JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 97, NO. E8, PAGES 13,319-13,345, AUGUST 25, 1992

Aeolian Featureson Venus' Preliminary Magellan Results

RONALDGREELEY, • RAYMOND E. ARVIDSON,2 CHARLES ELACHI, 3 MAUREEN A. GERINGER,• JEFFREY J. PLAUT,3 R. STEPHENSAUNDERS, • GERALD SCHUBERT, 4ELLEN R. STOFAN,• Emc J.P. THOUVENOT,3's STEPHEND. WALL)• ANDCATHERINE M. WEITZ•

Magellan synthetic aperture radar data reveal numerous surface features that are attributed to aeolian, or wind processes. Wind streaksare the most common aeolian feature. They consistof radar backscatterpatterns that are high, low, or mixed in relation to the surfaceon which they occur. A data base of more than 3400 wind streaksshows that low backscatterlinear forms (long, narrow streaks)are the most common and that most streaksoccur between 17øS to 30øS and 5øN to 53øN on smoothplains. Moreover, most streaksare associatedwith depositsfrom certain impact cratersand some tectonicallydeformed terrains. We infer that both of these geological settingsprovide fine particulatematerial that can be entrainedby the low-velocity winds on Venus. Turbulenceand wind patternsgenerated by the topographicfeatures with which many streaksare associatedcan account for differencesin particle distributionsand in the patternsof the wind streaks. Thus, some high backscatterstreaks are consideredto be zonesthat are swept free of sedimentaryparticles to expose rough bedrock;other high backscatterstreaks may be lag depositsof densematerials from which low-density grains have been removed (densematerials such as ilmenite or pyrite have dielectric propertiesthat would producehigh backscatterpatterns). Wind streaksgenerally occur on slopes < 2 ø and tend to be oriented toward the equator, consistentwith the Hadley model of atmospheric circulation. In additionto wind streaks,other aeolianfeatures on Venusincludg[ yardangs(?) and dune fields. The Aglaonicedune field, centeredat 25øS, 340øE, covers~1290 km': and is locatedin

90øE,anejecta coversflow ~17,120channel •n•the in aAglaonice valley betweenimpactIshtar crater. Terra The andMeshkenet MeshkenetduneTessera. field, located Wind at streaks67øN, associatedwith both dune fields suggestthat the dunesare of transverseforms in which the dune crestsare perpendicularto the prevailingwinds. Dunes on Venus signal the presenceof sand-size (~60 to 2,000 gm) grains. The possibleyardangs are found at 9øN, 60.5øE, about300 km southeast of the crater Mead. Although most aeolian features are concentratedin smooth plains near the equator,the occurrenceof wind streaksis widespread,and somehave been found at all latitudesand elevations. They demonstratethat aeolian processesoperate widely on Venus. The intensity of wind erosionand deposits,however, varies with locality and is dependenton the wind regime and supply of particles.

1. INTRODUCTION using (primarily) cycle 1 Magellan radar data (Figure 1). Aeolian,or wind-related,processes on the surfaceof Venus Aeolianfeatures include possible (1) dunefields, (2) yardangs have been debatedfor more than two decades,and many (wind-erodedhills), and (3)various types of wind streaks investigatorspredicted that aeolian features would eventually (surface patterns of contrasting radar backscattercross be found (reviewed by Greeley and Arvidson [1990]). sections). We describe these aeolian features and their Althoughimages of the surfacereturned from SovietVenera characteristicsas seenon Magellan radar imagesand assess landersand measurementsof near-surfacewinds suggested the geologicalsettings and properties of the surfacein which localmodification of the surfaceby wind, definitiveevidence they occur. We alsodiscuss the possiblemodes of formation for more widespreadaeolian activity was not observeduntil of the mostcommon aeolian features (wind streaks),drawing the Magellan mission[Saunders et al., 1991]. Preliminary on terrestrialexamples, Martian analogs,and resultsfrom analysesof Magellan radar imagesrevealed several regions wind tunnelsimulations. We thenconsider the relationships where wind-related features are abundant, as well as other between aeolian features and patterns of atmospheric isolatedoccurrences [Arvidson et al., 1991]. circulation on Venus. For thisreport, about 44% of the surfaceof Venushas been searched in a reconnaissance mode for wind-related features 1.1. Background

1Departmentof Geology, Arizona State University, Tempe. Wind-relatedfeatures observed on planetary images provide 2Departmentof Earthand Planetary Sciences, Washington direct evidence for the interactionof the atmospherewith the University, St. Louis, Missouri. surface. The presenceof depositionalaeolian features, such 3Jet Pro pulsion Laboratory, Pasade na, California. 4Departmentof Earthand SpaceSciences, Institute of as dunes,shows areas where particles capable of movement Geophysics and Planetary Physics, University of California, by the wind occur and gives indicationsof weathering Los Angeles. processes.The identificationof the type and orientationof 5CentreNational de la RechercheScientifique, Toulouse, aeolianfeatures provides clues to the physicalproperties of France. surfacematerials where they occurand the wind directionat the time of their formation. Assessmentof their age Copyright 1992 by the American GeophysicalUnion. providesinsight into pastwind regimesand climates. Wind streaksare amongthe mostcommon aeolian feature Papernumber 92JE00980. observedon planetarysurfaces. They occuron Earth,Mars, 014 8 -022 7/9 2/92 JE- 0098 0 $05.00 Triton, and Venus. On Earth, wind streaks are surface

13,319 13,320 GREELEY ET AL.: AEOLIAN FEATURESON VENUS

180 195 210 225 240 255 270 285 300 315 330 345 0 15 30 45 60 75 90 105 120 135 150 165 180 l' '"l""l"",'"',"" l''" l''"l''"l''"l''"l''" l''"l''"l"" l""l''"l''"l'"'l'"' f""l''",''",""l''"l''" l"",'"'l''",""l'"'l''"l""l'"'l''"l''"l'"'l''"l""l""l''" l""l'"'l''"l""l'"'l''"l""l''"l''"l''"l''"l''"l''"l''"l''" I""I'"'I''" I""I'"'I''"I''"I'"'I''" l""l''"l''"l''"l""[' '" I""I'"'I

?::•::' ) Fan Bright •-•-•:.....I•'•'H-:':T'A

-15 -15

-3O "•:•'"'•"'•.•,...•,•.•...• '• •------'•-•::•,..•,,•,-"•...... '.•:• -3O

-45 -45

-6O -6O

•.•.:•:•.•,•

Fig. 1. Map of Venus showingorientation of wind streaksequal-area by latitude and longitude "bin" and their distribution. Symbols are given in the center of each bin, or positionedwithin the bin to maximize legibility. Arrows indicate inferred downwind direction. Also shown are the location of dune (D) and yardang (Y) fields. Regionsnot yet analyzedon F-BIDRs includelongitude -30 ø through50 ø (superiorconjunction) and longitudes -160 ø through330 ø (digital basemap from U.S. GeologicalSurvey).

patternsin which loosematerials commonly <1 m thick are 2h). Wind streaks of several types also occur on Mars distributedby sediment-movingwinds. Typically,they are (Figures 2e and 2f), as reviewedby Greeley et al. [ 1992]. associatedwith wind patternsand turbulencegenerated by Thomaset al. [ 1981] derived a classificationof Martian wind topographicfeatures such as smallhills. On Earth and Mars, streaksbased on (1) their upwindsources (sediment deposit or wind streaksare visibleon opticalimages because of albedo topographicobstacle), (2) their albedo contrast(bright or contrastsrelated to particle size or compositionand to dark) in relation to the backgroundsurface, and (3) special exposuresof bedrock. Wind streaksare alsovisible on Earth morphologic or compositional features. Some, termed on radarimages [Greeley et al., 1989;Saunders et al., 1990], variable features,appear, disappear,or changetheir size, wherefactors such as bedrock exposures and sedimentcover shape, and orientation on time scalesof weeks to years resultin contrastingradar backscattercross sections. Wind [Thomas and Veverka, 1979, etc.]. Martian wind streaks streakson Earth rangein lengthfrom a few centimetersfor rangein lengthfrom a few centimetersat the Viking landing small sand drifts behind rocks to more than 15 km for sites to 115 km for a dark, plume-shapedstreak in the patternsdeveloped in the lee of hills and small mountains Mesogaearegion [Veverka et al., 1976]. (Figures2a-2d) and in associationwith impact craters (Figure Regardlessof type,mode of formation,or planetwhere they

Fig. 2. (Opposite)Typical wind streaks on Earth,Mars, and Venus; arrows indicate prevailing wind direction. (a) Amboy,California, optical image showing cinder cone (460 m in diameter)and dark streak. Prevailing wind is fromthe west(left); generalbackground consists of pahoehoebasaltic lava flowsand is mantledwith windblown sand(whim areas). Area of streakis darkbasalt swept free of sanddue to windturbulence shed from flow aroundthe cone(from Greeley and Iversen [1987]; air photo AXL-26K-36 taken January 10, 1953). (b) Seasatradar image (revolution882) of Amboy,California, showing radar-bright streak corresponding to dark areas on Figure2a thatis relativelyfree of windblownsand. Dark areas correspond toconcentrations of sand and (extending toward the top of thepicture from the cone) basalt pebbles and cinders. Radar illumination from the bottom. (c) SIR-Aradar image of the Altiplano,Bolivia, showing radar-dark streaks as longas 15 km formedin associationwith hills (bright features).Contrasts in radarbackscatter cross sections result from differencesin sandmantles, dune forms, and vegetation,all relatedto erosionand deposition patterns generated by windflow around the hills. Prevailingwinds arefrom the Pacific Ocean to thewest (left) (SIR-A: DT-31). (d) SIR-Aradar image of linearstreaks southeast of LaskarGan, Afghanistan,formed predominantly in sedimentarydeposits, including playa silts andclays. Prevailingwinds are from the northeast (marked with an arrow), which funnel through a gap in a lowridge and then spreadout to the southwest.Contrasts in radarbackscatter cross section evidently result from differences in the distributionof sediments(SIR-A: DT-35/36). (e) Darkwind streaks on Marsin thePhoenicis Lacus region; prevailing wind is from the southeast(left). These and similar dark wind streaksare consideredto result from GREELEYET AL.: AEOIJANFEATURES ON VENUS 13,321

.

