JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 97, NO. E7, PAGES 11,623-11,662, JULY 25, 1992
AtmosphericEffectsOn EjectaEmplacement
PETER H SCHULTZ
Departmentof GeologicalSciences,Brown University,Providence,RhodeIsland
Laboratory experimentsallow the investigationof complex interactionsbetween impacts
and an atmosphere.Although small in scale,they can provide essentialfirst-orderconstraintson
the processesaffecting late-stageballistic ejecta and styles of ejecta emplacementaroundmuch
larger craters on planetary surfaces.The laboratory experimentsinvolved impacting different
fine-grainedparticulatetargetsunder varying atmosphericpressureand density (different gas
compositions).During crater formation, ballistic ejecta form the classiccone-shapedprofile
observedunder vacuum conditions. As atmosphericdensity increases(for a given pressure),
however, the ejecta curtain bulges at the base and pinchesabove. This systematicchangein the
ejecta curtain reflects the combinedeffects of decelerationof ejecta smaller than a critical size
and entrainmentof theseejecta within atmosphericvorticescreatedas the outward moving wall
of ejecta displacesthe atmosphere.Additionally, a systematic change in emplacementstyle
occursas a functionof atmosphericpressure(largely independentof density):contiguousejecta
rampartsuperposing
ballisticallyemplaceddeposits(0.06 to 0.3 bar); ejectaflow lobes(0.3 •to
0.7 bar); and radial patterns (>0.8 bar). Underlying processescontrolling such systematic
changesin emplacementstyle were revealedby observingthe evolutionof the ejecta curtain,by
changing target materials (including layered targets and low-density particulates),by varying
atmosphericdensity,by changingimpact angle, and by comparingthe ejecta run-out distances
with first-order models of turbidity flows. Three distinct ejecta emplacementprocessescan be
characterized.Ejecta rampartsresult from coatsetclastssorted and driven outwardby vortical
winds behind the outward moving ejecta curtain. This style of "wind-modified" emplacement
representsminimal ejecta entrainmentand is enhancedby a bimodal size distributionin the
ejecta. Such "eddy-supported
flows" are observedto increasein run-out distance(scaledto crater
size) with increasingatmosphericpressure.By analogywith turbidity flows, this scaleddistance
should
increase
asR1/2for a givenatmospheric
pressure
anddegree
of entrainment.
Ejectaflows
with •nuch greater run-out distancesdevelop as the turbulent power in atmosphericresponse
windsincrease.Suchflows overrunandscourthe innerejectafacies,therebyproducingdistinct
inner and outer facies. The degreeof ejecta entrainmentdependson the dimensionlessratio of
drag to gravity forces acting on individual ejecta and the intensity of the winds createdby the
outwardmoving curtain.Entrainmentincreaseswith increasingatmosphericdensityand ejection
velocity (crater size) but decreaseswith ejecla density and size. The intensity of curtaingeneratedwinds increaseswith ejection velocity (crater size). The dimensionlessdrag ratio
characterizingthe laboratoryexperimentscan be applied to Mars since the reducedatmospheric
density is offset by the increasedejection velocities for kilometer-scale events. For a given
crater size (ejection velocity) and atmosphericconditions,a wide range of nonballistic ejecta
emplacementstyles could occur simply by varying ejecta sizes even without the presenceof
water. Alternatively, the onset crater diameter for nonballistic emplacementstyles can reflect
the range of ejecta sizespossiblefrom the diversemartian geologic history (massivebasaltsto
fine-grained aeolian deposits). Scaling considerations further predict that ejecta run-out
distances
scaledto cratersizeon Mars shouldincrease
as R1/2;hencelongrun-outflows
dependenton crater diameterneed not reflect depth to a buried reservoirof water. On Venus,
however, the dense atmospheremaximizes entrainment and results in ejecta flow densities
approachinga constantfraction of the atmosphericdensity. Under such conditions,ejecta runout distancesshoulddecreaseas R'1/2.
1. iNTRODUCTION
impacts on planets without atmospheres[Gault et al., 1968;
Laboratory experiments have been used to establish
fundamental processesand phenomenology associatedwith
Copyright 1992 by the AmericanGeophysicalUnion.
Oberbeck,1975;Gault and Wedekind,1977, 1978].Not every
detail observedin the laboratory,however,can be directly
applied,that is, laboratoryexperimentsrarely provide direct
simulationsof planetary-scale(10 kin) collisions. Yet, such
experiments
providean essentialperspective
for the complex
Paper number 92JE00613.
processesoperatingover a wide range of scalesin time and
0148-0227/92/92JE-00613505.00
length in three dimensions.This advantagenot only allows
11,623
11,624
SCHULTZ:ATMOSPIIERICEFFECTSOF FJECrA ,F3itn_••
testing first-order interpretationsand computationalcodesbut
also revealscontrollingvariablesthat can be then incorporated
into testable physical models of selected processes or
phenomena. With this philosophy, a wide range of
experimentsallow the exploration of the possiblerole of an
atmospherein modifying the emplacementof impact crater
ejecta.
Earlier studies of ejecta dynamics focused largely on
explosion craters [Sherwood, 1967; Wisotski, 1977]. Large
backpressures
createdby chemicalexplosionsdrive early-time
excavation,therebycomplicatingdirect applicationsto impact
cratering[Herr, 1971]. Impact craters,however,are produced
by the mechanical transfer of energy and momentumto the
target. Melt, vapor, comminution,and crater excavationare all
in responseto shocks created by the collision. This is the
basis for using particulate targets to investigatelate-stage
crater excavation:passageof intenseshockwaves transformsa
solid target into a particulate assemblage[see Gault et al.,
1968]. Introductionof an atmosphereshouldnot significantly
modify this fundamentalaspectof crater excavation,except
under extremeconditionsleading to projectiledisruptionprior
to impact [Melosh, 1981; Schultz and Gault, 1985] or
interactionswith the trailing wake [Schultz, 1990a].
Atmosphericinteractionsseparateinto early- or late-time
processes. Early-time interactions reflect the energy
partitionedto the atmosphereprior to and just after contact
during the compressionstage of crater growth. Late-time
interactionsconcern the final stagesof crater formation (2/3
complete) and ejecta emplacement.Early-time processesmay
affect late-time interactionsif shocks during first contact
temporarily heat and drive away the atmosphere. Such
interactions are often modeled as an intense cylindrically
expandingshock associatedwith the impactor [lvanov et al.,
1986] or a point source explosion [Jones and Kodis, 1982;
Vickery, 1986], both of which contain an assumedfraction of
the original impactor energy. Both approaches,however,
exaggeratethe early-time atmosphericresponseby ignoring
the role of the early-time crater cavity in containing and
redirectingwake-heatedatmosphericgasesor impact-generated
vapor into an upwarddirectedjet, rather than a hemispherical
explosion[Schultzand Gault, 1979, 1982, 1990]. Experiments
exploring the more complex early-time couplingbetweenthe
collision and atmosphereat high impact angles(>45') reveal
that only a small fraction of energy availableactuallydirectly
coupleswith the ambient atmospherearound the crater, even
for easily volatized targets[Schultz, 1988]. Consequently,this
paper emphasizeslate-time interactionswith an atmosphere,
which is assumed to resemble ambient conditions; a future
paper will explore this assumptionin more detail.
Late-time interactionscomprisemost of crater excavation
and ejection:one half of the ejectedmassleavesthe cavity after
the crater has achieved
85%
of its final
excavation
diameter.
Schultz and Gault [1979] exploredthe responseof ejecta to an
atmosphere by incorporating a first-order model of crater
growth with aerodynamic deceleration of individual ejecta.
Such an approachpredictedthe unrealisticconclusionthat most
ejecta below a critical size for a given crater diameter (and
atmosphere) never leave the crater. Consequently, they
concludedthat ballistic shadowingduring ejectionmust lessen
the full effect of aerodynamic drag. Hence calculationsand
predictions based only on trajectories of isolated particles
[e.g., O'Keefe and Ahrens, 1982] will not reproduceobserved
phenomena;nevertheless,they provide a useful upper limit for
aerodynamicdrag effects.Additionally, a distinctionwas made
between ballistic ejection and nonballistic emplacement.
Nonballistic emplacement represents a two-stage process
involving aerodynamicdecelerarionto near-terminal velocity
and then entrainmentin atmosphericturbulencecreatedby the
outward moving wall of ballistic ejecta. Consequently,
conditions leading to nonballistic styles of emplacement
dependon a critical ejectasize that dependson cratersize (i.e.,
ejection velocity), ejecta size, and atmosphericpressure(i.e.,
density).Schultzand Gault [1979] comparedthis critical ejecta
size for cratersin the laboratoryand on Mars, Earth,andVenus.
A subsequentstudy [Schultz and Gault, 1982] explored
atmosphericeffects on ejecta dynamics in the context of a
major terrestrial impact. Laboratory experimentsprovided a
basis for assessing
phenomenathat would stressthe terrestrial
environment, possibly leading to massive extinction of life
species. While Alvarez et al. [1980] emphasizedlofting and
long-term suspensionof finer ejecta fractions, Schultz and
Gault [1982] proposed that such ejecta would be largely
entrainedin near-rim ejecta flow and deposition.More critical
to global environmental stress would be the effects of
hypervelocity ejecta (particularly from oblique impacts) that
escapesignificantdecelerationat early times due to the finite
scale height of the atmosphere and ballistic shadowing;
nevertheless,enormousquantities of thermal energy will be
depositedin the upperatmospherefar from the impactduring
teentry.
Although such laboratory experiments were performed to
investigate complexities of ejecta dynamics, they also
produced distinctive ejecta depositswith striking similarities
to Martian craters [Schultz and Gault, 1982, 1985; Schultz,
1986, 1989]. The presence of an atmosphere, then, was
proposed to be an essential ingredient for diversity of
emplacementstyles on Mars [Schultz, 1989], in additionto the
presence or absenceof water [Carr et al., 1977; Gault and
Greeley, 1978; Greeley et al., 1980; Mouginis-Mark, 1979].
With this perspective,fluidizedejectalobesdo not necessarily
indicate the presenceof water but only that the flow properties
resembleda fluid. Becausesystematicchangesin eraplacement
style with atmosphericpressureparalleledchangesin cratering
efficiency and crater shape, energy lost to crater excavation
appeared to be expressed in the complexities of ejecta
dynamics and ejecta emplacement.Such energy lossescould
reflect a variety of early- or late-time processes. Hence
companion papers first emphasized atmospheric effects on
crater scaling [Schultz, 1990a, 1992] and crater shape[Schultz,
1990b;Schultz,
Atmospheric Effects on Crater Shape,
submittedto Icarus, 1991] before addressingstyles of ejecta
eraplacementin greaterdetail.
This paper first reviews the effects of impactor,target,and
atmosphericenvironmenton ejecta facies morphologyin the
laboratory. The effect of these variables on ejecta curtain
evolution next allows correlationwith the dynamicsof ejectaatmosphere interactions. Ejecta dynamics and ejecta
morphologyare then consideredin termsof processes
of ejecta
emplacement, thereby forming a basis for physical models
(analogousto turbidityflows) developedin the final discussion
section.Analysisof the laboratoryresultscan then be applied
to ejecta emplacementstyles on Mars where the range in
lithologiesand atmosphericconditionsallows a uniquetest for
the derived models.
2. EreCT^ MORPHOIX•Y
Impact environment. Ejecta morphology systematically
changeswith increasingatmosphericpressure.For compacted
pumice targets, a contiguous ridge or "rampart" begins to
appearabouta craterradiusfrom the rim for pressures
aslow as
0.06 bar and becomesvery well deftnedat pressures
of 0.25 bar
(Figure la) The rampartbeginsto break into individualejecta
lobes above 0.4 bar, and these lobes can extend over 6 crater
radii from the craterrims (Figure lb). In severalexperiments,a
scoured,faint rampart can be identified about a crater radius
from the rim even as other characteristicemplacementstyles
dominate.A narrow moatlike depressiontypically encirclesthe
SCHULTZ.'ATMOSPHERICEFFECTSOF EIECTAEMI'I.ACEMENT
Fig. 1. The effectof atmospheric
pressure(density)on ejcctacmplaccment
for
hypervclocity
impactsinto compacted
pumice.(a) At 0.25 bar, a distinctive
ejecta
rampart(arrow)encircles
the craters.(b) At 0.5 bar,lobesof ejecta(arrow)extend
over four radii from the crater rim. A narrow scouredmoat often developa
adjacentto the raisedtim. (c) Near 1 bar, a radial patterncharacterizes
the inner
11,625
ejectadeposits
with distalejectathinningwithoutidentifiable
boundaries.
Targetsin
Figureslb and lc had a thin veneerof dry temperasprinkledon the surface.All
impactswere0.635cm aluminumspheres
impacting
from5.0 to 5.4 km/sunderan
argonatmo6phere.
Scalebarscorrespond
to 10 cm.
raisedcraterrim for pressures
above0.4 bar. Impactsundera 1 ejecta lobes. At higher pressures,the surface was visibly
bar atmosphereproduce an extensive,radially scouredejecta scoured.Nevertheless,containersplaced at different distances
depositwith a morepronounced
rim depression
(Figurelc). In from the crater collectedonly minisculeamountsof ejecta
all cases, the amount of ejecta fallback and fallout is minor
fallout [Schultzand Gault, 1982].
since remnantsof the projectile and sinteredtarget material
The use of different gas compositions (helium, air,
remain clearly visible on the floor. Subtlelineationsin the nitrogen,argon,and carbondioxide)allowedthe atmospheric
preimpact target remain clearly preservedat pressuresbelow density to be varied while a constantpressureis maintained
about0.6 bar up to the continuous
faciesandeventhroughthe (seeTable 1). Figure2a summarizesthe dependence
between
11,626
SCHULTZ:ATMOSPItERIC
EFFECTSOFEJECTAEMPLACEMENT
ejecta morphology and atmosphericpressure.The onset and formation appears to be transferred to turbulent ejecta
emplacement.Table 2 provides a listing of empirical data
style of ejecta emplacement did not appear to vary
significantly with density. Impacts into a helium atmosphere shownin Figure 2.
A seriesof experimentsexplored the effect of a stratified
at 0.06 bar, which correspondsto an atmosphericdensity
0-01po(Pois thedensity
of airat 1 arm,1.293x 10'3 gJcm3), atmosphere.It was not possible to create a gradient in
produceda faint contiguousrampartthat wasnot enhancedby a atmospheric
pressure,but it was possibleto producea gradient
CO 2 atmosphereat the same atmosphericpressure(0.12 Po). in density. Sublimation of dry-ice blocks surrounding•the
The much denser CO2 atmospherealso did not affect the targetcreateda high-densityvapor layer abovethe surfacethat
transition to more complex ejecta morphologies at higher was containedby a 5 cm barrier enclosingthe taxget.The
pressures.Nevertheless,ejecta rampartstended to persist to presenceof the higher-densitylayer of CO2 in air (factorof 1.5)
enhanced the production and size of the contiguousejecta
higher atmosphericpressuresof helium.
Separate contributionssummarize atmosphericeffects on rampart.This trend was less evident for high-velocityimpacts
cratetingefficiency[Schultz,1990a] andcratershape[Schultz, producinglarge cratersrelative to the CO2 gas layer thickness.
Shadowgraphs
madeduringimpactdemonstrated
that the CO2
1990b]. Figure 2b correlatessuch atmosphericeffects with
layer was neither removed nor appreciablydisturbedby an
ejecta emplacement style. As ejecta emplacement style
becomesless ballistic and more turbulent,crateringefficiency impact-inducedair shock at late times [Schultz and Gault,
relative to vacuum conditions is reduced. In addition, the crater
1982]; consequently,the importanceof the density gradient
diameter-to-depthratio is reducedas the atmosphererestricts appears to decrease as the scale of the crater appreciably
late-stage lateral crater growth. Hence energy lost in crater exceedsthe scaleof the CO2 layer.
In summary,ejecta morphologybecomesincreasinglymore
complexwith increasingatmosphericpressurebut is relatively
TABLE 1. AtmosphericProperties
independentof atmosphericdensity for a given pressure.A
contiguous rampart characterizes impacts into compacted
ConstituentDensity* Viscosity*Sound
Speed
Ratio
of
pumicefrom 0.06 to 0.25 bar. The developmentof the rampart
(P/Po)
(g/Po)
(C/Co)
Specific
Heats
is enhanced by the presence of a density gradient in the
Air
1.00
1.00
1.000
1.4
atmosphereprovided that the thicknessof the higher-density
Hdium
0.138
1.06
2.915
1.66
layer is not significantly less than a crater radius. Extending
Nitrogen
0.969
0.97
1.009
1.4
Argon
1.38
1.21
0.964
1.66
these empirical and phenomenological observations to
Carbon dioxide
1.52
0.80
0.782
1.2
different planetaryenvironmentsrequiresfurther understanding
*Density
andviscosity
values
aregiven
withrespect
toair:Po= 1.29x 10'3g/crn'3; of the underlyingprocessesrevealedby varying impactorand
go = 183micro-poises;
co = 331m/s.
target propertiesas well.
