2 x particle flux preventa rigorous quantitative 10- cm/yr. assessment. The first study considered is that of Sharp (1949), in which ventifacts inthe Fromexamination of a wind-abraded Big Horn Mountains, Wyoming are described. dacite boulder on one of the volcanic domes Rocks of several types were abraded (granite, at Mono craters, California, and the estimated gneiss, andquartzitic sandstone) and Sharp age of the domes, Williams (1981) obtained a recognizes at least fourseparate periods of rate of abrasion of -lob3 cm/yr.

TABLE 3. Rates of aeolian abrasionon Earthderived fromventifacts

LOCALITY RATE MATERIAL REFERENCE

Big Mountains,Horn Wyoming -I 0- ' cm/yr gneiss Sharp949) (1

AnnapolisValley, Nova Scotia -1 0- cm/yrquartzite, sandstone Hickox (1959)

Western Egypt 1 to 2 x iu" cm/yr McCauley et al. (1979)

I>_ Mono Craters, California -1 0- cm/yr dacite Williams (1 98 1) Whitewater Wash, California "6 x 10" Sharpbrickcm/yr (1 964) -5 x 10" cm/yrhydrocal 38

Sharp (1964) measured abrasion rates of site are about 40% of the geostrophic wind relatively soft materials at Garnet Hill, Cali- speeds (Leovy andZurek, 1979). The 25-30 fornia over an eleven year period. He found m/s observed wind speeds thusimply a anabrasion rate of -6 x 10- cm/yrfor u, = 62-75 m/s. This range is consistent with common red brickand -5 x 10- cm/yr theupper dust cloud speed of 50 m/s ob- forhydrocal, a gypsum cement.Further served over, VL-1 when the near-surface studyat the site (Sharp, 1980) showeda winds of 25-30 m/s were recorded (James and substantialincrease in abrasion rate, due to Evans, 1981). For a height of 10-12 cm above the increased sandsupply upwind of the the surfaceand a u, = 70m/s, we would abrasiontest site caused byflooding of the expectthe typical saltating particle to be Whitewater River. moving at -30% of u,, or -20 m/s (Fig. 12).

The five cases described above have a range of ratesfrom to x6 10- cm/yr, The most difficult parameter to assessis andrepresent widely varying environments the flux of saltating particles. The movement from cold and wet to hot and dry. These of particles at VL-1 would notbe greatly ratesare comparable to thoseinferred by inhibitedby the presence of rocks. Even McCauley et al. (1 981) andprovide a basis thoughrocks are common at VL-1, Moore for comparingrates and ages of surfaces et al. (1977) found that fine-grained material described. covers most of the surfaceand the rocks occupy less than 25% ofthe total area. On Earth, comparable a rock concentration 6.2 RATE OF WIND ABRASION ON MARS would affect aeolian activity.Depending on the size andspacing of the rocks,particle Rates of wind abrasion atthe VL-1 siteon movement could be enhanced in areas of high Mars can beestimated from saltation fluxes turbulenceand retarded in wind ‘shadows.’ and wind frequencies (Table 4) and suscepti- Because of the lowmartian atmospheric bilities to abrasion forpotential martian density, however, only surfaces with a rough- surface materials (Table 5). Basalt and hydro- ness, zo, greaterthan 1 cm would greatly cal were considered as targets. Basalt isthe affectparticle motion. The value of zo at most likely materialpresent on the martian VL-1 is only 0.1-1.0 cm (Sutton et al., 1978); surface.Hydrocal wasused as ananalog to hence, from the standpoint of particle move- indurated clay which might also be on Mars. ment,the VL-1 siteshould be the same as Several impactors were considered:quartz, the ‘sand only’ case of Pollack et al. (1976). basalt, basaltic ash, and aggregates composed Therefore,the saltation fluxequation of of finer grains. All particles were -100 pm White (1979) can be used to calculate Q for indiameter, the size most easily moved on VL-1 (Table 4). Since wind speeds measured Mars (Fig. 3). For the calculation, abrasion of at VL-1 rarely exceed threshold and then only rocks was considered for aheight of 10-12 slightly,a value of u* 20% larger than u* cm above thesurface, corresponding tothe was used for the calculation. Between 10-11 size of many rocks at the VL-1 site. Impacts cm above thesurface, q, the flux per unit were considered to occur normal to vertically area, is about 5% of Q, theflux per unit oriented rock surfaces due to the flat trajec- lane width (Fig. 18). tories of saltating particles on Mars.

In order to use the appropriate values for The remainingparameter required to susceptibility to abrasion,particle velocities calculate an aeolian abrasion rate is f, the for measured winds at the VL-1 site must be frequency of saltation-strength winds. Al- determined. The winds measured at the VL-1 though we have only limitedmeteorological 39

TABLE 4. Factors involved in estimating rates of aeolian abrasion at VL-1

- ~ ". .~

FACTOR VALUE NOTES ___ - -. . ."