.

erosion.Area sho• is 30 by 43 • (Viking •biter 459A79). • •right wi•d s•e•s i• •e •es•ria Pla•um, M•s; prevail•g wi•ds •e •om •e not.west (up•r left). Bright wi•d s•e•s •e •ought to be dust depsited i• ß e lee of topgraphic fea•es, •rhaps under stable atmosphericcondition. •ea show• is a•ut 5• km across (Viking •bier 453•5). (g) Rad•-bright wi•d see• o• Venus at 23.9øS,345.1øE. S•e• is a•ut 10 • Io•g, ' is associatedwi• a small hill, •d occurswi• •e "p•a•lic ½oI1•" associatedwi• the impactcrater C•son (see Eigure 19); w•d flow at the time of streak fo•atio• is i•fe•ed to have been from •e south (left). Radar illumi•afio• •om •e top (Magell• F-MIDR 25S345). (h) Wolf Creek •pact crater,Austria, showingassociated s•d depsi• (•ight •eas) •d erosion• •eas (d•k zones);ridges •e li•e• dunesp•allel to •e wi•d; prevailing winds •e from •e east (left); •ea shown is -6 • by 9 • (Commonwealthof Aus•alia photograph,•illiluna 13,322 GREELEYET AL.: AEOLIAN FEA7URF3ON VENUS

occur,there is near-unanimousagreement that wind streaks !N HADLEY representthe prevailing wind directionat the time of their CELL formation. As such,they can be usedas local "wind vanes" to map near-surfacewinds and have been used to assesslocal, regional,and global patternsof atmosphericcirculation on Mars [Sagan et al., 1972; Thomas and Veverka, 1979; i Greeleyet al., 1992]. The discoveryof wind streakson Venus (Figure 2g) as describedby Arvidsonet al. [ 1991] affordsthe opportunity to learn about the interactionof the atmosphereand surface, both for the identificationof sedimentscapable of being moved by the wind and in mapping near-surfacewinds. Mapping winds is especially important becauseof the paucityof observationalconstraints on the circulationin the ALLEY lower atmosphere[Schubert, 1983]. For example,Doppler trackingof Venera probesand landers[Marov et al., 1973; Antsibor et al., 1976; Keldysh, 1977; Kerzhanovichet al., 1979; Moroz, 1981; Kerzhanovich and Marov, 1983] and trackingof the Pioneer Venus probes[Counselman et al., HALLEY i MEANWESTWARD ZONAL CELL i SUPERROTATION 1979, 1980] providedestimates of zonaland meridional wind i I speedsas high as severalmeters per. second to an altitudeof ! ~10 km. Surface winds measuredby Veneras 9 and 10 I [Avduevskiiet al., 1976]were 0.3 to 1 rns -1. Theobserved is motionsin the lowestscale height are sluggishbut appearto be neither mainly meridional nor zonal. However, the Fig. 3. Sketch of possible circulation patterns in the available measurements are inadequate for even atmosphereof Venus. The mean zonal velocity h is a westward approximatingthe patternsof lower atmosphericcirculation superrotation. The magnitude of h increaseswith height above [Schubert,1983]. Above 10 km, zonalwind speedsincrease the surface. The meridionalHadley circulationmay not extendto the poles. Centersof convergenceand divergencein the diurnal monotonicallywith altitude,and the dominantcirculation is Halley and anti-Halley circulationsmay not occur at noon. The a westwardzonal superrotation [Schubert et al., 1980]. diurnal circulationshave other flow componentsnot shownhere. Althoughthe lower atmosphericcirculation of Venus may be temporallyand spatiallycomplex, it is usefulto consider is conceivablethat such changescould be detectedwith two end-membermodels of the circulation(Figure 3). One is multipleMagellan observations over time, as are plannedin a Hadley circulationthat redistfibutessolar energy absorbed futuremapping cycles. Moreover,as notedby Saunderset in the lower atmosphereand at the groundnear the equator al. [1991] strongest,winds were predictedto flow to the [Stone, 1974, 1975; Kdlney de Rivas, 1973, 1975; Schubert west,away from the solarnoon longitude, and to flow down et al., 1980; Rossow, 1983; Schubert, 1983]. This direct hill. Thesepredictions can be addressedwith Magellandata. meridional circulationis symmetricabout the equatorand involvesequatorward surface winds, upflow over the equator, 1.2. Methodology polewardwinds aloft, and downflow at high latitudes. The long-term, zonally averaged circulation of the deep Observations for Venusian aeolian features included searches atmosphereof Venus may resemblea Hadley circulation. of syntheticaperature radar (SAR) images,assessment of The existenceof a Hadley cell in the lower atmosphereof surfaceproperties and elevationswhere features were found, Venus can be evaluated using Magellan imagesof wind and correlationsof aeolianfeatures with local geology.The streaksin thisstudy. Should a Hadleycell exist,wind streaks searchfor wind-relatedfeatures was conducted using F-BIDRs might also provideinformation on the latitudinalextent of (full resolutionbasic image data record), F-MIDRs (full the circulation. resolution mosaiked image data record), and C1-MIDR A secondpossible model of the lower atmosphereon Venus (mosaikedimage data record, compressed once) (see Saunders involvesdiurnal circulation. In this model (Figure3), there et al. [1990, this issue] and Pettengill et al. [1991] for is downflow over the subsolarregion just abovethe surface, explanationof dataproducts). The bestspatial resolution on flow towardthe antisolar region near the surface, upflow over F-BIDR and MIDR imagesis 150 m/line pair (represented the antisolarregion just abovethe surface,and flow toward by 75 m/pixel). F-BIDR prints throughorbit 1319 were the subsolarpoint aloft [Dobrovolskisand Ingersoll,1980; examinedfor small(<10 km) features(Figure 1). Aeolian Covey et al., 1986]. This is an "anti-Halley cell" above features>10 km wereassessed on F-MIDRs andC1-MIDRs; which(in the loweratmosphere) is a simple,thermally direct consequently,because not all data havebeen assembled as subsolar_to_antisolarcirculation pattern, with upflowin the mosaics,some larger features may not yet be recognizedin subsolarregion (Halley cell, Figure3). Wind streaksin this the areasanalyzed for smallerfeatures using only F-BIDRs. modelwould be orientedaway from the warm,dayside of The distributionof aeolianfeatures was plotted on a global Venusand toward the coolernight region. This circulation scaleand correlated with localand regional geologic settings patternwould be moredifficult to detectusing wind streaks to placeconstraints on the possiblesources for the deposits than a Hadley cell becausethe movementof the subsolar associatedwith thefeatures. pointover the surface(1 Venussolar day = 117 Earthdays) In identifyingaeolian features such as wind streaks,it is wouldchange wind streakdirections with time. However,it importantto note that BIDRs and MIDRs havebrighiness GREEI F•Y ET AL.: AEOLIAN FEAT[IRKS ON VENUS 13,323 valuesthat are proportionalto specificradar backscatter cross modes,respectively [Pettengill et el., 1991]. Emissivityand section(o0 which is crosssection per unit areadivided by the Fresnelreflectivity of geologicmaterials are closely tied to averagevalue for therelevant incidence angle and convened to the dielectric constant and thus allow model-dependent decibels). An understandingof the behavior of o0 as a separationof the effectsof roughnessand dielectric properties function of incidenceangle and look azimuth is critical to on the SAR signatures [Tyler et el., 1991]. These interpretationsof the radar appearanceof aeolian features. observationsprovide informationon the physicalproperties The incidenceangle for Magellan data variessystematically of surface materials. with the latitude,from a value of 43ø at the periapsislatitude Magellanaltimetric data enabled assessment of topographic (10øN) to a value of 18ø at the northpole. In general,in this control of aeolian features. Pettengill et el. [this issue] rangeof incidenceangles, radar wavelength-scale roughness derived a global topographicmap of about 5 km spatial appearsto dominatethe cross-sectionvalues of the aeolian resolutionand <100 m vertical resolution. For our analyses featuresdiscussed here, with secondarycontrol by topography of wind streaks, the elevation of each streak was determined anddielectric properties. However, the identificationof wind and the local slopes were assessed, using a bilinear streaksin the SAR images at high northern and southern interpolationof the topographicdata pointswithin a 10-km latitudesmay be moredifficult, because subtle differences in radius of the streak origin. However, these statisticsare small-scaleroughness are more difficult to detect at small based on a preliminary global topographic map which incidenceangles. Biota and Elachi [ 1981, 1987] haveshown containssome errors. As shownin Figure 4, the azimuthof that the azimuth of the radar viewing geometry may also the maximum slope ([3, in the downslopedirection, i.e., substantiallyaffect the visibilityof aeolianbedforms such as "dip" direction)and the amount(magnitude) of the slopein sand dunes. Continuing analysis of data from future degreesfrom the horizontalwere determined. Finally, the mappingcycles of Magellan with different incidenceangle angle0') betweenstreak and slopeazimuth was determinedin and look azimuthprofiles will allow a completeassessment order to assesswhether streakstend to be orientedupslope, of the streakpopulation on Venus. downslope,or randomlywith respectto slope. In additionto the high-resolutionspecific backscatter cross section data obtained by the Magellan SAR, estimatesof 2. WtND STREAKS surface emissivity, Fresnel reflectivity, and rms slope (at greaterthan wavelength scales) with ~10 km resolutionwere Wind streaks of several forms have been found on Venus. obtainedby the radar systemin its radiometricand altimetric Although it is tempting to derive a formal classification,

ridgeor strea fracture ,•.•-•'• Transverse ../•....r•. .• ,,,,•....,,,2" • :'-:'•:'/,,'7 ragged '".'••) ..••"

Transverse both smooth are fan streaks

DOWN-WIND fracture DIRECTION • N ridgeor

slo , **-,,, a '

Fig. 4. (a) Pl•fom sha•s •d terms appliedto Venusi• wind stre•s; a is •e az•u• measuredfrom north to the s•e• in •e infe•ed downw•d direction. (b) P•meters usedin •alysis: a is stre• length (m•imum) p•allel to infe•ed w•d direction,b is s•e• wid• (maximum)nomal to leng•, c is dimeter of l•dfom with which stre• is associated(or •e averagewid• of the ridge or •ench for tr•sverse stre•s), • is azimuthof stre• in the infe•ed downwinddirection, • is azimuthof te•ain in the downslopedirection, •d T is minimum•gle betweenstreak az•u• •d slo• az•u• (5180ø), A-A' •dicates the l•e along which rad• backscattercross section profiles were obtained for some stre•s. 13,324 GREELEYET AL.: AEOLIAN FEATURESON VENUS

