The Role of Rim Slumping in the Modification of Lunar Impact Craters
Transcription
The Role of Rim Slumping in the Modification of Lunar Impact Craters
VOL. 84,NO. B6 JOURNALOF GEOPHYSICAL RESEARCH JUNE 10, 1979 TheRoleof RimSlumping in theModification of LunarImpactCraters MARK SETTLE AND JAMES W. HEAD III Departmentof GeologicalSciences,Brown University,Providence,RhodeIsland02912 Wall failurehassignificantlyalteredthe structureof virtuallyall large,fresh-appearing lunar craters. Terraceblocksexposedupona crater'sinteriorwallsare interpretedto be sectionsof the transientcavity rim that slumpedinto thecavityduringthe terminalstagesof craterformation.Impactexcavationcavities have beenreconstructed by restoringthe innermostterraceblock exposedwithin a craterto its inferred originalpositionat the cavityrim and accountingfor the volumeof materialthat slumpedinto the cavity. Critical modelassumptions include(1) the radialvariationof topographynearthe initial cavityrim crest, (2) the structureof the failure surfacealong whichterraceblocksslumpedinto the cavity,and (3) the geometricshapeof the initial cavity.This terracerestorationmodel has beenapplied to 12 freshlunar craterswith observedrim crestdiametersDo rangingfrom 19 to 137km. Up to an initial cavitydiameter Dt of 30 km, reconstructed cavitydepthsare comparableto or greaterthan cavitydepthsextrapolated from nonterracedcratersof lessthan 15 km in diameter.For a wide range of model parametersthe reconstructed depthsof impactcavities15-30 km in diameterare significantly greaterthan depths predictedby small-cratermorphometry,implyingthat cavitydepthsinferredfrom depth/diameterratios observedfor smallcratersmay substantially underestimate the depthof excavationof impactcratering eventsin this sizeclass.Reconstructed cavitydepthsfor Di > 70 km, however,are consistently lessthan cavity depthsextrapolatedfrom smallercraters.This indicatesthat the morphometrictransitionfrom small,relativelyunmodified,bowl-shaped craters(Do < 15 km) to large,terraced,saucer-shaped craters (Do > 80-90 km) cannotbe solelyattributedto rim-slumpingmodification.Ejectafallbackand basement reboundalsoplay a role in modifyingimpactcratercavities;however,the mannerin whichthe volumeof fallback ejecta and basementrebound material varies with increasingcrater size is unknown.The discrepancybetweencavitydepthsextrapolatedfrom smallcratersand thoseobtainedfrom the terrace restorationmodelsuggests that impactexcavationcavitiesbecomerelativelyshallowerat largerdiameters. However, this cannotbe conclusivelydemonstrateduntil the effectsof reboundand ejectafallback are quantitativelyaccountedfor. INTRODUCTION Malin andDzurisin,1978].The purposes of thisstudyare to analyzethe process of rim slumpingand to measurethe effect of terraceformationon the shapeof largelunar craters.Preliminaryresultshave beenpresented previously[Settleand The shapeof lunar impact cratersvariessignificantlywith increasingcrater size.Craters with rim diametersDo lessthan 12-15 km are characterizedby an average ratio of depth/ diameterof 1:5. The depth/diameterratios of larger craters Head, 1976, 1978]. Knowledgeof thedepthof impactexcavation cavitieswould consistently decreasewith increasingcratersize,rangingfrom 1:8 for craters 20 km in diameter to 1:30 for craters with diametersof 140km [Pike, 1977a].This morphometrictransition is accompaniedby a distinctivevariation in crater morphology. Large cratersexhibit complexterraced walls, floor hummocks,centralpeaks,andflat floors[SmithandSanchez, 1973;Howard, 1974; M. Cintala and J. W. Head, manuscript provideimportantinformationon the maximumdepth of materialexposed withina crater'sejectadeposit (termedejecta sampling depthbyHeadet al. [1975])andtheinitialgeometry of very largecratersand basins.Presentestimatesof the maxi- mumdepthof ejectaexcavated by basin-sized impactssuchas lmbrium rangefrom 30 km to 200 km [Head et al., 1975; in preparation, 1979]whichareinferredto haveformedduring Dence et al., 1974]. the terminal stagesof crater formation [Gault et al., 1968; Dence, 1968; Quaide et al., 1965]. The frequencywith which terraces, central peaks, and flat floors occur in fresh lunar craters increaseswith increasingcrater size [Cintala et al., 1977], suggestingthat modification phenomenasuch as rim slumping,floor rebound, and ejectafallback becomeincreasingly pervasivein larger craters. Wall terracesare prominent featuresin large lunar craters. The mechanismof terraceformation is qualitativelyunderstoodto be a slumpingprocessin whichsegments of the rim of the initial crater cavity are translateddownward and inward [e.g., Shoemaker, 1962; Guest and Murray, 1969; Mackin, 1969]. Rim slumping increasesthe rim crest diameter of a crater and decreasesits depth, thereby reducingthe depth/ diameterratio. Although reboundand fallback may play important rolesin cavity modification,severalinvestigatorshave proposedthat rim slumpingmay be primarily responsiblefor Differences in the observedmorphologyof craterwallssuggestthat the mechanismand the scaleof wall failurevary with crater sizeand substrate[Cintala et al., 1977]. Many craters 15-20 km in diameterare characterized by cuspaterims and containso-called'scallop'featuresat the baseof their exposed interior walls (Figure la). Scallopdepositsare characterized by a distinctivesurfacetextureconsisting of a seriesof closely spaced, crescent-shaped ridges of low topographic relief (termed 'swirl texture' by Smith and Sanchez[1973]). The the transitionin lunar cratbrmorphome•rythat occursat of sheets of material that each maintained RIM SLUMPING MECHANISMS: MORBHOLOGICAL EVIDENCE Variationsof SlumpingMechanism With Crater Size arcuate out•:•ne ofthehead scarp associated with each scallop depositis responsible for the cuspateappearance of the crater rim (Figure la). The arcuate,ridgedappearanceof scallop depositssuggests that the initial cavityrim slumpedas a series some coherence diametersof 10-20 km [QuaMeet al., 1965;Gault et al., 1975; and movedin relationto oneanotherduringthe their descent en masseinto the cavity [Smith and Sanchez,1973;Cintala et Copyright¸ 1979by theAmericanGeophysica! Union. al., 1977]. Paper number 8B 1092. 0148-0227/79/008B- 1092501.00 3081 3082 SETTLEAND HEAD: RIM SLUMPINGAND LUNAR IMPACT CRATERS Fig. la. The crater Dawes(diameterDo = 18.4 km) possesses a cuspaterim outline and scallopslumpmassesat the baseof its interior walls (right sideof crater). The surfacemorphologyof scallopdepositsconsistsof a seriesof arcuate, tightly spacedridgesof low topographic relief. Wall terracesinitially appearin freshlunarcraters20-30 km addition, theoretical calculationsby Ullrich [1976] indicate in diameterandoccurwith increasing frequency in larger-sized that upward movementbeneaththe baseof a crater cavity may craters.All freshlunar cratersgreaterthan 70 km in diameter occurcompletelyindependentlyof rim collapsephenomenaas containdistinctiveterracedwalls[Cintalaet al., 1977].Craters the result of stress-waveinteraction. Unfortunately, there is with terracedwallsgenerallyexhibitpolygonalrim outlines very little morphologicalor theoreticalevidencethat can be [Quaideet al., 1965].Terracesare characterizedby scarps usedto determinethe subsurfaceconfigurationor inward exfacingthecenterof a craterand by relativelyfiat tops(ledges) tent of terrace block failure surfaces. arrangedin a stair-step mannerfromfloorto rim crest(Figure The transition from one style of rim slumpingto anotheris lb). The gross morphology of wall terraces indicatesthat gradational and occursover a range of crater sizes.Several individual terrace blocks possessed considerablecoherence craters contain some combination of scallopsand terraces, duringthe slumpingevent.In anyparticularsectorof a crater indicatingthat more than onetype of failure mechanismoperwall,contacts between terrace ledges andadjacent headscarps ated within a singlecrater (for example, note the structural are generallysubparallel,indicatingthat individualterrace contrast between the northeast and southwest walls of Lablocksslumpedalong a seriesof imbricatefailure surfaces. lande, 4.5øS, 8.7øW). Comparisonof the style of slumping However,the locationand morphology of the leadingedges within craters 15-50 km in diameter generallyindicatesthat (toes)of terraceblocksnearthecratercenters areobscured by with increasingcratersizethe failure surfacemigratesoutward fallbackand impactmelt deposits (Figure lb). The frontal beyondthe cavity rim crest,engulfingincreasinglylarger poredgesof terraceblocksmay have beenpartiallydestroyed tions of the initial cavity rim. owing to convergenceat the crater center and/or central Nature of TerraceFailureSurfaces rebound of the crater floor. Dence[1968]hassuggested that theslipsurfaces involvedin If individual terracesbehavedas perfectlycoherentblocks terrace slumpingextend to the center of the crater and that duringa rim-slumpingeventand if subsequent degradationof centralpeaksdevelopaspartof thecollapse process. However, the crater wall was insignificant, the observed structure of centralpeakscan be foundin cratersthat do not possess terraceswould representthe actualconfigurationof the terrace terracedwalls(for example,Diophantus, 27.6øN,34.3øW).In blocks at the conclusionof the slumpingevent. However, SETT•_EAND HEAD: RIM SLUMPINGAND LUNAR IMt'ACT CRATERS --- 3083 ;:;• ';•;,..• ...... ........ .,.. ..;,... ...:• ........."*:::!'"• ..... .. j.;:.... -}. .::•:.. %%'"*;'½":qL.*'* *:'"'*" ß ::;. ,/....-- .. ß ,.,.... .. :. .... ........:.. ;....