Wind velocity u measured atVL1 25-30 mls threshold strength (and strongest) winds measured at height of VL-1 anemometer (Hess et al., 1977; J.A. Ryan, personal communication, 1981)

Threshold velocity u 1.25 mls for D = 100 pm, atmospheric surface *t P pressure of 10 mb, 215K (Greeley et al., 1980)

Shear velocity u* 1.5 m/s 20% above threshold

Freestream velocity u, 70 m/s based on u = (O.~)(U,) given by Leovy and Zurek (1979)

Wind frequency f 10-4 frequency of winds 25-30 m/s (J.A. Ryan, personal communication, 1981)

Flux Q 1.3 x 10" g/cm=sec calculated from White (1979) for u,=l.Sandu=1.25 *t Flux q at 10-12cm height 6.6 x g/cm2 -set at 10-12 cm above surface, q = (0.05) Q, shown in Figure 18

Particle velocity at 10-12 cmheight 20 m/s average particle velocity at height of 10-12 cm, vp = 0.3 u, (from Fig. 12)

Particle size Dp 100 pm particle size most easily moved by lowest strength wind (from Fig. 3)

Impact angle 90" for simplicity in the calculation, only 90" impacts were considered datafrom VL-1, an upper valueis - theejecta and rims of the small cratersin (J.A. Ryan, personal communication). the vicinity of VL-1 would also have been heavily erodedand/or buried unless they The rate Of abrasion can be werevery young.However, there is nomodel for vL-l from Equation surfaceof rock production recentor impact andTable 4. Typicalrates at VL-1 for basalt cratering flux that would allow the vL-l targets are 3 x lo-' m/cm2 'set, which is and orbiter observations adthe high calcu- equivalent to x cm/yr.These rates lated abrasion rate all to be correct. are enormous and, if allowed to operate for any reasonable length of time,result in We consider several possible solutions drasticmodification of therocks at VL-1, tothe discrepancy betweenobservations of muchmore than is observed. Furthermore,the surface and the calculated abrasion rate. 40

TABLE 5. Estimated ratesof aeolian abrasion at VL-1

Abrasion Rate Abrasion Rate Particle Sa* gm/cm2 TargetSa* Particle s cmlyrt

Basalt quartz 2 x 10-4 1 x 10-10 2 x 10-3 Basalt basalt 2 x 10-4 1 x 10-10 2 x 10-3 Basalt ash 4 x 10-5 3 x 10-1' 3 x 10-4

Hydrocal 1 quartz x 10-4 6 x 10-l' 2 x 10-3 Hydrocal basalt 5 x 10-4 3 x 8 x lo-' Hydrocal ash 3 x 10-4 2 x 10-10 5 x 10-3

Rate = Susceptibility Sa x flux q x wind frequencyf * From MRED experiments ? Direct conversion from mass loss; doesnot take into account differencesin abrasion resulting from varying impact angles

In order of increasing likelihood,these are: changes in Pa about the current value. How- (1) climaticconsiderations, (2) surface his- ever, themeasurements of Greeley et al. tory, and (3) factors dealing with the targets, (1980) indicate that uq 0: Pa -In , as shown particles,and particle mobility on Mars. in Figure 3. And as illustratedin Figure 3 and Equation 33, the frequency of aeolian Climatic Effects activitydepends very sensitively on u*~, that is, atthe current epoch it is only the The estimated rates of aeolian abrasion given relatively infrequent, highest wind speeds here assume present-day wind strengthsand that set particles into motion. Thus if Pa is frequencies and other meteorologicalcondi- less thanabout 1/4 the current atmospheric tions.Yet the frequency of aeolian activity pressure, no aeolian activitymay occur; if on Mars may have varied considerably over it isan orderof magnitude larger, aeolian itshistory due to possible oscillatoryand activitymay occur almost all the time. secular changes in atmospheric pressure. The oscillatory changes arise fromastronomical Solar and planetaryperturbations cause variations in orbital and axial characteristics, the eccentricity of Mars' orbit and the obli- whereas secular changes result from long term quity of its axis of rotation to undergo quasi- geochemical processes. periodic oscillations on a time scale of lo5 to lo6 years and theorientation of its axis to Theatmospheric pressure at the surface, precess with a period of 175,000 years, with Pay affects the frequency of aeolian activity theeccentricity and obliquity varying be- due to thedependence of both the wind tween extremes of 0 to 0.14 and 15 and 35", speed distributions function and the threshold respectively (Ward, 1974).It hasbeen sug- wind speed, on Pa. According to argu- gested that a large amountof COz , the mentsadvanced by Pollack (1979),the for- principal atmospheric constituent, is currently mer is expected to change littlefor large contained in the subpolar regolith - about an 41