+
0.635
helium
Radial
cm aluminum
'
into compacted pumice
ß air (nitrogen)
•
i
0.635
cm aluminum
into compacted pumice
O argon
1.0
[] carbon dioxide
ß
+ helium
0.8
Flows
-_
ß air (or nitrogen)
0.6
O argon
[] carbon dioxide
Rampart
Ballistic
1.5
-2.2
km/s
[
1.5
-2.2
km/s
]
+[]
Radial
Flows
(©)
-I•
0.02
z
1.0
• 0.6
+
•
ß
i
a
0.1
O:2+
Rampart
Ballistic
(©)
I 3-6km/s
]
LU
i i
i
i
i
0.04 0.06
i i i
0.1
i
0.2
i
i
•
i
•
,
3-6km/s
1
0.4
i •
0.2
0.4 0.6 0.81.0
©
+
((9)
ATMOSPHERICPRESSURE(bars)
Fig. 2a. Effect of atmospheric
pressureandcomposition
on ejectamorphologyfar
0.635 an aluminum spheresimpacting compactedpumice at low and high
velocities.Atmosphericdensityat a given pressuredoesnot appearto be as
important.Transitionalor exampleswith both facies are positionedbetween
categories.Ballistic facies representgraduallydecreasingejectathicknesswith
distancefrom rim, characteristicof vacuumconditions.Rampartejecta facies
indicatesthe formationof a contiguous
ridgeon top of ejecta.Long run-outflow
lobes(flow style)andradialscouting(radialstyle)occurunderhigheratmospheric
pressures.
Radialfaciesdo not appearto developundera heliumatmosphere.
Use
of a thin layerof dry eggtempera(parentheses)
appearsto enhanceflow andradial
ejectamorphology.
0.1
b
,
Ballistic
i
i
I
Rampart
Flow
Radial
EJECTA MORPHOLOGY
Fig. 2b. Relationbetweenejcctamorphologyandcratex4mg
efficiencyreferenced
to
the expectedvaluefor the sameimpactorundervacuumconditions.
Increasing
atmospheric
pressuredramaticallydecreases
crateringefficiencyin particulate
targetswith small(or low density)grainsizes[seeSchultz,1992].This reductionin
craterexcavation
is expressed
in increasingly
nonballistic
ejectamoxphologies
from
rampartto radialstyles(Figure1). Empiricaldataaregivenin Table2.
SCHULZ: ATMOSPHERICEFFE•S OF EJEC'rAEMPI,ACEMENT
11,627
TABLE 2. EmpiricalData for Figure2
Impactor
Velocity
Atmosphere
Pressure
Type
Crater
Ejecta
Morphology
(M/m)/(M/m)v
B/(Rmp)
B/(Rmp)
Rmp
Rmp
Rmp
Rmp
Rmp
0.343
0.283
0.230
0.228
0.20
0.195
0.158
N/A
Round
821121
821209
821123
821204
821128
821126
821205
Vo
2.33
2.22
2.05
2.21
2.15
2.13
2.21
vi
2.33
2.20
2.02
2.21
2.14
2.11
2.20
821125
2.04
1.94
air
F
821207
2.13
1.98 0.50
CO2
Rmp
0.151
821206
1.98
1.96
0.84
helium
F
0.133
0.0632
0.063
0.125
0.125
0.25
0.25
0.50
0.50
air
CO2
air
helium
helium
air
helium
821124
1.99
1.81
1.0
air
RA
0.144
820509
1.97
1.79
1.0
air
F
0.160
810511
6.09
6.09
0.0006
air
B
1.0
890504
871205
2.00
5.04
2.00
5.04
0.020
0.033
helium
air
B
B
0.356
0.558
890509
1.52
1.51 0.0474
CO2
B
0.325
890506
1.9
1.9
helium
B
0.326
890508
1.77
1.75 0.078
argon
B
0.260
890505
1.70
1.69
0.088
air
B
0.294
871212
810509
810531
820516
820521
810214
810505
871211
871210
871221
810212*
810532
810211*
871209
890933
900904
890932
871213
820518
810506
5.24
6.09
5.44
6.00
5.54
6.50
5.34
5.06
5.13
5.52
6.5
6.09
6.8
4.95
5.40
4.67
5.41
5.47
5.86
6.09
5.24
5.99
5.36
5.80
5.34
6.28
5.16
5.04
5.09
5.25
6.07
5.73
6.15
4.89
4.86
0.131
0.118
0.160
0.250
0.250
0.25
0.25
0.25
0.48
0.50
0.50
0.64
0.75
0.76
0.789
0.888
0.92
0.92
0.94
0.95
helium
argon
nitrogen
argon
CO2
argon
argon
helium
helium
nitrogen
argon
nitrogen
argon
helium
argon
nitrogen
nitrogen
nitrogen
helium
argon
B/(Rmp)
B/(Rmp)
Rmp
Rmp
Rmp/F
Rmp/F
Rmp
Rmp
Rmp
Rmp
F
F/RA
F
Rmp
F/Rmp
F/RA
F/RA
Rmp/RA
Rmp/F
F/RA
0.335
0.468
0.247
0.255
0.220
0.287
0.327
0.302
0.327
0.286
N/A
0.245
0.194
0.251
0.182
0.197
0.240
0.278
0.169
0.174
810213*
6.6
5.84 0.95
ar•on
RA
0.149
4.95
4.97
5.78
5.37
0.078
All experimentsgiven in Table 2 used 0.635 crn aluminumspheresimpactingcompactedpumice;launch
(Vo) and impact(vi) velocitiesare given in unitsof kilometersper second. Atmosphericpressureis given in
bars. Ejecta morphologiescorrespondto ballistic (B), rampart(Rmp), flow (F), and radial (RA). Cratering
efficiency is given by displacedtarget mass (M) divided by projectilemass(m) and is scaledto expected
valuesundervacuumconditionsbasedon derivedscaling[seeSchultz,1992]. Asterisksindicateimpactsinto
targetswith a thin veneerof dry tempera.
Impactor variables. The style of ejectaemplacementfor a 30' from the horizontalat 0.25 and 0.9 bar. At 60ø, the ejecta
given environment and target does not change significantly morphologyis visible at both atmosphericpressures.At 0.25
with impactor size, velocity, or density. Figure 3 shows the bar (Figure 4a), the rampartis slightly offset downrangewith
formation of the distinctiveejecta rampart at impact velocities considerable
asymmetryin the radialclumps.At 0.9 bar (Figure
of 0.2, 2.0, and 6.0 km/s. The other characteristicpatterns 4b), however, the rampart borders a butterfly pattern of
shown in Figure 1 also develop at higher atmospheric continuousejecta, a patternmore characteristicof much lower
pressures,but the ejecta rampart tends to persist to higher angle impacts(i.e., 5•) undervacuumconditions[seeGault and
atmospheric pressures for lower-velocity impacts (same Wedekind, 1978]. The crater in Figure 4b also displays a
projectilesize). Suchresultsfurtherindicatethat the distinctive downrange flow-lobe. Projectile fragments commonly are
ejecta patterns are not the result of a projectile-induced air strewndownrangeon top of the emplacedejecta.At 30ø impact
shock but must reflect a responseof ballistic ejecta moving angles (Figure 4c), ejecta depositsform a complex deposit
through an atmosphere.
fanning downrange at both atmosphericpressures.At the
Nonvertical impacts result in a similar pressure-dependent higher atmosphericpressure(Figure 4d), the ejecta is more
change in ejecta morphology for compactedpumice targets. restricted along the downrange axis. As was illustrated and
Figure 4 contrastssuch changesfor imvact angles of 60ø and discussedby Schultz and Gault [1982], ejecta from oblique
11,628
SCHULZ:
ATMOSPHERIC
EFFE•SOF•A
EMPL•CEMENr
Fig. 3. The effectof impactvelocityon theformation
of theejectarampartfrom aluminumspheres
wereusedin Figures3b and3c. Hencethe formationof the
(a) 0.2 km/sto (b) 2 km/sto (c) 6 km/sundera common
atmospheric
pressure
of ejectarampartis not controlled
by shockwavescreatedin theatmosphere.
Scale
0.25 bar.The projectilein Figure3a is a 2.5 cm nylonsphere,whereas
0.635 cm barscorrespond
to 10 cm.
impacts are drawn by strong atmospheric flow trailing
ricocheted projectile material. It should be noted that the
cratersformed at 30 ø exhibit an oblong shape,perpendicularto
the trajectory. Again such a shape is more characteristicof
lower angleimpacts(~100) undervacuumconditions.
Target variables. Significant effects of the environment
on ejectaemplacementoccuronly for targetswith grain sizes
sufficiently small to be aerodynamically decelerated at
laboratoryscales[see Schultzand Gault, 1979]. Table 3 lists
the different characteristicsof various targets used in this
study.Pumiceprovidedthe smallestmodalparticlesize (80 gm)
but had to be compactedin order to minimize porosity.Without
compaction air pockets and cohesion resulted in ejecta
clumpingwhich resultedin decreasedaerodynamicdrag effects.
For comparison, sand targets with different characteristic
particle sizes were used. Particle size ranges were selected
through use of U.S. standard sieves with mesh openings
denotedby number. Sand passingthrougha U.S. no. 24 sieve
are hereafterlabeled 24 sand, and sandpassingthrougha no.
140 sieve but blocked by a no. 200 sieve are labeled 140-200
SCHULTZ:ATMOSPHERIC
E•S
a
OFEIECrA EMPLACE•
11,629
b
Fig. 4. Effectof impactangleon ejectaemplaeement
under0.25 bar (left) and0.9 patternsarepartlydestroyed
by the effectsof residualdownrange
airflowcreated
bar(right).Figures4a and4b contrast
theeffectsat 60øwiththeeffectsat 30øfrom by the obliquelyimpingingwakeandricochet.Nevertheless,
an ejectarampart
thehorizontal
shownin Figures4c and4d. At highimpactangles,thecontiguousexistsat 0.25 bar. Both low and high pressures
resultin oblongcraterswith
rampartis slightlyoffsetdownrange
at 0.25 bar (Figure4a). Undera 0.9 bar maximum
dimensions
perpendicular
to thetrajectory.
Thisresponse
toimpactangle
atmosphere,
a butterflypatternemerges
with maximumaxesperpendicular
to the is more characteristic
of lower angleimpacts(10ø) undervacuumconditions.
impactor
trajectory
(Figure4b). Sucha pattern
is typicalof muchlowerangle(5ø) Impactvelocities
at thetargetwereapproximately
4.7 km/s(Figure4a) 5.5 km/s
impactsunder vacuumconditions.A rampart-bordered
lobe also extends (Figure4b), 5.6km/s(Figure4c), and5.5 km/s(Figure4d). Ambientatmosphere
downrange.
At 30' impactangles(Figures4c and4d), the characteristic
ejeeta wasargon.Scalebarscorrespond
to 10 cm.
TABLE 3. Target Properties
In•rnal
Type
Grain
Size,
[tm
Pumice
(compacted)
81 (25)
Density, Porosity,
•
Angle
of
•cm3
Friction
•ercent
,
1.52
39
-80'
angles of internal friction. The irregular particle shapes of
pumice resulted in friction angles sufficient to preserve the
postimpact crater profile, whereas the much lower internal
angles of friction of sand and microspheretargetsresulted in
rim collapse, particularly at higher ambient pressures [see
Schultz, 1990a, b, 1992].
sand. Size distributionsof these target materials are illustrated
in Schultz (1992). Targets composedof 140-200 and 24 sand
exhibited larger modal particle size (125 gm and 450 gin,
respectively),thereby providing a wide range in drag effects
due to particle size. Since particle density also should affect
Figure 5a illustratesan impact into 140-200 sand only. In
general,the developmentof distinctiverampart-borderedejecta
facies persists for impacts into sand to higher ambient
pressures,but at 1 bar, the radial scouredpattern of grooved
inner ejecta is clearly established.Experimentsperformed at
equivalentpressuresof both 0.25 bar and 1 bar but contrasting
demities (helium and carbon dioxide atmospheres)produced
virtually identical ejecta patterns.Thus the trends previously
described for pumice targets also occur for slightly coarser
grained sand targets.Impacts into 24 sand, however, failed to
produceany significantchangefrom near-vacuumconditions.
Aerodynamic drag force relative to gravity increases if
either particle size or particle density (for given impactor
conditions)is decreased.Figure 5b illustratesan impactcrater
aerodynamicdrag, targetsof low-density(0.7 g/cm3)
formedin a targetcomposed
of low-density
(fie= 0.7 g/cm
3)
Micro, heres
100
0.5
-40
20'
140-200 sand
24 sand
125
457
1.55
1.70
40
35
32 ø
30'
i
* Grain
size
issizeofejecta
where
cumulative
fraction
Coy
weigh0
iscoarser
than50%; pumiceexhibitsa bimodaldistribution
(seeSchultz[1992]for complete
distribution).
• Density
ofconstituent
grains
forpunflee
is2.5g/cm';
microspheres,
0.7g/cm';
140-200
and24silica
sand,
2.6g/cm
•.
microsphereswere also tested.Table 3 revealsthat the pumice, microspheres(median size of 100 gm) with a carbon dioxide
microsphere,and 140-200 sand targetsall exhibit comparable atmospherenear 0.9 bar. Such a target and atmosphereshould
porosities.These three targets,however,exhibit very different exhibit aerodynamicdrag effectsgreaterthan even pumice.The
11,630
SCHULTZ:ATMOSPHERICEFFECTSOF EJECTAEMPLACEMENT
resulting ejecta depositsexhibit a very well defined rampart
similar to but extendingto greaterdistancesfrom the rim than
for sand(Figure 5a). In addition,the distal ejectaare deposited
in lobes with splayedtermini extendingmore than 10 crater
radii from the rim (Figure 5c). The craterrim in this example
was lost through rim/wall collapse immediately after
formation.
Particle sizes in pumice exhibit a bimodal distributionwith
the greatestfractioncenterednear 81 gm and a secondarypeak
near 25 gm. Hence the pronouncedcontiguousrampart might
reflect differential sorting during emplacement.Mixtures of
pumice with coarse sand (16 and 24) enhancedthe bimodal
distribution
and resulted in a distinctive
modification
of the
crater with exaggerationof the ejecta rampart(Figures6a and
6b). A pumicecomponentassmallas 8% by weightmixedwith
24 sand produced progressively smaller craters and larger
ramparts with increasing atmosphericpressure.In contrast
with pumice-only targets,however, the rampart-stylepattern
dominatedthe emplacement
styleto a pressureof 1 bar andwas
evidentat pressures
as low as 3 mbar. Sievingthe ejectafacies
revealedthattherampartwascomposed
primarilyof thecoarser
24 sand fraction, whereas ejecta emplaced uprange was a
relatively unsortedpumice/sandmixture veneeredwith pumice
(Figure 6c). Furtherdissectionof the targetbeyondthe rampart
revealed
that the red coarse sand fraction
formed
a distinctive
deposit thinning with distance and resembling unmodified
ballistic deposition.Consequently,it appearsthat a wide range
or bimodal distribution in ejecta (target grain) sizes enhances
the development of the ejecta rampart over a wide range of
environments. The limiting atmospheric pressure and the
minimum pumice (fine-grained)componentrequired for onset
of this style, however, remainsto be established.
The effects of layered pumice and sand further help to
identify processes responsible for the distinctive style of
ejecta emplacementbut may have limited relevanceat broader
scales. Impacts into targets with pumice overlying sand
b
producedcratersresemblingimpactsinto sand alone, even at 1
bar atmosphericpressure.Impacts into a 1 cm layer of sand
overlying compactedpumice, however, produceddramatically
different results.At pressuresof 1 mbar, a bowl-shapedcrater
was formedthat resembledan impactinto pumicealone(Figure
6d). Impactsunder increasingatmosphericpressures(above30
mbar) produceda nestedcraterappearance:a raised-timcraterin
the pumicesubtargetsurrounded
by a ridge of sand(Figure 6e).
The sand layer between the crater rim and sand rampart was
removed during impact with radial grooving evident in the
exposed pumice substrate.The diameter of the sand rampart
relative to the pumice crater increased with increasing
atmosphericpressure. The process responsible for the sand
rampartwill be examinedin more detail in subsequent
sections.
A thin veneerof dry egg temperapaint on the pumicetargets
appearedto enhance the developmentof ejecta flow lobes at
higher atmosphericpressures.The temperaprovided a useful
stratigraphicmarker but also appearedto ignite when mixed
with the hot wake gases.Consequently,carbonatetargetswere
also used in order to examine the possibleeffects of impact
vaporization. High-frame-rate imaging clearly documented Fig.5. Effectof changing
particle
sizeandparticle
density
in thetarget.
(a)
vapor release and allowed the estimateof the energy in the Characteristic
ejecta
rampart
created
byimpacting
0.635cmaluminum
spheres
into
vapor cloud [seeSchultz, 1988;Schultzand Gault, 1990]. 140-200sand(125p.
m)at2km/sunderO.5bar(air)formsclosetothecraterr
Undervacuum
conditions,
thereleased
vaporclouddid not part
ofwhich
has
collapsed
during
the
final
stages
offormation.
(b)Results
ofusing
(100
avery
lowparticle
density
(0.7g/cm
3)relative
create any of the ejecta flow patterns. Under modest microspheres
3 gm)exhibiting
to sand(2.5 g/crn ). The low angleof internalfrictionresultsin extensiverim/wall
atmospheric
pressures
(0.25bar),neither
therampart
northe collapse
after
crater
excavation.