•:• •;' :;•? ?.. .... .-;,.:?- .;•;:; ..':::•-:;*. .? .? ./ .... ,...I:. -'•1..: .... :•,-: . •:•..:::; ..... -......... ,,. , .... . .:., .... ß......... ...... ; •: ½•.... •:..: ;-:• •:•: .,..•...... ....** .•:.:..•--.:•:•::: ß ..... .::½ .... ,..... ., ... . , :;'*' ;.:;* ", ...-,..... .7•:...,, • .•:..•.......,:, ,•..j;* .... • ,>.... .. '*.,•:,. Fig. lb. In contrastto the crater Dawes (Figure la) the craterTimocharis(D0 = 34 km) is characterizedby a polygonal rim outline and terraced crater walls. Terrace ledgesare situated at progressivelyhigher elevationswith increasingradial range. Both scallop deposits and wall terraces are interpreted to be sectionsof the rim and walls of a crater's initial excavationcavity that slumpedinto the cavity during the terminal stagesof crater formation. The structure and morphology of wall terracesindicatethat terraceblocksmaintainedconsiderablecoherenceduring the rim slumpingevent. terrace blocks consist of fractured and brecciated crustal mate- rial, and it is unlikely that theseblockswould slump as perfectly coherent massesin the absenceof a lubricating agent such as groundwater [Sharpe, 1938]. Consequently, talus movementalong head scarpsduring and after slumpingmay modify the original configuration.In many fresh craters the boundary between the head scarp above a terrace and the actual ledgeforming the top of the terraceblock is sharply delineated and locally linear. Furthermore, in certain fresh craterssuchasAristarchus,terraceledgesexhibit a concentric surfacetexture similar to ejecta depositsbeyond the crater's rim crest. If extensivemass wasting had occurred, then the contacts between terrace ledgesand head scarps would be M orphometric evidencealso indicates that the structure of terracedwallswithin freshcratershasnot beenseverelyaltered by talus movement. In general, the inclination of the face scarpsof individualterraceblocksincreases with increasing rangefrom a crater'scenter.For example,Figure2 displays the variation of wall slopealong the east wall of the crater Timocharis (Do = 34 km) as a function of normalized radial range.The positionof terracescarpsis indicatedby vertical arrowsin Figure2; terraceledgesare areasof relativelylow inclination.The maximumslopeof terracefacescarpson the eastwall of Timocharisvariesfrom 24ø to 32ø with increasing range.If mass-wasting processes had extensivelymodifiedthe terracedwalls of Timocharis,it is unlikelythat the inclination irregular,and primary ejectatextureson terraceledgesshould of terracescarpswould vary in this manner,but rather terrace have been destroyed.On the basisof this morphologicalevi- scarpslopeswouldall beapproximately constantandequalto dence we conclude that modification of the terraced walls of the angle of reposeof fragmental crustal material. Similar fresh cratershas been relatively limited and that the observed variationsin terrace scarpslope are observedin other fresh configurationof terraceswithin fresh cratersis representative craters (Figure 5; R0 is crater rim crest radius). of the wall structureat the end of the main slumpingevent. If the face scarpsof individual terrace blockshave not been This conclusion is supported by the unmodified nature of severelydeformedin the latter stagesof a slumpingevent,as impact melt depositsfound on crater walls and floors [Hawke arguedabove, then the head scarpabovea particularterrace and Head, 1977]. ledge representsa vestigeof the terrace failure surface.The 3084 SETTLE ANDHEAD:RIM SLUMPING AI•D LUNARIMPACTCRATERS manner in which the shape of the slumping failure surface varied with depth. The fact that the slopeof terracescarps increasesas a functionof radial range(as shownin Figure 2) impliesthat terraceblockshave slumpedalong curvedfailure surfaceswhich are highly inclined near the original ground T I MOCHBB I S SCARP CREST _ surface andlesssteeply inclined at greater-depths. Timingof theEvent Various features on terraced walls are interpreted to be impact melt depositsemplaced during the cratering event [Howard and Wilshire, 1975;Hawke and Head, 1977], including (1) smooth dark pools of material perchedon terrace ledgesand (2) lavalike flows with and without well-drained channels and levees. Flow features and cracks associated with thesemelt depositsindicatethat the material was molten and behavedin a fluid mannerat the time of emplacement.SuperCREST position of melt depositson terracedwalls indicatesthat the Fig. 2. Variation of wall slopealongthe easterninteriorwall of main slumping event must have occurred during the latter Timocharis as a function of normalized radial range. Timocharis stagesof the crateringevent.Gault et al. [1968] usedthe term possesses a flat floorthat extendsto a rangeof 0.34craterradius.Wall modificationstageto describeshort-termand long-termcrater scarpsarerepresented by areasof highinclinationto theleftof vertical modificationprocesses.However, we use the term in a more arrowsdenotingscarpcrests.Note that wall slopedecreases near the restricted sense,as the modification stage of the cratering crater's rim crest. event, to-refer to those processesoperating in the terminal stagesof the event which modify the shape of the transient maximum slope observedalong a head scarpshould thus be crater cavity. approximatelyequivalentto the inclinationof a planar surface locally tangentto the original failure surface.If the imbricate IDEALIZED TERRACE-SLUMPING PROCESS failure surfacesseparatingindividual terrace blocks can be In order to developa methodfor restoringterracesto their consideredapproximatelyparallel in the vicinity of the initial preslump positionsit is necessary to formulatea conceptual cavity rim (seeFigure 3), then the inclinationsof head scarps process.On the basisof the situated at progressivelylower elevationsshould reflect the model of the terrace-slumping observations citedabovewe envisionthat a major sectionof the cavity rim slid down into the cavity as a collectionof discreteblocksseparatedby a seriesof failure surfacesin the IDEALIZED TERRACE SLUMPING EVENT terminalstagesof the crateringevent(Figure 3). Thesefailure surfaces werecurvedzonesof rupturethat wereapproximately INNERMOST FAILURE parallel in the vicinityof the initial cavity rim; their location SURFACE INTERSECTS and structurein the vicinityof the cavity floor are unknown. CAVITY RIM-.•_ /• GROUND SURFACE AT INITIAL CAVITY RIM WA Failure may have occurredas the result of (1) mechanical I I 0.5 BFINOE/BIM I I I I 1 0 BFIDIUS instability of thecavitywallsunder theinfluence ofgravity • I AFTER J •/ ?/?/?'/ [Quaideet al., 1965;Gaultet al., 1975;Melosh,1977],(2) latestagereorientationof the flow field within the targetmaterial, resulting in vortical downward and inward motions in the FAILURE SURFACEvicinity of the cavity rim [Maxwell andMoises, 1971;Ullrich, 1976], or (3) some combination of these two mechanisms [Ullrich et al., 1977]. Although the actual mannerin which failure occursis unknown,materialalong the failuresurfaces OBSERVED CRATER RIM was probablydeformedextensivelyin an irreversiblenonelastic fashion.We assumea plasticmode of failureduringthe slumpingprocess[seeMelosh, 1977],implyingthat the terrace blockspossessed somecombinationof cohesiveand frictional •'--OUTERMOST TERRACE BLOCK CRATER RIM SCAR• shearstrength. Analyticalsolutions to slopestabilityequations wereinitiallydevelopedby Sokolovski[ 1965]for plasticfailure /? TERRACE STRUCTURE by ignoringthe weightof the slopematerial [seeHarr, 1966; ...9.•? * Scott, 1963]. These solutionsindicate that (1) failure occurs alonga logarithmicspiralslipsurfaceif the materialpossesses frictional strengthand (2) failure occursalong a circulararc slip surfacein purely cohesivematerials.Civil engineering Fig. 3. Schematicdiagramof idealizedterrace-slumping eventde- techniques for determiningslopestability,suchasthe Swedish scribedin the text. Terrace blocksare inferredto be segments of the circular arc methodand the methodof slices,also employ rim and wallsof the initial cratercavitythat slumpedalonga seriesof circulararcsandlogarithmicspiralsto represent slumpfailure imbricate, curved failure surfacesduring the latter stagesof crater surfaces [Wu, 1976]. It seems likely that terrace blockswere formation.The subsurfaceextentand configurationof terracefailure surfacesare unknown.Terracescarpsobservedwithin freshcratersare translatedalonga failuresurfacethat couldbe approximated interpretedto be vestigialremnantsof terraceblock failure surfaces. by oneof thesegeometricshapes. • • SETTLEAND HEAD: RIM SLUMPINGAND LUNAR IMPACT CRATERS TERRACE RESTORATIONTECHNIQUE Reconstructionof the initial crater cavity is accomplished by a two-stepmethod.In the first stepthe radiusof the initial cavity is inferred by determiningthe point of intersection between(1) a polynomial equation representingthe craterward extensionof observedrim topographyand (2) a model failure surfacewhich is tangent to the face scarp of the innermostexposedterrace (Figure 4). The face scarp of the TERRACE 3085 RESTORATION MODEL A.[ INFERRED CAVITY RIM MODEL FAILURE SURFACE INCLINED • ATANGLE0 AT CAVITY RIM TRANSL/•,TION PATH : P AT .Y ALFIT • '"•/%/EXTRA OLED POE NOMI OF CAVITY RIMWA LL -------.•.•.-• ./TO EXTERIOR CRATER TOPOGRAPHY (MODELFAILURE SURFACE)• :.: --"c---...