TABLE 1. Parameters used in Data Base to Describe Wind Streaks crater deposit. However, for most wispy streaksit is not on Venus possibleto determinewhich endis the apparentsource, nor is it feasibleto determinea meaningfulorientation (azimuth) Principalparameter Sub-type because of their meanderous character. Contrastsin radarspecific backscatter cross section between Planimetric shape Fan the streak and the backgroundenable wind streaksto be Linear identified.Bright streaks are brighter than the background on Transverse whichthey occur (Figures 5a and5e), darkstreaks are darker Wispy thanthe background(Figures 5b, 5d, 5g and 5h), andmixed Radar reflectivity Bright streaks(Figure 5c) have both bright and dark components Dark (generally a bright interior and a dark "halo", set on a Mixed backgroundof intermediatespecific cross section). All wispy Landform origin Cone streaksare radar-dark (Figure 5h). Nearly all linearstreaks are Hill radar-dark,most fan-shapedstreaks are radar-bright,and Crater wansverseragged streaks are nearly equally radar-dark, -bright, Ridge and -mixed (Figure 6a). However, somelinear streaksand Trench transverse-raggedstreaks occur in multiple setsand createa Planimetric measurements Length brightand dark patternin which it is impossibleto separate Width dark streakson a brightbackground from brightstreaks on a Azimuth darkbackground (the "zebra"effect, Figure 5f). Landform diameter (or width) The landform with which the wind streak is associated is Terrain consideration Geologic setting alsoconsidered in thedescriptions where appropriate. Cones Elevation are conicalin topographiccross section and commonlyhave Local slope direction summit craters; hills are more rounded, typically lack Local slope magnitude summitcraters and include domical crosssections; craters are circulardepressions and may showevidence of modestraised See text for explanations. rims;trenches are linear features of negativerelief; and ridges sucha derivationwould be prematureuntil the full rangeof are linear featuresof positiverelief. The small widths of possibilitiesis known upon completion of mapping by trenches and ridges make it difficult in most cases to Magellan. Table 1 showsthe parametersthat appearto be distinguishbetween the two on SAR images. Somewind importantin describingVenusian wind streaks.A database streaksoccur on otherwisefeatureless plains and are not is being compiled which includes these parametersand associated with obvious landforms. descriptionsof the terrain,topography, and surface properties The parameters shown in Table 1 were used in the in which the streaksoccur. Each streakis "tagged"in the assessment for each wind streak found. Measurements database by thelatitude and longitude of its inferred(upwind) (Figure4b) weremade for themaximum length and width of point of origin, usingbest availablepositional information. the streak (length was taken to be the axis of the streak Aeolianfeatures in someareas occur as multiplestreaks. For oriented parallel with the inferred wind direction), the suchareas, one entryis madein the database, along with an diameterof the landform(somewhat arbitrary for somehills estimate of the total number of individual streaks. The andcones that mergewith the surroundingplain) or widthof valuesof length,width, etc., given are estimatedto represent the trenchor ridgewhere determination was possible. The the set, but the estimatesare not based on a rigorous streakazimuth (i.e., degreesfrom north) in the downwind statisticalanalysis. direction was also measured. The terrain in which streaksand otheraeolian features occur was assessed for generalgeology, 2.1. Wind StreakDescription usingpreliminary units definedby Saunderset al. [1991]. The shapeof the streakin planform is consideredto be a Backscatter cross sections were obtained in lines across primarydescriptive characteristic. Five shapesare commonly selectedstreaks, perpendicular to theiraxes. found: fans, linear streaks,wispy streaks,transverse-ragged Following the method outlined above, more than 3400 streaks,and transverse-smoothstreaks (Figures 4 and 5). streakshave been identified on Venusthus far in theanalysis Fan-shapedstreaks have a variety of outlinesand are often of Magellandam. associatedwith landformssuch as smallhills (Figures5a-5c). Transversestreaks typically occur in setsalong fracturesor 2.2. Wind Streak Distribution ridgesoriented perpendicular to the inferredwind direction, and may be either ragged (serrated)or smoothin outline The distribution of wind streaks was assessedin relation to (Figures 5d and 5e). Linear streaks(Figures 5f and 5g) type, streak length, latitude, elevation on Venus, and the typicallyare more than 20 timeslonger than their width and slope and slope direction(downslope) of the surfaceson often occur in sets of a half dozen or more similar streaks. which they occur. Distributionsof the azimuthsof wind Wispy streaksare wavy, meanderouspatterns that vary in streakswere also determined for comparisonswith modelsof width alongtheir length(Figure 5h). Wispy streaksare often atmosphericcirculation. These distributions include the associatedwith ridgesand impact craters. Many of the wispy estimatedtotal numberof streaksin areaswhere multiple streaksassociated with ridges are parallel to (and have the featuresoccur except for thedistributions by elevation,slope, same length as) the ridge. Wispy streaksassociated with andazimuth. As shownin Figure6a, darklinear streaks are impact cratersoccur in setsof a half dozenor more and form the most common, whereas dark transversesmooth streaks a meanderouspattern approximately radial to the crater or are the least common,with only one havingbeen found. 5km , !.-O'.km

'"

ß -.. ..

Fig. 5. Venuswind streaks(arrows indicate inferred downwind direction). (a) Radar-brightfan-shaped wind streak 10.5 km long associatedwith a small hill in easternNiobe Planitia, centeredat 36.5øN, 174.6øE (MagellanF-BIDR 1194). (b) Radar-darkfan-shaped wind streakabout 10 km long associatedwith a smallhill centeredat 29.4øN,57øE (Magellan MRPS 40983). (c) Radar-brightand -dark (mixed) fan-shaped wind streakin the Carsoncrater area, centered at 23øS,344.9øE. Area shownis about25 by 36 km (MagellanF-MIDR 23S345). (d) Transverseragged wind streak(radar-dark) associated with a ridge systemin southernLeda Planitia, centeredat 37.5øN, 65.5øE;area shownis about44 by 64 km (MagellanMRPS 38883). (e) Transversesmooth wind streak (radar-bright)associated with a ridgein GuineverePlanitia, centered at 26.2øN,331.4øE; area shown is about39 by 57 km (MagellanF-MIDR 25N333). (f) Multiple linear streaksin the vicinity of Mead crater,centered at 15øN, 65øE;area shown is about44 by 64 km (MagellanMRPS 37877). (g) Multiplelinear streaks (radar-dark) in western A.phrodite,centered at 0.9øS,71.1øE; area shown is 82 by 120km (MagellanF-MIDR 00N070). (h) Radar-dark wispy streak in easternSedna Planitia, centeredat 37øN, 2øE; area shownis about 87 by 128 km (Magellan C1-MIDR 30N009). 13,326 GREELEY ET AL.: AEOLIAN FEATURESON VENUS

14oo 12oo • Dark • Dark 12oo I--I Bright 1000 I--I Bright

i•i lOOO -:'•Mixed n-• 800 :.,:...:•Mixed 0• 800 o o 600 n- 600 a n- b rn 400 • 400

z 200 200 o , I , o ,law[_ Transv. Tmnsv. Fan Wispy Ridge Hill Cone Trench Crater ragged smooth STREAK TYPE STREAK ORIGIN Fig. 6. (a) Histogramsof wind streakson Venusby shapeand radar brighiness. (b) Histogramsof the type of' landforms with which streaks are associated.

Figure 6b showsthe distributionof radar-dark,-bright, and location of the altimetry data acquisition(nadir) and the -mixed backscatter cross section and the landform with which location of SAR and radiometry data acquisition. they are associated.Most radar-darkstreaks (mostly linear Consequently,where gaps in Magellandata occur, they affect forms) occurin associationwith ridges,whereas most radar- data at different locationson the planet. Where no altimetry bright streaks are found with small hills and cones. data are presentfor a streak,the streakwas not includedin Assessmentof strengthlengths shows that dark linear streaks elevationdistributions. Moreover, where altimetry data were are the longest(>100 km), whereasmost bright fan-shaped missing in the slope calculation region, the streak was streaksare <10 km long. omitted in both slope magnitude and slope azimuth Figure 7 showsthat streaksoccur over a broad range of distributions. Because of these considerations, the latitudeswith peaks in the latitude bands23øS to 30øS and distributionsshown in Figure 8 contain considerablyless 23øN to 30øN. Many streaksoccur in clustersassociated than the total number of streaksin the data base. Finally, with ejecta depositsfrom impact cratersin plains east of where comparativeglobal distributionsare shown,the two Alpha Regio, southernGuinevere Planitia, and in eastern curvesare normalizedto haveequal areas. AphroditeTerrae (Figure 1). No assessmentwas made of the Figure 8a showsthe distributionof elevationsfor streaks distributionof streakswith longitudebecause the Magellan relative to 6051 km (the referenceelevation on Venus, taken coverageby longitude was incompleteat the time of this from the center of the planet) and the distribution of all study. elevationsdetermined by the Magellanaltimetry experiment Figure 8 shows distributionsof streaks with elevation, [Pettengillet al., this issue]. Resultsshow that streaksform slope magnitude, and slope-streakangle. Because many at nearlyall elevations. Figure 8b showsthat moststreaks streaksare >10 km long, the elevation,slope, and anglemay form on surfacesof low slope, equal to or only slightly change as a function of distancealong the streak. For a steeperthan averageslopes on Venus. The distributionof given spacecraftorbit thereis a variableoffset betweenthe the smallestangle betweendownslope direction and streak

0.25 -

• all elevation data ", ...... streaks • 0.2 , , Z ,

o ß , , o o.15 ß o

LU o.1

• 0.05 o z o

-o.o5 ' I I ' I ' I ' I -90-64-53-44-36-30-23-17-11 -5 0 5 11 17 23 30 36 44 53 64 90 South North -4 -2 0 2 4 6 LATITUDE (degrees) ELEVATION (kin above 6051 ) Fig. 7. Distributionof streaks(of all types)by equal-areabands Fig. 8a. Distribution of streaksby elevationon Venus; curves of latitude. are normalized to unit area under each. GRF•LEY ET AL.: AEOLIAN FEATURES ON VENUS 13,327

0.25 70 globalslopes streak slopes ••Z0.2 oO0.15 • 0.1

z0.05 0 0.5 I 1.5 0 15 30 45 60 75 90 105 120135 150 165 180 DOWNSLOPE ANGLE SLOPE-STREAK ANGLE (degrees) (gamma, in degrees) Fig. 8b. Distribution of streaks by slope magnitude at 10 km Fig. 8c. Distributionof streaksin relationto slopedirection scale; curves are normalized to unit area under each. = 0ø = downslope,7 = 180ø = upslope). direction(Figure 8c) revealsthat streaksform at nearly all fan-shapedstreaks associatedwith small cones, although anglesto the local slope,with only a slightpreference for there are also some transverse-smoothbright streaks downslope orientation. However, in someplaces there is a associatedwith ridges. Comparisonsof F-BIDRs and F- correlationwith local slope. For example, streaksin the MIDRs from both cycle 1 and 2 showedno apparentchanges Ovdaregion tend to be orientedupslope, as discussed later. in the streaksduring the eight monthsbetween acquisition of Figure9 showsthe distributionof streakazimuths in the spacecraftdata. northernand southernhemispheres. Streaks in the northern hemispheresuggest formative winds predominandy from the 2.4. Wind Streak Formation northtoward the equator (azimuths mainly between-120 ø and 250ø). Very few wind streaksin the southernhemisphere Studyof wind streaksin the planetarycontext began with have azimuths between -90 ø and 270 ø, indicating a their discoveryon Mars in the early 1970s [Saganet al., preponderanceof inferred wind directions toward the equator. 1972, 1973]. Various investigationshave been made of Martian features[Veverka et al., 1977], terrestrialanalogs 2.3. TemporalChanges [Greeleyand Iversen, 1986], wind tunnelsimulations liversen and Greeley, 1984], and atmosphericconditions [Veverka et Duringthe second cycle of Magellanradar mappin g, 51 al., 1981] in an attempt to understandthe formation and orbitsof data with the sameleft-looking geometry as those evolution of wind streaks. in cycle 1 wereacquired. Using data from theseorbits, the Most wind streaks on Earth and Mars are associated with northernNavka Region was studied to determineif anyof the topographicobstacles and form in responseto wind patterns streaksfound in cycle 1 hadchanged. The regioncovers 10 ø and turbulencedeveloped around the obstacles.Dark (low to 30øN, and 329ø to 336øE longitude. Most of the streaks opticalalbedo) Martian streaksrepresent either bedrock areas foundin thisregion during cycle 1 areradar-bright and -mixed sweptfree of looseparticles or lag depositsof coarse•rains