For
this
target,
the
ejecta
rampart
develops
at
flowmorphology
wasparticularly
enhanced.
Henceimpact-larger
distances
from
therim.
(c)Basal
flows
create
much
longer
run-out
ofthin
generatedvolatilesalone will not createejecta flow; ejecta
lobes
with
distinctive
splayed
termini.
Theregion
shown
inFigure
5cis
atmospheric
interactions
are necessary.
The enhancement
of beyond
theboundary
ofFigure
5b(indicated
bythearrow).
Hence
both
rampart
run-out resultingfrom powderedtempera,however,appearsto andflowemplacement
styles
wereobserved.
Theuseoflow-density
microspheres
reflect added turbulencecreatedby its fine size or exothermic increases
theeffects
ofaerodynamic
drag.
Impact
velocity
inFigures
5band5cwas
reaction.
4.8 km/swithan ambientatmosphere
of carbondioxideat 0.94 bar.
SCHULTZ:ATMOSPHERICEFFECTSOF EJECTA -EMPLACE•T
11 631
.
EJECTA
SORTING
2.0
1.5
_
-
rampart
SAMPLE
BULK
1.0
0.5
...................... ?.?:.:•
........................
_buik•
•rejecta_
1
lO
lOO
1ooo
SIZE (microns)
Fig. 6. Effect of mixingred-stained
24 sand(mediansize of 450 gm) with 8%
pumice(mediansizeof 81 grn)by weight.The development
of theejectarampar,.
is
enhancedwith concentrated
coarsefractions(darkerred component):(a) low-angle
illuminationenhancing
the rampartand (b) high-angleilluminationrevealing
distribution
of lightercolored
pumice.Dissection
of ejectafaciesrevealsa mixture
of pumiceand sandinsidethe rampartwith a dustingof pumiceon the surface.
Beyondthe rampart,the coarsersandfractionsdominate.(c) Sortingof ejectasizes
as referencedto the preimpactbulk particlesize distribution[seeSchultz, 1992].
TABLE 4. Effectof TargetTypeand Atmospheric
PressureonEjecta
EmplaeementStyle in LaboratoryExperiments
The coarseningin the rampartis believedto reflect terminaldepositionof larger
clastssuspended
by windscreatedduringcratergrowth.(d ande) Effect of layered
targetsin the presence
of an atmosphere
for 0.635 cm aluminumspheres
launched
at about2.2 km/s.Impactsinto pumiceoverlainby a layerof 20 sandunder1 mbar
(Figure(mr) and 30 mbar (Figure6e) producecontrastingresultsdue to scouring
actionby impact-generated
winds.Experimentsusingpumiceoverlayingsanddid
not createthispatternevenat a 1 bar atmosphere.
Scalebar corresponds
to 10 cm.
3. EIECTA CURTAIN EVOLUTION
The contrast in evolution of the ejecta curtain under a
vacuum and with an atmosphereis illustrated in Figure 7a
Pumice
(compacted)
-0.05
0.06-0.3
0.3-0.7
>0.8
througha paired sequenceof high-frame-ratephotographs
(400
Pumice(compacted)
-0.01
0.04-0.4
0.4-0.7
>0.7
framesper second(fps)). Both impactswere producedby 0.635
with dry tempera
cm aluminum spheres impacting at about 1.6 km/s into
Microspheres
<1 mbar
>2 mbar
>0.1
*
compactedpumicewith the left andright sequences
occurringat
140-200 sand
<0.09
>0.1
->1.0
24 sand+ pumice(8%)
<0.02
-0.05
....
less than 1 mbar pressureand 1 bar, respectively.The ejecta
24sand
allP1r
......
curtain for the impact at 1 bar remainsrelatively undisturbed
throughout crater growth, although a small disturbance is
* Atfullatmospheric
pressures,
radial
pattern
ismasked
bythinveneer
from
basalejectaflow.
observedto move progressivelyup the curtain with time. The
• Pstands
for_pressures.
ejectacurtain under 1 bar (right side) progressivelyincreases
from 57' at 2.5 ms after impactto 66' at 10 ms. In contrastthe
In summary,the characteristicstylesof ejecta emplacement ejectacurtainunder 1 mbar (left side) decreasesfrom 44' to 40'
were most affectedby targetparticle(ejecta) size or particle overthe sametime interval.At latertimes,the curtaindevelops
densityfor given impactorconditions(Table 4) but could be a distinctivebulge at its basewith a flared uppercurtain.The
enhanced
by increased
entrainment
or turbulence.
only significant difference for impacts at much higher
Ballistic
Pumice
(uncompacted)allP•
Rampart
......
Lobes,
Radial
11,632
SCHULTZ:ATMOSPHERICEFFECTSOF EJECTAEMPLACE•
velocities (6 km/s) is the enhancementof the upward moving
kink in this curtain at early times [see Schultz and Gault,
1982]. This kink reflects an eddy exhibitingoutwardairflow at
top and inward flow at bottom.
Impacts into mixtures of sand and pumice (Figure 7b)
graphically demonstratethe effects of differential air drag on
different size ejecta. At high atmosphericdensities,the coarser
size fraction retains the undistorted funnel-shaped curtain
profile, whereas the fine size componentcreates a separate
curtain characteristicof an impact into this componentalone
or equivalently under vacuum conditions.The two curtains
merge, however, at the base. Consequently,aerodynamic
sortingduringballisticejectionand flight may not resultin
aerodynamicsortingduring deposition,exceptfor very late
stage fallout.
In order to directly compare the sequentialgrowth of the
ejecta curtain, individual frames from the film record were
digitally imaged. Digital subtraction of successiveimaged
frames through a computer left an image for only those
portions of the ejecta curtain that changedthroughtime. Two
approaches
were used.First, individualframes(•) were selected
suchthatJ•+l = 2J•in order to enhancedifferencesat late times
when relative motions decrease(Plate l a). Second, changes
over a constant time interval
of 5 ms were examined
at the end
of eachJ• in order to examinevelocity changeswith both time
and position. Different colors were assignedto different time
intervalsfrom cool colors (black, blue) at early times to warm
colors (yellow, red) at late times (Plate lb). Consequently,
continuousgrowth of the curtain is depictedin Plate l a for the
intervals 7.5-15 ms, 15-30 ms, 30-60 ms, 60-120 ms, and
120-240 ms, whereas changesin growth at 7.5, 15, 30, 60,
120, and 240 ms over the 5 ms interval are shown in Plate lb.
Additionally, crater profiles taken after impact were scaledto
the film record and digitally superimposed.The successive
sequenceshownin Figure7 can now be comparedmoredirectly
over a wide rangeof atmosphericpressures.
Plate l a reveals the systematicchange in ejecta curtain
evolutionfor four different impacts,all at about 1.5 km/s but
with progressivelyincreasing atmosphericpressure(top to
bottom). As the crater forms, the ejecta curtain retains its
conical shapeat all atmosphericpressuresas indicatedby the
base of the undistorted
curtain
tied to the raised crater rim
(superimposedprofile). After crater formation, however, the
base of the curtain bulges outward while the upper portion
converges inward. At 1 bar, the curtain evolves into a basal
ejecta flow clearly revealing the nonballistic style of
emplacement.The companionsequence(Plate lb) documents
the effect of pressureon the velocity of the curtainat a given
time as indicatedby the width of eachcolor slice. During crater
growth, the outward velocity of the curtain rapidly decreases
until only a sliver remainsat the base.Justprior to the end of
crater formation,the width of the color slice correspondsto an
outward curtain velocity of about 50 cm/s. After crater
formation under vacuum conditions, the color slices increase in
width with time, thereby indicating an increase in curtain
velocity due to higher velocity ballistic ejecta comprisingthe Fig. 7a. Comparisonof impactsby 0.635 cm aluminumspheres(.2 kin/s) into
(right)conditions.
Each
curtain. With increasingatmosphericpressure,ejecta curtain pumiceundervacuum(left) andfull (1 btr) atmospheric
conditions
producea •teeperejects
velocities at the bulge not only increase but appear to be frameis 5 ms apart.Impactsunderatmospheric
greater than under vacuum conditions. Additionally, these plumethat evolvesinto a bulbousprofile at late timesowingto aerodynamic
images reveal the inward flow well above the crater at late decelerationof ejectaparticles.
times. Hence the presence of an atmosphere appears to
decelerate and redirect the curtain above, but enhances advance
near the surface.
Comparisonsof ejecta curtain growth for pumice and 140200 sand targets are shown in Plate 2. Here, constanttime
intervals (at the same successivesample times shown in Plate
lb) permit focusingon the rate (as well as the form) of growth.
Plate 2a contrastscurtain evolution at 0.25 bar for sand (top)
and pumice (bottom) under atmospheresof helium (left) and
carbon dioxide (right). Impacts into fine sand produce a
slightly less bowed curtain with less irregularitiesat the same
atmospheric pressures. The former phenomenon can be
attributedto the slightlylarger ejecta size for sandtargets;the
latter, to the narrowerrange in particle sizesfor sandrelative to
pumice. Plate 2 demonstratesthat the observedatmospheric
effectsare not theresultof someuniquepropertyof pumice.
The effect of gasdensityis illustratedin Plate 2b by the use
of helium and air. For reference,ejecta curtain evolutionfrom
SCHULTZ:ATMOSPtIERIC
E•S
OFEJECTAE•CEMENT
11,633
Fig.7b.Effect
ofdifferent
particle
sizes
ontheevolution
oftheejecta
curtain.
A under
a vacuum
andatmosphere,
respectively,
asshown
inFigure
7a.Resulting
mixture
of coarse
16sandandpumice
undera 1 baratmosphere
(air)resulted
in crater
resembled
example
inFigure
6a.Scale
barscorrespond
to 10cm.
bothundistorted
anddistorted
components
of theejectacurtainresembling
impacts
Fig.7c.Effectof a layerof coarse
20 sand
overcompacted
pumice
onejectasame
undisturbed
conical
shape,
whereas
thefinerpumice
fraction
resembles
the
curtain
evolution
under
vacuum
(leftside)andatmospheric
(rightside)cortditions.
shape
produced
by impacting
intopumice
alone(right).Resulting
craterunder
Undervacuum
conditions,
theclassical
conical
ejectacm•ainadvances
wellafter vacuum
conditions
isshown
in Figure
6d;crater
at 1 barresembled
example
in
craterformation(left).Undera 1 bar atmosphere
(air), the coarsesandformsthe Figure6e.
impacting140-200sandundervacuumconditions
is shownin pressurecan neverthelessbe seenby the relative sizes of the
the upperleft. Ejectacurtainangleand curtaindistortionare final craters,as discussedby Schultz [1990a, 1992].
At early times (within 7.5 ms) the conical ejecta curtain
generally unchanged, but more detailed examination
shape typifying vacuum conditions remains evident at all
pressures,but higher atmosphericpressuresresult in higher
comparable
ambientdensity.Althoughnot shown,impactsin angles(with respectto the horizontal).Figure 8 documentsthis
a CO2 atmosphere
producethe oppositetrend.Thusair drag, evolution as a function of both time and atmosphericpressure
revealedhereby the effectof atmospheric
density,controlsthe for sand and pumice targets.The curtain angle for both sand
growthof the ejectacurtain.The separate
effectof atmospheric (140-200) and pumice targets under vacuum conditions
demonstratesthat curtain growth under high pressuresof
helium (~1 bar) resemblesgrowthin air at lower pressurebut
11,634
SCHI•TZ: ATMOSPHERICEgFEC-'rsOF EJECWA•CEMENT
a
b
Plate 1. Computer-digitized
fdm recordshowingevolutionof ejectacurtainunder (red). Atmospheric
conditionsrangefrom 0.125, 0.25, 0.50, to 1.0 bar of air with
differentatmospheric
conditions.
(a) Overallgrowthbetween7.5-15ins (purple), impactvelocitiesof about1.5 km/s(0.635cm aluminumspheres).
Widthof color
15-30ms(blue),30-60ms(green),
60-120ms(yellow)and120-240
ms(red).(b) slicesin b indicate
theoutward
velocity
of theejectacurtain
thatdecreases
during
Incrementaladvanceover a constanttime intervalof 5 ms endingat about7.5 ms cratergrowthbut increases
afterwards.
(black),15 ms (purple),30 ms (blue),60 ms (green),120 ms (.veliow),and 240 ms
SCHULZ: ATMOSPHERIC
EFFECTS
OF•A
He
s
EMPLACEMENT
11,635
co 2
s
C•
He
.\
P
P
air
Vac
S
P
S
P
Plate2. Comparison
of atmospheric
pressure,
density,
andparticle
sizeonejectaunder
a pressure
of 1 barof air(right,topandbottom)
andhelium
(left,bottom).
curtain
evolution.
(a) Effectof atmospheric
density
ata givenpressure
(0.25bar) Topleftsequence
shows
theevolution
oftheejecta
curtain
for140-200
sand
under
for 140-200
sand(toppair)andpumice
(bottom
pair)bytheuseof a heliumvacuum
conditions
asa reference.
Thecolorcoded
stages
ofcurtain
evolution
are
atmosphere
(left)andcarbon
dioxide
atmosphere
(right).
(b)Effect
ofatmospheric
thesame
asinFigure
8.Scale
barcorresponds
to10cm.
density
onejecta
curtain
evolution
forimpacts
intosand(top)andPumice
(bottom)
11,636
SCHULZ: ATMOSPHERICEFFECTSOF F_JEC•AEMPLACEMENT
90
I
90
i
I
I
I
i
i
I
I
I
I
I
-
/- P/Po
=1
•.) 60
60
- _+
.... ......
©
Helium
,,,._
P / Po=0.12
30
30
(9 Vacuum
,
_
Air
+ Helium
No. 140-200 Sand
No. 140-200
I
lO
Sand
I
lOO
O.5
Time (ms)
1.0
Pressure (bars)
90
90
I
I
I
i
i
I
I
i
i
i
I
i
60
60
Helium
30
(9 Vacuum
ß Air
Pumice
+ Helium
I
Pumice
I
I
a
lO
lOO
Time
(ms)
I
I
t
I
I
I
I
0.5
b
I
I
1.o
Pressure
(bars)
Fig. 8. Contrasting
ejectacurtainanglesasa function
of time,atmospheric
pressure,evolution.Moreover,curtainanglein pumiceis greaterthan in sand.Both
atmospheric
density,andtargetparticlesize.(a) Time evolutionof the ejecta responses
shouldbe expected
if atmospheric
density(aerodynamic
drag)controls
curtainfor 140-200san:l(top) and pumice(bottom)targets.Undervacuumejectacurtainangle.(b) Evolutionof ejectacurtainangle25 ms (+2 ms)after
conditions,
thecurtainanglefor sandandpumicegradually
decrease
fromnear50ø impactasa function
of atmospheric
pressure
for air andheliumusingsand(top)
to43ø.Asatmospheric
pressure
increases,
ejection
anglealsoincreases.
Comparison
andpumice
(bottom)
•rgets.Ejectacurtain
angleismoreaffected
in pumice
thanin
of a heliumatmosphere
at 1 bar with air at 0.125 bar revealsa very similar sand.Useof heliumhaslitdeeffect.Impactvelocities
wereall about2km/s.
decreasesonly slightly with time (from 50ø to 42ø). The drag. Under high atmospheric
densities,a basalejectasurge
additionof an atmosphereincreasesthe curtain angle. Use of develops and advancesoutward at velocities that exceed the
helium allows the comparisonof the effect of atmospheric ballisticejeetacurtainundervacuumconditions.
pressureand densityand revealsthat atmosphericdensityis the
4. •A
EMPL•C•M•r
I•OC•SES
controlling variable for both pumice and sand:helium at 0.84
bars exhibits the same curtain evolution as air at 0.125 bar,
both conditionsresulting in nearly identical densities(0.116
and 0.125 air at STP, respectively). Figure 8b further
The preceding descriptions focused on the variables
controlling the observed ejeeta patterns and ejeeta curtain
establishes the difference between helium and air at common
profiles in different atmosphericenvironments.This scction
times (25 ms). At high atmosphericpressuresthe curtain reviewsspecificobservations
and experimentsthatcan provide
continues to move outward at the base after crater formation but
clues for possible processesproducing contrastingejeeta
ceases above the surface. Comparisonsof the vacuum and emplaeementstyles in the laboratory. Possible processes
nonvacuum eases (Plate 1) reveal that the base of the curtain unrelatedto craterformationare consideredbeforeexamining
under high pressuresmoves outward at a velocity higher than the originsfor the contiguous
ejeetarampart,ejectaflow lobes,
the curtain under vacuum conditions.Consequently,a base- andradial scourpattern.
surge-like
motiondevelops
afterthecraterfinishes
forfi•ing. Nonimpact processes. Could the various ejeetapatterns
The nature and cause of this motion will be discussed in a
and ejeetacurtainmodificationsreflect disturbances
createdby
subsequentsection.
launchdesign,the preimpaetpassageof the projectilethrough
In summary, an ensembleof ballistic debris composesthe an atmosphere,or the release of gasestrappedin the target?
ejeeta curtain during crater growth even at high atmospheric Several procedures and observations minimize or constrain
pressuresand densities. Both particle size and atmospheric such possibilities.
densityaffect the shapeand evolutionof the ejeetacurtainafter
Muzzle blast by gasestrailing the projectilecan occur in
crater formation, thereby indicating the role of aerodynamic both two-stagelight-gasguns and conventionalpowderguns.