•(CORRELATION COEFFICIENT>_0.9 MODEL FAILURE SURFACE ..'"'"' • innermost terrace block is assumed to be a remnant of the rim wall of the initial cavity. The modelfailure surfaceconformsto a specificgeometricshapeandrepresents the hypotheticalpath alongwhichthe initial cavityrim wall slumped.Strictlyspeaking, the translationpath of the cavityrim wall is not a 'failure surface,'since the wall of the initial cavity is an unbounded (free) surfaceduringa slumpingevent.However,owingto the imbricate nature of the terrace failure surfaces and the as- B.I INFERRED CAVITY RIM MODEL INITIAL CAVITY PRIOR TO SLUMPING sumptionthat deformationprincipallyoccurredalong failure surfaces,this translationpath shouldhaveparalleledthe innermostfailure surface(Figure 3). If this failure surfacecan be represented by a particulargeometriccurve,suchasa logarithmic spiral or a circulararc, then the translationpath of the cavity rim wall shouldbe representedby a similarlyshaped •'"ASSUMEDCAVITY GEOMETRY [•E•[•[• VOLUME =• NO BULKING curve. The model failure surface is constrained to be inclined CAVITY VOLUME ASSUMED DEPTH at a particular angle0 at the rim of the initial cavity. In the Fig. 4. The cavity reconstructionmethod developedin this study secondstepof the reconstructionmethod a geometricshape suchas a coneor paraboloidis assumedfor the initial cavity. consistsof two steps:(a) The radiusand elevationof the initial cavity rim are determinedby restoringa crater's innermostterrace upward Thedepth ofthecavity isinferred bydetermining a specific along amodel failure surface (dotted line) and extrapolating the radial geometric surface forwhichthevolume of material forming . trend ofexterior crater topography (dashed line)inward toapoint of the rim andupperwall of theinitialcavity(verticallyhatched intersection. (b) Cavitydepthis determined by assuming a general area in Figure 4) is equalto the volumeof materialpresently geometricshapefor the cavity and accountingfor the volumeof cavity situated between the crater floor and the bottom of the initial cavity(crosshatched area in Figure 4). This modelassumes no net bulking (i.e., densitychange)of material involvedin the slumpingprocess. We now examinethe assumptions involvedin definingthe terrace restorationtechnique: 1. Caoity rim topography. A crater'spresentlyobserved rim crestand exterior depositsrepresentthe unslumpedportion of its initial cavity rim. It is assumedthat the topographic structureof the initial cavity rim can be approximatedby extrapolatingpresentlyobservedradial topographictrendsinward toward a crater'scenter.Preslumprim topographyis specifiedby a polynomialequationfit by the methodof least squaresto the presentlyobservedexteriortopography.In relatively flat regionsthis polynomialfittingprocedureis applied between1.0 and 2.5Ro.However, in areasof highly variable preexistingtopography(for example, highland terrain) the fittingprocedureis appliedbetween1.0and a minimumrange of 1.5Ro.The purposeof this polynomialequationis to scribemajor topographictrends.Thereforethe equationwas contrainedto be the lowest-order polynomialexpression that _ achieved a correlation coefficient of 0.90 or better with the observedexteriorcratertopography(a samplingof graphical resultsis presentedin Figure 10). It was found that polynomialsof third degreeand lower orderwereable to satisfy this criterionfor all cratersoccurringon mare surfacesor at mare/highlandboundaries. Highlandcraterscharacterized by exteriorswhich couldnot be represented by a third-degreeor lower-orderpolynomialexpression (with correlationc > 0.9) were not consideredfor further analysis. 2. Initial cavityrim wall. The facescarpand ledgeof the innermostterraceare assumedto correspondto the rim wall and rim crestof the initial cratercavity(Figure3). The structure and morphologicalcharacteristics of wall terraceswithin wall and rim material (vertically hatched area) which is presently situatedbeneatha crater'sfloor (crosshatchedarea). Model assumptions are discussedin detail in the text. A specificexample of this cavity reconstructiontechniqueis provided in Figure 7. fresh craters indicate that individual terrace blocks maintained considerablecoherenceduring the slumpingevent, implying that deformationprincipallyoccurredalongfailure surfaces.If the upper portion of the innermost terrace block was not extensivelydeformed during the slumpingevent, the slump path of the cavity rim wall should parallel the innermost terrace failure surface. Under these circumstances the inclina- tion of the rim wall slumppath at the cavity rim crest(angle0 in Figure 4a) is approximatelyequivalentto the inclination of the innermostterracefailure surfacenear the initial cavityrim (angle0 in Figure3). The modelfailuresurface corresponding to the translation path of the cavity rim wall is therefore constrainedto be tangentto the face scarpof the innermost terraceand to be inclinedat a specificangle0 at the cavityrim (Figure 4). Measurementsof rim scarp slopeswithin terraced craters can be used to estimate the inclination of terrace block failure surfacesnear the initial cavity rim. The maximumslopeobservedon scarpsbeneatha crater'srim crestshouldapproximate the inclination of the outermost failure surface at the groundsurface.Crater wall slopeis presentedas a functionof normalizedradial rangein Figure 5 for the 12 largeterraced cratersselectedfor analysis(Table 1). Wall slopewas measured over horizontal distancesof severalkilometersusing topographiccontour data from Lunar TopographicOrthophotomaps.Maximum valuesof wall scarpslopeoccur near the observedcrater rim at rangesof 0.8-1.0R0 and generally range from 25ø to 45ø. Localizedmass-wasting processes would tend to reducethe 3086 SETTLE AND HEAD: RIM SLUMPING AND LUNAR IMPACT CRATERS 6O resolutionphotographsthe flanksand toesof individualrock slidescanbe identifiedon smallcraterwalls[Howard,1973]. Downslope talus movementwill reducethe curvatureof the ._.5O initialcavitywallsanddecrease thedepthof theinitialcavity. DELISLE LRMBERT However, in comparisonto larger cratersin which the initial cavityhas beencompletelydestroyed,we will considersmall freshcratersto be relativelyunmodified. LRNGBENUS MROLEB PEIBCE PICRRD PLINIUS TIMOCHRRIS THEOPHILUS KING LR PEROUSE SKLOOONSKR The radialvariationof interiorcratertopography can be describedby the equation dowhere y is the elevationabove a crater's floor, do is the ob- servedrim crest-to-floor craterdepth,r is radialrangefrom a crater's center, and R0 is the observed rim crest radius of a 0.5 BRNGE/BIM crater,all in meters.A conicalcratershapewouldbe representedby a = 1 in (1), whereas a paraboliccratershapewould 10 CBEST BRDIUS be representedby a = 2. The observedvariation of interior Fig. 5. Variation in wall scarpslopeas a function of normalized radial rangefrom a crater'scenter.Wall slopewasdeterminedimmediately beneaththe crestof terracescarps.Slopemeasurements performed on several different sides of a crater have been combined in this graph, accountingfor the data scatterfor individualcraters.Note the generalincreasein wall scarpslopewith increasingradial rangewithin individual craters. Rim scarpslopeswithin thesefresh lunar craters generally vary from 25ø to 45ø. elevationasa functionof radialrangewithinfivemorphologicallyfreshcraterswith Do< 15 km is shownin Figure6 (see Table 2). The shapeof thesecratersis intermediatebetweena conical(a = 1) andparabolic(a = 2) geometry.Localizedwall failurehasprobabl'yreducedthe initial curvatureof the walls, andit is likelythat theinitialcavitiescorrespond moreclosely to a parabolicshapethan to a conicalgeometry.However, resultswill be presentedfor differentcavityshapes.We note initial slopeof a crater'srim scarp.Furthermore,failuresur- that Dence[1973]hassuggested that the initial cavitiesof imfacesnearera crater'scentermostlikelyintersected the exte- pactcraterspossess parabolicshapeson the basisof fieldstudrior groundsurfaceat slightlyhigheranglesthan the out- ies of terrestrialimpactstructures. ermostfailuresurfaceowingto the curvednatureof the failure 4. Angleoffrictionduringterraceslumping. To determine surfacesand the generalincreasein elevationtowardsthe rim the logarithmic spiralfailuresurfaceappropriate to plastic crest.Thereforeinnermost terracefailuresurfaces mayhave failurein frictionalmaterials,it is necessary to specifythe intersected thecavityrim at an anglesomewhat greaterthan angleof friction4 of materialinvolvedin theslumping process 45ø. An angleof 60ø is consideredto be a reasonableestimate [Sokolovski, 1965;Scott, 1963].Soil mechanics experiments of the inclination of the model failure surface at the rim crest conductedat the Apollo landingsitesindicatethat the lunar of the initialcavity(angle0 in Figure4). However,cavity regolithpossesses an averageangleof frictionrangingfrom reconstruction resultswill alsobe presentedfor 0 = 45ø. 38ø to 42ø [Mitchellet al., 1973,Table 8-V]. Analysisof 3. Initial cavityshape. The shapeof theinitialcavitycan bouldertracksat the Apollo 17 landingsitesuggests a wider be approximated fromthecharacteristics of small,fresh,non- rangeof 280-50ø ([Mitchellet al., 1973,Table 8-111];seealso terracedlunarcraters(1 km < Do< 15km).Thestructure of resultsof theApollo16penetrometer experiments [Mitchellet thesesmallcratersprovidesthe bestapproximation of tran- al., 1972,Table 8-VIII]). Melosh[1977]hasarguedthat the sient cavityshapepresentlyavailable,althoughsomewall strengthof lunar crustalmaterialsshoulddecreasesignififailurehasoccurred[WoodandAndersson, 1978].In high- cantlyduringa large-scale crateringevent,and thusthe effecTABLE1. Characteristics andLocations of LunarCratersAnalyzed bytheTerrace Restoration Model Average Crater Diameter, km Age Substrate