15

: :.:.: rr 15 .:.:,.:.:. :•:•: :• i½ii•:: i:i:': i:i:i LU ..... :!:i

• lO :i:!: :.:.ii•! ii:i! !i:i: •: •Ji:::.:i: :•: :.:jii :i:i', :i•: :i:!! :i:i .:.: m 5 iii:'• • •?: '•'"'• •:•:

z :3 '•i• !i:.•.:!:i!• i•ii• i:::•- 5 .....>::•...... >.•.: •'• • •i:: •::•i •:::::::: • • • ;:•• ::• • •?:• o ...... o 60 120 180 240 300 360 0 • 120 180 240 • 360 AZIMUTH (degrees) AZIMUTH (degrees) (0 ø = North) (0ø=North) Fig. 9a. Distribution of streaksby azimuth (inferred downwind Fig. 9b. Distributionof streaksby azimuth(inferred downwind direction)in the northernhemisphere. direction)in the southernhemisphere. 13,328 GREELEYET AL.: AEOLIANFEATURES ON VENUS from which smaller, brighter particleshave been deflated. terrestrial wind streaks, such as those shown in Figures BrightMartian streaksare consideredto be dustdeposited in 2a-2d, are due primarilyto roughnessdifferences. On Venus, the lee of topographicobstacles to the wind, perhapsunder regionalcontext often providesinsight into the responsible conditionsof atmosphericstability [Veverka et al., 1981]. mechanism. We considerroughness differences related to The albedocontrast of Martian streakscan be explainedby sediment cover to explain the radar contrast for most depositsof sedimentas thin as a few microns[Thomas et al., Venusian wind streaks, although in some cases such 1981, 1984;Lee, 1984]. Under mostcircumstances, deposits interpretation is ambiguousand differences in dielectric this thin wouldbe penetratedby radarenergy, and the streaks constantmay be involved. would not be seenon Magellanimages. On the otherhand, Figure 10 showsa regionof Venuswhere insight may be the estimates of sediment thickness for the Martian streaks gainedinto the thicknessof windblownmaterial forming are only lower limits, andthey could be muchthicker. radar-darkstreaks. It showsa radar-brightcrater outflow What are the requirementsfor wind streaksto be visibleon deposit(inferred to be rough)overlain with severalradar-dark radarimages? The backscattersignatures of aeolianfeatures wind streaks. The most plausibleexplanation is that the may indicateone of severalpossible modes of origin, some windblown material forms a smooth,homogeneous layer of which can be assessedquantitatively or by analogy to overthe flowsand absorbs some of theradar energy, leading featureson other planets(Table 2). At full resolution,the to a decreasein backscatterfrom the underlying flow. Using ability to detectfeatures by theircontrast in radarbackscatter a simplemodel of a homogeneouslayer overa roughsurface is limited primarily by coherentnoise ("speckle"). The (neglectingsurface refraction effects and assumingthat the minimum detectablecontrast ratio in Magellan SAR images layerhas a low dielectricconstan0, the change in backscatter allows identificationof streaksin which radar specificcross (Ao) dueto the overlyinglayer is givenby sectiondiffers from the surroundingmaterial by ~1 dB or AO- 8.7 H more, with betterthan 67% confidence.This correspondsto ~10% changein returnedpower. In general,for Magellan cos0 L radar,energy returned from a surfacedepends on (1) surface in which 0 is the incidenceangle, L is the penetrationdepth slope relative to the incomingradiation at the scale of the of the material, and H is the thickness[Elachi et al., 1984]. SAR resolution(for Magellan,--150to 300 m, dependingon For the area shown in Figure 10, 0 = 43 ø and AO for the latitude), (2) surface roughnessat the scale of the SAR streaksranges from 2 to 6 dB. Assuminga losstangent of wavelength (for Magellan, 12.5 cm) averaged over a 0.005 to 0.01 [Campbelland Ulrichs, 1969], this leadsto a resolutionelement, and (3) the complexdielectric constant of penetrationdepth of ~0.6 to 1.2 m and a corresponding the material. For Venusian wind streaks to be visible in minimumlayer of particlescomposing the dark streakof 0.1 Magellanimages as a consequencesolely of slopesrequires a to 0.7 m thick, well within the range for wind streaks physically improbable(in some cases,unrealistic) surface. observed on Earth but thicker than the minimums inferred for Thus,the radarcontrast between streaks and the surrounding Mars. Consequently,the Venusianfeatures may form over terraincould result either from differencesin roughnessor in much longer time scalesthan Martian features,which can dielectric constant. Radar discrimination of observed changein as little as ~38 days.

TABLE 2. Wind Streaks: Hypothesesand Tests

StreakType Possible Origin Tests

Bright High reflectivity 1. Mineralogy Emissivity, reflectivity anomaly. 2. Bulk density(welding) Roughness 1. Rough deposit RMS slope value (if "fractal"). 2. Microdunes RMS slopevalue (if "fractal"). 3. Scouredsurface (bedrock or lag) Association with obstacles; "window" effect. 4. Nondeposition(assuming smooth deposit "Shadow"effect possiblein streak setting? off streak) Radar penetrationwith enhancedscatter from subsurface interface 1. Smooth depositsoverlying rough substrate Consistent with necessaryconditions? (inclination angle, dielectric, etc.) Dark Low reflectivity 1. Low densitydeposit (soil, pumice) Emissivity, reflectivity anomaly. 2. Lossy material Emissivity, reflectivity anomaly. Smoothness 1. Soil filling in or covering roughness Possible sources, gradational contacts. elements 2. Smooth sheet of rock Superposition,morphology (margins). 3. Exhumation of smooth surface Association with obstacles; "window" effect. 4. Nondeposition(assuming rough depositoff "Shadow"effect possiblein streak setting? streak) GREEI•EYET AL.: AEOLIAN FEATURESON VENUS 13,329

Fig. 10. Part of westernNiobe Planitiashowing radar-dark streaks (arrows) superposed on radar-brightcrater outflow deposits.The outflowsare associatedwith the 42-km-diarnetercrater Manzolini. Radar-darkmaterial formingthe streaksis estimatedto be 0.1 to 0.7 m thick;area shown is centeredat 26.5øN,93.5øE (Magellan C1- MIDR 30N099).

.

Terrestrialanalogs. Wind streakson Earth can provide insight into wind streak formation. Most wind streaks associated with landforms such as hills, raised-tim craters, VORTEX CORE and othertopographic obstacles can be relatexlto a turbulent "SHADOV•ZONE wind patternknown as a horseshoevortex [Greeleyet al., 1974]. As shown in Figure 11, wind flow separationand reattachment results in distinctive patterns of sediment ATTACHMENTPOINTOF :' erosion and deposition related to the geometry of the landform and the shear stressexerted on the surfaceby turbulent winds. Wolf Creek impact crater in Australia [McCall, 1965] exemplifies wind erosion and deposition patterns around a raised-rim crater (Figure 2h); sand depositionoccurs in a zoneupwind from the rim and as two HORSESHOE REVERSE FLOW VORTEX trailinglinear dunes downwind from the crater. The lee zone of the craterlies beneaththe merging"cores" of the trailing vorticesand is sweptfree of sandto exposebedrock. This samebasic pattern is seenat the Amboylava field, California Fig. 11. Diagram of wind flow aroundtopographic obstacle such [Greeleyand Iversen, 1986], where wind flow arounda cinder as a small hill. A "horseshoevortex" wraps around the hill, conecreates patterns of sanddeposition and sand-freezones; creatinga zone of turbulenceand high surfaceshear stress in the this pattern is visible on both optical and radar images wake of the hill. Material is preferentially eroded in this zone. (Figures2a and2b). Rising turbulentwind componentsalong the outer edgesof the vortex coresmay causepreferential deposition, creating a "halo" Wind streaksof severalforms and in a varietyof seuingson of particulate material around the eroded zone [from Greeley, Earth are visibleon Seasatand shutfieradar images [Greeley 19861. 13,330 GREELEYET AL.: AEOLIANFEATURES ON VENUS

%--,-.'-.; -'-:..

.:•:•.;•...-...... '"'"

4E:?..::?,:-:;•ß.•.. ,...... :::: •;;•:•...... ß"½½:•t;i:•:,;•'•:,•;•!(:•':'•i½:•½•:•:.:;:; :...... ]:;•:•::.,...... :.::; .:.. 'f,:..',•:;•.4•:•:•::: :::-...•.:½;•". ::•,':;;:.:•!•,•;,.;•-•,:•..:';::&.. .•;:;; •:;•::;:,% ';;?•.....-'½'.z ...... • •::,:;;4::.,, •:• .: ...... "':'.•:::.; ;•: ;::•,, ...•;: •.'.. 3...:.3,•:;;•:•;:•::.•,%:.•:.:•4s•;:•,•R:;:½::;•:.:;•;2•:•%,•::-:'.;.f

Fig. 13. Radar-brightfan-shaped streaks resulting from erosion. Dark haloes around bright streaks are inferred to be concentrationsof particulate material swept from the bright zonesby wind. Area shownis centeredat 22.3øN,332.1øE and is ~47 by 60 km (Magellan F-MIDR 20N334).

run. Patterns of erosion are defined as zones where sand has been removed,exposing the bare (dark) wind tunnelfloor. Thesezones correspond to highwind shear relative to therest of the floor and are pointsof flow reattachmentas relatedto the horseshoevortex flow field (Figure 11). Summaryfor Venusianwind streaks. Complexitiesin the Fig. 12. Venuswind tunnelresults for flow over dome-shapedinterpretation of backscattersignatures make it difficult to hill; flow is from left to right; area shownis about30 by 100 assessthe erosional vs. alepositionalorigin of aeolianfeatures cm: (a) beforethe run, surfacewas covered with fine quar• sand, on Venus (Table 2). Results from studies of terrestrial (b) during the run, erodedzones show as dark areaswhere the wind tunnel floor (black) becomes exposed; note the dark analogsand wind tunnel simulationsshow that erosional erosional "collar" that reflects turbulence associated with the zonesare expectedin the wake of topographicobstacles to horseshoevortex shown in Figure 11, and (c) after the run, the wind as a consequenceof flow reattachment and showingcontinued erosion and th6 merging of trailing cores acceleration of the wind. With time, on surfaces manfled from the horseshoe vortex to form an eroded streak downwind with loose sediments, these zones would be scoured free of from the hill. By analogy with wind streaksvisible on radar images,the areascovered with sandwould be relativelyradar- sedimentsto exposebare bedrock. In termsof radar-visible dark, and eroded areas would be radar-bright, assumingthat streakson Venus,these zones generally would have brighter underlyingbedrock were rougherthan sediment-coveredareas. radar backscatter cross sections than the surrounding, sediment-manfledterrain, resulting in radar-brightstreaks et al., 1983, 1989; Elachi et al., 1982; Saunders et al., similarto the Amboy streakin the Mojave Desert. 1990]. In most cases, contrastsin radar backscattercross It shouldbe notedthat radar bright streaks, as erosional sectionresult from depositsof windblownsand (radar-dark) features,are considered to be equivalentto opticallydark wind and exposuresof bedrock(radar-bright) that form patterns streakson Mars. For example,the radar-brightfan-shaped related to the local wind flow field and which are consistent streaksshown in Figure13 are consideredto be theresult of with regional-scalewind regimes. erosion. The terrain surroundingthe streak is relatively Wind tunnel simulations. Laboratory experimentsalso smooth and radar-dark, suggestinga mantling deposit, provideclues to thecomplex flow of windsover and around probablypart of the parabolicejecta collar from Aurelia topographicfeatures and indicate where zones of erosionand crater. Outside this area, the surfaceis characterizedby a depositionoccur [Greeley, 1986]. The VenusWind Tunnel reticulatepattern. We observethat the same reticulate pattern (VWT) was designed to study the physics of particle is foundwithin the fan-shapedstreak, suggesting that the movement in the Venusian environment and to model manflingdeposit has been removed by winderosion enhanced erosionand depositionaround landforms [Greeley et al., by turbulentflow aroundthe smallcone. 1984]. To supportthe analysisof Venusianwind streaks,a Someradar-bright streaks could represent deposits of high series of tests was conducted in which wind flow was reflectivitymaterial. For example,the long, radar-bright assessedover stylizedhills. The modelhills were placedon streaks originating from small bright hills shown in the floor of the wind tunnel, the floor was mantled with a Figures14a and 14bcould be materialeroded from the hills 0.5-cm layer of loosesand, and the wind speedwas setjust anddeposited downwind. Candidate materials are illmenite, abovethreshold for thesand entrainment (-0.5 cms-l). pyrite, or other high-densityminerals that have high Runs were continued until most of the sand was removed. dielectricconstants, as suggestedby Pettengillet al. [1982, Figure 12 showsVWT resultsbefore, during, and after one 1983] and Garvin et al. [1985] for someregions on Venus. GREELEYET AL.: AFDLIAN FEA• ON VENUS 13,331

•-" J •.:• t /' ,••.- • '..'\"...... •::::'::-:

.., t .:) ' ..,

. •,

, .• ,...... : , •.