order of magnitude more C02 than is present Over the lifetime of Mars, a few bars of in theatmosphere (Fanaleand Cannon, COz - that is, about 300 times the amount 1979). At times of high obliquity, theannual- now in the atmosphere - may have been out- ly averaged temperatureat high latitudes gassed from the planet’s interior (Pollack and increases and hence a larger amount ofC02 is Black, 1979). Conceivably, much of the placed intothe atmosphere and a smaller outgassed C02 remained inthe atmosphere amount is taken up bythe regolith.The duringthe early historyof Mars. However, oppositeoccurs at times of lowobliquity, due to the possible formation of carbonate with theC02 being storedin permanent rocksand the developmentof the regolith, polar caps. the atmospheric pressure may have more or less monotonicallydeclined with time over Measurements of C02 absorption onto the last few billion years (Pollack and Yung, fine-&ained dust suggest that the atmospheric 1980).In this case, the level of aeolian pressure could rise to values of about 20 mb activity may have been substantiallyhigher (3 times thecurrent value) at times of the in its past than at the present time. We can highest obliquityand could decrease to place anupper boundary on the amount of values of less than 1 mb at times of the lowest enhancementby supposing the global dust obliquity, due to the formation of permanent storms occurred all thetime, but allowing C02 polar caps (Fanaleand Cannon, 1979; enough time between successive storms for a Toonet al., 1980).Under these circum- given one to decay enough to permit the next stances, no aeolian activity would be expected to occur. in thelatter situation and substantiallya elevated rate of aeolianactivity - perhaps From these considerations of the climatic as much as an order of magnitude - would be history of Mars, aeolian activity and abrasion expected in the former situation. However, it could have been as much as an order of mag- should be remembered that an enhanced level nitude larger in the past than they are today. of aeolian activity may result mostly from the Thus, not only do climatic considerations fail lowering of the threshold wind speed and thus to accountfor the discrepancybetween the involves chiefly lower wind speeds, for which preservation of the surfaceand the present erosionoccurs at adiminished rate. Never- aeolian abrasion if holocrystallinematerials theless, global duststorms will be more fre- were involved, these considerations drive the quent and may involve wind speeds compar- result in the opposite direction. able to today’s values due to dust-wind feedback relationships. Surface History The precession of the axis of rotation causes a change in the relationshipbetween Theassumption is made that the surface of seasons in a given hemisphereand the posi- ChrysePlanitia observed fromorbit is very tion of Mars in its orbit. At present, summer old because of the presence ofabundant solstice in the southern hemisphere occurs at impactcraters that were formedmillions to atime when Marsis close toits perihelion hundreds of millions of years ago. These position. This relationship may be responsible craters and the surface on which they occur for the apparently greater aeolian activity in could be preserved if they had been protected the southern hemisphere than in the northern from the aeolian regime by a mantling deposit forthe presentepoch. If so, thenorthern and then recently exhumed, as proposedby hemispheremay have hadgreater aeolian Sagan et al. (1977) and Binder et al. (1977). activity half a precession cycleearlier Much of Mars appears to be mantled by de- (-100,000years) and more generally this posits of unknown origin, as first proposed by difference will vary in an oscillatory fashion. Soderblom et al. (1973) based on Mariner 9

I 42 results. Thecharacteristics of the mantling preservation of the cratered surface. However, deposits suggest tomost investigations (re- high resolutionViking Orbiter images show viewed byCarr, 1981) that they are of thatmany areas of Mars preserve small, aeolian origin andprobably resulted from ancient impact craters, and it seems unlikely deposition of fine-grained materialsfrom that all such areas couldbe explained by dust storms. exhumation because thereare no known ‘sinks’ that could accomodate the volume of deflatedsediments. Further, the large calcu- Viking Orbiter imagesshow manyparts lated abrasion rates at VL-1 would imply that of Mars that have been mantled and exhumed, strippingoccurred at that site inthe very as shown in Figure 37. The impact crater in recent. (-lo6 years) past, .perhaps unreason- this figure has been partly exhumed, note that ably recent. ratherfine detail has been preserved in the ejecta field. Were itnot for the fact that part of the craterremains buried, it is unlikely Material Properties that the crater and surrounding surfacewould havegiven any evidence of having been Ratesof aeolian abrasion arepartly deter- mantled. mined by the material properties of both the targetand the windblown particles. Despite Parts of Chryse Planitia, siteof the all evidence tothe contrary, perhaps the Viking Lander 1 , also show evidence of having bedrockmaterial on Marsis not basalt,but been mantled and partly exhumed, as noted someother extremely resistant rock. How- byGreeley et al. (1977b). Furthermore, the ever, we determinedthe susceptibilities to deposits of fine-grained debris observed from abrasion for a wide variety the Viking 1 lander (Fig. 5) have been de- of common rocks (Table 2), and even the most resistant yield a scribed as drifts of aeolian deposits that have very high rateof abrasion for Mars. Thus, beeneroded (Mutch et al., 1976). These thebedrock wouldhave to besomething deposits could be the last remnant of a much completely foreign to terrestrialexperience; more extensive mantle(Sharp and Malin, frompetrological considerations this seems 1983). highly unlikely.

On Earth,exhumed surfaces formerly blanketedwith aeolian deposits canreveal Agents of Abrasion features of pristine form. Figure 38 shows a basaltic plain in Iceland that was buried by The calculated aeolian abrasion rate is based loess to a depth of several meters. The area is on theassumption of anunlimited and un- currentlyundergoing deflation andthe restricted sand supply. If there is little sand in basaltic flows are being exhumed.Fine thesystem, abrasion will occuronly at a structure on the scale of centimeters, such as greatly reduced rate. There is a considerable pahoehoe ropes, has been preserved bythe amount ofsand sized material on Mars,as loess and despite theextensive aeolian activity shown by the IRTM (Christensen, 1983) and requiredfor deflation, remains in excellent as inferredfrom the presence ofdunes and condition. dune fields (Tsoar et al., 1979). However, the sandsupply atthe Vikingsites could be Thus,mantling of the cratered basaltic limited. It is also possible that the free move- plains ofChryse Planitia andsubsequent ment of sand is inhibited by the presence of exhumation millions or hundreds of millions cohesion in the uppermost layer, as perhaps of yearslater could explain boththe high inferredby the structure of theslump in a current calculated rateof abrasion andthe drift at VL-1 (Binder et al., 1977), or by the 43

Figure 37. Martianlandscape showing aeolian mantle being stripped away, exhumingcraters (Viking frame 438SOl). The width of the frame is approximately 75 km.