SCHULTZ:ATMOSPHERIC
EFFECTS
OFF3F.
CrA EMPLACE•
This effect was minimized in theseexperimentsby the use of
thin mylar diaphragmsmountedboth at the exit port from the
launch tube and at the entranceport in the impact chamber.A
mylar diaphragm is necessaryfor experimentswith the light
gas gun in order to prevent atmosphericgasesfrom entering
the launch tube. A further benefit, however, is the containment
of muzzle blast. For example, the self-luminousionized wake
left behind the projectile at high velocities(>5 kin/s) exhibits
fine-scale turbulencethat remains unmodified (see Schultz and
Gault, 1982), thereby confirming the absenceof a launchrelated blast disturbance.Failed experimentsproducing large
muzzle blast break the mylar and producevisible effects; such
occurrencesare excludedfrom this analysis.
Turbulencecreatedby the projectile air shockrepresentsa
secondpossiblenonimpactprocessthat might modify ejecta
eraplacement. Several observations, however, indicate that
this effect is minor. First, air shock effects did not modify a
dry-ice vaporcloud suspended
abovethe impactsurface[Schultz
and Gault, 1982]. Second, air tracers (threads) showed the
passageof the air shockwell abovethe impactbut alsoshowed
rapid equilibration.Third, the characteristic
ejectapatternsand
curtainevolutionoccurfor impactsat subsonic(30 m/s) as well
as supersonicvelocities. Signatures of the projectile shock
wave and trailing wake appearto exist in early-timeturbulence
in the ejecta curtain [Schultz and Gault, 1979, 1982], in
crateringefficiency [Schultz,1990a, 1992], and in cratershape
[Schultz, 1990b]. Such signatures,however, do not extend to
late-time evolution of the ejecta curtain when near-rim
eraplacementoccurs.
Compactedpumice exhibits nearly the same density (1.5
g/cm3) andporosity(-40%) as 140-200sandbut haslower
permeability. Consequently, sudden-impact-released
gases
trappedin pore spacesmight play a role in modifying the
ejectionprocess,but severalobservations
do not supportthis
suspicion.Normal procedureinvolvesslowly evacuatingthe
impact chamber(20-60 rain) prior to introducingdifferent
gases.Slow evacuationpreventslarge contrastsin pressure
betweenpore spacegas and ambientpressures.
The effective
removalof pore-containedgas is believedto be demonstrated
by the closesimilarity in ejectaplume evolutionfrom earlytime to late-timebetweensandandpumiceat low atmospheric
pressures.
The effectof impact-released
gasesat low ambient
pressures
shouldbe more pronounced
than at high pressures
owingto the largepressuredifferential.
Any possibleeffectof gassuddenlyreleasedfrom the target
shouldbe amplifiedfor impactsinto volatile-richtargets.Lowdensitystyrofoamspheresof varying sizeswere placedacross
target surfacesof volatile-rich carbonates(dolomite) and
pumice. High-frame-rate photographsrevealed that these
tracerswere blown away for impactsinto carbonatetargetbut
remainedin place for pumice targetsuntil entrainedin the
ballistic flow of ejecta. In both experiments,impact occurred
under similar ambient conditions following evacuation at
comparablerates. The use of carbonatetargets did not,
however, result in enhanced ejecta entrainment. Finally,
pressuregaugestraversedby the expandingejecta curtain
indicateda decrease(not an increase)in pressureimmediately
behindthe ejecta curtain.
In summary,the distinctive ejecta eraplacementsequence
and curtainevolutionare believedto reflect a processrelatedto
ballistic ejecta interacting with the atmosphereand not
phenomena
relatedto muzzleblast,atmospheric
shockwave,
projectilewake,or sudden
releaseof gastrappedin thetarget.If
atmospheric
gasfilling targetporespacedoesplay a role, then
it mustbe expressed
asviscousdragactingwithinthe cratering
11,637
following observations. First, contiguous ramparts are most
pronouncedfor a limited range of atmosphericpressures,above
which ejecta flows are observed(Figures 1 and 2). Second,the
expression and maximum radial extent increases with
increasing aerodynamic drag controlled by the atmospheric
density, drag coefficient, target particle size and density, and
ejectionvelocity (i.e., crater size). Third, rampartscan develop
even without significant modification of the ejecta curtain
shape (i.e., at low densities, Plate 2). Fourth, impact into
targets where grain size radially decreasedfrom the impact
point did not produce the rampart pattern. Fifth, impacts into
sand-over-pumicetargets (Figure 6d) produce sand ramparts
outside a sand-clearedzone in a manner analogousto rampart
formation.And finally, off-verticalimpacts(Figure4) produce
a rampart-bordered,butterfly ejecta pattern.
The
first
three
observations
indicate
that entrainment
of
fine ejecta plays an important role. Too much entrainment
resultsin the onsetof more complex ejecta morphologies;less
entrainment suppressesthe complex ejecta morphologies,
even at high pressures.Mixtures of relatively small fractions
of fine-sized pumice with coarser sand produce analogous
rampartlike ridges at very high atmosphericpressures(Figures
6a and 6b) but not in a vacuum. The fourth observation
indicates that rampart formation is a late-stage processand
requires finer fractions. The fifth observation suggeststhat
impact-related,outward moving winds may play a role. The
sand rampart in sand-over-pumicelayered targetsdevelopsby
accumulation of the coarse sand rather than uplift of the
substrate. This interpretation is consistent with observations
of the basalejectasurge(Plate 1). The low-shearstrengthof the
sand permitted its radial displacementwell beyond the rim of
the crater formed in pumice substrate.Finer pumice ejecta were
piled on the crater-facingside of the rampart.Finally, the sixth
observation indicates that rampart formation is tied to
eraplacementof ejecta,not to vapor releaseor wake effects.
While Plate 1 suggeststhat outward winds result from
impacts in an atmosphericenvironment,the formation of the
rampart in pumice targets appearsrelated to a depositional
process.Profiles of the ejecta deposits(Figure 9a) show that
the rampart occurson top of the continuousejecta.The rampart
can be seen to crossejecta clumps, and obstaclesplaced near
the rim were found to suppressthe developmentof the rampart
downrange(Figure 9c).Thus the rampart doesnot result from
preemplacementscouring of the preimpact surface but from
deposition. Figure 6c revealed that coarse ejecta are
concentrated in the ejecta rampart with finer fractions
dominating the inner continuous deposits. This pattern of
deposition is consistent with ballistic ejection and
eraplacementof the coarsefraction but modified eraplacement
of the finer fraction. Moreover, it is consistent with the onset
of rampart formation without significant distortion of the
ejecta curtain (Plate 2).
Although perhaps counterintuitive, the ejecta rampart
increasesin distancefrom the rim with increasingatmospheric
pressure.The rampart typically becomesmuch more subtle,
however, as the turbulentejecta overrunsand scoursthe inner
facies (Figure 9b). Thus ejecta ramparts and ejecta flow
morphologiesare not exclusionaryprocesses.
Ejecta flow. At higher atmosphericpressures(0.4-0.7 bar),
the contiguous rampart developed in compacted pumice
increasesits diameter and commonly breaks into tonguesof
ejecta (Figure lb). Ejecta rampartsmay be brokenor over-run
by these flow lobes under transition conditions.The ejecta
lobes extend as far as six radii from the crater rim and reflect
depositionfrom a ground-huggingflow witnessedin the film
record. This turbiditylike flow behavior can be illustratedby
Ejecta rampart formation. Contiguousejecta ramparts the effect of topographicbarriers on its advance. A cratercharacterizecraters formed under lower atmosphericpressures facingscarp2 cm high constructed
in the targetwasovemm by
over a wide range of ambient gas densities and impact individual flow lobes. Consequently, the ejecta exhibited
velocities. The causative process must account for the fluidlike behavior even in the absence of water.
flow field.
11,638
SCHULTZ:
ATMOSPHERIC
EFFECTS
OF•EJEC'FA
EMI:q.ACEMENT
P-0.25
bars (Ar)
b
rampart
P-0.94
scour
bars
(Ar)
zone
..
Fig. 9a. Comparison
of ejectaprofilesfor impactsintocompacted
pumiceat 0.25
bar and 0.94 bar of argon.The ejectarampartat 0.25 bar developson top of the
eraplaced
ejecta(top).At higherpressures,
scouring
createsa moatadjacentto the
rim (bottom).Ballisticallyejectedparticlesare entrainedand transported
to large
distances,
thereby
producing
a thinned
ejectadistribution.
Profiles
havea twofold
verticalexaggeration.
Ejecta flow did not develop in either 140-200 sand or in
sand-pumice mixtures. Moreover, it was not particularly
enhancedby the use of easilyvolatizedtargetmaterial.The use
of low-density microsphereswith comparable ejecta sizes,
however,resultedin both an inner ejectarampart and distal runout lobes (Figures 5b and 5c). The distal ejecta lobes form a
distinctive divergent pattern resemblinga cauliflower (Figure
4c). Consequently,the developmentof ejecta flow seemedto
require only a sufficient ambientpressureand a target having
small (<80 Ixm) or low densityconstituentgrains.
The developmentof ejectaflows thereforeappearsto reflect
a three-stageprocessof ejecta deceleration,entrainmentwithin
turbulent vortices, and run-out as a ground-huggingdensity
flow. Such a descriptionmirrors the observedbasal flow of
ejecta that outpaces ballistic ejecta for identical impactor
conditionsin the absenceof an atmosphere(Plate 1). While the
formation of the ejecta rampart reflects a broad-scalevortex
pattern trailing behind the wall of ejecta at lower pressures,
ejecta flows representejectakinetic energybeing convertedto
atmospheric convective turbulence near the rim and the
formationof a turbidity flow.
Ejecta scour (radial pattern). At the highest ambient
pressurespossiblein the vertical gun (i.e., 1 bar), material is
still ejected in ballistic trajectoriesduring crater formation,
thereby indicating response to the collisional shock and
rarefactionin the target. Emplacementof ejecta, however,now
is severely modified by an intense, tight eddy forming an
Fig. 9b. Comparison
of ejectarampartformedin compacted
pumiceat 0.25 bar
(top)and0.9 bar (bottom)of argon.The ejectarampartat 0.25 baris closeto the
craterrim andrestson ejectadeposits,
asshownby theprofilein Figure9a. At 0.9
bar, the rampartextendssignificantlyfartherbut is mutedby scouring.This
scouringalsoproduces
a distinctivenear-rimmoat(innerarrow).Impactvelocities
were5.0 and5.4 km/s,respectively.
Scalebarscorrespond
to 10 cm.
preserved beyond the ejecta rampart at lower atmospheric
pressures,thesetexturesare muted or lost at high pressuresdue
to the intense scouring action witnessesin the film record.
Thus high atmospheric pressures do not prevent crater
formation but result in nearly completetransferof energy from
ejecta kinetic energy (arrestingejecta curtain advancesoon
after crater formation) to turbulent power within the
atmospherewhich scoursthe surroundingsurfaces.
Crater and ejecta dimensions. The changein emplacement
style with atmosphericpressureappearsto reflect increased
entrainment of the atmosphere with ballistic ejecta at
interparticle scales and increased intensity of turbulence at
broaderscales.In order to extend theselaboratory-scaleresults
to kilometer-scalecraterson Mars, ejecta entrainmentand the
expandingdonutlike toroid. This vortex scoursballistically emplacementprocessneed to be quantified.
The diameter of the ejecta rampart relative to the crater
emplacedejectanear the rim, therebycreatinga characteristic
moatsurrounding
the raisedrim (Figures9a and9b). Entrained diameter appears to increase with increasing atmospheric
ejectaarecarriedanddispersed
manycraterradii fromtherim by pressurefor a given compositionand for the same impactor
this process.The long run-out lobes in Figure 5c further conditions (Figure 9). The specific physical processes
underscorethe process.Such distal ejecta deposits,however, affecting crater growth during excavation, however, are
are characteristicallythin and principally composedof very different from processescontrolling ejecta emplacementafter
fine grained ejecta. While subtle preimpact textures remain excavation. Consequently, observed crater dimensions
SCHULZ: ATMOSPHERICEFFT.
L•S OF FJECrA EMPLA••
11,639
the impactortrajectoryand perpendicularto the cameraliterally
bisected the cratering event. Such configurationsare termed
quarter-spaceexperimentsand reveal the pattern of circulation
inside the ejecta curtain.
The quarter-spaceexperiments graphically demonstrated
ballistic ejection and unobscuredtransientcavity growth well
after passageof the shockwave (Figure 12). The impact point
is necessarily offset from the window (about a projectile
diameter away) and results in ejecta streaming along the
plexiglass without atmospheric distortion. Nevertheless, the
sequenceof views revealsthat the baseof the ejectacurtainis
carried outward by the rolling vortex inferred from stereo
photography. Once the vortex no longer can mobilize and
supportthe ejecta,it collapsesto form a rampartwhile smaller
ejecta entrained in the rolling vortex help to identify its
continuedoutward advance.After the ejecta curtain dissolves,
suspendedejecta above the crater simply fall directly to the
surfacewithout evidencefor any thermalbuoyancyeffects.
The pressure-controlledonset of different emplacement
stylesyet the density-controlled
run-out distancessuggestthat
pressure differential created in the basal vortex controls the
onset of a particular emplacementstyle while the degree of
entrainmentin the vortex controlsthe run-outdistance.Ejecta
entrainmentdependson the relative role of aerodynamicand
gravitational forces:
d/g= (3/8)(CDp%2/tSga)
(1)
where d is the decelerating force, g is gravitational
acceleration,CD is the drag coefficient,and ve is late-stage
ejection velocity for an ejected particle of density /5 and
diameter a. The abscissain Figure 10, then, can be viewed
Fig. 9c. Effect of obstacles
on ejectarampartformationin compacted
pumice.A
vertical barrier placedin the preimpacttargetinterruptedthe advancingejecta
cuecainand suppressed
the formationof the rampart.A small scarp (top) also
affectedrampartformation.Impactvelocity (0.635 cm aluminum)was 5.2 km/s
with an atmospheric
pressureof 0.25 bar (argon).Scalebar corresponds
to 10 cm.
Illuminationis from the upperright.
equivalent to equation (1) with all variables constant,
exceptingdensity. At laboratoryscales,the small size of the
momentumof the impactor.
relative viscosities listed in Table 1 reveal that such a case
ejecta(<80 Ixm) and low velocityof ejectaat late stage(-50
cm/s)resultin low Reynoldsnumbers(Re) approaching
unity
wherethe dragcoefficientvariesas (1/Re)•ø.For givenimpactor
and target conditions,the range in densityand gas viscosity
controls the variation in CD. The complex conditions and
principally serve as reference. Crater size under vacuum distributionof shapesprecludeestimatingCD directly.In the
conditionsactuallyprovide a more relevantcomparisonsince limiting casewhereo• is unity, the role of (d/g) reducesto a
this dimension can be related to the initial energy and dependence
on only viscosityof the ambientatmosphere.
The
Figures10 and 11 permitcontrasting
the effectsof ambient wouldstill remainconsistent
with the trendsin Figure10.
pressureanddensityon the diameterof the ejectarampartand
The onsetof nonballisticejecta emplacementalso can be
crater for two different impact velocitiesbut identicalimpactor viewed analogous to conditions necessary for aeolian
size. Although the data are not shownin dimensionless
form, transport.Greeley and Ivcrsen (1985) examinedthe threshold
pressureand density representthe only remainingrelevant friction speedsnecessaryfor particle motion on Earth. For
controllingvariables.Atmosphericpressuredoesnot appearto particle sizes comparableto pumice, the thresholdfriction
affect rampart diameter for different constituentatmospheres speedwas20 cm/s,a valuecomparable
to theoutwardvelocity
(Figure 10a). Increasingatmosphericdensity (Figure 10b) of the ejectacurta.;n
as it advances
beyondtherim (Plates1 and
systematically increasesrampart diameter with decreased 2). Entrainmentin a vortex, however,was found to be
scatter. Since atmosphericdensity affects aerodynamicdrag, independent
of viscouseffects(i.e., particlediameterand wind
entrainmentof ejecta in the atmosphericresponseappearsto speed); rather, the pressure difference created in the vortex
control the observedrun-out. Parallel effects of atmospheric lofted and entrainedparticles.Wind velocitiesare createdand
conditionson rampart and crater diameterfor sand targetsand controlledby the pressuredifferencein front of and behind the
more complex stratigraphiesand mixtures are shown in Figure ejecta curtain;consequently
atmospheric
pressurerather than
11. Table 5 providesa listing of empiricaldata usedin Figures densitycontrolsthe onsetof rampartformation.As notedby
10 and 11.
Controlling processes. The film record provides critical
information about ejecta emplacementprocesses.First, highframe-rate photography from cameras positioned above the
impact reveal that the region inside the ejecta curtain is visible
throughoutcrater growth. Only after crater formation does the
crater cavity become obscured.At this time, ejecta well above
the impact surfacerapidly convergeinward and appearto be
drawn downward. Second, stereo imaging by a side camera
further reveals an upward motion behind the leading wall of
ejecta. And third, a thick plexiglasswindow placed parallel to
Greeleyand Iversen[1985], however,viscousdrageffectsdo
becomeimportantas particlesizeanddensityincreasebeyond
a criticalvalue, that is, as the laboratorycratersare scaledto
much larger Martian craters.