:(•, , ,•:.•":i,,

•f•. 14•. •e •o•n•-Mes•enet dunefield, centered•t 67.7•N, 9O.•oE;•e• shown•s •7 b• •7 km (M•ell• M•PS •9•4).

Becauseof theirrelatively high density,they may form "lag" 3. DUNES deposits from which lower-density particles have been removedby the wind. Wind tunnelexperiments to simulate Two possibledune fields have been identified on Venus, Venus show that lag depositscould form under Venusian one centeredat 25øS, 340øE and the other centeredat 67øN, conditions[Greeley et al., 1991]. Suchpreferential wind 90øE. Bright wind streaksare associatedwith both dune winnowingwould be expectedon Venusin the wake of the fields and indicate that the dunes are oriented transverse to the hills seenin Figures14a and 14b, and the hills couldalso be prevailingwinds. The first dunefield, initiallydescribed by the source of the radar-reflective material. Arvidsonet al. [1991], is about100 km northof the impact Radar-darkstreaks are moredifficult to explainin termsof craterAglaonice and covers an areaof-1290 km2. We wind flow patterns.In general,we considerradar-dark streaks designatethis the Aglaonicedune field (Figure 16). The to represent deposits of sedimentsthat have low radar Aglaonicedunes range in lengthfrom 0.5 to 5 km; however, backscatter cross sections. Most radar-dark streaks on Venus becausethe dunesare dominatedby specularreturns on the are associatedwith ridgesand trenchesand thusprobably radarimages, their spacing cannot be determinedaccurately. represent places where sedimentsare concentrated;for The orientationof the dunesand nearby wind streakssuggest example,"gaps" in ridgescould funnelwindblown material winds toward the west. into narrow corridors. Radar-dark streaks on Venus associated Duneson Earth resultfrom the accumulationof saltating with "point"features such as hills couldalso be explainedas particles;as discussedby Bagnold[1941], "sand"size grains depositsof sediments.Although runs have not beenmade in (-60 to 2000 gm in diameter)are commonlytransported by VWT, previousruns made at 1 atm underlow wind speeds the wind in saltation.Smaller (i.e., "dust")grains are moved and long durationshave resultedin long depositsin the predominantlyin suspension,and largergrains (i.e. granules obstacle wake (Figure 15). Under these conditions, a and gravels) move in "creep." Despite the differencein "shadow zone" protected from the wind extended far atmosphericdensity between Earth and Venus, approximately downwind, whereasthe entire surfaceof the model was eroded the same mode of transportby size distributionoccurs of loose sediments. Such featuresproduced in the wind [Iversenand White,1982]. Consequently,one would expect tunnelare considered to be analogousprincipally to theradar- the duneson Venus also to be composedof "sand"size darklinear streaks seen on Venus(Figure 5g). material,i.e., grainsmoved in saltation. 13,332 GRFZLEY ET AL.: AEOLIAN FEATURES ON VENUS

Fig. 14b. Enlargementof areaindicated in Figure14a, showingdunes and radar-bright streaks.

The Aglaonicedune field is in theso-called "crater-farm", an field appearto originatefrom small (-200 m), radar-bright areaof relativelyabundant impact craters. Consequently, the cones. The streaksmost likely consistof the same high surfacein this area is expectedto be a sourceof sand-size/J radarreflectivity material as the cones. The sharpbrighiness materialfrom the ejectagenerated by the impactevent. The boundaryin the middle of the image shownin Figure 14a dunefield is locatedin an "outflow"deposit extending ~250 does not appear to be associatedwith any topographic km north from Aglaonice. Although the origin of crater change,suggesting that the brighinessis associatedwith a outflow featureson Venus is unclear,they couldresult from change in dielectric constant. Measurementsacross the turbulentlyemplaced ejecta and/or outflow of lava [Phillips boundaryshow a changein backscatterof 6 dB (a factorof 4 et al., this issue]. Regardlessof the process,at Aglaonice, in returned power) which, in turn, correspondsto a outflowmaterial evidently has been reworked by the wind to significantchange in the dielectric constant. If the low- form both dunes and wind streaks. Similar aeolian features reflectivityareas are assumedto havea dielectricconstant of havenot beenidentified at otheroutflow deposits thus far in 2 to 3, the brightarea would have a dielectricconstant of 5 to the analysesof Magellandata. The Aglaonicedeposit may 11. We suggestthat the brightarea is composedof materials have been initially less consolidatedor may be more thatare of a differentcomposition than the surfaceon which weatheredthan other crater outflow deposits. theyoccur. The northern dune field (termed Fortuna-Meshkene0is The most likely source for the material forming the located in a valley between Ishtar Terra and Meshkenet Fortuna-Meshkenetdunes is debris from the surrounding Tessera.It covers~17,120 km 2 andhas -40 radar-brightregions of complexlydeformed tessera. The N-S trending linear wind streaksthat occurwithin the field (Figure14a). low-lyingregion between the tesseraeappears to haveserved The orientation of the dunes and the wind streaks in the as a trap for weathereddebris. Some materialalso may be southernpart of the field indicate a southeastto northwest derivedfrom a nearbyparabolic halo crater. A similarvalley wind flow thatshifts to a westwardflow in thenorthern part to the east contains no visible aeolian features. of the field. Two brightstreaks near a 12-km-diametercrater Based upon look angle effects of terrestrialdunes using northof the dunefield alsosuggest westward wind flow. The airix)me,Seasat, and shuttle imaging radar images, dunes are dunesrange from 0.5 to 10 km in length,are •0.2 km wide, bright in low look angle radar imagesbecause of quasi- and havean averagespacing of 0.5 km. The spacingof the specularreflections from smoothdune faces that are near- dunesincreases toward the westernand eastern margins of the normal to the radar beam, where the incidenceangle at the valley. dune face is zero [Blom and Elachi, 1981, 1987] as shownin The bright wind streaksin the Fortuna-Meshkenetdune Figure 14c. At Magellanwavelength (12.6 cm), a radarecho GREELEYET AL.: AF_•LIANFEATURF3 ON VENUS 13,333

Fig. 14c. Radarimage of the dunefield in the Gran Desierto,Sonora, Mexico; areashown is ~25 by 15 km; radar "look direction"is from the southwest(left) towardthe northeast(right) (Seasatimage revolution 1312). from a duneis possibleonly whena sandsurface several betweenthe dunesthemselves. One possibilityis that the wavelengthson a sideis nearlyperpendicular to theimaging duneslie on a roughbasement. If thisis the case,the rough radar. Becausewindblown sand on Earth has an angle of basementwould showas brighton the radarimage, whereas reposeof about33 ø, radarbackscatter from duneslip facesis the smoothdune faces would appear dark. In this case,the possibleonly at lookangles less than 33 ø wherea duneslope dunes need not be near-normal to the radar illumination. is normal to the incident beam [Blom and Elachi, 1987]. Anotherpossibility is that the duneshave a lower dielectric Assuminga similarslope geometry for Venusiandunes, only constantthan the surroundingterrain so they appeardarker. thosedunes viewed at look angles-33 ø and with slip faces In either case, the periodic image brightnesspattern is orientedapproximately perpendicular to theradar illumination consistentwith imagesof sanddunes on Earth. will yield a radar backscatteron Magellan images. The Aglaonicedune field hasan incidenceangle of ~34ø, andthe 4. YARDANGS dune slopes are oriented perpendicular to the radar illumination,thus satisfyingthe necessaryslope geometry Yardangsare streamlinedhills that resultfrom wind erosion for viewing. The Fortuna-Meshkenetdune field wasviewed of rockand indurated sediments. A field of possibleyardangs with an incidenceangle of ~22ø, and most,although not all, has been identified on Venus at 9øN, 60.5øE, about 300 km of the dunesare orientednearly perpendicularto the radar southeastof the crater Mead (Figure 17a). The region illumination. But this dune field also has faces oriented surroundingMead containsthe highest concentration of wind- parallel to the radar illumination. Studiesby Blom and related features on Venus identified to date and is discussed in Elachi [1981, 1987] have shown that dune faces that are not more detail in section 5.1. The yardanglike featureson near-normalto the radarillumination will not returna quasi- Venusconsist of setsof slightlysinuous, parallel ridges and specularreflection to the radar. Instead,the dunesbecome grooves.The featuresaverage 25 km longby 0.5 km wide, invisiblewhen not imagednear-normal. This would imply with spacingbetween the ridgesranging from 0.5 km to 2 thatin orderto be interpretedas dunes,other scattering effects km. Unlike wind streaks,they have well-definedboundaries must be involved, such as change in the roughnessor and do not originatefrom topographicfeatures, such as hills. composition(i.e., dielectric constant)across the dunesor The proposedyardangs occur in two sets,each composed of 13,334 G•Y ETAL.: AEOIJANFEA• ON VENUS

Fig. 14d. Enlargementof some dunes in Figure14a. Thedunes on Earthin Figure14c and the Venusian dunes shownhere are both displayed at thesame scale. The imageis centeredat 67.7øN,90ø5E. about a hundred individual features. Both sets indicate a and dust can be producedfrom impact crateting, volcanic northeast-southwestwind regime. eruptions, and weathering by chemical and physical Yardangsoccur in many desertregions on Earth and in processes. Mechanical weathering of rock and particle some areas of Mars (Figure 17b). Yardangs commonly generationon Venus also may occur with the formationof develop on relatively soft depositsthat are sufficiently tesseraterrain, ridge belts,coronae, and rift zones. Figure 1 cohesiveto retainsteep slopes. Lakebed sediments and some shows that the distribution of aeolian features on Venus is volcanicash deposits are typicalmaterials in which yardangs not random;rather, they appearto be associatedwith certain form [McCauleyet al., 1977]. On Earth, yardangsoccur in impactcraters and sometectonically deformed terrains. To clustersoriented parallel to the prevailing winds which illustrate theseassociations, we discussthe Mead and Carson formed them but are.also controlledby structuralfeatures, craterareas and deformedterrains in TellusRegio and Ovda. suchas joints, and erosionalpatterns such as streamvalleys. The shape of yardangsis the result of several factors, includinglithology, structure, wind flow field, surrounding topography,and the supplyof agentsof abrasion[Ward and Greeley,1984]. The possibleyardangs on Venus suggestthe presenceof relativelyfriable depositsthat have been subjected to erosion and thatagents of erosionby the wind havebeen active in the pastor are currentlyactive. The sourceof the materialthat -! formedthe possibleyardangs is interpretedto originatefrom the formationof Mead crater. The yardangsindicate that continuingwinds in this region have reworked the Mead depositsover time. 5. CORRELATIONSOF AEOLIAN FEA• WITH GEOLtX;Y

The association of wind streaks and other aeolian features with local and regionalgeology can provideinformation on Fig. 15. Wind tunnel results for flow over raised rim crater, thepossible sources of windblownmaterial. In general,sand generatinglong depositionalfeature in wake of crater. GREELEY ET AL.: AF.DEIAN FEA• ON VENUS 13,335

Fig. 16. Aglaonicedune field, centered at 24.8øS;area shown is ~78 by 180km. Thisdune field, indicated by the specularpattern at "A", is locatedwithin an outflow associated with theAglaonice impact crater. Radar-darklinear streakssweep across the area, suggesting winds from the east (right) toward the west (left). If thiswind orientation is correct,the proposeddunes would be transverseforms (Magellan MRPS 34032).