Delisle 26 E mare King 75 C highland Lambert 30 E mare Langrenus 135 C mare/highland La Perouse Madler 80 27 Ic• C highland mare/highland Peirce Picard Plinius 19 24 43 E E E mare mare mare Ic• C highland mare/highland C mare DataSource 30øN, 35øW 5øN, 121øE 26øN, 21øW 9øS, 61øE LTO39B 1, B2 LTO65CI,C4, D2, D3 LTO40A3, B4 LTO80B4, C 1, D2 N39øE, E30øS, S47øW, N56øW N2 øE, S29øE, W 19øS,N 17øW 11øS,76øE 11øS,30øE LTO81D2 LTO78C2,79D1 N0øE,N54øE,E5os E23øN,W28øN border border Sklodowska Theophilus l l6 95 Cross-Section Directions Location N23øE, S23øE, WI2øN SI6øW, S41øW 18øN,53øE 15øN,55øE 15øN,24øE LTO44D4 LTO62AI, A2 N26øE, W33øS,N43ow E9øS,S20øE,W33øS,N 19ow LTO60BI, B2, 42C4 N4øE,E41øS,W35øS 18øS,96øE LTOI00AI, A2 W27øS 11øS,26øE LTO78C2 N26øE,NI3øW LTO40B2, B3 N49ø E, N90øE, W25øS, W39øN border Timocharis 34 27øN, 13øW Crater selection criteria arediscussed inthetext. Structural data forindividual craters were compiled from Lunar Topographic Orthophoto- maps (LTO's) along a variety ofcrater cross sections. Crater ages asgiven byI'Vilhelms and McCauley [1971] andI'Vilhelms and EI-Baz [1977]. SETTLE ANDHEAD:RIM SLUMPING ANDLUNARIMPACT CRATERS tive angleof frictionof crustalmaterialduringterraceslumping may be much less than valuesdeterminedby static __ measurements. It is assumed here that values of 5 ø and 50 ø brackettherangeof frictionanglescharacterizing lunarcrustal materialsduringterraceslumping.Bothangleshavebeenem- -- BOBEL (N] BOBEL (SW} CBUCH¾ •E) CAUCH¾(NW} OESEILLIGN¾ISE} DESEILLIGN¾INN;q} ployedin restoringterraceblocksalonglogarithmicspiral -7 KUIPEB •NE) failure surfaces. -- •_• 3087 KUI PEF• IN) SRF•RBHRI {WSW} SRF•RBHRI {NNWI Data Sample _ The cavityreconstruction modelwas appliedto 12 fresh craters 19-137 km in diameter. Crater morphometricdata were compiledfrom Lunar TopographicOrthophotomaps (LTO, preparedby NASA andthe DefenseMappingAgency; l: 250,000scalewith a nominal 100-mcontourinterval) using an electronic digitizing board (accuracy, +0.25 mm) and a Hewlett-Packard9830A minicomputerfor data storage.The size,age,and background terrainof craterswithinthe dataset are documentedin Table 1. Craters were selectedfor analysis • -- • -- 0.5 F•RNGE/BIM CBEST 1.0 F•RE)IUS Fig. 6. Radial variationof interiortopographywithin five fresh lunarcraterswith Do < 15 km (Table2). In orderto comparethe on the basisof (1) freshmorphological appearance (i.e., pris- interiorstructureof different-sizedcratersdirectly,theelevationabove tine ejectatexturesand we!l-defined terracedwallscharacter- a crater's floor is normalized to rimrto-fioor crater depth, and range ized by steeprim scarpsand nondissected terraceledges[see from a crater'scenteris normalized to rim crestradius in this graph. Pohnand Offield,1970;Head, 1975]),(2) craterspossessingInterior cratergeometryis concavewithin the centralportionsof these and becomesincreasinglyconvex(in crosssection)as the rim depth/diameterratiosand rim height/diameterratiosrepre- craters crestis approachedfrom a crater'sinterior,accountingfor the general sentativeof other freshcratersof similarsize[seePike, 1977a], S-shapedtrendof the combineddata. The referencelinesa = 1 anda (3) the availabilityof topographicdata (topographicmaps = 2 correspondto idealizedconicaland parabolicshapes,respectively structure ofthese craters atranges presentlyexistonly for areason or adjacentto the ground (see( 1)). Noteihatthebowl-shaped tracksof the Apollo 15, 16, and 17 missions[Kinder, 1975]). of lessthan 0.6R0correspondsmorecloselyto thea = 2 referenceline. Craters with diameters smaller than 70 km in Table 1 were formedin Copernicanor Eratosthenian times(all lessthan •,3 b.y. old). Owingto the scarcityof youngcratersgreaterthan 70 km in diameter, a few Imbrian-agedcratersinferred to be youngerthanor contemporaneous withtheOrientalecratering eventare includedin the data sample(Table 1). PositiveBouguergravity anomaliesinterpretedto result from isostaticstructuraladjustments areexclusively associated with larger(Do > 200 km) and older craters[Phillipset al., 1976].Craterswithinthe datasampleareconsistently smaller and youngerthan craterscharacterized by suchpositivegravity anomalies, implyingthat thecratersanalyzedherehavenot experienced major structui'almodifications due to long-term isostaticcompensation. RESULTS Type Case The northeastpart of the crater Timocharis(Do -• 34 km) providesan illustrationof how the initial (preslump)cavityof a large lunar crater is reconstructed by the model described above. As shown in Figure 7a the radial trend of exterior cratertopographyobservedbeyondthe northeastrim crestof Timochariscan be representedby a polynomialequation of secondorder (dashedline in Figure 7a, determinedby least squaresfit with correlationc >_0.9). The hypotheticalslump pathof the initial cavityrim scarpis represented by a circular arc in Figure 7a. This circulararc is assumedto parallel the TABLE2. FiveRelatively YoungMareCraters LessThan15kmin Diameter Selected forShape Analysis Rim Crest Crater Borel Rim-toFloor Diameter Depth Do,km do,km 4.9 1.10 Location do/Do Age 0.22 E (DataSource) 22.4øN,26.4øE C ross-Section Directions N8øw, S41øw (LTO42C2) Cauchy 12.2 2.69 0.22 C 9.6øN, 38.7øE Deseilligny 6.2 1.28 0.21 E 21.1øN, 20.6øE N25øW, S41 øE Kuiper 6.8 1.64 0.24 C 9.8øff, 22.7øE N40øE, W2øN Sarabhai 7.5 1.80 0.24 E 24.8øN,21.0øE NI4øW, W6øS (LTO61A3) (LTO42Cl) (LTO76D2) N47øw,E5øN (LTO42B4) Crater ages havebeendetermined fromthe1:1,000,000 scale U.S.Geological Survey mapseries for thelunarnearside. These craters possess circular rimoutlines, concave interior walls,smallflatfloors, anddepth/diameter ratios characteristic ofsmall fresh lunar craters [Pike,1977a]. Rim-to-floor depth hasbeenmeasured betweena crater'smaximumrim crestelevationandminimumfloorelevaton.Mor- phological andmorphometric evidence indicates thatrim-slumping modification ofthese small craters is relatively limited in comparison to largerterraced craters. Theinterior structure of these craters, measuredalongcross-sections specified in thetable,is displayed in Figure6. ß 3088 SETTLEAND HEAD: RIM SLUMPINGAND LUNAR IMPACTCRATERS TIMOCHARIS ] IA INFERREDCAVITYRIM which the volume of material initially situatedabove the terraced wall of the observedcrater (vertically hatchedregion in Figure 7b) exactly equals the volume of material presently situated above the model cavity and below the floor of the observedcrater (crosshatchedregion in Figure 7b). The rimto-floor depth observedalong the northeastcrosssectionof Timocharis is 2940 m, whereasthe rim-to-floor depth of the model parabolic cavity is 9350 m. In comparison,a depth/ diameter relationship describingthe averageshapeof small (Do < 15 km) fresh lunar craters (do = 0.196Dd'ø• [Pike, 1977a]) predicts a rim-to-floor depth of 4780 m for a crater cavity with an initial diameter of 23,640 m. Thus for the set of "•--....• " EQUATION SECOND-DEGREE POLYNOMIAL •& FIT TO EXTERIOR ..-"""'---.•'__ CRATER TOPOGRAPHY CIRCULAR ARC .-' MODEL FAILURE SUR•.." FACE SCARP OF...'"'" "'""'"""' INNERMOST TERRACE.....'•-• •- •OBSERVED CRATER i VERTICAL I modelparameters employed in Figure7 thecavityreconstruc- I HORIZONTAL- tion methodproducesan estimateof Timochariscavitydepth that is significantlygreater than the depth predictedby extra- "•J'•' /•,/E polating small-crater depth/diameter ratios to crater cavities INFERRED CAVITY RIM INITIAL CAVITY • greater than 15 km in diameter. PRIOR TO SLUMPING IFFFF .......• ,.,,.,,• iiiii !!!!! i iiiii:,"•-- •,,,, ,,,, ,,,, ,,,, ,, ,,,, ,,,,;, "11• ' - Resultsfor Entire Data Sample Inspectionof the interior structureof largecratersreveals OBSERVED CRATER TOPOGRAPHY that the positionof ,terraceledges,the numberof terraces PARABOLICCAVITYGEOMETRY CAVITY 131:: DEPTH ESTIMATE ON TERRACE RESTORATION VERTICAL INITIAl BASED _ ,OR,ZONTALMODEL exposed, and the inclinationof terracescarpsmay be quite differentin differentsectorsof a singlecrater.In addition,the radialvariationof exteriortopography mayvarysignificantly in different directions [SettleandHead,1977].Average cavity dimensions canbedetermined byapplying theterracerestoratibntechnique to several cross-sectional profiles at individual Fig. 7. Type case example of the cavity reconstructionmethod craters.In somecases,suchas Theophilusand Sklodowska, schematicallyillustratedin Figure 4 as applied to a northeastcross- topographic data are only availablefor certaincratersectors; sectionalprofile of the crater Timocharis. (a) The position of the in othercases,suchas King, preexisting topographyis excavity rim prior to slumpingis determinedas the point of intersection tremelyvariable,andno low-order,monotonically decreasing betweena model failure surfacewhich is tangentto the facescarpof express. ion canbe fit to the topography lyingbethe innermostt•e•rrace (dottedline corresponding to a circulararcwith polynomial 0 = 60ø) and a second-degree polynomialequationwhichdescribes yondcertainsection•of the craterrim. Theseconsiderations radial topographictrendsbeyon•dthe presentlyobservedrim crestof restricted thenumber of cross-sectional profiles thatcouldbe Timocharis(dashedline). For this setof modelparametersthe terrace examined at certain craters. restoration method indicatesthat rim slumpingenlargedthe TimEstimatesof initial cavity rim diameterDt and cavitydepth ochariscratercavityby 40%. (b) Cavity depthis determinedasthe rim- to-floordepthof a paraboliccavityfor whichthevolumeof material below the original ground surface, d,t, for the entire data initially