Different ejecta emplacement styles in laboratory
experiments,then, reflect different processesthat largely
dependon the atmospheric
response(turbulenceor winds) and
the degreeof entrainmentof ejecta into this response.The
variousstyleshave not been placed into a taxonomicmatrix,
as is commonlydonefor Martiancraters,in orderto emphasize
underlyingprocessesand to acknowledgethat suchprocesses
11,640
SCHULTZ:ATMOSPHERICEFFF.
L•S OF .EJECTAEMII•CEMENT
2.0
2.0
I
0.635
I
cm AI
0.635
w
1.5
cm AI
. air
Pumice
Pumice
I-
I
v-2
km/s i+He
I
+HeI
v---2 km/s
-
w
w
x
1.5
_RAMPART
w
x •
[]
1[]0021
+
RAMPART
x
I
-2.0
+
1.0
CRATER
•-'--
I
I
-1.0
0.0
-; .0
1.0
2.0
km/s
0.635
cm AI
I
-1.0
2.0
w--5.5
CRATER
0.0
I
1.0
I
+He
Pumice
RAMPART
RAMPART
• CO2
1.5
1.5
•
+
ß
•
©
¸
CRATER
1.0
-
1.0
i
-2.0
a
-1.0
••
CRATER
-
I
0.0
i
1.0
LOGPRESSURE
(bars)
-2.0
b
-1.0
I
0.0
1.0
LOGDENSITY
(p/Pair)
Fig.10a.Comparison
oframpart
andcrater
dimensions
asa function
ofpressure
Fig.10b.Comparison
of datain Figure10abutasa function
of atmospheric
for different
gascompositions
at 2 km/s(top)and5.5km/sCoottcaxQ.
Althoughdensity.
Therampart
diameter
nowexhibits
a consistent
increase
withdensity,
crater
dimensions
show
littledifference
asa function
of gascomposition,
rampartthereby
suggesting
thatentrainment
resulting
fromaerodynamic
dragplays
a role.
diameter
exhibits
noapparent
change
buthasincreased
scatter
withhigher-density
Linesrepresent
visualfitsfor datafor reference.
gasesresultingin largerramparts.
may not scaleto muchlargerdimensions
in a simpleor direct ballisticejectacurtainandentrainment
of ejecta.A fundamental
way.Moreover,differentcombinations
of processes
canleadto assumption
mustbe madeat the outset:the crateringflow field
multipleemplacement
stylesarounda singlecraterresulting
in andeventualballisticejectionfrom thecratercavityreflectthe
differenttaxonomicclasseswithoutchangingthe underlying mechanicalresponseof the target to the impact shockand
cause.Table 6 summarizes
the variousemplacement
styles consequent rarefaction waves from the free surface as
observed
in the laboratory,
specificexamples
citedin the text, summarizedby Gault et al. [1968]. In other words,material is
observed
processes,
controlling
variables,
andinterpretation. ballistically excavated,not simply driven by explosive
5. DISCUSSION
The variousstylesof ejectaemplacement
bear striking
similaritiesto ejectafaciessurrounding
Martiancraters.Such
similarities
havegreateruseif theycanbe incorporated
into a
testablemodel of the emplacement
process.The following
discussiondevelopssuch a working model and explores
possibleimplications
for understanding
ejectaemplacement
on
Mars, Earth, and Venus.
backpressures.
Suchan assumption
can be justifiedby the
observedevolution of the ejecta curtain, an observation
discussed
in previoussections.
Althoughbaclc•essures
created
by the impingingwake trailingthe projectilemay augment
cratergrowth[Schultz,1990a, 1992], thisprocessis assumed
to be a second-order
effect modifyingearly-timeejecta.
Consequently,
growthandejectionis a systematic,
not chaotic
processunder atmosphericconditionsjust as it is under a
vacuum.
Interactionbetweenejectaand atmosphere
can be divided
Emplacement
model. Threedifferentstylesof late-stage into two time scales.Early-timeimpactprocesses
occurwhen
ejectaemplacement
with increasingatmospheric
pressurecan the crater has achievedonly 40% of its final diameter. As
be recognized
fromthelaboratory
experiments:
rampart,multi- reviewedpreviously[SchultzandGault,1982],a complexair
lobed, and radial. These styles correspondto processes shockmustdevelop
in response
tojettedmaterialandanearlyreflecting the increasinglyintenseatmospheric
responseto time supersonic
ejectaplume(Figure13a). Suchinteractions
crater formation resulting in increasedmodification of the appear
to berestricted
to withinthef'trstmillisecond
of impact,
SCHULTZ:ATMOSPHERIC
EFFECTS
OFEIECTAEMPLACEMENT
2.0
i
i
0.635
cm
AI
/' air I
Sand
(No.
140-200) l []C02
J
•c
w
1.5
w
•
_x- --•"
.....
•
•_x-
__•- --RAMPART
•
CRATER
1.0
11,641
front of the ejecta curtain, thereby creatinga strongprecursor
wind. Low-density styrofoam spheres placed across the
preimpact surface documentedpassageof this vapor cloud
front. Most of the vapor, however, is partly containedin and
redirected by the growing transient cavity [see Schultz and
Gault, 1979, 1982; Melosh, 1982; Schultz and Gault, 1990].
Consequently,the early-time vapor cloud resemblesan upward
directedjet exhaustfor high-angle impacts [see Schultz, 1988].
In the presence of an atmosphere,this upward moving and
expandingcloud is retarded both by the ambient gas and the
incoming wake trailing the projectile. Such processes,
however, are all early-time phenomenaand do not appear to
affect later-stageejecta eraplacementprocessesat laboratory
scale.
At low impact angles (<30ø), the early-time jetting
component
and vapor cloud travel rapidly downrange,thereby
I
becomingdecoupledfrom the initial ejectaplume and growing
1.o
-1.o
o.o
-2.0
transient cavity [Schultz, 1988; Schultz and Gault, 1982,
1990]. Under low atmosphericpressuresthe resulting cloud
expands hemispherically while traveling downrange. Under
2.0
I
I
high atmosphericpressures,the vapor cloud, early-time jet,
v --- 2 km/s
and ricochetedprojectile fragmentscombinein a tight fireball
0.635
cm AI
that scoursthe downrangetarget surfaceprior to eraplacement
by late-stageejecta.
o mixture: 92% No. 24 and 8% pumice
The evolutionof the later stageejectacurtainand subsequent
• layered: No. 20 sand over pumice
ejecta emplacementare illustrated in Figure 13b. Becausethe
•c
w
evolution of the late-stagecurtain is similar from subsonicto
1.5 hypersonic impact velocities, the observeddistortion of the
w
RAMPART
ejectacurtainis interpretedas a responseof the atmosphereto
the outwardmovingwall of ejectacoupledto the effect of drag
on individual ejecta as they leave the cavity. Specifically, a
partial vacuum develops behind the outward moving curtain,
but recovery to ambient conditionsoccurs as the atmosphere
CRATER
rushes through (and over) its thinnest upper portion.
Moreover, airflow respondingto the inclined ejecta curtain is
b
I
I
analogous to a well-known hydrodynamic problem of flow
impingingon an inclinedplate. In the frame of referenceof the
-2.0
-1.0
0
1.0
moving plate, the redirected airflow develops compressed
streamlines as it passes over the top, thereby producing
LOG PRESSURE (bars)
augmentedvelocity and pressure.If the curtain is viewed as a
flexible plate, then a bowedprofile shoulddevelop.Behindthe
Fig. 11. (a) Comparison
of rampartandcraterdimensions
for 140-200sandas a
functionof atmospheric
pressure.
Craterdiameter
increases
at higheratmosplseric curtain, a rolling circulationpatterndevelopswith inward flow
at top and outward flow at the base. The intensity of this
pressures
due to cratercollapseand the effectsof projectilewake blast. (b)
Comparison
of atmospheric
effectsonrampartformedin sand-pumice
mixturesand circulation(wind velocity) shoulddependon the velocityof the
layercdsandtargets,examplesof which are shownin Figures5b and 6c,
outwardmoving curtain at late times, which in mm dependson
respectively.
the ejection velocity. As depicted in Figure 13, then, these
recoverywinds entrain and mobilize ejectalandingbehindthe
a small fractionof the 100 ms requiredfor craterformation]n ballistic ejecta curtain. The circulation will be sustained,
laboratory experiments.The wake trailing the projectile however,only as long as a pressuredifferential exists.
With increasingatmosphericpressures,turbulenceincreases
becomescontainedand compressed
within the transientcavity.
A short-lived ionized ball of gas (<0.5 ms) evolves into a and sustains long run-out distances of individual flows.
toroid that interacts with the ejecta plume (0.2-0.5 ms). Eventually, the intensityof the entrainedatmosphereat higher
Sufficientlysmallejectabecomeentrainedin this toroidandare pressureserodesthe crater rim, scoursballistically emplaced
drawninto the cavity, therebypartly obscuringthe craterfloor. inner ejecta deposits, and dispersesthe distal ejecta. The
of this processcan be inferredfrom the deepscour
At later times (0.5-2 ms), vortices develop along the ejecta effectiveness
curtain.Suchvorticesdependon atmospheric
densityandmach annulussurroundingthe rim and the reducedejecta thickness
number and are interpretedas drag-inducedflow analogousto shownin Figure 9a. Erosionalscouringis further indicatedby
turbulence created by relative air flow (created by upward the basal ejecta flow, which exceedsballistic velocities (Plate
moving ejecta in the plume) moving acrossa flat plate (the 1), the radial grooving characteristicof the near-rim (<2R)
outwardmoving ejecta plume). This summaryof early-time ejectadeposits,and the roundingof the inner facies.
The scenarioin Figure 13 depicts two different modes of
interactionsin the context of inferred/observedphenomenais
provided for completeness.It is emphasizedthat such eraplacement.After the crater has finished forming, ballistic
interactionsat laboratoryscalesoccurover a time representing ejectaform the inclined, outwardmoving curtain.Near the rim
only 1% of the timefor craterformation;hencethereshouldbe (<0.5 R), impactingejectacreatea debrisflow that is limited in
advanceat laboratory scalesowing to low potential energy.
litfie effect on late-stageemplacement.
a
Early-timeimpactprocesses
also includeevolutionof any
impact-generated
or liberated vapor. Vertical impactsinto
volatile-rich targets(dry ice and carbonates)under vacuum
conditionsreveal that a portionof the vaporcloudexpandsin
This first phase, however, resembles qualitatively
experimentalsubaqueous
debrisflows [e.g.,Hampton, 1972].
Flow separation
in thelee of suchadvancing
debrisflow creates
a turbulent cloud of debris mixing with the ambient fluid;
11,642
SCHULZ: ATMOSPHERIC
EFFE•S OFFJE•A EMPLA••
TABLE 5. EmpiricalDatafor Figures14 and 15
Atmospheric
Impactor
,
Environment
Target
Crater
Roun
d
2r,cm
Mass,
[•
Vo
vi
Pressure Type
Type
DC
ER
821121
0.635
0.376
2.33
2.33
0.0632
air
P
14.5
21.6
821209
0.635
0.376
2.22
2.20
0.063
CO2
P
13.2
18.9
821123
821214
0.635
0.635
0.376
0.376
2.05
1.92
2.02
1.90
0.125
0.125
air
air
P
S
11.6
18.0
20.9
21.9
871212
0.635
0.376
5.24
5.24
0.131
helium
P
17.7
29.8
810531
850611
850609
850615
820516
0.635
0.635
0.635
0.635
0.635
0.376
0.376
0.376
0.376
0.376
5.44
5.35
4.73
4.40
6.00
5.36
5.18
4.38
4.08
5.80
0.160
0.24
0.25
0.25
0.25
nitrogen
argon
argon
argon
argon
P
P
P
P
P
17.5
17.5
14.1
17.4
16.9
31.7
32.7
28.1
29.1
29.5
820521
0.635
0.376
5.54
5.34
0.25
CO2
P
16.9
29.5
810214
810505
0.635
0.635
0.376
0.376
6.50
5.34
6.28
5.16
0.25
0.25
argon
argon
P
P
17.5
15.6
27.1
30.9
821128
821126
0.635
0.635
0.376
0.376
2.15
2.13
2.14
2.11
0.25
0.25
helium
air
P
P
10.9
11.9
23.8
821127
0.635
0.376
2.24
2.16
0.25
CO2
P
10.4
20.9
821223
820508
0.318'
0.159
0.149
0.0056
5.08
1.92
5.08
1.73
0.25
0.25
helium
air
P
P
12.2
3.6
24.4
5.8
821218
0.635
0.376
2.10
2.09
0.25
helium
S
17.4
22.4
821217
0.635
0.376
2.02
1.95
0.25
CO2
S
17.0
22.4
811151
0.635
0.376
1.94
1.89
0.25
air
S
17.7
22.5
871211
0.635
0.376
5.06
5.04
0.25
helium
P
16.5
29.8
821213
0.635
0.376
1.92
1.83
0.50
air
S
16.8
23.3
810210
0.635
0.376
3.20
2.99
0.50
argon
P
13.3
21.8
821205
0.635
0.376
2.21
2.20
0.50
helium
P
10.4
19.3
821207
0.635
0.376
2.13
1.98
0.50
CO2
P
9.9
18.5
820507
0.159
0.159
2.14
1.74
0.5
air
P
3.4
5.8
871221
0.635
0.376
5.52
5.25
0.5
nitrogen
P
14.9
29.2
871210
0.635
0.376
5.13
5.09
0.48
helium
P
16.6
29.2
871209
0.635
0.376
4.95
4.89
0.76
helium
P
14.0
27.9
840501
840502
0.635
0.635
0.376
0.376
6.04
5.98
5.34
5.29
0.92
0.92
argon
argon
P
P
14.2
14.6
34.4
33.8
820518
0.635
0.376
5.86
5.78
0.94
helium
P
14.0
26.8
871213
820552
0.635
0.635
0.376
0.376
5.47
2.10
4.97
1.82
0.917
0.99
nitrogen
CO2
P
S
15.2
18.5
27.0
23.8
18.9
820553
0.635
0.376
2.19
2.16
0.99
helium
S
19.5
25.4
811149
0.635
0.376
2.04
1.85
1.0
air
P
18.0
23.2
860822
0.635
0.376
1.68
1.68
0.029
air
PS
17.4
21.7
860821
0.635
0.376
2.00
1.95
0.25
air
PS
15.6
21.6
860825
0.635
0.376
1.92
1.74
1.0
air
PS
11.2
21.6
820551
820550
820548
0.635
0.635
0.635
0.376
0.376
0.376
2.20
2.22
2.19
2.19
2.17
1.99
0.062
0.25
1.0
air
air
air
S/P
S/P
S/P
12.4
8.5
6.9
21.7
21.2
19.3
In Table 5 the impactvelocityvi has beencorrectedfor dragthat reduceslaunchvelocityvo. Pressureis in bars. Targettypesinclude
compactedpumice(P), no. 140-200 sand(S), no. 24 sandmixed with 8% Coyweight)pumice(PS), and and layer over compactedpumice
(S/P). Rim to dm crater diameter(DC) and "rampart"diameterare (DR) in centimeters.Asterisksindicateimpactsinto targetswith a
thin veneer of dry tempera.
enoughturbulencegeneratesa turbidityflow. In the laboratory
impact experiments,recoverywinds createdby the temporary
vacuum behind the ejecta curtain enhance an analogous
turbidityflow: the greaterthe pressuredifferential,the greater
the winds,the greaterthe powerin the turbulence.
This analogy
permits exploring the various modes of emplacementand
factorscontrollingejecta run-out in the contextof a physical
model.
Run-out by flow of emplacedejecta in the absenceof a
fluidizing agent(e.g., entrainedair, water,or releasedgas) can
energy created by the ejecta curtain moving through the
atmospheregoes to zero and where St is the run-outdistance
following this turbulent influx. Here the effect of basal shear
drag on the run-out is ignored. Under limited turbulence,So
shouldcorrespondto the distancewhere the suspended
load is
suddenlydeposited,therebyforming the ridge of coarserejecta
on top of the ballistically emplaced ejecta deposit. Under
enhancedturbulence,
run-outwill be extendedby a distance(St)
until energyin the flow has fallen below a critical value. As
discussed in more detail below, this distance could be further
augmentedby an autosuspensionprocess [Bagnold, 1962,
1963; Pantin, 1979] in which the moving ejecta cloud creates
S = So+ &
(2) more turbulence,therebyincreasingits power.
The distanceSo dependson the wind velocities(w) generated
whereSOis the distancetraveledwherethe supplyof turbulent by the ejectacurtainmoving at a velocity, Vc(X/R). At a given
be divided into two components:
SCHULTZ:ATMOSPHERICEFFE•S OF •A
E•CEMENT
11,643
So= (l/2)(w2[g)= kR
(3)
This result predicts that rampart distancescaled to crater size
should be approximately constant.BecauseR is calculatedon
the basis of late-stage gravity-controlled growth, it is
implicitly assumedthat it refersto craterradiushad it formedin
a vacuum(i.e., material flow createdby the impact without any
modification due to presence of an atmosphere). At low
atmospheric densities (low Reynolds numbers) turbulence
generatedby the moving ejecta curtain shouldbe minimal, as
graphically illustrated in Plate 2b for impacts under an
atmosphereof helium. Ejecta rampartsneverthelessform above
a critical pressure,and Figure 14a revealsthat equation(3) is
consistentwith this model of entrainmentand depositionfrom
curtain-generatedwinds.