protection of the sandfrom the wind by a basalt grains, and basaltic ash - all of about dust fallout deposit. The dust would be diffi- 150 pm in diameter and all striking the same to move because Of its grain size target type at the same velocity. Quartz and and the corresponding large '*t needed to basalt grains are roughly comparable in their initiate movement (Christensen, 1982). ability to abrade.Although basaltic ash Anotherimportant factor is the composi-appears to be less effective thanquartz or tion of the abrasion agent. Figure 35 shows Sa basalt by a factor of about 2, rates of abrasion values based on abrasionby quartzsand, based on basaltic ash arestill much higher Figure 38. Basalt lava flows in Iceland, northwest of Aswa volcano, were buried by windblown silt to a depth exceeding 2 m and are currently being exhumed by deflation of the silt; primary flow features as small as a centi- meter remain on the exhumed surfaces, evidencedby the pahoehoe ropes seen inthe foreground. Mushroom-shaped features to the left of figure on horizon is remnant of aeolian siltnot yeteroded.

than would be reasonable, based on the age other areas on Mars based onthe global of the cratered surface. However,as discussed circulationmodel of Pollack et al. (1981), inSection 4, aggregates of finematerial the frequency of geostrophic winds exceeding are very inefficient as agents of abrasion. At 75 m/s is extremely low. particlespeeds lower than about 15 m/s, Conclusions aggregates form a veneer on the target surface (Greeley et al., 1982).Abrasion does not Calculations of present rates of aeolian abra- occuruntil velocities greater than 15 m/s sionyield unreasonably high values forthe areachieved. Average particlespeeds in this Viking landing sites when based on impacting range on Mars require freestream wind speeds grains composed of holocrystalline grains or in excess of 75 m/s. The maximum inferred basaltic ash. However, the surface displays an geostrophic winds measuredat the Viking impactcrater distribution that indicates its Landersite are only about 62-75m/s. Al- age to be several hundred millionyears old. though windspeeds higher thanthose mea- Thepresence of the craters and blocks that sured atthe landing sites arepredicted for are relatively unmodified on a surfaceof great 45 age, argue against a continuous high abrasion found to be very inefficient agents of abrasion rate.Although there aremany uncertainties atthe low windspeeds typical for Mars. and assumptions in the analysis, we consider However, some sand of holocrystalline grains the best explanation for this discrepancy to are undoubtedly present, as evidenced by the be a combination of factors dealing with the presence of dunes on Mars. Thus, over the particles and the surface history, at least for the region surrounding the Viking Lander1 long period since the craters were formed, we site. A model has beenproposed (Greeley, consider thatthere would have beenmore 1979; Greeley and Leach, 1978)which erosionthan is apparent.Thus, given the suggests that fine particles(- few pm) derived photogeological evidence formantling, we from the frequent dust storms and the rapid also consider the surface was covered with a attrition of larger grains would aggregate to blanket of sediments and effectively shielded form' sand-size grains. Such aggregates are from erosion. ACKNOWLEDGMENTS

Thiswork beganin I974 whenit became parameters under Earth and martian environ- apparentthat knowledge of rates of wind ments. Usinga design by R. Leach, in I977 erosionwas essential to understanding the J. Burtfrom the California Institute of surfacehistory of Mars. Because ofthe Technology fabricated a particle collector and complexityof the problem, it wasdecided carried out a series of tests on particle flux. to useteama approach. Experiments to Subsequent analyses of the results led to re- determinethe susceptibility to abrasion for design of the collectors; the final collectors different materialsbegan withfabrication weretested using a smokewind tunnel at of a device to simulate aeolian abrasion under the University of California (Davis) to deter- terrestrial and martian conditions by R. Leach mine the most effective shape for collecting andthe instrument shops at NASA-Ames the particles with least disruption to airflow. ResearchCenter. Several different machines Datawere then systematically collected in weretested, each contributing solutions to the wind tunnel, first by M. Niemiller and M. different parts of theproblem. During the Plummer, University of Santa Clara, then by earlyphase W. McKinnon, then a student at K. Malone, a NASA PlanetaryGeology the California Institute of Technology, carried summerintern. Particle velocity measure- outtests partly to determinethe procedure ments were made in wind tunnel tests using for usinglaboratory devices for abrasion adevice modified by R. Leachand the studies. He was followed by S. Williams (then NASA-Ames instrument shops from a similar a NASA PlanetaryGeology summer intern) deviceused by the US,Forest Service to who initiatedsystematic abrasion studies of measurewindblown snow particles. Experi- rock.Concurrently, R.Leach conducted mentswere conducted by R.Leach, K. experiments on theerosion of the aeolian Malone, and R. Leonard. Particle trajectories particles. In I9 75 D.Krinsley began SEM andvelocities were determined by 3. White studiesof both the particlesand abraded and J. Schulzfrom high speed motion pic- rocks to compare their surfaces with geologi- turesobtained bythe PhotographicTech- calmaterials thathad been abraded under nologyBranch atNASA-Ames. Computa- natural conditions in order to verify the use tionalmodels of particlemotion were of laboratory devices. The final machine used developedby 3. White. J. Marshalldeter- in most of the abrasion studies is based on a mined the manner in which geological mater- design used at the Argonne National Labora- ials abrade on the microscopic scale. tory for material testing, and was the result We thank Jack McCauIey for his review ofefforts by T. McKee and D. Hamilton, ofthis manuscript and helpful discussion both of Arizona State University. S. Williams regarding wind erosion. and G. Stewartcompleted the series of abrasion experiments as part of their graduate studies. AI1 phases of the work weresupported by the Planetary Geology Program, National Parallel with the investigations of aeolian AeronauticsandSpace Administration, abrasion,wind tunnel experiments were Washington, D.C., throughAmes Research conducted to determinevarious particle Center. REFERENCES