formir•g the cavity rim and walls (vertically hatched area) sample(Table 1) are presentedin Figures8 and 9 for a circular equals thevolumeof materialpresently located abovethebaseof the arc and logarithmicspiral failure surface,respectively.Both cavity and below the observedcrater floor (crosshatched area). hypotheticalfailure surfaceswere constrainedto be inclined at an angle of 60ø at the cavity rim (angle 0 in Figure 4); a failure surface along which the innermost terrace block parabolic initial cavity was employedin determiningcavity slumpedinto the initial cratercavity.This circulararc is con- depthin both Figures8 and 9. An angleof friction4 of 50ø strained to be tangent to the face of the innermost terrace wasusedin determining thelogarithmic spiralfailuresurfaces. blockand to be inclinedat a predetermined angle(0 = 60ø) at Examples of reconstructed (preslump) cavities determined by the rim of the initial cavity. The model circular arc failure restoringterracesalong circulararc failure surfacesare illustratedin Figure 10for specificcrosssections withinthe craters surface(dotted line, Figure 7a) is extrapolatedupward,and the polynomialequationrepresenting exteriorcratertopogra- Delisle(Do= 26 km),King(Do= 75 km),andLangrenus (Do phy is extrapolatedinward to a point of intersectionwhichis = 137 km). taken to be the radius of the preslumpcavity. Along the In both Figures8 and 9 the orderof the polynomialexnortheast Timocharis cross section the observed crater i'im pressionused to representthe radial variation of exterior crest(5840-melevation)occursat a rangeof 16,770m (l.0Ro) from the centerof the crater and the ledgeof the innermost terraceis situatedat a rangeof • 10,000-m(0.60Ro)and 4100-m elevation.This terraceis restoredby the methodillustratedin Figure7a to an initial positionat a rangeof I 1,820m (0.70Ro) topographyin differentdirectionsis designated by different symbols: circlessignifyfirst-order polynomials, diamonds signify second-order polynomials,and trianglessignifythirdorder polynomials.Higher-orderpolynomialsindicatethat rim elevation increases at a greaterratewithdecreasing radial and an elevation of 6915 m. range. Therefore if the interior structure of each crater was Once the initial radius of Timocharisis specified,cavity reasonablyuniform, then, in general,greater volumesof depthprior to slumpingcan be determinedby accountingfor slumped rim materialwouldbeestimated bythehigher-order equations, andlargerestimates of initialcavity thevolume q•wallandrimmaterial thatslumped intothe polynomial cavity. Cavit3•'depth iscalculate• through aniteration process depthwould be preferentiallyassociated with higher-order by assuming a generalized geometricshapefor the initial cav- polynomial expressions. However,thedatain Figures8 and9 ity suchas a parabola(a = 2 in (1)). Cavity depthprior to indicate thatthereisnoconsistent relationship between cavity slumpingis then equivalentto the deptl•of a paraboloidfor depthestimates andtheorderof thepolynomial representing SETTLEAND HEAD: RIM SLUMPINGAND LUNARIMPACTCRATERS CIRCULAR ARC FAILURE - SURFACE // - e=6o ø . 3089 / _ - // PARABOLIC CAVITY (õ=2.) doi // SKLODO t // <{ i,9) 8 ' TIMOCH/•RIS -f •7C:: g / • / / /I LANGRENUS ,o ,KING ELISL •4 • p,c•.•// 2 IO 20 30 CAVITY 40 Rl• 60 CREST DIA•ETER 80 IOO 120 (km) Fig. 8. Cavity dimensionsprior to slumpingd•t•rmin•d by th• cavity r•constructiont•chniqu• •mploying a circular arc mod•l •ailur• suffac•with 0 = 60ø and a paraboliccavityshape.For th• purposeso• data presentation th• abcissascal• is'discontinuousat a cavity diameter o• 40 kin. Solid symbolsd•not• th• order o• th• polynomial •xpr•ssion fit by th• m•thodot l•astsquares to •xt•riorcratertopography alongspecific cratercrosss•ctions (circles, first-order polynomial; diamonds,s•cond-ord•rpolynomial;triangles,third-orderpolynomial).Th• data scatter•or individualcratersprincipally r•sults from azimuthal variationsin interior wall structur• and •xt•rior crater topography(s• Figur• 10 for •xampl•s). Solid r•r•nc• lin•s r•pr•s•nt th• av•rag• observeddimensionso• •r•sh lunar craters[Pike, 1977a,b]. Th• dashedr•r•nc• lin• is an •xtrapolation ot th• d•pth/diam•t•r r•lationshipobservedfor smallcraters(Do < 15 kin). Not• that th• av•rag• d•pths o• r•construct•dcavitieswith D• < 40 km ar• situatedabov• th• dashedlin• r•pr•s•nting small-cratermorphore,try, whereasth• av•rag• d•pths ot r•construct•dcavitieswith D• • 70 km ar• l•ss than cavity d•pths predictedby th• smallcrater depth/diameter relationship. theexteriortopographic surface. Thisis dueto theextensiveLangrenus(Di = 100km) cavitiesare 70-60% of cavitydepth variability of interior and exterior topography at individual craters(see Figure l0 for specificexamples). A comparisonof Figures8 and 9 indicatesthat estimatesof cavity dimensionsprior to slumping determinedfor the two failure surfaces are quite similar. The apparent depth/rim diameter (d,/Do) relationship observed for small (Do < 15 km), fresh, nonterraced lunar craters is representedby a dashedline in both figures[Pike, 1977b]. For theseparticular setsof model parameters,estimatesof averagecavity depth for individual cratersare uniformly greater than the depth of a comparable-sizedcrater cavity predicted by small-crater morphometry(dashedline, Figures8 and 9) up to a cavity rim diameter of approximately 40 km. For example, in Figure 8 the ratio of averagecavity depth basedupon the terracerestoration model to cavity depth predicted by small-crater morphometryrangesfrom 1.9 for Picard (Dr - 16 km) to 1.6 for Timocharis(D• = 25 km); in Figure 9 the ratio variesfrom 1.8 for Picard (D• = 16.5 km) to 1.4 for Timocharis (D• - 25 km). Reconstructedcavitiesgreater than 70 km in diameter, however,are characterizedby initial depthswhich are significantly lessthan crater cavity depthspredictedby small-crater estimatesbased upon small-cratermorphometry.Similar results are obtained by employinga logarithmicspiral failure surface (Figure 9). Malin and Dzurisin [1978] have estimatedpreslump cavity depth by employinga simplifiedreconstructiontechniquein which (1) a crater's presently observedfloor diameter is assumed to correspondto the initial cavity diameter and (2) slumpedwall material situatedbeneatha crater's floor is assumedto be containedwithin a disc-shapedregion. Although this reconstruction model does not account for the structure of terrace failure surfacesor the concaveshapeof impact excavation cavities,estimatesof cavity depth reportedby Malin and Dzurisin [1978] vary with cavity diameter.in a manner generally analogousto the resultspresentedin Figures8 and 9. Parameter Tests The sensitivityof the terrace restorationmodel to certain criticalassumptions canbeexamined by varyingthevaluesof key parametersand comparingresultingestimatesof initial cavitydiameteranddepthwithcavitydimensions displayed in morphometricrelationships.For example,in Figure 8 the Figures 8 and9. Parameter testing results are•presented in averagedepthsof the restoredTheophiluS(D• - 78 km) and Figures 11, 12, and 13. 3090 SETTLEAND HEAD: RIM SLUMPINGAND LUNAR IMPACTCRATERS I I d.i _ LOGARITHMIC SPIRAL FAILURE SURFACE e=6o" - ½ = 5o" PARABOLIC CAVITY (õ=2_.) I I d•i _ la.i - /? , TIMOCHARIS ;6•AARD •/•'""•AMBERT -,/ _ Di I0 i I 20 30 I I 40 I 60 CAVITY RIM CREST DIAMETER i 80 I 120 (kin) Fig. 9. Cavitydimensions priorto slumpingdetermined by thecavityreconstruction methodemployinga logarithmic spiralmodelœailurc surœacc with0 = 60ø anda paraboliccavityshape(symbolsandr½œcrcncc linesarcdescribed in Figure 8). Logarithmicspiralœailur½ surœac½s havebccnspecifiedutilizinga $0ø angleoœfriction whichis representative oœthe frictionalstrength oœncar-surœacc lunarrcgolithmaterial.Notethatcavitydimensions determined by terracerestoration alonglogarithmicspiralœailurc surœaccs (0 = 60ø, ½ = 50ø) arc quitesimilarto numericalresultspresentedin Figure8 for terracerestorationalongcirculararc œailurc surœaccs (½= 60ø;paraboliccavityshapeassumedin both cases). Initial cavitydimensionsfor terracerestorationalonglogarithmic spiral failure surfacesusing a 5ø angle of friction are presentedin Figure I I. Such low friction anglesmay more accuratelycharacterizethe dynamicstrengthof geologicalma- terialsduring a large-scaleimpactcrateringeve,nt[Melosh, 1977]. Other model parameters(0 = 60ø, parabolic cavity geometry)havebeenleft unchanged.Figure I I showsthat this set of model assumptionsproducesreconstructedcavitieswith depth/diameterratios that are comparableto ratios observed for small (Do < 15 km) fresh craterswithin the 15- to 30-km range of cavity diameter. Cavities larger than 30 km fall below the dashed reference line. The difference between recon- structed cavity depth and cavity depth predicted by smallcrater morphometryincreaseswith increasingcavity size.The averagedepth of the restoredTheophilus(D• = 83 km) and Langrenus(Dr = 105 km) cavitiesis •,50% of the cavity depth predicted by the extrapolateddepth/diameterrelationshipfor small nonterraced craters. Another critical parameterin the terracerestorationmodel is the assumed inclination of the model failure surface at the rim of the initial cavity (angle0 in Figure 4a). In Figure 12, reconstructedcavity dimensionsare presentedfor terrace restoration along circular arc failure surfaceswhich intersectthe initial cavity rim at an angle of 45ø (all other parameters remain as describedin Figure 8). The value of 45ø is an averageestimateof maximum rim scarpslopepresentlyobservedwithin large lunar craters and