As atmospheric density increases, turbulence increases
(Plate 1), and the distanceto rampart formationor run-out of
individual lobes increases. In the absence of continued influx,
the headof the densitycurrentcreatedby ejectaentrainedin the
turbulentatmosphere
behindthe ejectacurtainshouldinitially
continue to advance outward at a horizontal velocity u (if
frictional forces along the surfaceare considerednegligible).
Under these conditions, conservationof energy gives the
following relation:
1/2po
u2=(Pe- po)gh
Fig. 12. Quarter-space
experimentin whicha plexiglasssheetallowsonlyonehalf
wherePois the ambientdensity,andPeandh are thedensityand
heightof the turbulentejectacloud,respectively.This density
cloudshouldtravel a distancegiven by
of the craterto form and revealsvorticesbehindthe ejcctacurtain.Arrows indicate
circulationpattern(left) and hollow coreof outwardmovingveerrex
(right) created
by windsbehindthe advancingwall of ejectafrom a 1.6 km/simpactunder0.5 bar
atmosphere
(air). Vorticestemporarilyentraincoarserejectathataredeposited
in a
rampartas vortex winds fall below a threshold.V-shapedpatternabovecrater
represents
scouringof plexiglassby ballisticejectaduringcraterformation.Scale
bar corresponds
to 10 cm.
(4)
St= (112)(u2[g
)
(Sa)
= [(pdpo)- 1]h
(5b)
The height of the turbulentejectacloud at any distancefrom the
crater rim scaled to crater radius should increasewith ejecta
thickness,
whichin turn scalesas (gR)1/2 as discussed
by
TABLE6. Summary
ofEjecta
Emplacement
Styles
andProcesses
inLaboratory
McGetchin et al. [1973] and Schultz and Gault [1979].
Consequently,equation(2) now can be writtenas
Experiments
Style
Rampart
Figure
Observation Contelling
Interpretation
Examples
Variables*
la, 3,4a,4b, reworked
ballistic
a,P
eddy
suspen5, 6, 9
cjectaby windsbe-
coarscrclasts
eddy-driven
turbidity
cur-
surge
rent of entrain-
ed cjec•a
over-rides inner
ballistic facies
(andramparts)
Radial
lc
intense
basalvortices a, p,Rv
(6a)
= [Fe(g[R)
112
+ k]R
(6b)
sionof
hind ejectacurtain
Flowlobes lb,4c,5b,5c, deposition
from
a, p,Rv
9b
ground-hugging
x = {[(Pe/Po)
- 1][g[R]
112
+ k}R
or
scouringof
inner facies,
large run-out
where F e is defined as an entrainmentfactor. The distance
traveledbeforeforminga rampartthereforeis augmented
by a
factor dependent on the degree of ejecta entrained in the
recovery wind circulation.
The degreeof entrainmentdependson the decelerationof
individual ballistic ejecta as given by equation (1). Two
limiting conditionscan be derived.The first considersthe drag
coefficientto be constant;the secondassumes
thatCD ~ 1/Re:
depositionof
thin veneer in
outer facies
Fe- poRIbga
* Variables
correspond
toejecta
size
a,atmospheric
pressure
P,atmospheric
Fe~ PRø'Slõga2 broad
range
inR•
density
p, andcrater
radius
hadtheeventoccurred
in a vacuum
Rv.
stage of crater growth (x/R = const),Vc(X) is taken to be a
constant vc for a given crater size. Since the laterally
advancing ejecta curtain simply reflects the horizontal
componentejecta trajectories,w shouldincreaseas vc and
narrowrangein Re
(7a)
(7b)
where g is the viscosityof the atmosphereintroducedthrough
the Reynoldsnumber(Re = pva/•t), anda andõ are the sizeand
velocity of individual ejecta fragments.Velocity in equation
(1) and in the Reynoldsnumberhas been replacedby the
approximation
ve2~ gR, as requiredfor late-stage
gravitycontrolled ejection velocities. The eddy-driven rampart
hence(gR)1/2,whereR is theradiusof thegravity-controlled
transientcrater and g is the gravitationalacceleration[Post,
1974; Schultz and Gault, 1979; Housen et al., 1983]. If the
rampart ridge representsdepositionwhen the wind velocities
reducebelow a critical value necessaryfor suspension,
then the
distanceis simply given by:
distancepredictedfor different rangesof Reynoldsnumbers
becomes
x/R~ (po/õa)(R/g)
112
~ per1/2
for CD = COnst,
(8a)
11,644
SCHULTZ:ATMOSPHERICEFFECTSOF F,JECTAEMPLACEMENT
EARLY-TIME
PHENOMENA
[?
'••'••11onized
wake
'•
•
Eiecta cloud •-
High-velocity
\
•-'"•
.....
•
....
incandescent
eiecta
...................................
Fig.13.Processes
affecting
ejecta
eraplacement
atearlyandlatetimes
asinferredballistic
ejecta
curtain.
Targets
resulting
ina widerange
ofejecta
sizes
(factor
of
fromhigh-frame-rate
imaging,
changing
targetandatmospheric
conditions,
and 100)resultin a twocomponent
curtain:
largersizefractions
define
theclassic
special
experimental
designs
(quarter-space
experiments
andlayered
targets).
(a)At conicalprofLle
observed
undervacuum
conditions;
smallerfractions
become
earlytimestheionized
waketrailing
theprojectile
fillsthegrowing
impact
cavity. entrained
in anejecta
cloud(seeFigure
To).Theejecta
cloudisturbulent
atsmall
Interactions
between
theionized
wake
andhigh-velocity
ejecta
create
a turbulent
scales
butreflects
anatmospheric
circulation
atlarge
scales
induced
bytheoutward
disturbance
thatmovesup the curtain.(b) At latetimesevolution
of the ejecta movingejectacurtain.
curtainis largelycontrolled
by response
of theatmosphere
to theoutward-moving
x/R~ (l•p/•a)m(R
i/4/Sa)
~ (l•p)mRU4
Figures 14b and 14c show the range of possibilitiesgiven
by equation(8) for different Reynoldsnumbersand reveal that
the proposed scaling of the rampart is generally consistent
forCo ~ (1/R.)1/2,and
with the observations.Figure 15 correlates ejecta run-out
distanceswith formationof the crater cavity as revealedby
xlR~ (I.t/Sa2)(1/g)
1/2~ I.t
(8c) crater profile (diameter/depth). Greater run-out distances
characterizesmaller values of diameter/depth,i.e. arrested
forCO~ (1/Re)whereit is assumed
thatF,•(g[R)112
>> k andthat crater growth. Consequently,energy expendedin reducing
8, a, andg areconstant.
Equation(8a) accounts
for theobserved craterdiameteris neverthelessexpressedin enhancedturbulent
dependence
betweenrampartdiameterand atmospheric
density power of ejectaflows.
for given impactorsize and velocity revealedin Figure 10.
As atmospheric pressure increases, the intensity of
Small residual offsets, however, for different atmospheric turbulencebehind the ejecta curtain increases.The resulting
loadand
compositionssuggestthat a viscosity-dependent
drag eddy-drivenflow overridesdepositionof the suspended
coefficient(introduced
by a dependence
on the Reynolds channelizes into individual lobes (Figures lc and 5c).
number)alsomay apply.
Divergent flow within the flow head createscontinuedvortical
(Sb)
SCHULTZ:ATMOSPH]•IC •S
LATE-TIME
OFEIF.C•A EMIx..ACE•
11,645
PHENOMENA
Inwardfolding
..........
.-:..,'-'-.
-:-'--'...;•i
-'?;......
-•i•i•
air flow
/
'"•"•;-'-",•i?--'..:----...-'...
-';,,f•
......
•
•
•,..
!•'"•"-'•"'•'••••••.,".;>•i!•.-'.'.-.:•'.--'.:-.::..-;.'-.-'-•'"""':•'
...................
'•"•.-'•
-' ,
'
-
•...........................
"•'••••••'"•½"
--•'••"••':'''"
..................
Z"-"• --'
"•;•;-----•? cIosur.
'•'"•i•i•
, ""••••
;';'
•:"'""'"
'
•'
"•"••••:•;••l.'.'"•
'"••"•''•" .............
,,• .....
'•:;-::•:-'.'•.
.......
•
'•
'"
. •,
'
,,..•........
,"•"-':•----:•:•••,.'•:'
-";,•:"-.--:•.;.•.....•:•..:•••
.......'•'•-:•
"'I---•.
-'"'
Rampart
Fig. 13. (continued)
Pantin [1979] found that turbidity currents must either
motion, mixing with local material; eventually a rampartdecelerate, thereby depositing their load, or accelerateinto
bordered terminus forms as turbulent energy dissipatesßThe
autosuspension.
Conditions leading to autosuspension
were
observed ejecta lobes in the laboratory no longer grow
proportionallyand perhapsindicatean autosuspension
process given as
described by Baghold [1962, 1963] and Pantin [1979].
Autosuspension
is a feedbackconditionwherein suspended
material causesmotion, therebyproducingfurther turbulence
and more suspension.
In rivers and certainavalanches,gravity
andslopecontroladvanceof the densitycurrent.For impactsin
an atmosphere, movement down the radially thinning
continuousejectadepositspartly controlthis process.Gravity
also plays an indirect (but perhaps more important) role
throughthe impactingejectaat the baseof the advancingejecta
curtain. Since the ballistic ejecta curtain increasesin velocity
with increasing distance from the rim (Plate l a),
autosuspension
should be a natural consequenceas leading
ballisticejectaadd kineticenergyto the flow.
>1
(9)
whereex is an efficiencyfactor,[3is the slope,u is the flow
velocity, and vt is the fall velocityin the entrainedflow. The
efficiency factor represents the degree of excess power
available for suspendingthe load relative to the power
expendedalong the base. As the overridingturbiditycurrent
runs off the continuousejecta, • may be replaced by an
effective-gradient factor reflecting the increased outward
velocityin the ejectacurtainwith distance.Sucha factoralso
would go to zero, however, as the ejecta curtain thins
sufficientlyßThe basal surge observed at laboratory scales
11,646
SCHULTZ: ATMOSPHERIC•S
1.5
I
OF EJE•A EMPLA•
Degree of Entrainment
i
0.0
(D
c-
-0.2
.,.,•
•.0
-0.3
-0.4
i
-0.5
o::: 0.5
o
CD•, (1/R½)
1/2
nimal Drag
x
/
Curtain-wind
Driven Flow
-0.6
0.5
J
I
1.0
1.5
'
-0.5
-0.4
'
-0.3
'
-0.2
,
-0.1
I
0.0
i
0.1
0.2
Log(•p)1/2
R1/4
Log CraterDiameter(vacuum)
Degree of Entrainment
0.0
,0.1
0.2
0.3
,0.4
C D '" constant
I
0.3
0.4
I
i
0.5
0.6
i
0.7
I
0.8
i
0.9
1.0
Logp R]/2
Fig.14.Theeffect
ofcrater
size
and
atmospheric
conditions
onrun-out
ofejectashould
depend
on(pp)l/2Rvl/4.
Asshown,
viscosity
and
density
are
referenced
to
ramparts.
(a) Distance
fromcrater
rimtorampart
asa function
of crater
diameterairatonebarwithcrater
radius
givenin centimeters.
(c)Higher
relative
Reynolds
hadtheimpact
formed
in a vacuum.
DataareforverysmallReynolds
numbersnumbers,
allwithabout
thesame
value,
thereby
result
ina constant
drag
coefficient
withminimal
drag(viscous
effects).
(b) Comparison
of rampart
run-out
distance(equation
(Sa)).Arrows
indicate
direction
of additional
corrections
fordatawith
scaled
tocrater
diameter
(vacuum
conditions)
asafunction
ofentrainment
asgiven slightly
lower
Reynolds
numbers,
thereby
having
slightly
higher
drag
coefficients.
inequation
(8).Entrainment
depends
onthedragcoefficient,
which
alsodepends
on Density
(p)isreferenced
to•ir at1barwithcrater
radius
given
incentimeters.
See
theReynolds
number.
Figure14bconsiders
a rangein Reynolds
numbers
over Table5 forlisting
ofempiric•l
data.
which
itisassumed
that
CO~ (1/Re)
1/'2
. From
equation
(Sb),
therun-out
distance
indicatesthat a large ex due to increasing turbulence, under 0.25 bar pressure at much higher angles (60 ø from
entrainment,and small vt all can contributeto excessiverun-
out. The possiblerole of increasedturbulenceis illustratedby
the effect of addinga thin veneerof temperaover compacted
pumice,whereasthe role of entrainmentandsmall vt is clearly
demonstrated
by the long distallobesfor impactsinto a target
of low-densitymicrospheres
(Figure5c). In the latterexample,
the fanlike
terminus
characterizes
sudden termination
of
channelizedautosuspended
flow. Conversely,the absenceof
flowlike lobesin coarsergrainedtargetsor mixtures(Figure 6)
reflect fall velocitiesof coarseejectathat are too high to meet
this condition.
horizontal) relative to vacuum conditions (5ø). The ionized
wake and point of impact clearly establishthat the impact
angle has not changed [see Schultz and Gault, 1982]. The
enhancementof ejecta asymmetry can be attributedto three
causes.First, the developmentof an ionized wake trailing the
projectileat higher atmosphericpressurescreatesan early-time
turbulent barrier for uprange ejecta [see Schultz and Gault,
1982]. Gases within the projectile wake are observed to
continueto impinge on the impact cavity even after contact.
These two nonaxisymmetric processes result in an
exaggerationof the eraplaced ejecta• pattern at modest
Obliqueimpacts. The variousmodesof ejectaemplacernent atmosphericpressures.Second,despitea relatively symmetricshould not fundamentallychange with impact angle: eddy- appearing crater and ejecta distribution for impacts into
suspendedand autosuspended
flows resultingin rampartsor particulates between 15 ø and 60 ø, there is considerable
in the evolutionof the ejecta•curtain
undervacuum
lobes should and do (Figure 4) develop. Nevertheless, a asymmetry
distinctiveasymmetryand even the butterflypatterndevelop conditions[e.g., see Gault and Wedekind,1978; Schultz and
SCHULTZ:ATMOSPtIERIC
•S
Diameter/Depth
4
i
0.0
,0.1
0.2
OFEJEC•AFAVlln•ACEMENT
11,647
both the outwardmoving ejecta curtain and the ejectaparticle
moving throughthe atmosphere.
Scaling laboratoryresults. Impact cratersin the laboratory
can be scaled over many orders of magnitude through
dimensionlessscaling parametersas discussedby Holsapple
and Schmidt[1982]. Their derivedprincipal scalingparameter
can be viewed as Froude-scalingwhereinertial forcesare ratioed
to gravity forces. An explicit assumptionis made that details
of the transferof energyfrom impactorto targetat early times
can be ignored at late times when gravity controls crater
growth.The use of looseparticulatetargetsis implicitly based
on this assumptionwhere the initial passageof the shockwave
converts solid target material into broken debris. Shockbroken target material may be further comminuted by the
shearingactionresultingfrom the redirectedmaterialflow field
o
-0.3
[e.g., Schultz and Mendell, 1978] or perhaps vaporization
[Wohletz and Sheridan, 1983]. Froude-scalingpredicts that
late-stage ejection velocities and time should scale as the
O.5
O.6
O.7
square root of linear dimension and gravity as derived in
various
studies [e.g., Post, 1974; Schultz and Gault, 1979;
Log Diameter/Depth
Housen et al., 1983] and as usedin derivationsin the present
Fig. 15. Rampartrun-outdistanceobservedin laboratorycratersscaledto the
study. Table 7 explores the implications of this scaling
observed
craterdiametercCanpared
with crateraspectratiofor compacted
pumice
principle on late-stage ejection velocities and representative
targets.Compacte•pumicewas foundto preservethe final cratershape,whereas
peak pressuresover a wide range of crater sizes on Mars. For
craters
in sand
collapsed
'uhmediately
afterformation.
Bothatmospheric
pressure
anddragwerefoundto •ffect cratershape[Schultz,1990b].The correlation
here reference, an impact producinga 16-cm-diametercrater in the
suggests
thatthedevelopment
oftheejecta
rampart
isrelated
tothesame
proceaseslaboratorywith late-stageejectionvelocitiesof 50 crrdswould
and that energylost to craterformationis represented
in atmospheric
turbulence correspondto velocitiesof 77 rrdsfor a 10-kin-diametercrater
leadingto greaterran-outof the ejectarampart.SeeTable 5 for listingof empirical (apparent diameter of transient crater) on Mars in order to
maintain the same relative radial distributionof ballistic ejecta
(without atmosphericeffects).
Interactionsbetweenejecta and the atmospherealso require
Gault, 1979, 1985]. The downrange
ejectacurtainadvances
at a
scaling
inertial forces to viscous friction forces, that is, the
higher velocity and •containsmore ejecta than the uprange
curtain.
Consequefitl•,
bothejecta
rampart
andrun-out
lobes Reynoldsnumber, in order to extendthe model. It is desirable
shouldextendto greaterdistances
in the downrange
direction. to have the Reynoldsnumbermatchedat laboratoryand actual
but this is not possible for impacts in the laboratory
Third,oblique
impac
tsresultin lowdownrange
ejection
angles scale
since two interaction scales must be considered. The movement
andhigh uprangeanglesundervacuumconditions.