Adler, W.F., 1974.Erosion of fused silica by glass Goudie, AS., R.U. Cooke, and J.C. Doomkamp, beads, in Erosion, Wear,and Interface with 1979. The formationof silt from quartz dune Corrosion, Am. SOC.Test. Mater., Spec. Tech. sand by salt-weathering processes indeserts, Publ. J. And Environments, 2, 105-112. Adler, W.F., 1976a. Analysis of particulate erosion, Greeley, R., 1979.Silt-clay aggregates on Mars, Wear, 37, 345-352. J. Geophys. Res., 84, 6248-6254. Adler, W.F., 1976b. Analytical modeling of multiple Greeley, R., J.D. Iversen, J.B. Pollack, N. Udovich, particleimpacts on brittlematerials, Wear, 37, and B. White, 1974. Wind tunnel simulations of 333-364. lightand dark streaks on Mars, Science, 183, Andrews, JS., 1931. Percussion figures, BOC.Phys. 847-849. SOC.London, 43, 18-25. Greeley, R., B.R. White, J.B. Pollack, J.D. Iversen, Arvidson, R., E. Guinness, and S. Lee, 1979. Differ- and R.N. Leach,1977a. Dust storms ofMars: entialaeolian redistribution rates on Mars, Considerationsand simulations, NASA Tech. Nature, 278, 533-535. Memo., XU 78423. Bagnold, R.A., 1941. The Physics of Blown Sand and Greeley,R., E. Theilig, J.E. Guest, M.H. Carr, H. Desert Dunes, Methuen, London. Masursky, and J.A. Cutts,1977b. Geology of Binder, A.B.,R.E. Arvidson, E.A. Guinness, K.L. ChrysePlanitia, J. Geophys. Res., 82,4093- Jones, E.C. Morris, T.A. Mutch, D.C. Pieri, and 4109. C. Sagan,1977. The geologyof the Viking Greeley, R., and R.Leach, 1978. A preliminary Lander 1 site, J. Geophys. Res., 82, 4439-4451. assessment of theaffects of electrostatics on Blake, W.P., 1855. On the grooving and polishing of aeolian processes, Rept. Plan. Geol. Bog. 1977- hardrocks and minerals by dry sand, Am. J. 1978, NASA '"-79729,236-237. Sci., 20, 2nd series, 178-181. Greeley, R., R. Papson, and J. Veverka, 1978. Crater Briggs,G.A., W.A. Baum,and J. Barnes, 1979. streaks in the Chryse Planitia region ofMars: Viking Orbiter imaging observations of dustin Early Viking results, Icarus. 34, 556-567. themartian atmosphere, J. Geophys. Res.,84, Greeley,R., R.N. Leach, B.R. White, J.D. Iversen, 2795-2820. and J.B. Pollack,1980. Threshold windspeeds Carr, M.H., 1981. The Surface of Mars, Yale Univer- for sand on Mars: Wind tunnelsimulations, sity Press, New Haven, Conn. Geophys. Res. Lett., 7, 121-124. Christensen, P.R., 1982. Martian dust mantling and Greeley, R., B.R. White, J.B. Pollack, J.D. Iversen, surfacecomposition: Interpretation of thermo- and R.N. Leach,1981. Dust storms of Mars: physicalproperties, J. Geophys. Res., 87, no. Considerations and simulations, Geol. SOC.Am. B12,9985-9998. Spec. Atbl. 186, 101-121. Christensen, P.R., 1983. Aeolian intercrater deposits Greeley, R., and P.D. Spudis,1981. Volcanism on on Mars: Physical properties and global distribu- Mars, Rev. Geophys. space Phys., 19, 13-41. tion, submitted to Icarus. Gieeley, R., R.N. Leach, S.H. Williams, B.R. White, Cutts, J.A., 1973. Wind erosion in the Martian polar J.B. Pollack, D.H. Krinsley,and J.R. Marshall, regions, J. Geophys. Res., 78, 4211-4221. 1982. Rate of wind abrasion on Mars, J. Geo- Cutts, J.A., and R.S.U. Smith, 1973. Eolian deposits phys. Res., 87,no. B12,10,009-10,024. and dunes on Mars, J. Geophys. Res., 78, 4139- Greeley, R., S.H. Williams, and J.R. Marshall, 1983. 4154. Velocities of windblownparticles in saltation: Fanale, F.P., and W.A. Cannon,1979. Mars: C02 Preliminarylaboratory and field measurements, absorption and capillary condensation of clays - ROC.11th Int. Congress on Sedimentology significance for volatile storage and atmospheric (in press). history, J. Geophys. Res., 84, 8404-8414. Guinness, E.A., R.E. Arvidson, D.C. Gehret, and L.K. Finnie, I., 1960, Erosion of surfaces by solid parti- Bolef, 1979. Color changes at the Viking landing cles, Wear, 3, 87-103. sites over a course of a Mars year, J. Geophys. Res., 84,no. B14,8355-8364. \