Theseregions were chosen on thebasis of theabundance and contrast,radar-bright streaks in the area have emissivity varietyof streaksand are unique, rather than typical, regions valuesas high as 0.827. Fresnelreflectivity values (corrected on Venus. For each area, we have assessedthe local for the effectsof diffusescattering [see Pettengill et al., this geology,possible sources of particulatematter, and the issue]) generally have values that are close to the unit developmentof thewind streaks. complementof emissivity,with a range of values from 0.107 on brightstreaks to 0.138 on darkstreaks. The overall 5.1. Associationof AeolianFeatures With Impact Craters loweremissivity for theregion suggests that if thedeposit is Mead Crater. Mead (Figure 18a andPlate 1) is 275 km in fine-grained,then differences in mineralogymay account for diameterand is the largestpreserved impact crater found on thehigher dielectric constant. Venus. It is characterizedby two tings,a radar-brightfloor, CarsonCrater. Carson(Figure 19a andPlate 2) is one of and little apparentejecta [see Phillips et al., this issue; severalimpact craters on Venusthat is surroundedby a low- Schaber et al., this issue]. The area surroundingMead backscatter,low-emissivity parabolic halo [Phillips et al., includesmotfled radar-bright and -darkplains and has a large this issue]. The radar-darkhalo is superposedon plainsunits concentration of wind streaks. Most of the streaks in the composedof lobatevolcanic flows. Within the halo and in regionare darklinear and dark transverse-ragged forms. A the immediateregion (within a 500-kin radiusof the crater), few radar-darkwispy streaksare found west of Mead and numerouswind streaksare found that appearto be either a abouta dozenradar-bright fan-shaped streaks occur northeast directresult of the impactprocess or a resultof subsequent and southeast of the crater. Most streaks form in the lee of redistributionof fine-grainedmaterials. The streaksare ridgesin thelow-lying plains, although some streaks form in locatedin low-lyingplains west of Alpha Regioand include the lee of small hills (Figure 18b). Streaksrange in length radar-brightfan, -darkwispy, and -dark linear forms. Streaks from a few kilometersto > 100 km. Streakssurrounding rangein lengthfrom tensof kilometersto > 100 kin. Many Mead indicatewind flow towardthe equator,at leastfor the of the streaksform in the lee of small cones,although some time when the streaks were formed. In addition to the wind streaks form in associationwith ridges. Most streaks streaks,the field of possibleyardangs described in section4 indicatewinds flowing towardthe equator,although others is locatedsoutheast of the crater. Immediatelyeast of Mead indicate flow in random directions. is a gap in Magellandata causedby superiorconjunction. Emissivityvalues on the parbolichalo rangefrom 0.778 to Consequently,the full extentof aeolianfeatures in thearea is 0.830, comparedwith an average value of 0.844 for the not known at this time. surroundingterrain. Valuesof 00, emissivity,reflectivity, The surfacearound Mead crater appears to be blanketedwith and rms slope for some streaksin the Carson area are fine particlesthat were probablyproduced at the time of includedin Table 3. As in the Mead region,the dark streaks impact.In addition,the crater is surroundedby a faintradar- show higher reflectivitiesthan the bright streaksand the dark halo, visible in the emissivity data (Plate 1). The surroundingplains. Emissivityvalues appear to be strongly generalarea shows lower radar backscatter and emissivity influencedby the positionof the streakrelative to the low- values (-20.8 riB, 0.804, respectively)than the average emissivityparabolic halo. (-15 dB, 0.860,respectively) for theVenusian surface imaged Although Mead and Carson are unusualin their high at the sameincidence angle (45 ø in the first mappingcycle). concentrationof aeolian features, wind streakson Venus are Moreover, some of the lowest emissivity values in the most commonly found near impact craters. Most streaks regioncorrespond to concentrationsof wind streaks.For around craters form within 5 to 6 crater diameters of a crater example,the radar-dark region associated with the streaks near or cratercluster, indicating that impact-producedparticulate the center of Figure 18 has an emissivityof 0.808. In materialmay be locallysignificant. However, the amount of 13,336 GREELEY ET AL.: AEOLIAN FEATURES ON VENUS

• 20 km

.

Fig. 17a. Venusyardangs, centered at 9øN,60.7øE; area shown is -200 km by 200km (MagellanMRPS 37879).

finematerial produced by impactson Venusis notthought to be significantwhen averaged over the entire globe [Garvin, 1990]. Parabolichalo craters,such as Carsonand Mead, most commonly have adjacentstreaks, with over 70% of these craters having associatedaeolian features. The occurrenceof streaksaround impact craters that are surrounded by parabolasor darkhaloes and around the diffuse deposits thoughtto be "failed impacts"[Phillips et al., 1991, this issue]may indicate relatively young regions that have not yet been"homogenized" by theVenus environmen[ The impact craters where wind streaksare found may representthe ß .. youngestcraters on the planet, indicatingthat the streaks may be usefulas stratigraphicmarkers. We interpretthe

. . high correlationof streakswith impact craters,and the generallack of streaksin otherregions, to indicatethat the impactprocess is the mostefficient producer of particulate matter on Venus. From observationsand analyses of aeolian features associatedwith Mead,Carson, and other large impact craters and from considerationsof the propertiesof the surfaces where aeolian featuresoccur, we suggestthe following scenario:Prior to impact,a bow shockwas produced in the Venusianatmosphere by the incomingbolide. Becauseof the highdensity of the atmosphere,such a bow shockwould Fig. 17b. Yardangsand mesasin the westemThatsis region of be capableof producingsubstantial turbulence where it Mars; area shown is -190 by 190 km (Viking Orbiter frame interceptedthe surfaceand probablywas responsiblefor 44B37). generatingand dislodging weathered debris and injecting sand GREELEY ET AL.: AFDLIAN FEATURF3 ON VENUS 13,337

Fig. 18a. Mageilanradar image, centered at 15øN,59.1øE, of the Fig. 18b. Detail of linear streaks northeast of Mead. area northeastof the impact crater Mead (lower left cornerof Interfingering bright and dark streaks complicate the image). Modificationof the surface,indicated by wind streaks, interpretationof depositionalversus erosional origins; area extends as much as 150 km from the crater. White box outlines shownis centeredat 15øN, 60.4øE (Magellan MRPS 39821). the locationof Figure 18b. Numberson image correspondto Table 3 (MagellanMRPS 39820). obstacles. The radar propertiesof the streaks(Table 3) and dust into the atmosphere. Some of the dark patches indicatethat thebright streaks have higher radar backscatter describedby Phillips et al. [1991, this issue]as "failed crosssections than the surroundingplains, whereas the dark impacts"could represent particulate material both from the streaks have lower backscatter cross sections than the disruptedbolide and from weatheredmaterial dislodged surroundingplains. In addition,the darkest streak areas show locally. Thosebolides reaching the surface-generatedejecta emissivityvalues-0.02 lower than the surroundings, of a wide rangeof particlesizes; together with the material suggestingslightly higher dielectric constants for the less raisedby thebow shock,a symmetricalhemisphere of debris darkstreaks. The appearanceof boththe bright and dark areas expandedfrom the pointof impact. As the masslofted into in the SAR is therefore probably a result of roughness the atmosphere,fine particleswere caught by theprevailing differences,rather than differencesin dielectricconstant. In easterlywinds and distributedtoward the westto form the thisregion, the darkand bright streaks behind the ridges are radar-darkparabolic collar. Althoughmany wind streaksat interpretedto representadjacent areas of deposition(dark Mead crater (and to some degree at Carson)are oriented streaks)and scour(bright streaks). westward,some are randomlyoriented and may reflectlocal Hestia-Rupes/OvdaRegio is alsotectonically deformed and turbulence,or formation at a time not associatedwith the has a concentrationof streaks(Figure 21). Wind streaks impact. extenduphill from tesserablocks, indicating upslope winds.. 5.2. Associationof Aeolian Features With Tectonically As detailed by Arvidson et al. [this issue], the streaks delineate an elevation contourboundary in which streaks DeformedAreas occurwest of the boundaryat lower elevations.To the east Wind streaks and other aeolian features are found in some (at higherelevations) the plainshave enhancedbackscatter areasthat have been tectonically deformed. For example,the crosssections, and as discussedby Arvidsonet al. [1991], area southwestof Tellus Regio, shown in Figure 20, this variation may be related to elevation-dependent contains abundantwind streaks that do not appear to be weatheringreactions. Above a criticalelevation (6054 km related to an impact crater (the nearest impact crater, [Pettengillet al., 1983]) the plainsare inferredto be bright Voynich,is > 1000km away).Consequently, the particulate becauseof the presenceof high dielectricmaterials which material associatedwith the wind streaks was probably maybe stableat thelower temperatures and pressures found generatedfrom the complexlydeformed, rugged terrain of at higherelevations. The streaksdeveloped because winds TellusRegio. The streaksin thisregion include radar-dark segregatedthese materials within the elevationzone where linear streaksand radar-brightand -dark fan-shapedstreaks. theweathering reaction occurs. Wind streaksrange from ~ 10 to > 50 km in length. All of Wind streaks are also associated with a few other the streaksare in low-lyingplains immediately adjacent to a tectonicallydeformed regions, such as Alpha and Laima zoneof complexlydeformed terrain that extendsfrom the tesseraeand in someridge belt regions.Tesserae and ridge mainbody of the highlands.Streaks in thisregion tend to beltshave been fractured and uplifted, resulting in mechanical form in the lee of ridges in the plains, with some dark erosion that may have produced sufficient fine-grained material apparentlyaccumulating behind topographic material to form aeolian features. On the other hand, no 13,338 G• ET AL: AF•LIAN FEATUR• ON VENUS

100 km

Plate 1. Image data with emissivityvalues shown in color overlay in the vicinity of Mead crater (circularpurple area in upper right). Emissivityvalues in the centralpart of Mead are as low as 0.702, while a large region around the crater has a mean emissivityof 0.804, which is roughly 0.05 less than a typical plains surface. Values < 0.788 are shown as violet; >_0.840 are shown as red. aeolian features have been identified around coronae and rift largeamounts of volatiles,and may be relativelyrare [Garvin zones,both of which are characterizedby extensivetectonic et al., 1982; Itead and Wilson, 1986]. Alternatively, deformation. However, both rift zones and coronae are far volcanicparticulate matter may be easily welded and thus less complexly deformedthan tesseraeand, thus, may not more difficult to rework with the wind. producesufficient sedimentto form aeolian features. In On mostplanets, more particulate matter is generatedwith addition,these terrains lack the steepouter slopes of tesserae age as surfacefeatures weather and erode. In general,a that may enhancemass wasting and the generationof local relativelylarge amount of particulatematter on Venusseems winds. Aeolian featuresmay be more commonlylocated in to indicatea youngerage, as in thecase of theparabolic halo the lee of ridge belts due to their unique topographic craters. Arvidson et al. [this issue] have observedthat lava characteristicsserving as an obstacleto the wind, ratherthan flows tend to becomemore "homogenized"over time. This theirassociation with significantmechanical erosion. processmay be relatedto somematerial binding process on Wind streaks form least frequently in associationwith the surface. For example,laboratory experiments show that volcanic features. Streaks,however, do form in the wake of in the high-temperature,high-pressure environment of some volcanic cones, but in many casesthe particulate Venus,particulate material tends to "coldweld" and becomes materialmay also have been derivedfrom nearbyimpact more difficult to move by the wind [Marshall et al., 1991]. craters.Explosive eruptions on Venusare thoughtto require Therefore,rexent impacts (i.e., Carson),regions of ongoing GREELEYET AL.: AEOLIAN FEATURF_SON VENUS 13,339 tectonic activity (i.e., Tellus), or elevated regions with uniqueweathering characteristics (i.e., Ovda)may be someof the relativelyfew siteson the planetwhere loose material is presentand is movedby thewind.