probably representsa minimum estimate of the inclination of the model failure sur- face at the initial cavity rim. The averagedepthsof restored cavitiesare 1.4-1.2 timesgreaterthan cavitydepthspredicted by small-cratermorphometryover a rangeof cavitydiameters from 17 to 26 km in Figure 12. Restoredcavitiesgreaterthan 40 km in diameteragain plot belowthe referenceline. Average cavity depthsfor Theophilus(Dr = 81 km) and Langrenus(D• = 101 km) are approximately60% of cavitydepthspredicted by the small-cratermorphometricrelationships for equivalent cavity diameters.Similar resultswere obtained by restoring terracesalonglogarithmicspiralsurfaceswith 0 =45 ø for • = 50ø and a paraboliccavity geometry(data not shown). Therefore decreasing0 to 45ø has the effect of reducingrestored cavity depths(compare Figures 8 and 12). In order to evaluatethe degreeto whichmodel resultswere influenced by the assumedshape of the initial (preslump) cavity,terraceswere restoredalongcirculararc failure surfaces with 0 = 60ø (as in Figure 8 results),and initial cavity depth was determinedusinga cavity geometryintermediatebetween a cone and a paraboloid. This cavity shapeis explicitly describedby a radial variation of interior cavityelevationproportional to r•'" (equivalent to a = 1.5 in (1)). This cavity geometrycloselycorrespondsto the observedshapeof small freshcratersat rangesof 0.0-0.6Ro(seeFigure6). As shownin Figure 13, this intermediatecavity geometryproducescavity depth estimatesthat are 2.3-1.8 times greater than cavity depthspredictedby small-cratermorphometryover a rangeof cavity diametersfrom 15 to 30 km. As observedin the previous cases,the ratio of restoredcavity depthsto cavity depth esti- SETTLE AND HEAD: RIM SLUMPING AND LUNAR IMPACT CRATERS INITIAL CRATER CAVITIES RECONSTRUCTED BY TERRACE 3091 RESTORATION /VIVE WNW /-.-• DELISLE / x / Do=26 km / VERTICAL I HORIZONTAL = •- ....._•"/ ssw $W KING VERTICAL= ,5 HORIZONTAL Do--75 km :.'.3 wsw SE LANGRENUS Do= 135 km \ • VERTICAL 5 HORIZONTAL 2 CIRCULARARC FAILURE SURFACE(9--60") PARABOLIC CAVITY GEOMETRY Fig. 10. Comparisonof preslumpcavitystructureinferredby the cavityreconstruction method(dashedcross-section lines)with observedcraterstructure(solidcross-section lines)for threedifferent-sized craters.Terraceshavebeenrestored alongcirculararc failuresurfaces (0 = 60ø), anda paraboliccavityshapehasbeenassumed (samemodelparameters as employedin Figure 8). Discrepancies in cavity radius and depth determinationsfor differentcrater crosssections principallyresultfrom azimuthalvariationsin wall structureand exteriorcratertopography. mates based upon small-crater morphometry decreaseswith increasingcavity diameter. Restoredcavity depthsfor Theophilus(D• = 78 km) and Langrenus(D• = 100km) rangefrom 80% to 75% of the cavity depths predicted by small-crater morphometricrelationshipsfor equivalentcavity diameters. In summary, greater estimatesof initial cavity depth are producedby (l) larger valuesfor the angleof friction of lunar crustalmaterialduringa slumpingevent(•b),(2) largervalues for the inclination of the model failure surface at the initial cavity rim (0), and (3) a more conicalgeometryfor the initial excavationcavity. However, model estimatesof initial cavity dimensionsdo not vary in a linear fashionwith changesin theseparameters.The greatestvariation in model resultswas produced by an order of magnitude reduction in the assumed value of the angle of friction during a slumpingevent (compare Figures I l and 9); the smallestvariation in model results was producedby a 25% decreasein the assumedinclination of the model failure surface at the initial cavity rim (compare Figures 12 and 8). DISCUSSION AND CONCLUSIONS The depth of excavationof an impact crateringevent is the maximum depth below the original ground surfaceat which the target material is forcibly dissociatedand laterally displacedduring the excavationstageof crater formation. A zone of plasticdeformationextendsbeyondthe maximumdepthof excavation in which material is permanently displacedbut individual particleshave maintained their relative positions. Rock samplesrepresentingthe maximumdepthexcavatedwill most likely fail to be ejectedbeyond the crater rim and will probablybe mixedinto the brecciadepositfillingtheexcavation cavity. The maximum depth of material actually ejected from the initial cavity (i.e., ejectasamplingdepth;seeHead et al. [1975]) may be significantlylessthan the maximum depth of excavationof the crateringevent owing to the excavationof large volumesof material which are not transportedbeyond the cavity rim. In small-scalelaboratory impacts into sand targets a significantfraction of crater depth is producedby 3092 SETTLE AND HEAD: RIM SLUMPING AND LUNAR IMPACT CRATERS LOGARITHMIC SPIRAL FAILURE SURFACE 0=60; / /// _ // 5' /// PARABOLIC CAVITY (õ=2.) - // I / i doi / SKLODOWSKA • • / z :D LANGRENUS • • //// •LLA N*[•"•J •THEOPHILUS o _z 6 o en 4 /// _ PEROUSE MADLER F ; o PI L •,.c•1•.••.....P""us J_ m Di I I0 20 :50 CAVITY 40 RIM CREST 60 DIAMETER 80 1 I00 i 120 (km) Fig. 11. Cavitydimensions prior to slumpingdetermined by thecavityreconstruction technique employing a logarithmicspiralmodelfailuresurface(0 = 60ø) anda paraboliccavityshape(symbolsandreference linesaredescribed in Figure 8). Model parameters are equivalentto thoseutilizedin Figure9 exceptthat a frictionangleof 5ø hasbeenemployedin specifying logarithmicspiralfailuresurfaces.Suchlow frictionanglesmaybe morerepresentative of the actualfrictional strengthof lunar crustalmaterialsduringan impactcrateringevent.Comparisonof cavitydimensions determined by restoringterracesalonglogarithmicspiralfailure surfaces(this graphand Figure9) indicatesthat smallervaluesof the frictionangle4•producelargerestimates of preslumpcavitydiameterand shallowerestimates of initial cavitydepth. compressionof substratematerial [StUffieret al., 1975]such that the depthof excavationof the crater,as definedabove,is actually lessthan the observedcrater depth. However, since the density and strengthof lunar crustal materials should increasewith crustaldepth [T6ksozet al., 1974; Talwaniet al., 1974;Toddet al., 1973],compression will probablyaccountfor only a small portion of the total volume of the excavation cavitiesformed by large-scalelunar impacts [Head et al., 1975].At this scalethe depth of excavationof the cratering eventshouldbeapproximatelyequalto the depthof the initial crater cavity. If rim slumpingis the primarymechanism of cavitymodification, initial cavity depthsinferredby restoringslump terracesto their originalpositionsand reconstructing the excavation cavity prior to slumpingshouldserveas approximate estimatesof the depth of excavationof individualcratering events. Initial cavity dimensionsdeterminedby the terrace restorationtechniquedeveloped in thisstudyaremodeldependent, in that differentcombinationsof assumedparameters producevaryingestimates of initialcavitydiameteranddepth. tions of model parameters have several common features (summarizedin Figure 14). In all cases,reconstructed cavity depth is approximatelyequal to or somewhatgreaterthan cavitydepthpredictedby small-cratermorphometryoverthe 15- to 30-km rangeof cavitydiameters.Over the 30- to 70-km range of cavity diametersthe differencebetweenreconstructed cavity depth and cavity depth basedupon the extrapolated depth/diameterrelationshipfor small lunar cratersdecreases to a point at which the inferred depthsof restoredcavities fall belowthereference linerepresenting small-crater morphometry(Figures8, 9, l l, 12, and 13). Finally, the depthsof reconstructedcavities greater than 70 km in diameter are consistently lessthan cavity depthspredictedby small-crater morphometricrelations(Figure 14). Althoughsmall(Do < 15km), fresh-appearing lunarcraters do not appearto have experienced wholesalerim collapse, localizedwall failure processes have undoubtedlymodified theirinitial morphometryto somedegree.Thereforethe initial depth of excavation of small nonterracedcraters should be somewhatgreaterthan the presentlyobserveddepthof these The exact structure of terrace failure surfaces and the exact craters.If the depth/diameterratio of impactexcavationcavshapeof large-scaleexcavationcavitiesprior to slumpingare itiesis constantovera widerangeof cavitysizes(for example, not known, and thereforeit is presentlyimpossibleto specifya 1 km < D• 1000 km) and if rim slumpingis the principal setof boundaryconditionsthat will uniquelyconstrainthe ter- mechanismof cavity modificationin craterswith D• > 15 km, racerestorationmodel.Nevertheless,cavityrestorationresults thenit is reasonable to expectthat reconstructed cavitydepths presentedin Figures 8, 9, 11, 12, and 13 for variouscombina- shouldbe situatedabovethe referenceline representing the SETTLEAND HEAD: RIM SLUMPINGAND LUNAR IMPACTCRATERS 3093 doi -15 CIRCULAR FAILURE ARC SURFACE e=45' PARABOLIC CAVITY (õ=2.) i i doi • io • 9 - /// G TIMOCHARIS ._•/LA PEROUSE•I rn 4 •• -5 INIU _/• // PICARO _L io i I 20 •0 I 40 CAVITY • I I 60 RIM CREST DIAMETER I 80 Di I I I00 • I 12o- (kin) Fig. 12. Cavity dimensionsprior to slumpingdeterminedby the cavity reconstructiontechniqueemployinga circular arc failure surface(0 = 45ø) and a paraboliccavity shape(symbolsand referencelinesare describedin Figure 8). Model parametersare equivalentto thoseutilizedin Figure 8 exceptthat the inclinationof the model failure surfaceat the cavity rim (0) wasconstrainedto be 60ø in Figure 8. Rim scarpspresentlyobservedwithin freshlunar cratersare