Henry, RM., 1975. Saltation on Mars and expected Leovy, C.B., and R.W. Zurek,1979. Thermal tides lifetime of Viking 75 wind sensors, NASA Tech. andmartian dust storms: Direct evidence for Note. 0-8035, 39 pp. coupling, J. Geophys.Res., 84, 2956-2968. Hess, S.L., R.M. Henry, C.B. Leovy, J.L. Mitchell, Lobeck, A.K., 1939. Geomorphology, McGraw-Hill, J.A. Ryan, and J.E. Tillman, 1976. Early meteo- New York, 731 pp. rological resultsfrom the Viking lander,2 Malin, M.D., 1974.Salt weathering on Mars, J. Science, 184, 1352. Geophys. Res., 79, 3888-3894. Hess, S.L., R.M. Henry, C.B. Leovy, J.A. Ryan, and Marshall, J.R., 1979.The simulation of sand grain J.E. Tillman, 1977. Meteorological results from surfacetextures using pneumatica gun and the surface of Mars: Viking 1 and 2, J. Geophys. quartzplates, Ph.D. thesis, Univ. College, Res., 82, 45594574. London. Hickox, C.F., 1959.Formation of ventifacts in a McCauley, J.F., 1973. Manner 9 evidence for wind moist, temperate climate, Geol. SOC.Am. Bull., erosion in the equatorial and mid-latitude regions 70, 1489-1 490. of Mars, J. Geophys.Res., 78, 41234137. Huffman, G.G., and W.A. Price,1949. Clay dune McCauley, J.F., C.S. Breed, F. E1-Baz, M.I. Whitney, formation near CorpusChristi, Texas, J. Sed. M.J. Grolier, and A.W. Ward, 1979. Pitted and Pet., 19, 118-127. flutedrocks in the Western Desert of Egypt: Hunter, W.A., 1979. Ventifactson Garnet Hill, Viking comparisons, J. Geophys.Res., 84, unpublishedsenior thesis, Dept. of EarthSci- 8222-823 1. ences, Cal. State Polytechnic Univ., Pomona, McCauley, J.F., C.S. Breed, and M.C. Grolier, 1981. 223 pp. The interplay of fluvial, mass-wasting and eolian Iversen, J.D.,J.B. Pollack, R. Greeley, and B.R. processes in the eastern Gilf Kebir region, Annals White, 1976.Saltation threshold on Mars: The of the Geological Surveyof Egypt, 11, 207-239. effectof interparticle force, surface roughness, McLaughlin, D., 1954. Volcanism and aeolian deposi- and low atmosphericdensity, kurus, 29, 381- tionon Mars, Bull.Geol. SOC. Amer., 65, 715- 393. 717. James, P.B., and N. Evans, 1981. A local dust storm Mills, A.A., 1977. Dust clouds and frictional genera- inthe Chryse region of Mars: Viking orbiter tionof glow discharges on Mars, Nature, 268, observations, Geophys.Res. Lett., 8, 903-906. 614. Johnson, K.L., J.J. O’Connor, and A.C. Woodward, Moore, H.J., R. Hutton, C. Scott, C. Spitzer,and 1973.The effect of theindenter elasticity on R. Shortha,1977. Surfacematerials of the the Hertzian fracture of brittle materials, Roc. Viking landingsites, J. Geophys.Res., 82, Royal SOC.London, A, 334, 95-1 17. 44974523. Kawamura, R., 1951. Study on sand movement by Morsi, S.A., and A.J. Alexander, 1972. An investiga- wind, Inst. of Science and Technology, Tokyo, tion ofparticle trajectories in two-phase flow Report 5,95-112. systems, J. FluidMech., part2, 55, 193-208. King, I.C., 1936. Wind-faceted stonesfrom Marl- Mutch, T.A., R.A. Arvidson, A.B. Binder, F.O. Huck, borough, New Zealand, J. Geol., 44, 201-213. E.C. Levinthal, S. Liebes, E.C. Morris, D. Knight, C.G., M.V. Swain, and M.N. Chaudhri, 1977. Nummendal, J.B. Pollack,and C. Sagan,1976. Impact of small steel spheres on glass surfaces, Fineparticles on Mars: Observationswith the J. Mat. Sci., 12, 1573-1586. Viking1 lander cameras, Science, 194, 87-91. Krinsley, D.H., J.H. Fink,and R. Greeley,1980. Neilson, J.H., and A. Gilchrist, 1968. Erosion by a Explosive volcanism: Possible source ofaggre- stream of solid particles, Wear, 11, 111-122. gate formation on Mars, EOS, nuns. Am. Geo- Oh, H.L., K.P.L. Oh, S. Vaidyanathan, and I. Finnie, phys. Union (abst.), 61, no. 46, 1138. 1972. On the shaping of brittle solids by erosion Kuenen, Ph.H., 1960.Experimental abrasion 4: and ultrasonic cutting, Proceedings of the Sym- Eolianaction, Journal of Geology, 68, 427- posium onthe Science of Ceramic Machining 449. andSurface Finishing, US. Natl. Bur. Stand. Lawn, B., and R. Wilshaw, 1975. Review - indenta- Spec. Atbl., 348, 119-132. tion fracture:principles and applications, J. Peterfreund, A.R., 1981. Visual and infrared obser- Material Science, IO, 1049-1081. vations of wind streaks on Mars, Icarus, 45, no. 2,447467. 49