Introduction
of an atmosphere
resultsin verticalejectioncurtainangles, of the curtain of ejecta throughthe atmosphereestablishesone
therebyamplifyingaSymmetries
in emplacementdistribution. scale of interaction; the interaction of individual ejecta
At high atmospheric
pressures
and low impactangles,the comprising this curtain with the atmosphere establishes
emplacement
proceSS
becomesdisruptedby two additional another.In laboratorystudiesof dust avalanches[cited by
processes:projectile ricochet and effects of vaporization. Melior, 1978], sufficiently high Reynolds numbers allow
Impactanglesless•an 30øresultin ricochetof largefractions Froudescalingprovidedthat the relativedensityof the flow and
of theprojectile
at highvelocities
[GaultandWedekind,
1978; ambient medium is maintained. For this scale interaction, the
Schultz
andGault,1990]Asgraphically
revealed
in highframe- laboratory impact experimentsprovide a reasonablemodel
ratephotography
[see.Schultzand Gault, 1982], the downrange sincethe Reynoldsnumberis typically well over 1000 and the
relative densities approachwhat might occur in impacts.
Interactions at grain-by-grain scale, however, are clearly
ricochet createsa strongturbulentwake and wind that draws
considerable
ejectadownrange.Thermaleffectsfrom this wake
and from impact-induced vaporization create significant
thermalturbulencebehindthe downrangeejecta.As a result,
projectileand targetmaterialare drawndownrange,even at late
violated
but
can be accommodated
to first
order
if
such
interactions are significant only in establishing onset of
entrainmentin the flow. A similarapproachwas suggested
by
times.At laboratoryscales,thesetwo processes
significantly Mellor [1978], and the relevantscalingparametercan now be
given by the dimensionlessdrag parameter (equation (1)),
disruptthe eraplacementprocess.
which
6. PosSmLE IMPLICATIONS
Althoughthe laboratoryimpactcratersare clearlyat a small
scale,they neverthelessreveal phenomenaand processesthat
can be scaledto the planets.Futurework will explorespecific
aspectsin greater detail; the following discussionconsiders
selectedtestablepredictionsfor ejectaemplacementstyleson
Mars with implicationsfor the Earth and Venus. Interactions
betweenlate-stageejectaand the atmosphere
requireevaluating
a variety of dimensionlessscalingrelationsdepictingrelative
forces in order to maintain similarity betweenthe laboratory
"model"and planetaryscaleprocesses
[e.g.,Langhaar, 1951].
The relevantscalingratios includethe Froude(inertia:gravity
forces), Reynolds (inertia:viscous friction forces), Euler
(pressure:inertiaforces), and Weber (inertia:surfacetension
forces) numbers.These scaling ratios must be consideredfor
establishes
critical
conditions
for
entrainment
as
discussed
by Schultzand Gault [1979].
Euler scalingrequiresmaintainingthe sameratio of pressure
forcesto inertialforces(P/•v2) for dynamicandkinematic
similarity. This scaling requirementis often implicitly used
[e.g., Chabai, 1977; Holsapple, 1979] to argue that the
advancingejecta plume overwhelmsambientpressureat large
scales,thereby minimizing atmosphericeffects on scaling.If,
however,ejecta sizesare sufficientlysmall, aerodynamicdrag
can result in considerabledynamic interactionsthat require
replacing static pressure with a dynamic pressure term.
Increasing
cratersize(i.e., •ve2) requires
thatthepressure
differential(and resultingwindsor turbulence)mustincreasein
order to maintain the same scalingratio.
Entrainment of ejecta by winds and turbulencefurther
requiresreconsideringthe Froude,Reynolds,and Euler numbers
at the scale of individual particles.These ratios are combined
11,648
SCHUL•
ATMOSPHERIC•S
OF•A
••
TABLE 7. Effect of CraterSize (Relativeto 0.5 inn)
Apparent
Excavation
Observed
DiameterDiameter
1
Ejecta
o.•
0.6
1.0
5.0
lO.O
•0.0
100.0
1.2
6.2
17
•3
166
Maximum Median
1
i
i
1.4
3.2
4.$
10.0
14.0
1.6
4.6
7.4
22.0
34.0
?
?
?
?
?
Peak
Shock Dynamic Atmospheric
1
'l
3
50
160
•00
s20o
2
10
20
lOO
2oo
1.0
0.7
0.$
o.oo2
1Observed
rim-• diameter
DOisderived
from
excavation
apparent
diameter
DEbyassuming
that
DO--1.25
DE.For
craters
largerthan5 kin,rim/Wall
failure(slumping)
enlarges
thecrateranadditional
33•, thatis,DF = 1.33Do.
2Scales
approximately
asDE1/2forcrater
excavation
diameter
DE.
3Scales
asDE2/3'
4Depends
onUthology:
aeolian
dust
(<16
!tm),
loess
(60Itm),
moving
sand
(200
I•m),
fluvial
transport
(~<10ram),
shock-broken
basali(0.1-20cm).
5Scales
approximatedy
asDE1'7forthe
same
stage
ofcrater
growth.
6Scales
as(ejecta
velocity)
2forgiven
atmospheric
pressure
(nobaRistic
shadowing).
7Assumes
exponential
decrease
ofatmospheric
density
with
altitude
•esults
inadecreasing
atmmphcric
effect
as
exp(-•DF/H)), wherek = 0.66,DF is finalcratersize(for33%erdargement
of excavation
crater),andH is atmospheric
scale
heightof 10 kin.
i
I
i
100km
•
10.0
I
MARS
.......
..C.a.r.•.?.
a.c.e..
o.u.s.
...............
.•. ................................................
chondrite
10km
D
1.0
•
0.1-
•-
0.01
1 kin
--Surface
tension
for
i mm
water
drop
=6Xi0bars
0.001
0.2
0.4
0.6
TRANSIENT RADIUS/CRATER
0.8
1.0
RADIUS
EARTH
VENUS
i
•
1000
i
i
i
_•...0.1
•.•e2p•
in•z.?•
•rC.h•t•
t•ri=t
.e
..........
•
_
100,000
•.Oiivine-bronzite
chondrite
................
.•.....
100
km
'•.
10,000
10km
1000
•
lO
•
I
..
-
.,o.
.:.:.:.e.
.......
. . ..........
I
0.2
0.4
0.6
0.8
100
1.0
TRANSIENT RADIUS/CRATER RADIUS
Fig. 16. Dynamicpressure
experienced
by cjectaencountering
an ambientMartian pressures.
Shock-weakened
ejectashouldbe furtherfragmented
dtuingejection(or
(top),tenmu• (bottomleft scale),andVenusJan
(bottomrightscale)atmosphereduringreentryif theyescapethe impactregion).Notethatliquidwatershouldbe
at diff•t
stagesof cratergrowthfor diffe•nt diameterexcavation
craters.This atomizedand dispersedas cratersexceedabout5 km in diameteron Mars.
idealizedrepresentation
is basedon scalingejectionvelocities% calculated
frt•n Moreover,dynamicpressures
experienced
by ejectafrmn largecraterson Earth
numericalcodesdescribedby Orphal •t al. [1980] to largercraterswith the (henceVenus)shouldexceedthe strengths
of shock-weakened
debris,thereby
assumption
that v• ~ (gD)112at the samestageof growth.Differentstagesof augmenting
entrainment.
growthprovidea measureof the fractionof ejecta subjectto high dynamic
SCHULTZ:ATMOSPHERIC
EFFF.
t•S OFFJF.
t-'FA•CEMENT
11,649
TABLE 8. ParticleSizesThat CanBe Suspended
in WindsCreatedby anOutwardMovingEjectaCurtainon Ma•s
Crater
diameter
1,km
Critical
Particle
Sizefor
Onsetof Eraplacement
Style
4
Wind
Excavation
Final
Velocity
2,
(A•arent) (Rim-Rim)m/s
0.5
1.0
5
10
25
50
0.6
1.2
6
17
42
83
12
17
38
54
86
121
Suspended
Particle
Size
3,•
54
77
170
240
380
540
Rampart, Flow,
mm
!
2
10
20
50
100
mm
0.1
0.2
1
2
5
10
Radial,
•n
50
100
2OO
400
1000
2000
1Diameter
isbased
onapparent
diameter
ofexcavation
crater.
Final
crater
diameter
requires
a25%increase
toprovide
the
observed
xim-rimdiameterfor craterssmallerthan5 ]cm.Craterslargerthan10 krnincludean additional33% enlargement
dueto rim/wall collapse.
2Values
provide
af'•rst-order
rneasure
ofrecovery
winds
and
iscalculated
onthebasis
ofthehorizontal
velocity
component
of ejectaascratarapproaches
80% of its final size.
3 Size
corresponds
toparticles
withterminal
velocities
that
match
thecalculated
wind
velocity
foranatmospheric
density
at
thesurface(assumed
CD = 1.5).
4 Sizes
arebased
ondirect
extrapolation
frcrn
laboratory
experiments
(equation
(10))and
represent
combined
effects
of
dragdeceleration
on ejectaandentrainment.
TABLE9. Approximate
Onset
Diameters
forEraplacement
Styles
inDifferent following crater formation serves to underscorethe fact that
Lithologies
large ballistic massesdo not overwhelm the atmospherebut
can be aerodynamicallydeceleratedthroughturbulenceacting
Onset Diameter
Airfall
Windblown Fluvial
Basalt
• on the constituentparticles.
On the basis of such scaling considerations and the
St•le
(<100
tun) (100-300
•.m) (0.4-1
nun)
crn laboratory results, processes affecting late-stage ejecta
Rampart
50m
0.25km
0.75lun
5km
Flow
0.5km
2,5km
7.5km
(50kin) emplacement on Mars can be considered.Theformation of
Radi•!
1km
5km
15km
(100km,) ejectarampartsis interpretedas the least degreeof entrainment,
Based
ondirect
extrapolations
fromlaboratory
experiments
(equation
(10)).
and onsetof this style shouldbe constrained
by equation(1). If
t Atvery
large
sizes,
firrite
scale
height
oftheatmosphere
becomes
important; craters are too small, ejection velocities and curtain-induced
derived
onset
diameters
(inparentheses)
areincluded
onlyforcompleteness.
vortices are too low to modify ballistically emplaced ejecta.
Appearanceof rampart-styleemplacementshouldbe dependent
when drag forcesdue to aerodynamicdrag are scaledto gravity on local substrate properties. Equation (1) predicts that
forcesas indicatedin equation(1). In order to maintaina given comparableejecta entrainmentin the laboratoryand on Mars
ratio of drag-to-gravity forces for larger ejecta particles, requires
however,dragforcesmay exceedtheir internalstrength.Figure
16 provides a summaryof the maximum expecteddynamic
Rst= (COr/COM)(pZJpSt)(aM/aœ)(Rœ)
(10)
pressuresfor ejecta encounteringan atmosphereat different
stages of crater growth for Mars, Earth, and Venus. Such where the L andM subscriptsrefer to laboratoryand Martian
conditionsare idealizedsincecomponentejectatrajectoriesand conditions, respectively. Equation (10) indicates that
"attack angle" within the ejecta curtain changewith time [see substratesleading to millimeter-sizedejecta would have onset
0.5 km (CD.= 25; CD. = 0.8;
Schultz and Gault, 1979] and since a dynamically evolving diameters(rim-tim) approaching
to 60mbof air;aL = 80I.t).•As
craters•exceed
boundarylayer must developin the advanceof ejecta curtain. PLcorresponding
Nevertheless,Figure 16 establishesthe possiblerole of further 1 km, gravity-scalingdominatesand minimum ejecta sizestend
ejecta disruption and potential for entrainment in the to approachthe averagesize of the componentmineral grains
atmosphericresponseto ejecta curtain. Since water is often in the impacted substrate, that is, the inherent flaw size
proposedto accountfor the apparentfluidity of Martian ejecta discussed by Curran et al. [1977]. Impacted lithologies
[e.g., Gault and Greeley, 1978; Greeleyet al., 1980; Carr et al., characterizedby windblownmaterials,however,shouldexhibit
1977], the ratio of inertial to surface tension forces (the Weber significantly smaller onset diameters of nonballistic
number) also becomesrelevant during ballistic ejection. For emplacementstyles.
illustration, ejected water dropletssmaller than 1 mm across
Table 8 provides first-order estimates of particle sizes
would be dispersedand atomized due to dynamic pressures potentially entrainedby winds createdby an advancingejecta
exceeding
theirstrength
(6 x 10'4 bars)overmostof crater curtain
excavationfor craterslarger than about 1 km in diameter.Such
ejecta would subsequentlybecome particularly sensitive to
aerodynamicdrag and entrainment.Moreover, volatization of
this componentshould further enhanceentrainmentof solid
ejecta. In this view, distantejecta run-out is enabledby water
acting as finely dispersedejecta,rather than as a "splash."
for different
size craters on Mars. The outward curtain
velocity is calculatedon the basisof the horizontalcomponent
of ejecta velocities at late stagesof growth in a particulate
target (see Plate lb). For this calculation,an ejecta velocity of
50 cm/s corresponds
to late-stagegrowthof a crater 16 cm in
apparentdiameter[seeSchultz, 1992]. Ejectaparticlesreduced
to terminal velocities are readily entrained in winds and
Theinteraction
between
an atmosphere
andanensemble
of turbulence,
thereby
producing
nonballistic
ejectaeraplacement
ballistic particulatesor water at much greater scalescan be
illustratedby airbornefire-fighters.Over 6 tonsof materialcan
be droppedduring a singlepassat 100 m/s. Aerodynamicdrag
actingon individualparticlesresult in rapid decelerationof the
entire mass, and fire-fighting techniquesrequire low-altitude
release of the extinguishing medium in order to lessen
disruptionand dispersal.Delivery of water results in a similar
response. This analog for a parcel of the ejecta curtain
styles. Such conditionscan be met for 1- to 2-km-diameter
craters only if the substrateis dominatedby airfall deposits
with particle sizes less than 100 I.tm. Such substrateswould
resultin craterejectawith a radial scourpatternand basalejecta
flow (Plate 1, bottom) based on extrapolation from equation
(10). Craterssurroundedby contiguousejectaramparts(Figure
l a) would be expected in lithologies characterizedby much
coarserdebris (e.g., sand size).
11,650
SCHULTZ:
ATMOSP•C EFFECWS
OFF.JEC'•A•C'EMENT
Fig. 17. Comparison
of cjcctaemplacemcnt
stylesaroundsmallcraters(<5 kin) in
contrastinglithologieson Mars.(a) Three-kilometer-diameter
crater formed in a
basalticplain (13.8'N, 151.$'W; Viking frame 693A03) westof OlympusMons
0eft) and in an 0.8-kin-diametercraterin sedimentarydepositsnear the northern
highland/lowland
border(47'N, 346'W; Viking frame458B71). The craterformed
in basalt resemblescratersof comparablesize on the Moon, whereasthe crater
formed in sediments exhibits long ejecta run-out lobes (arrows). From
extrapolationof laboratoryexperiments,the ballisticejectaernplacement
style at
left is consistent
with modalejectasizesexceeding10 cm, whereasthe multilobed
patternat rightis possiblefor a targetcomposed
of fine-grained(<100 grn) debris.
Scale bar indicates1 kin. (b) Five-kilometer-diametercrater in mottled phins
(43.4'N, 134'W; Viking frame 9B35). The anomalouslylarge secondaries
around
such a small primary crater are believed to indicate a porous, fine-grained
lithology.Impactsinto uncompacted
pumiceproduceunusuallylargeejectaclumps
that are relatively unaffectedby aerodynamicdrag. Such a lithology would be
consistentwith gentlyaccumulated
air fall deposits,whichalsocanbe inferredfrom
mutedterrainsand craters.Scalebar corresponds
to 10 km.
Figure 6 revealed the importance of finer constituent
fractions (less than 10%) entraining large ejecta.
Consequently,in the absenceof other mobilizing substances,
lithologies leading to bimodal ejecta size distributions(e.g.,
basalts over sediments) may produce craters surroundedby
contiguousejecta ramparts. Substratescharacterizedby wellsorted finer material (wind, fluvial deposits) should exhibit
craters with ejecta emplacement styles progressing
systematicallyfrom rampart styles at the smallestdiametersto
more fluid lobate emplacementstyles as turbulenteddies and,
eventually, the autosuspensionprocess create greater run-out
distances of channelized flows (see Table 9). As this process
increasesin importance, the continuousejecta depositsnear
the rim shouldshow increasedevidencefor scouring,including
traversingradial grooves and near-rim depressionscreatedby
the strongtoroidal winds. Further variety in ejecta morphology
can result from variations in target porosity, as indicatedby
impactsinto uncompactedpumice.
SCHULZ: ATMOSPIIERICEH:EC•S OF E1EC•AEMPLA••
11,651
Fig. 18. Comparison
of ejectaeraplacement
stylesat differentsizesfor impacted
lithologieslikely composed
of (a) lava and(b) sediments.