typicallyinclined at anglesof 45ø or less(seeFigure 5). Comparisonof cavitydimensionsdeterminedby restoringterracesalongcirculararc failuresurfaces(thisgraphand Figure8) indicatesthat smallervaluesof the angle0 producelargerestimatesof initial cavity diameterand shallowerestimatesof preslumpcavity depth. observeddepth/diameterrelationsof smallcraters[seeH6rz et al., 1976]. This in fact is the case over a range of cavity diametersfrom 15 km to at least 30 km for a wide variety of model parameters.On the basis of these model results we concludethat (l) initial cavitiesexcavatedby impactcratering events in the I km < D• < 30 km size range are morphometrically similar and (2) the observedshallowmorphometry of terraced craters with rim crest diameters Do of lessthan -,•40 km can be accountedfor by rim-slumping modification of their initial excavation cavities. For impact crater cavitiesgreater than 70 km in diameter (equivalentto Do = 80-90 km), however,the averagedepth of restored initial cavities is significantlylessthan cavity depth estimatesextrapolated from small-cratermorphometry. This discrepancymay in part be due to (1) an inability to recognize innermost terrace blocks which were destroyed as they slumped,(2) partial settlingof exteriortopographyaroundthe crater rim as it reachedits presentconfiguration,or (3) assumedvaluesof model parametersas discussed in the previous section.There is little morphologicalevidencewithin the domical and hummocky terrain forming the flat floors of large craters which suggeststhat additional terrace blocks were initially situated at ranges less than the presentlyexposed innermost block. Furthermore, there is no consistentrelation- ship betweencrater size and the order of the polynomial expressionrepresentingexterior crater topography which might indicate that the rim topographyof larger craters is partially 'deflated' [Settle and Head, 1977]. Rather, the discrepancy between reconstructedcavity depth and cavity depthspredictedby small-cratermorphometryobservedfor Dt > 70 km is interpretedto be the result of a changein the depth/diameterratio of initial excavationcavitiesformed by suchlarge impacteventsand/or a changein the relativeimportanceof rim slumpingin modifyinginitial cavitiesgreater than 70 km in diameter. In either case we conclude that rim slumping cannot solely account for the difference between depth/diameterratioscharacterizing small,relativelyunmodifled lunar cratersand the observeddepth/diameterratios of craterswith rim crestdiametersof greater than 80-90 km. Two other factors may contribute to the modification of impactexcavationcavities:(l) fallbackof craterejectaand (2) reboun__d of basementmaterial. At terrestrialimpact craters suchas Brent and Meteor Crater, Arizona, lensesof highly shockedbrecciacontainingclastsfrom all target formations overliethe allocthonousbrecciadepositfillingthe cratercavity and have been interpreted as fallback [Dence, 1968; Shoe- 3094 SETTLE AND HEAD: RIM SLUMPING AND LUNAR IMPACT CRATERS - do' CIRCULAR FAILURE // ARC -15 SURFACE e=6o* SKLODOWSKA o/ // NON PARABOLIC CAVITY (&l.5) KIN6 // LANGRENUS•• // iTHEOPHILUS • io / _TIMOCHAR•..) / _ LA PEROUSE z oc T z • 5 _••DELISLE MADLfR Si////// • - PICARD / / •PEIRCE /// t - _ _ i i0 20 i 50 40 CAVITY RIM CREST I 60 DIAMETER i 80 i00 120 (km) Fig. 13. Cavitydimensions priortoslumping determined bythecavityreconstruction technique employing a circular arcfailure surface (0 = 60ø) anda nonparabolic cavity shape (symbols andreference lines aredescribed inFigure 8).Model parameters are equivalent to thoseutilizedin Figure8 exceptthatcavityshapein thisca,'eis assumed to beintermediate between a coneanda paraboloid (corresponding to a = 1.5in (l)). Comparison of reconstructed cavitydimensions presented inthisgraphandFigure 8 indicates thatmoreconical cavity geometries produce larger estimates ofinitialcavity depth,aswouldbeexpected. Underthesemodelconditions theaverage depthofreconstructed cavities oflessthan40kmin diameter isgenerally twiceaslargeascavitydepths predicted bysmall-crater depth/diameter ratios(dashed line). maker,1963].Roddyet al. [1975]haveestimated thatapproxiIn thepast,depth/diameter ratioscharacterizing smallfresh mately 4% of the total massejectedfrom Meteor Crater, lunarcratershavebeenemployed to estimatethe depthof Arizona, was redepositedwithin the crater as fallback. Grieve excavation of largelunarcratersandbasins[e.g.,Denceet al., et al. [1977]havepresented evidence suggesting thatthe per- 1974;Moore et al., 1974;H6rz et al., 1976].This methodof centageof moltentargetmaterialremainingwithinterrestrial estimating impactexcavation depthassumes that the depth/ impactcratersincreases withcratersize,producing a coherent diameterratiosof excavation cavitiesproduced by lunarimmelt sheeton the floorsof largercraters.Unfortunately,ero- pactsareessentially constantfor all craterslargerthanseveral sionhasdestroyed majorsegments of mostterrestrialimpact kilometers in diameter. It further assumesthat the observed structuressothat it is difficultto estimateaccuratelythe total discontinuityin craterdepth/diameterratiosat Do -• 15 km is amountof fallbackand melt initially depositedwithin terres- theresultof a significant increase in theabilityof modification trial craters[Dence,1971,Table 1]. The actualvariationof the processes to reshapethe initial cavitiesof largercraters.This percentageof ejecta which is redepositedwithin a crater as a studyhasdemonstrated that rim slumping cannotsolelyacfunctionof cratersizeis unknown.Similarly,althoughmor- countfor a transitionfroman excavation cavitypossessing a phologicalevidenceindicatesthat the heightandareaof cen- characteristicsmall crater depth/diameterratio to the obtral peakswithinlunar cratersincreasewith increasing crater servedmorphometryof lunar impactcratersgreaterthan 80 size [Wood, 1973;Allen, 1975],the volumeof substratemate- km in diameter.Additionalinvestigation of the volumetric rial whichis actuallytranslatedabovethe floor and wallsof an ,significance of basementreboundand ejectafallbackin reimpactexcavationcavityby the reboundprocess is unknown. shapingthe initial cratercavityis requiredin order to deterWe arepresentlyanalyzingthesetwocratermodification proc- minetheactualvariationof initialcavitydepthwithincreasing essesin order to estimatequantitativelythe extentto which cavitysize.However,in the absence of quantitativestudiesof they havealteredthe structureof initial cavitiesformedby the effects of these factors we note that the difference between large impact events. initialcavitydepthspredicted by small-crater depth/diameter SETTLE ANDHEAD:RIMSLUMPING ANDLUNAR IMPACT CRATERS 3095 MODIFICATIONOF INITIAL CAVITY SHAPE PRODUCED BY RIM SLUMPING aO ß• EXTRAPOLATED DEPTH/DIAMETER RELATIONSHIP ß OBSERVEDFOR SMALL (D•< 15kin) LUNAR CRATERS n,, :3 03 ß 15 . 1::3 •Z SKLODOWSKA • ."m- .-r-LANGRENUS E DEPTH/DIAMETER •3 RELATIONSHIP :ii!iii!i!ii!i!i::!i!i!•i?:•i• ....... :::::::::::::::::::::: ......................... OBSERVED DEPTH/DIAMETER RELATIONSHIP FORLARGELUNARCRATERS 50 IOO RIM DIAMETER 150 (krn) Fig.14. Summary ofterrace restoration model results. Databars represent therange ofaverage values ofcavity depth anddiameter obtained forindividual craters employing thevarious model parameter combinations presented inFigures 8, 9, 11,12,and13.Results forSklodowska arcbased uponanalysis of a single cratercross section andshould bctreated circumspcctly. Thegeneral variation ofprcslump cavity depth withincreasing cavity diameter isrepresented bythedashed linc,which isanapproximate best fittoreconstructed cavity dimensions determined forawiderange ofmodel conditions. Dataforrestored cavities ofless than50kmindiameter arctoodensely concentrated tobcpresented individually. Scctext for interpretationof model results. ratiosand by the restorationof slumpterracesincreases with increasingcrater size (Figure 14). If rebound and fallback effectsdo not showcomplementary changes,this would indicatethat the shapeof impactexcavationcavitiesvarieswith tionmodelto an Imbrium-sized cavity.In comparison, cavity excavationdepthsestimatedby extrapolatingsmall-crater depth/diameter relations to cavities 620-970 km in diameter are more than a factor of 2 greater,of the order of 80-120 km increasing cavitysizeandthatthedepthof excavation of large (e.g., Moore et al. [1974];seealso Denceet al. [1974]).The cratersand basinscannot be inferredfrom depth/diameter resultsof this studyimply that basin-forming impactevents relationshipscharacterizingsmall freshlunar craters. did not excavatelunar mantlematerialat depthsgreater Rim-slumpingmechanismswithin basin-sizedcratercavities than60 km. However,it is necessary to notethat the cavity maybequitedifferentfromtheterrace-slumping process modifyingcraters20-300 km in diameter[Head, 1977].Furthermore,craterswith diametersof greaterthan200 km typically exhibitpeakringstructures insteadof centrallylocatedpeak mountains,implying that reboundmechanisms operating withinverylargecratercavitiesmaydifferfundamentally from the basementdilationprocessaffectingintermediate-sized craters.Thereforevariationsin cavitydimensions inferredby the terracerestorationtechniquepresented in this studycanonly be speculatively extrapolatedto basin-sizedcavities. An approximatelinear fit to the averagedimensionsof reconstructed cavitieswith D• > 50 km (dash-question mark reconstruction modelsappropriateto verylargelunarcraters may differ significantlyfrom the terrace restorationmethod presented here,owingto