Peterfreund, A.R., and H.H. Kieffer, 1979. Thermal Sharp, R.P., 1964. Wind-driven sand in Coachella infrared properties of the martian atmosphere 3, Valley, California, Geol. SOC.Am. Bull., 75, local dust clouds, J. Geophys.Res., 84, 2853- 785-804. 2864. Sharp, R.P., 1980. Wind-driven sand in Coachella Pettijohn, F.J., P.E. Potter, and R.Siever, 1972. Valley, California: Further data, Geol. Sac. Am. Sand and sandstone, Springer-Verlag,New Bull., 91, 724-730. York,618 pp. Sharp, RP. and M.C. Malin, 1983. Geology from the Pollack, J.B., 1979. Climatic change on the terrestrial Viking Landers on Mars: a second look, Bull. planets, Icarus, 37,479-553. Geol. SOC.Amer., in press. Pollack, J.B., R. Haberle, R. Greeley, and J. Iversen, Sheldon, G.L., 1970. Similarities and differences in 1976. Estimates of the wind speeds required for the erosion behavior of materials, Dans. ASME, particle motion on Mars, Icarus, 29, 395417. J. Basic Eng., 92, 619-626. Pollack, J.B., and D.C. Black, 1979. Implications of Sheldon, G.L., and I. Finnie, 1966a. On the ductile the gas compositional measurements of Pioneer behavior of nominally brittle materials during Venus for the origin of planetary atmospheres, erosive cutting, Dam. ASME, J. Eng. Ind., Science, 205, 56-69. Nov., 387-392. Pollack, J.B.,D.S. Colburn, F.M. Flasar, R. Kahn, Sheldon, G.L., and I. Finnie, 1966b. The mechanism C.E. Carlson, and D. Pidek,1979. Properties of material removalin the erosive cutting of. andeffects of dust particles suspended in the brittle materials, 7’rans. ASME, J. Eng. Ind., Martian atmosphere, J. Geophys.Res., 84, Nov., 393-400. 2929-2945. Sheldon, G.L., and A. Kanhere, 1972. An investiga- Pollack, J.B., and Y.L. Yung, 1980. Origin and tionof impingement erosion usingsingle parti- evolution of planetary atmospheres, Ann. Rev. cles, Wear, 21, 195-200. Earth Planet. Sci., 8,425487. Singer, R.B.,T.B. McCord, R.N. Clark, J.B. Adams, Pollack, J.B., C.B. Leovy, P.W. Greiman,and Y. and R.L. Huguenin, 1979. Mars surface composi- Mintz, 1981. A martian general circulation tion from reflectance spectroscopy: A summary, experimentwith large topography, J. Atmos. J. Geophys. Res., 84, 8415-8487. Sci., 38, 3-29. Smdey, I.J., and D.H. Krinsley, 1979. Aeolian sedi- Ryan, J.A., R.M. Henry, S.L. Hess,C.B. Leovy, mentationon earth and Mars: Some compari- J.E. Tillman, and C. Walcek, 1978. Mars meteo- sons, Icarus, 40, 276-288. rology: Three seasons at the surface, Geophys. Smeltzer, C.E., M.E. Goulden, and W.A. Compton, Res. Lett., 5, 715-718. 1970. Mechanism of metal removal by impacting Sagan, C., 1973. Sandstorms and eolian erosion on dust particles, 7’rans. ASME, J. BasicEng., Mars, J. Geophys. Res., 78,41554161. Sept., 639-654. Sagan, C., J. Veverka, P. Fox, R. Dubisch, J. Leder- Soderblom, L.A., M.C. Malin,J.A. Cutts, and B.C. berg,E. Levinthal, L. Quam, R. Tucker, J.B. Murray, 1973. Mariner 9 observations ofthe Pollack, and B.A. Smith, 1972. Variable features surface ofMars in thenorth polar regions, J. on Mars: preliminary Mariner 9 television results, Geophys. Res., 78,41974210. Icarus, 17,346-372. Sutton, J.L., C.B. Leovy, and J.E. Tillman, 1978. Sagan, C., D. Pieri, P.Fox, R. Arvidson, and E. Diurnal variations of the martian surface layer Guinness, 1977. Particle motion on Mars inferred meteorological parameters during the first 45 from the Viking lander cameras, J. Geophys. sols at two Viking lander sites, J. Atmos. Sci., Res., 82,4430-4438. 35, 2346-2355. Schmidt, R.A., 1977. A system that measures blow- Suzuki, T., and K. Takahashi, 1981. An experimental ing snow, Res. Pap. RM-194, US. Dept. of study ofwind abrasion, J. Geol., 89, 509-522. Agric., Washington, D.C., 80 pp. Thomas, P., and J. Veverka, 1979. Seasonal and Scott, D.H., and M.H. Carr, 1978. Geological map of secular variations ofwind streaks on Mars: an Mars, U.S. Geol. Survey, Misc. Geol. Inv.Map analysisof Mariner 9 and Viking data, J. Geo- 1-1083. phys. Res., 84,8131-8146. Sharp, R.P., 1949. Pleistocene ventifacts eastof Tillett, J.P.A., 1956.Fracture ofglass by spherical the Big HornMountains, Wyoming, J. Geol., indentures, Roc. Phys. SOC.,B69, 47-54. 57, 175-195. 50