6. CONSTR• ON MODELS OF ATMOSPHERIC CIRCULATION As discussedin section1, atmosphericcirculation near the surfaceof Venuscould be predominantly:(1) a Hadley cell with equatorwardsurface winds, (2) an anti-Halleycell with surfacewinds away from the subsolarregion, (3) a zonal circulationwith westwardsurface winds, or (4) something more complex. The global patternof wind streakazimuths may provide clues as to which of these models (if any) characterize the near-surface circulation. Figure 9 shows that many streaks are oriented with equatorwardcomponents. This is particularlyevident in the southwardorientations of streaksin the northernhemisphere regionsat (25øN,335øE), (10øN, 65øE), and (20øN 100øE)and in the northward orientations of streaks in the southern hemisphereat 20øS, 345øE. Northern hemispherestreak azimuthsare clearlyconcentrated around 180øE; 73% of the total population of northern hemisphere streaks have azimuthsbetween 90ø and 270ø . Similarly, 77% of the southernhemisphere streaks have azimuthsbetween 0 ø and 90øE and 270ø and 360øE. Northernhemisphere streaks are predominantly oriented southward whereas southern i i i i i i i i i hemispherestreaks are mainlyoriented northward. The streak orientations are consistent with a near-surface Hadley cell circulation, and accordingly, they provide observational support for the existence of such an atmospheric circulation pattern. This conclusion is preliminaryand shouldbe treatedcircumspectly until more completewind streakdata for the whole planetare obtained i i i i i i i i i i i i i and analyzed. Possibleobservational biases, such as those associatedwith imaginggeometry, need to be assessedmore carefully. For example, the orientation of Magellan spacecraftground tracks might result in preferentialradar detectionof streakswhich lie parallelto them. The latitudinal distributionof wind streaksmay contain informationon the strengthof the equatorwardnear surface winds in the Hadley cell or on the polewardextent of the Hadley circulation.There is a strongconcentration of wind streaksin the southernhemisphere between 17 ø and 36øS latitude (Figure 7). Wind streaks are more uniformly distributedwith latitudein thenorthern hemisphere, although there is a concentration of streaks between 23 ø and 30øN latitude(Figure 7). The areabetween 17øS and 36øSis 30% of the area of the southernhemisphere, but 57% of the southernhemisphere wind streaksare concentratedthere. Similarly, the area between 23øN and 30øN, 10% of the northern hemisphere area, contains 20% of northern hemispherewind streaks. The peaks in the latitudinal distribution of wind streaks between 23øS and 30øS and 23øN and 30øNtend to suggestthat equatorward Hadley circulation windsare strongestat theselatitudes. The broaddistribution of northernhemisphere streaks over all latitudes(Figure 7) suggestthat the Hadley cell in the northernhemisphere may extendall the way to the pole;however, the relativelysmall numberof streaksin the southernhemisphere poleward of 13,340 GREELEYET AL.: AF_DLIANFEATURES ON VENUS

Fig. 19a. Magellanradar image of the areaaround the impactcrater Carson, centered at 25øS,345øE, radar-dark halo associatedwith the clusterof craters,and the parabolic-shaped"collar" associated with Carson. Numberson image correspondto Table 3 (MagellanF-MIDR 25S345).

36øS impliesweaker winds or lesspoleward penetration of 7. SUMMARY AND CONCLUSIONS the southern hemisphereHadley cell. There is some sampling bias, however, becauseMagellan coverage is Wind streaks are the most common aeolian feature on incompleteover the southernlatitudes and southpolar area. Venus. More than 3400 have been identified in this study. The implicationsof Figure 7 for atmosphericcirculation They occurin a varietyof shapesand includeradar-bright, must be viewed with cautionat this stageof our analysis radar-dark,and mixed radar reflectivity forms in relationto the because the latitudinal distribution of wind streaks can also backgroundon whichthey occur. Most streaksare foundin be influencedby otherfactors that would vary with latitude relativelysmooth plains in latitudinalbands of 23øSto 30øS suchas the supplyof windblownmaterial. and 23øN to 30øN. Wind streaks tend to be oriented Streak orientationsshow no preferentialalignments that downwindtoward the equator,consistent with surfacewinds would confirm or deny the existence of other lower relatedto a Hadleycirculation in the lower atmosphere.Data atmospheremotions such as the anti-Halley circulation or the do notsupport Halley circulation patterns related to subsolar- westwardzonal flow discussedearlier. For example, streak antisolarheating contrasts, although observations are limited azimuthspoint eastward as frequently as they point westward. to test this possibility. Wind streaks are found at all Thus, the dominantcirculation of the cloudlevel atmosphere elevationson Venus. In contrastto pre-Magellanpredictions doesnot penetrateto the surfacewith sufficientstrength to of downslopewinds, most wind streaksare foundon gentle preferentiallyorganize the wind streaks toward the west. The slopes(< 2ø) andare randomly oriented with regardto slope. wind vector at a fixed point on the surfacerotates through Locallye'asin Ovda Regio, streaks can be oriented upslope. 360ø duringthe diurnalcycle of the anti-Halley circulation. Both erosionaland alepositionalaeolian features have been Accordingly,it is difficultto devisea testof the wind streak identified on Venus. The presence of possible dunes datathat would provide evidence for or againstthe anti-Halley (alepositionalfeatures) provides clues to thenature of someof cell. the sediments and their behavior in the aeolian environment. GREELEY ET AL.: AEOLIAN FEATURES ON VENUS 13 341

50 km

Plate 2. Image data for the samearea as above,with emissivityvalues shown in color overlay. Emissivityvalues in the vicinity of the cratersare typically 0.05 lower than thosein the surroundingregions. Values < 0.784 are shown as violet; > 0.844 are shown as red.

Sand dunesform only from sand-sizeparticles (60 to 2000 !.tmin diameter)which are transportedprimarily in saltation by thewind, regardless of planetaryenvironment [Greeley and lversen, 1985]. Consequently,if the featuresidentified in Figures 14 and 16 are dunes, they signal the presenceof sandgrains and processesthat produceparticulate material. Althoughthe identificationof yardangsis tentative,their possible presencealso provides insight into the aeolian regime on Venus. As features formed by wind erosion, yardangsshow that windblown particles are capable of erodingmaterial despitethe relatively low kinetic energy produce•by slow-movingwinds. In conclusion,the surfaceof Venusis characterizedby low rates of erosion,primarily due to the lack of water on the surface.Mechanical erosion through tectonic deformation in Fig. 19b. Detail of streaksnortheast of Carson. Note variation ridgebelts and regions of tesseramay producesmall amounts in backscatterfrom the coreto the edgeof the streaks. of particulatematter that can form aeolianfeatures. Volcanic deposits may also play a small role in producing fine depositionalprocesses on Venus. Data from Magellan's material on Venus: many streaksform in associationwith extended mission will be used to assess the backscatter conesof probablevolcanic origin. However, the primary characteristicsof aeolian features of different viewing contributionto the productionof particulatematter on Venus geometriesand to obtaina completeinventory of wind-related is from impact cratering. Aeolian features form featureson Venus. Most significantly,over the next several predominantlynear impact craters, especially those with years, Magellan will provide the opportunity to detect associatedejecta haloes or parabolas,or near dark deposits changesin aeolianfeatures or the formationof new features, thoughtto be "failed"impacts [Schaber et al., thisissue]. providing further information on atmosphere/surface Continuinganalysis of the backscattercross sectionsof interactions and the nature and evolution of surface materials aeolianfeatures will providefurther insight into erosional and on Ven us. ":'i'::•i•:.' •j•'• ,'i'.-....:.."..:. ..,.:.....:.;:.::• • ,:.• :..::.:.i::*?: • "'...... :.;..... '" ..... '"' * '*::: ' •:-','•'