scale-dependent variations in cavity modificationprocesses. Acknowledgments. The authorswishto thankMike Dence,Don Gault, Jay Melosh,Dick Pike,Keith Howard,andClark Chapman for thoughtful and constructivereviewsof an earlier versionof this paper.Thepatienceanddiligence of NancyChristy,LaurieRaymond, and Judy Botelhoin preparingthe manuscriptfor publicationis sincerely appreciated. A portionof thisstudywasconducted whilethe seniorauthorwasat theAir ForceGeophysics Laboratory.We gratefully acknowledge the supportof the laboratoryandthe supportof NASA under grant NGR-40-002-116. linein Figure14)wouldimplythat basinimpacteventson the scaleof Orientale(inferredDt • 620 km) and Imbrium(in- REFERENCES ferredD• • 970 km) excavatedof the orderof 35-50 km below Allen,C. C., Centralpeaksin lunarcraters,Moon,12,463-474, 1975. the lunarsurface,respectively. In a previousstudy,Headet al. Cintala,M. J., C. A. Wood,andJ. W. Head,The effectsof target [1975] evaluated various estimatesof the interior volume of characteristics on freshcratermorphology: Preliminaryresultsfor the moon and Mercury, Proc. Lunar Sci. Conf 8th, 3409-3425, theImbriumbasinandthevolumeof itsexteriorejectadepos- 1977. its in order to determinethe amountof primaryejectaex- Dcncc,M. R., Shockzoningat Canadiancraters:Pctrographyand cavatedby the Imbrium event. On the basisof thesevolume estimates,Head et al. [1975] concludedthat the maximum depthof excavationof Imbriumejectais of the orderof 27-40 km, in generalagreement with a cavitydepthestimateof 50 km inferredby extrapolatingthe resultsof the terracerestora- structuralimplications, in ShockMetamorœhism of NaturalMaterials, editedby B. M. FrenchandN.M. Short,pp. 169-184,Mono, Baltimore, Md., 1968. Dence,M. R., Impact melts,J. Geophys.Res., 76, 5552-5565, 1971. Dence,M. R., Dimensionalanalysisof impactstructures, Meteoritics, 8, 343-344, 1973. 3096 SETTLEAND HEAD: RIM SLUMPINGAND LUNAR IMPACT CRATERS Dence, M. R., R. A. F. Grieve, and A. G. Plant, The lmbrium basin and itsejecta(abstract),in Lunar.•Science V, pp. 165-167,The Lunar Science Institute,Houston,Tex:•'1974. Gault, D. E., W. L. Quaide, and V. R. Oberbeck, Impact cratering mechanicsand structures,in ShockMetamorphismof Natural Materials, edited by B. M. French and N.M. Short, pp. 87-99, Mono, Baltimore, Md., 1968. Gault, D. E., J. E. Guest,J. B. Murray, D. Dzurisin, and M. C. Malin, Somecomparisonsof impact craterson Mercury and the moon, J. Geophys.Res., 80, 2444-2460, 1975. Grieve, R. A. F., M. R. Dence, and P. B. Robertson, Cratering processes: As interpretedfrom the occurrenceof impact melts, in Impact andExplosionCratering,editedby D. J. Roddy, R. O. Pepin, and R. B. Merrill, pp. 791-814, Pergamon,New York, 1977. Guest, J. E., and J. B. Murray, Nature and origin of Tsiolkovsky crater, lunar farside, Planet. SpaceSci., 17, 121-141, 1969. Harr, M. E., Foundations of TheoreticalSoil Mechanics,McGraw-Hill, New York, 1966. Hawke, B. R., and J. W. Head, Impact melt on lunar crater rims, in ImpactandExplosionCratering,editedby D. J. Roddy, R. O. Pepin, and R. B. Merrill, pp. 815-841, Pergamon,New York, 1977. Head, J. W., Processesof lunar crater degradation:Changesin style with geologictime, Moon, 12, 299-329, 1975. Head, J. W., Origin of outer ringsin lunar multi-ringedbasins:Evidencefrom morphologyand ring spacing,in Impact and Explosion Cratering, edited by D. J. Roddy, R. O. Pepin, and R. B. Merrill, pp. 563-573, Pergamon,New York, 1977. Head, J. W., M. Settle,and R. Stein, Volume of material ejectedfrom major lunar basinsand implicationsfor the depthof excavationof lunar samples,Proc. Lunar Sci. Conf. 6th, 2805-2829, 1975. HiSrz, F., R. V. Gibbons, R. E. Hill, and D. E. Gault, Large scale crateringof the lunar highlands:SomeMonte Carlo modelconsiderations,Proc. Lunar Sci. Conf. 7th, 2931-2945, 1976. Howard, K. A., Avalanchemode of motion: Implicationsfrom lunar examples,Science,180, 1052-1055, 1973. Howard, K. A., Fresh lunar impact craters:Reviewof variationswith size, Proc. Lunar Sci. Conf. 5th, 61-69, 1974. Howard, K. A., and H. G. Wilshire, Flows of impact melt at lunar craters,J. Res. U.S. Geol. Surv., 3(2), 237-251, 1975. Kinsler, D.C., UserGuideto 1: 250,000ScaleLunar Maps, The Lunar Science Institute, Houston, Tex., 1975. Mackin, J. H., Origin of lunar maria, Geol. Soc.Amer. Bull., 80, 735748, 1969. M alin, M. C., and D. Dzurisin, Modification of fresh crater landforms: Evidencefrom the moon and Mercury, J. Geophys.Res., 83, 233-243, 1978. Maxwell, D. E., and H. Moises,Hypervelocityimpactcrateringcalculations, NASA Contract. Rep., CR-115350, 213 pp., 1971. Melosh, H. J., Crater modificationby gravity:A mechanicalanalysis of slumping, in Impact and ExplosionCratering, edited by D. J. Roddy, R. O. Pepin, and R. B. Merrill, pp. 1245-1260, Pergamon, New York, 1977. Mitchell, J. K., W. D. Carrier, W. N. Houston, R. F. Scott., L. G. Bromwell, H. T. Durgunoglu, H. J. Hovland, D. D. Treadwell, and N. C. Costes,Soil mechanics,sect.8, Apollo 16 Preliminary Science Report, NASA Spec.Publ., SP-315, 1972. Mitchell, J. K., W. D. Carrier, N. C. Costes, W. N. Houston, R. F. Scott,and H. J. Hovland, Soil mechanics,sect.8, Apollo 17Preliminary ScienceReport, NASA Spec.Publ., SP-330, 1973. Moore, H. J., C. A. Hodges, and D. H. Scott, Multi-ringed basins-Illustrated by Orientale and associatedfeatures,Proc. Lunar Sci. Conf. 5th, 71-100, 1974. Phillips, R. J., E. A. Abbot, J. E. Conel, T. V. Johnson,and R. S. Saunders,Isostaticadjustmentin lunar history(abstract),in Lunar Science VII, pp. 688-690, The Lunar ScienceInstitute, Houston, Tex., 1976. Pike, R. J., Sizedependence in the shapeof freshimpactcraterson the Moon, in Impact and ExplosionCratering,editedby D. J. Roddy, R. O. Pepin, and R. B. Merrill, pp. 489-509, Pergamon,New York, 1977a. Pike, R. J., Apparentdepth/apparent diameterrelationfor lunar craters, Proc. Lunar Sci. Conf. 8th, 3427-3436, 1977b. Pohn, H. A., and T. W. Oftield, Lunar crater morphologyand relative age determinationof geologicalunits, 1, Classification,U.S. Geol. Surv. Prof. Pap., 700-C, C 153-C 162, 1970. Quaide,W. L., D. E. Gault, and R. A. Schmidt,Gravitativeeffectson lunar impactstructures, Ann. N.Y. State,,icad.Sci., 123, 563-572, 1965. Roddy, D. J., J. M. Boyce,G. W. Colton, and A. L. Dial, Meteor Crater, Arizona, rim drilling with thickness,structuraluplift, diameter,depth,volume,andmassbalancecalculationsl Proc.LunarSci. Conf. 6th, 2621-2644, 1975. Scott, R. F., Principlesof Soil Mechanics,Addison-Wesley,Reading, Mass., 1963. Settle,M., and J. W. Head, The role of rim slumpingin the modification of lunar crater morphometry(abstract), in Lunar ScienceVII, pp. 794-796, The Lunar ScienceInstitute,Houston,Tex., 1976. Settle, M., and J. W. Head, Radial variation of lunar crater rim topography,Icarus,31, 123-135, 1977. Settle, M., and J. W. Head, Rim slumpingmodificationof lunar craters:Implicationsfor the depth of excavationof basin-sized impacts(abstract),in Lunar ScienceIX, pp. 1053-1055,The Lunar and Planetary Institute, Houston, Tex., 1978. Sharpe, C. F.S.,Landslides andRelated Phenomena: ,4Study ofik•ass Movementsof Soil and Rock, Columbia University Press, New York, 1938. Shoemaker,E. M., lhterpretationof lunar craters,in Physicsand Astronomyof the Moon, edited by Z. Kopal, pp. 283-359, Academic, New York, 1962. Shoemaker,E. M., Impact mechanicsat Meteor Crater, Arizona, in The Solar System,vol. 4, The Moon, Meteoritesand Comets,edited by B. M. Middlehurstand G. P. Kuiper, pp. 301-336, Universityof Chicago Press,Chicago, Ill., 1963. Smith, E. I., and A. G..Sanchez, Fresh lunar craters:Morphology as a function of diameter, a possiblecriterion for crater origin, Mod. Geol., 4, 51-59, 1973. Sokolovski, V. V., Staticsof QranularMedia, translatedby J. K. Lusher, Pergamon,New York, 1965. SttSffier,D., D. E. Gault, J. Wedekind, and G. Polkowski,Experimental hypervelocityimpact into quartz sand:Distributionand shock metamorphismof ejecta,J. Geophys.Res., 80, 4062-4077, 1975. Talwani, P., A. Nur., and R. L. Kovach,Implicationsof elasticwave velocitiesfor Apollo 17 rock powders,Proc. Lunar Sci. Conf. 5th, 2919-2926, 1974. Todd, T., D. A. Richter, G. Simmons,H. Wang, Unique characterization of lunar samplesby physicalproperties,Proc. Lunar Sci. Conf. 4th, 2639-2662, 1973. TiSksoz,M. N., A.M. Dainty, S.C. Solomon, and K. R. Anderson, Structure of the moon, Rev. Geophys.Space Phys., 12, 539-567, 1974. Ullrich, G. W., The mechanicsof central peak formation in shock wave crateringevents,Tech.Rep. 75-88, Air ForceWeaponsLab., Kirtland Air Force Base, N. Mex., 1976. Ullrich, G. W., D. J. Roddy, and G. Simmons,Numericalsimulations of a 20 ton TNT detonation on the earth's surfaceand implications concerningthe mechanicsof centraluplift formation,in Impactand ExplosionCratering,editedby D. J. Roddy, R. O. Pepin,and R. B. Merrill, pp. 959-982, Pergamon,New York, 1977. Wilhelms, D. E., and F. EI-Baz, Geologicmap of the eastsideof the moon, scale l: 5,000,000, Map 1-948, U.S. Geol. Surv., Reston, Va., 1977. Wilhelms, D. E., and J. F. McCauley, Geologicmap of the nearside of the moon, scalel' 5,000,000, Map 1-703, U.S. Geol. Surv., Reston, Va., 1971. Wood, C. A., Moon: Central peak height and crater origins.Icarus, 20, 503-506, 1973. Wood, C. A., and L. Andersson,New morphometricdata for fresh lunar craters,Proc. Lunar Planet. Sci. Conf. 9th, 3669-3689, 1978. Wu, T. H., Soil Mechanics, 2rid ed., Allyn and Bacon, New York, 1976. (Received M arch 5, 1977; revised June 21, 1978; acceptedOctober 12, 1978.)