Tilly, G.P., 1969. Erosion caused by airborne parti- Veverka, J., and C. Sagan,1974. McLaughlin and cles, Wear, 14. 63-79. Mars, American Scientist, 62, 44-53. Tilly, G.P., and W. Sage, 1970.The interaction of Ward, W., 1974. Climate variations on Mars 1. astro- particle andmaterial behavior in erosionpro- nomical theory of insolation, J. Geophys. Res., cesses, Wear, 16. 447465. 79,3375-3385. Toon, O.B.,J.B. Pollack, W. Ward, J. Burns,and White, B.R, 1979. Soil transport by wind on Mars, K. Bilski, 1980.The astronomical theory of J. Geophys. Res., 84, 46434651. climatic change on Mars, Icarus, 44, 552-607. White, B.R.,R. Greeley,J.D. Iversen, and J.B. Toulman, P., 111, A.K. Baird, B.C. Clark, K.Heil, Pollack,1976. Estimated grain saltation in a H.J. Rose, R.P. Christian, P.H.Evans, and Martianatmosphere, J. Geophys. Res.,81, W.C. Kelliher, 1977.Geochemical and mineral- 5643-5650. ogic interpretationsof the Viking inorganic White, B.R., and J.C. Schulz, 1977. Magnus effect in chemical results, J. Geophys. Res.,82, 4625- saltation, J. Fluid Mech., 81, no. 3, 497-512. 4634. Whitney, M.I., and R.V. Dietrich, 1973.Ventifact Tsoar, H., R. Greeley, and A.R. Peterfreund, 1979. sculptureby windblown dust, Geol. SOC. Am. Mars: the north polar sand sea and related wind Bull., 84, 2561-2582. patterns, J. Geophys. Res.,84, 8167-8180. Whitney, M.I., 1979. Electron micrography of miner- Tsuchiya, Y., 1969. Mechanicsof the successive al surfacessubject to wind-blasterosion, Geol. saltation of asand particle on a granularbed SOC.Am. Bull., part 1,90, 917-934. in aturbulent stream, Disas.Prev. Res. Inst. Williams, S.H., 1981. Calculated and inferred aeolian Kyoto Univ., Bull., 19, part1(152), 3144. abrasion rates: Earthand Mars, M.S. thesis, Ark. StateUniv., Tempe. 1. Report No. 2. GovernmentAccarion No. 3. Recipient's Gubg No. NASA CR-3788 4. Title and Subtitle 5. Repon Date- ABRASION BY AEOLIANPARTICLES: EARTH AND MARS March 1984 6. Performing Orpanirahn Code

~~ 7. Author(s1 8. Performing Organiution Aepon No. R. Greeley, S. Williams, B. R. White, J. Pollack, J. Marshall,and D. Krinsley 10. Work Unit No. 9. Performing Organiution Name and Address Planetary Geology .Group - Department of Geology I 11. Contract or Grant No. Arizons. State University .NSG-2284 and NCC2-60 TemDe, Az 85287 13. Type of Repon and Period Cover& 12. Sp3nsoring Agency Name andAddress Contractor Re-mrt Office of SpaceScience and Applicatlons

Netional Aeronautics and Space- A6ministration Wesnington, DC: 20546 15.Supplementary Notes R. Greeley:Arizona State University, Tempe, Arizona. S. Williams: Universityof Santa Clara, Santa Clara, California. B. R. White: University of California at Davis, Davis, California. J. Pollack: NASA Ames ResearchCenter, Moffett Field, California. J. Marshalland D. Krinsley:Arizona State University, Tempe, Arizona. NASA TechnicalMonitor: Joseph M. Boyce Summary Report

~ 16. Abstraa Estimation of the rate of aeolian abrasion of rocks on Mars requires knowledge of (1) particle flux, (2) susceptibilities to abrasion of various rocks, and (3) wind frequencies on Mars. Fluxes and susceptibilities for a wide range of conditions were obtained in the laboratory and combined with wind data from the Viking meteorologyexperiment. Assuming anabundant supply of sand-sized particles,estimated rates range up to 2.1 x cm of abrasion per year in the vicinity of Viking Lander 1. This rate is orders of magnitude too great to be in agreement with the inferred age of the surface based on models of impact crater flux. The discrepancy in the estimated rate of abrasion and the presumed old age of the surface cannot be explained easily by changes inclimate orexhumation of ancient surfaces. We consider the primary reason to be related to the agents of abrasion. Either windblown grains are in very short supply, or the grains are ineffective as agents of abrasion. At least some sand-sized (-100 pm) grains appear to be present, as inferred from both lander and orbiter observations. High rates of abrasion occur for all experimental cases involving sands of quartz, basalt, or ash. However, previous studies have shown that sand is quickly comminuted to silt- and clay-sized grains in the martian aeolian regime. Experiments also show that these fine grains are electrostatically charged and bond together as sand-sized aggregates. Laboratory simulations of wind abrasion involving aggregates show that at impactvelocities capable of destroying sand, aggregates form a protective veneer on the target surface and can give rise to extremely low abrasion rates.

~ 17. Key Words lsuggerted by Authorlr) I 18. DistributionStatement Mars Erosion Unclassified - Unlimited Venu s Abrasion Venus Earth Geology Earth Aeolian Eo 1ian SubjectCategory 25

19. SecurityOassif. (of this rewort] 20. Security Classif. (of this page1 21. No. of Pq 22. Rice Unc 12s s ified Unc lass if ied 50 A0 3 i -

For sale by the National Technical Inlormalion Servtce. Sprinelield. Vlrgrnla 22161 NASA-Langley , 1984