13,344 GRI•LEY ET AL.: AEO• FEATURF• ON VENUS

Acknowledgements. We wish to thank the following for their Greeley,R., and R. Arvidson,Aeolian processeson Venus,Earth contributions to this report: R. Blom and T. Fart for helpful Moon Planets, 50/51, 127-157, 1990. discussions, D. Ball for photographic support, E. Lo for Greeley,R., andJ.D. Iversen,Wind as a GeologicalProcess, pp. computational support, S. Selkirk for figure preparation, and 133-197, CambridgeUniversity Press, New York, 1985. S. Blixt for word processing.The Magellan Projectand parts of the researchdescribed here are carried out by the Jet Propulsion Greeley, R., and J.D. Iversen,Aeolian processesand featuresat Laboratory, California Institute of Technology,under contract Amboy lava field, California,Physics of Desertification,edited from the National Aeronautics and Space Administration, and by F. E1-Baz and M.H.A. Hassan, pp. 290-316, Martinus through contractsto individuals: JPL-958880 (Greeley) JPL- Nijhoff Dordrecht,Netherlands, 1986. 958496 and NAGW 1874 (Schubert), and JPL-957415 Greeley, R., and J.D. Iversen, Measurementsof wind friction (Arvidson). Support for Plaut was provided by the National speedsover lava surfacesand assessmentof sedimenttransport, ResearchCouncil under the ResearchAssociate Program. Geophys.Res. Lett., 14, 925-928, 1987. Greeley, R., J.D. Iversen, J.B. Pollack, N. Udovich, and REFERENCES B. White, Wind tunnel simulationsof light and dark streakson Antsibor, N.M., et al., Estimates of wind velocity and of Mars, Science, 183, 847-849, 1974. instrumentsdropped from Venera 9 and Venera 10, Cosmic Greeley, R., Y. Asmerom, and S. Barnett, Radar signaturesof Res., Engl. Transl, 14, 625-631, 1976. wind streaks: PreliminarySIR-A analysis,Lunar Planet. Sci., Arvidson, R.E., V.R. Baker, C. Elachi, S. Saunders, and XIV, 259-260, 1983. J.A. Wood, Magellan: Initial analysis of Venus surface Greeley, R., J. Iversen, R. Leach, J. Marshall, B. White, and S. modification, Science, 252, 270-275, 1991. Williams, Windblown sand on Venus: Preliminary resultsof Arvidson, R.E., R. Greeley, M. Malin, R.S. Saunders, laboratory simulations,Icarus, 57, 112-124, 1984. N. Izenberg, j.J. Plaut, and E. Stofan, Surface modificationof Greeley, R., P. Christensen, and R. Carrasco, Shuttle radar Venus as inferred from Magellan observationsof plains and imagesof wind streaksin the Altiplano, Bolivia, Geology, 17, tesserae,J. Geophys.Res., this issue. 665-668, 1989. Avduevskii, F.S., et al., Measurementof wind velocity on the Greeley, R., N. Lancaster, S. Lee, and P. Thomas, Martian surfaceof Venus during the operationof stationsVenera 9 and aeolianprocesses, sediments, and features,in Mars, University Venera 10, CosmicRes., Engl. Transl., 14, 622-625, 1976. of Arizona Press,Tucson, in press, 1992. Bagnold, R.A., The Physics of Blown Sand and Desert Dunes, Greeley, R., J.R. Marshall, D. Clemens, A.R. Dobrovolskis, 265 p., Methuen and Co., Ltd., London, 1941. and J.B. Pollack, Venus: Concentrations of radar-reflective Blom, R., and C. Elachi, Spaceborneand airborneimaging radar mineralsby wind, Icarus, 90, 123-128, 1991. observationsof sand dunes,J. Geophys.Res., 86, 3061-3073, Head, J.W., and L. Wilson, Volcanicprocesses and landformson 1981. Venus: Theory, predictions and observations,J. Geophys. Blom, R., and C. Elachi, Multifrequency and multipolarization Res., 91, 9407-9446, 1986. radar scatterometry of sand dunes and comparison with Iversen, J.D., and R. Greeley, Martian crater dark streaklengths: spaceborneand airborne radar images, J. Geophys.Res., 92, Explanation from wind tunnel experiments, Icarus, 58, 358- 7877-7889, 1987. 362, 1984. Campbell, M., and J. Ulrichs, Electrical propertiesof rocks and Iversen, J.D., and B.R. White, Saltation threshold on Earth, their significance for lunar radar observations,J. Geophys. Mars, and Venus, Sedimentology,29, 111-119, 1982. Res., 74, 5867-5881, 1969. K•lney de Rivas, E., Numerical models of the circulationof the Counselman, C.C., III, S.A. Gourevitch, R.W. King, atmosphereof Venus, J. Atmos.Sci., 31, 763-779, 1973. G.B. Loriot, and R.G. Prinn, Venus winds are zonal and K•lney de Rivas, E., Further numerical calculations of the retrogradebelow the clouds,Science, 205, 85-87, 1979. circulation of the atmosphereof Venus, J. Atmos. Sci., 32, Counselman, C.C., III, S.A. Gourevitch, R.W. King, 1017-1024, 1975. G.B. Loriot, and E.S. Ginsberg, Zonal and meridional Keldysh,M.V., Venus explorationwith the Venera 9 and Venera circulation of the lower atmosphereof Venus determined by 10 spacecraft,Icarus, 30, 605-625, 1977. radio interferometry,J. Geophys.Res., 85, 8026-8030, 1980. Kerzhanovich, V.V., and M.Ya. Marov, The atmospheric Covey, C., R.L. Walterscheid, and G. Schubert, Dissipative dynamicsof Venus accordingto Doppler measurementsby the tides: Application to Venus' lower atmosphere,J. Atmos.Sci., Venera entry probes, in Venus, edited by D.M. Hunten, L. 43, 3273-3278, 1986. Colin, T.M. Donahue, and V.I. Moroz, pp. 766-778, Dobrovolskis, A.R., and A.P. Ingersoll, Atmospherictides and University of Arizona Press,Tucson, 1983. the rotation of Venus. 1. Tidal theory and the balance of Kerzhanovich, V.V., Yu.F. Makarov, M.Ya. Marov, torques,Icarus, 41, 1-7, 1980. E.P. Molotov, M.K. Rozhdestvenskii, V.P. Sorokin, N.M. Elachi, C., et al., Shuttle imaging radar (SIR-A) experiment: Antsibor, V.D. Kustodiev, and V.I. Puchkov, An estimate of Preliminaryresults, Science, 218, 996-1003, 1982. the wind velocity and turbulencein the atmosphereof Venuson Elachi, C., L. Roth, and G. Schaber,Spaceborne radar subsurface the basisof reciprocalDoppler measurements by the Venera 11 imagingin hyperaridregions, IEEE Trans. Geosci.and Remote and Venera 12 spacecraft,Cosmic Res., Engl. Transl., 17, 569- Sens., GE-22, 383-388, 1984. 575, 1979. Garvin, J.B., The global budget of impact-derivedsediment on Lee, S.W., Mars: Wind streakproduction as related to obstacle Venus, Earth Moon Planets, 50/51, 175-190, 1990. type and size, Icarus, 58, 339-357, 1984. Garvin, J.B., J.W. Head, and L. Wilson, Magma vesiculationand Marov, M., V. Avduevskii, V. Kerzhanovich, pyroclasticvolcanism on Venus, Icarus, 52, 365-372, 1982. M.K. Rozhdestvenskii,N. Borodin, and O. Ryabov, Venera Garvin, J.B., J.W. Head, and S.H. Zisk, On the origin of higher 8: Measurementsof temperature,pressure, and wind velocity radar reflectivity surfaceson Venus (abstract),Lunar Planet. on the illuminated side of Venus, J. Atmos. Sci., 30, 1210- Sci., 16, 266-267, 1985. 1214, 1973. Greeley, R., Aeolian landforms: Laboratory simulations and Marshall, J.R., G. Fogleman, R. Greeley, R. Hixon, and field studies, in Aeolian Geomorphology, edited by W.G. D. Tucker, Adhesion and abrasion of surface materials in the Nickling, pp. 195-211, Allen and Unwin, Boston, Mass., Venusian aeolian environment, J. Geophys. Res., 96, 1931- 1986. 1947, 1991. GREELEY ET AL.: AEOLIAN FEATURES ON VFJqUS 13,345

McCall, G.J.H., Possible meteorite craters--Wolf Creek, and J. Russell, Geology and distribution of impact craters on Australia and analogs,Science, 123, 970-998, 1965. Venus:What are they telling us?,J. Geophys.Res., this issue. McCauley, J.R., M.J. Grolier, and C.S. Breed, Yardangs, in Schubert,G., General circulationand the dynamicalstate of the Geomorphology in Arid Regions, Proceedings 8th Venus atmosphere,in Venus,edited by D.M. Hunten, L. Colin, Geomorphology Symposium,edited by D.O. Doehring, pp. T.M. Donahue, and V.I. Moroz, pp. 681-765, University of 233-269, Allen and Unwin, Boston, Mass., 1977. Arizona Press, Tucson, 1983. Moroz, V.I., The atmosphereof Venus, SpaceSci. Rev., 29, 3- Schubert, G., et al., Structure and circulation of the Venus 127, 1981. atmosphere,J. Geophys.Res., 85, 8007-8025, 1980. Pettengill, G.H., P.G. Ford, and S.D. Nozette, Venus: Global Stone, P.H., The structureand circulation of the deep Venus surfaceradar reflectivity, Science, 217, 640-642, 1982. atmosphere,J. Atmos. Sci., 31, 1681-1690, 1974. Pettengill, G.H., P.G. Ford, and S.D. Nozette, Venus: Evidence Stone, P.H., The dynamics of the atmosphereof Venus, J. for high dielectric constantsin elevated regions (abstract), Atmos. Sci., 32, 1005-1016, 1975. Lunar Sci., 14, 602-603, 1983. Thomas, P., and J. Veverka, Seasonal and secular variation of Pettingill, G.H., P.G. Ford, W.T.K. Johnson,R.K. Raney, and wind streakson Mars: Analysisof Mariner 9 and Viking data, L.A. Soderblom, Magellan: Radar performance and data J. Geophys.Res., 84, 8131-8146, 1979. products,Science, 252, 260-265, 1991. Thomas, P., J. Veverka, S. Lee, and A. Bloom, Classification of Pettingill, G.H., P.G. Ford, and R.J. Wilt, Venus surfaceradio- wind streaks on Mars, Icarus, 45, 124-153, 1981. thermal emissionas observedby Magellan, J. Geophys.Res., Thomas,P., J. Veverka, D. Gineris, and L. Wong, "Dust"streaks this issue. on Mars, Icarus, 60, 161-179, 1984. Phillips, R.J., R.E. Arvidson,J.M. Boyce, D.B. Campbell,J.E. Tyler, G.L., P.G. Ford, D.B. Campbell, C. Elachi, Guest, G.G. Schaber,and L.A. Soderblom,Impact craterson G.H. Pettengill, and R.A. Simpson,Magellan: Electrical and Venus: Initial analysis from Magellan, Science, 252, physical propertiesof Venus' surface,Science, 252, 265-270, 288-297, 1991. 1991. Phillips, R.J., R.R. Herrick, R.E. Grimm, R.L. Raubertas,I.C. Veverka, J., C. Sagan,and R. Greeley,An unusualcrater streak in S arkar, R.E. Arvidson, and N. Izenberg, Impact crater Mesogaea,Icarus, 27, 241-253, 1976. distributionon Venus: Implication for planetaryresurfacing Veverka, J., P. Thomas, and R. Greeley, A study of variable history, J. Geophys.Res., this issue. features on Mars during the Viking primary mission, J. Rossow, W.B., A general circulation model of a Venus-like Geophys.Res., 82, 4167-4187, 1977. atmosphere,J. Atmos. Sci., 40, 273-302, 1983. Veverka, J., P. Gierasch, and P. Thomas, Wind streaks on Mars: Sagan, C.A., J. Veverka, P. Fox, R. Dubisch, J. Lederberg, Meteorological control of occurrenceand mode of formation, E. Levinthal, L. Quam, R. Tucker, J.B. Pollack, and Icarus, 45, 154-166, 1981. B.A. Smith, Variable featureson Mars: PreliminaryMariner 9 Ward, W., and R. Greeley, Evolutionof the yardangsat Rogers television results, Icarus, 17, 346-372, 1972. Lake, California, Geol. Soc. Am. Bull., 95, 829-837, 1984. Sagan, C., J. Veverka, P. Fox, R. Dubisch, R. French, P. Gierasch, L. Quam, J. Lederberg, E. Levinthai, R. Tucker, R.E. Arvidson, Department of Earth and Planetary Sciences, B. Eross, and J.B. Pollack, Variable features on Mars, 2. Washington University, One Brookings Drive, St. Louis, MO Mariner 9 global results, J. Geophys.Res., 78, 4163-4196, 63130-4899. 1973. C. Elachi, J.J. Plaut, R.S. Saunders, E.R. Stofan, Saunders,R.S., A.R. Dobrovolskis, R. Greeley, and S.D. Wall, E.J.P. Thouvenot, S.D. Wall, and C.M. Weitz, Jet Propulsion Large-scalepatterns of eolian sedimenttransport on Venus: Laboratory,4800 Oak Grove Drive, Pasadena,CA 91109. Predictions for Magellan, Geophys. Res. Lett., 17, 13 6 5- M.A. Geringer and R. Greeley, Department of Geology, 1368, 1990. Arizona StateUniversity, Tempe, AZ 85287-1404. Saunders, R.S., R.E. Arvidson, J.W. Head III, G.G. Schaber, G. Schubert,Department of Earth and Space Sciences,IGPP, E.R. Stofan, and S.C. Solomon, An overview of Venus University of California, Los Angeles,CA 90024-1567. geology, Science,252, 249-252, 1991. Saunders,R.S., et al., Magellan mission summary,J. Geophys. Res., this issue. (Received September27, 1991; Schaber, G.G., R.G. Strom, H.J. Moore, L.A. Soderblom, R.L. revised April 27, 1992; Kirk, D.J. Chadwick, D.D. Dawson, L.R. Gaddis, J.M. Boyce, acceptedApril 28, 1992.)