Cratersin the ridged
plainsof HespeHaPlanurnexhibitprogressively
morecomplexejectafacieswith
increasingsize (Figure 18a). The 10-kin-diametercraterat left (28.3øS,240.2øW;
Viking frame418S39)is surrounded
by a contiguous
ejectarampartextending
abouta crater radiusfrom the rim. The 15-kin-diametercraterat right (30øS,
241'W; Viking frame418S40) exhibitsan inner continuous
faciesout to a crater
radius with rampan-bordered
lobes extendingto two crater radii. Although
frequentlycitedasevidencefor increased
volatilesexcavated
fromdepth,thelarge
ejecta run-out is also expectedfor increasedatmosphericentrainmentdue to
ejectionvelocities
increasing
withsize• excavation
of fine-grained
debrisat depth.
Scalebar corresponds
to 10 kin. Cratersat highlatitudesexhibitgreaterrun-out
distances
withoutrampan-bordered
terrnini(Figure
18b).Availableresolution
often
preventsdetailedcharacterization
of ejectafaciesas around9-kin-diametercraterat
left (Viking frame676B13) but scouringof preexistingterrainsout to 10 crater
radii is common.The 8-kin-diameter
craterat rightexhibitsbothscouring
and
multilobedfacies(73øN,232øW;Vikingframe676B33).The innerfaciesextendto
1.2 crater radii from the rim, whereasthe outer continuousfacies extend over three
craterradii. Additionally,scouringextendsto nearly 11 craterradii with faint
rampart termini. The inner continuousfacies is believed to indicate ballistic
ernplacement
(possiblywith posfformationHow), whereasthe multilobedfacies
reHectsdeposition
froma turbiditylike
Howcreatedby atmospheric
entrainment
of
veryf'me-grained
(<100•zn)airfall
deposits.
Release
ofnear-surface
volati]es
by
frictional
heating
duetocolliding
ejecta
would
enhance
thisprocess.
Theouter,
scoured
faciesmayreHectthe effectsof early-time
vaporblast.Scalebar
corresponds
to 5 krn.
Even thoughthe variousdimensionless
scalingratiosmay
allow extendingselectedaspectsof the crateringprocessto
muchbroaderscales,the partitioningof impactorenergyat the
1-6 km/s velocities available in the laboratory might
fundamentallychangethe atmospheric
environmentaroundthe
impactat the 10-30 km/s on Mars or Venus,therebyaffecting
ejectaeraplacement.
In the laboratory,energytransferredto the
atmosphere
eitherby the projectilebow shockor duringimpact
was consumedduring the earliest stagesof crater growth as
discussed
in a previoussection.A first-ordercomparisoncan be
made between the response time of the atmosphere and
responsetime of target. Impactor energy is coupled to the
atmospherethroughthe accompanyingbow shock,high-speed
ejecta, and vapor cloud expansion.If it is assumedthat the
specific mode of energy transfer to the atmosphere is
eventuallylost at late times, then an analogycan be made with
strong explosions in an atmospherewith a source energy
assumed to be a fraction of the initial impactor energy.
Analytical expressions
have been derivedto treat this problem
[e.g., Taylor, 1950]. The intenseblast with an energy EA
rapidly expands hemispherically with atmosphericpressure
returningto ambientconditionson a time tE given by:
tE--(l'5x10' 6)E• 1/3Po1/2
(11a)
11,652
SCHULTZ:ATMOSPttERICEFFECTSOF F2ECI'AEMPLA••
Fig.19.Comparison
of eroded
ejectafaciesindicative
of contrasting
sizefractions.degraded
faciesbmpreserved
rampart.
Thissurvival
is alsoconsistent
withlarger,
(a andb) Ejectaramparts
generally
remain
asthelastsurviving
relictsof theejecta moreresistant
ejecta
clasts.
(c)Craters
in theCasius
fractured
plains(360N,2580W;
facies.
Figure19a(140N,159øW)
shows
survival
of distalejecta
rampart
yetlossof Vikingframe538A04)wheretheouterfaciesarenearlyuniformly
andcompletely
otherejectaflow textures.
Superposed
pedestal
craters
indicatethatregiononce removed(arrows).This erosionalstyleis consistent
with outerejectafacies
mayhavebeencoveredby a surfacelayer,nowlargelyremoved.
Contrastingcomposed
of f'mergrained
debriseraplaced
asa thinunitfroma turbiditylike
flow.
survival
of rampart
is consistent
witha lessmobRe
fraction
composed
of coarserFinergrained
ejectawouldbemoresusceptible
to aeolian
removal
under
present
debris.Scalebar corresponds
to 20 krn; •r•kingframe886A-11.Figure19b Martianconditions.
Scalebar
corresponds
to 10kin.
illustratesa smallercrater4 km x 5 km (7.4øS,313øW;Viking frame751A15) with
the atmosphere,respectively,under ambientconditions.If it is
with a distanceRE,
=
in
t,) assumedthat the energyfractioncoupledto the atmosphere
wherec andPo indicatethe ambientsoundspeedanddensityof
such a blast is 10% of the initial impactorenergy,then the
blastzonefor a laboratory
experiment
(mp= 0.376g andv = 5
SCHULTZ:ATMOSPHERIC
EFFF.•S OF•A
•CEMENT
11,653
.
.
..
.-..
.•?-.-'-:.:....
:'j
it:;'
•-•...ti....
Fig. 19. (continued)
west of Olympus Mons. The mode of ejecta emplacementis
essentiallyballistic with hummocky facies resemblingcrater
ejecta on the Moon. In contrast,the right half of Figure 17a
shows crater 0.8 km in diameter formed in distinctly
sedimentarydepositsnear the lowland/highlandcontact. This
smaller crater exhibits a nonballistic style of ejecta
emplacementwith long run-out lobes. From equation(10) and
Table 8, such an emplacementstyle at such a small diameter
would require in situ particle sizes less than 80 [t m
correspondingto loessor aeoliandust. Figure 17b illustratesa
further possiblecontrastin eraplacementstyle characterizing
lithologies at high latitudes. The bulbous inner ejecta facies
with a near-rim moat suggestshigh ejection anglesreflecting
affected.
aerodynamiceffects. The ubiquitouslarge secondarycraters,
Suchestimatesimplicitly assumethat the energyfractionis however, usually surroundmuch larger crater diameters(>30
coupledto the atmospherewithout any moderatingeffectsof km). On the basisof laboratoryexperimentswith uncompacted
the growingcratercavity. Experimentsreveal that the energy pumice, the contradictoryinferencesof strong aerodynamic
fraction coupled with the atmospherestrongly dependson effects yet large secondariescan be reconciledfor a surface
impactangle.At high impactangles(>45ø from the horizontal) composed of a high-porosity, fine-grained lithology where
a large fraction of the energy represented by impact large ejecta assemblagesare created.Impacts of suchclumps
vaporizationis containedandredirectedupwardas a jet rather into poroussubstratewill producelargecompression
craters.
than as a point-sourceatmosphericexplosion [Schultz and
The ridged plains on Mars are widely believed to be
Gault, 1979, 1982;O'KeefeandAhrens,1982].At lowerangles analogousto the basaltic maria on the Moon and shouldresult
kin/s) would be limited to about 8 cm in 0.04 ms. Hence a 30-
cm-diametercraterin a particulatetargetformingin about 150
ms would be largely unaffectedby the blast during late-stage
ejecta eraplacement. Early-time interactions with ejecta,
however, should be expected and are observed[Schultz and
Gault, 1979, 1982]. For comparison, a 10 km-diameter
(apparentexcavationdiameter)crater on Mars would take about
30 s to form, but the atmospheric
disturbance
would last only
abouta secondwhenthe blastfronthadexpandedto nearly100
km. Even if considerablevaporizationhad occurredwith 50%
of the impactor energy coupled to the atmosphere, the
equilibrationtime would increaseonly 23%. Consequently,
late-stage ejecta emplacementshould not be dramatically
(<30ø),the transferof energymay be coupledmoreefficiently in largest ejecta sizes for a given size crater. Rampartterminatedejecta depositsin the ridged plains of Hesperia
but is offset downrangefrom the crater [Schultz and Gault,
1984, 1990]. A future paper will assessin more detail the
relative efficiency of atmosphericcoupling as a function of
impact angle and vaporization.
Planum exhibit a general increase in run-out distance with
increasingcrater size (Figure 18a). At the smallestsizes(<5
kin), the rampartforms a contiguous
ridge on top of the inner
ejectadeposits.At largersizes,the rampartsform the terminus
Ejecta emplacement on Mars. Contrasting ejecta
eraplacementstyles at small crater diametersin contrasting for lobesextendingover (Figure18a, left) andbeyond(Figure
lithologieson Mars are shown in Figure 17. The left half of 18a, right) the inner continuousdeposits.In contrast,the
Figure17a showsa 3-kin-diameter
craterformedin lavaplains stippledplains at high latitudesmay be coveredby a fine-
11,654
•TZ:
ATMOSPHERIC
E•S
grained sedimentarycover as indicated by the high erosion
rates [Arvidson et al., 1979; Schultz and Lutz, 1988]. Figure
18b illustrates the more fluidized-appearingejecta typifying
craters formed in this lithology. Insufficient image resolution
commonlypreventscharacterizationof the emplacementstyle
of distalejectaaroundsmallercraters(Figure 18b, left), but the
bulbous inner facies and motfled outer depositsare suggestive
OF•EJECTA
EMIn.•CE•
Figure20a (e.g., Figures5b and5c). If the effectivedensity(Pc)
of the ejecta flow greatly exceeds the ambient atmospheric
density, (P o), then the flow can be characterized as an
atmosphere- or gas-entrained debris flow. This condition
should apply to Mars.
Figure 20b shows data for pristine craters on the ridged
plains of Lunae Planurn and on the sedimentary northern
plains. The upper limit of the ejecta run-out on Lunae Planurn
of a nonballisticemplacementstyle. The 8-km-diametercrater
D 1/2dependence
for eddy-driven
flow
at right (Figure18b), however,clearlyexhibitssuccessive
flow matchesthe expected
Horner and Greeley[1982] found
lobes extending to much greater distancesthan lobes around mechanism.For comparison,
comparablysizedcratersin the ridgedplains(Figure 18a, left). that the diametersof the outerejecta lobes aroundcraterslarger
asD 1-47.Whenexpressed
as run-out
Additionally, pristine examplesexhibit evidence for scouring than30 km increased
and thin distal ejecta lobes extendingas far as 11 crater radii distance from the crater rim and scaled to crater diameter to
from the rim (Figure 18c). In contrast with Figure 17b, match conventionin the presentdiscussion,the scaledrun-out
increases
asD0-47,againconsistent
with eddy-driven
secondarycraters are less frequent aroundcraters with such distance
flows. Because considerable local variation in coarse fractions
fluidized facies[seeSchultzand Singer, 1982].
The style and rate of destructionof their contrastingejecta can exist (basalt thickness, overlying sediments, etc.),
facies
are
also
consistent
with
the
inferred
modes
of
emplacement. The oldest rampart-bordered ejecta facies
typically exhibit nearly complete removal of the flowlike
textures yet preserve the distal rampart. If the rampart
representslarge clasts(centimeterand larger)entrained in an
eddy-drivenflow, then they would form a wall like barrier of
erosion-resistant
debris(Figures 19a and 19b). In contrast,the
oldest craters in inferred sedimentary deposits quickly lose
their outer facies (Figure 19c). At high latitudes, such craters
commonly evolve into pedestal craters as aeolian processes
first remove the surroundingwind-sensitivesubstrate.For
surface wind velocities of 2 m/s, wholesale removal of the outer
ejecta facies would suggestwell-sorteddebris with grain sizes
between 50 I.tm and 200 I.tm. Broad regions surroundingthe
impact are more resistantto aeolianreworking,possiblydue to
scouring from strong (300 m/s) vapor blast winds (Schultz,
1988), as well as depositionof very fine fractions(<10 Ixm)
behind autosuspended
ejecta flow closerto the crater.
The proposed role of atmospheric effects on ejecta
emplacement can be further tested for consistency by
comparingthe observedwith the expectedflow distancesas a
function of crater size and impactedlithology. From equation
(3), wind-driven ballistic emplacementresultingin a near-rim,
contiguousrampart shouldresult in a rampart distancethat is
proportional to crater diameter. As turbulent eddy-driven
entrainment becomes important (equation (8a)), rampart
distance scaled to crater diameter (D) should progressively
considerable dispersion should exist for craters in Lunae
Planurnascratersizedecreases
as shownin Figure20b. A factor
of 2 in average ejecta size could account for the observed
dispersionin rim-rampartdistancefor craterssmaller than 10
km in diameter.
Because run-out distance should be scaled to the
excavation crater, varying degrees of crater enlargementby
rim-wall failure near the simple-complex transition (3-8 km)
will offset data downwardto the right, therebyalso increasing
scatter
at smaller
sizes.
If
some
of
the
observed
scatter
indicates variations in atmosphericdensity through time, the
observed run-out distances indicate that such variations did not
exceed a factor of 2 or 3. Run-out distancesfrom craterslarger
than 20 km in diameterin the northernplains at high latitudes
closely match data for Lunae Planum, but smaller craters
exhibit enhancedrun-out distancesindicative of ejecta sizes a
factor
10 smaller in size.
It is importantto reemphasizethat interpretationof Figure
20b doesnot requirewater as a fluidizing agentbut only ejecta
entrained
in a turbulent
fluidlike
flow
with
a run-out
that
increases
withD 1/2.Impactcraters10 km in diameter
onMars
shoulddeveloprecoverywinds exceeding100 m/s, more than
capable of transportingimpacting ejecta. Since the present
Martian atmosphere
generates20 m/s winds [Hesset al., 1977]
and transportsfine-sizematerial, this scenarioshouldnot seem
unreasonable.
Addinglubricantssuchas atomizedwater,finely
broken ice, and even devolatizing clays should (and probably
does) contribute to the full range of observedmorphologies
acrossMars, but these lubricantsare not necessaryfor craters
increase
asD 1/2,providedthatPe>> Po. If autosuspension
occurs, then distal ejecta run-out will be sustained until in Lunae Planurn.
equation (9) is no longer satisfied. For Mars, autosuspended
A generalcomparisonof resultsfor differentatmospheric
flow will be controlled by substratepropertiesthat enhance and gravitational environmentsare shown in Figure 20c.
suspensionof debris in the flow (ex) or decreasethe fall Ballistic ejecta facies on the Moon and Mercury scalenearly
velocity (vt). Degassingvolatiles within the primary ejecta geometrically,that is, the outer limit of the continuousejecta
debris flow or entrainedby secondarymixing and turbulence increasesnearly linearly with craterdiameter.The inner ejecta
illustrateprocessesthat could enhancesuspension.
Decreased facieson Mars also scalegeometricallybut extendfurtherthan
fall velocities in the ejecta flow will occur for very fine on Mercury, even though it has the same gravity. This
flow due to
grained,entraineddebrisand for flowscontainingdevolatizing enhancedrun-out could indicatepostemplacement
fractions. For a given impacted target with essentially a volatiles as concludedby Horner and Greeley [1982] or to
constantratio of ex/vt, autosuspension
will be controlledby atmosphericentrainment. Under the extreme atmospheric
the flow velocity(u) andthe gradientfactor([5),eachof which conditionson Venus,entrainmentis maximized,and the ejecta
shouldscale as D 1/2 . If the ejecta flow is not furtherdrivenby flow densities Pe will approacha constantfraction of the
atmospheric
densityPo,whichis about0.1 g/cm
3.
precursorimpacting ejecta, then autosuspension
is largely surrounding
controlledby the initial flow velocity,that is, the eddy-driven From equation(6a) suchejecta-entrainedflows shouldresultin
mechanism.
run-out
distances
thatdecrease
asD'1/2if Pe= kpobutapproach
Figure 20a providesa qualitativesummaryof the relative
ejecta run-out distances expected for various modes of
emplacement. A given crater could exhibit multiple
emplacementstylesdependingon the ejecta size distribution
and crater size: an inner continuousballistic ejecta facies can
be overrunby eddy-drivenejectaflows, while sufficientlysmall
(or low-density) ejecta may continue to be entrainedin an
autosuspended
flow with large run-outdistancesas depictedin
a constant value as P e approaches P o (Figure 20c).
Nevertheless,flow separationand autosuspension
of finegrained ejecta with large run-out distancesbeyond the inner
faciesalsomight occur(Figure20a).
Modes of ejecta eraplacementon Mars. On the basis of
experiments,ejecta emplacementprocessesreflect the degree
of entrainment
of ejectain a turbulentflow createdby the ejecta
curtain moving through an atmosphere.Three emplacement
SCHULTZ:ATMOSPHERIC
E .I2;F.L"rs
OFEJECrAEM•CEMENT
11,655
lO
MARS
predicted
slopefromlaboratory
experiments
N. PLAINS
o
-suspended flo
0 coCb
0 O_•e
•o -'I,,o•ø o o
o
.•-.
o
0
[•".••2.-"
• D1/2
ß
ee ee
variation
duetocoarse
fractions
in
substrate
ands•mplc-complcx
transition
Onsetof rampart
formation,
Do
• •;_.•.a
-. -
(Gurtain-windsuspendedflow)
Do
•
!
RIDGED PLAINS
onsetof contiguous
cjcctarampart
11
I
i
I I I I J
Log Crater Diameter
,0
......
400
CRATERDIAMETER(km)
10
VENUS
Air-Entrained
Row
"-.....Pe=kpA
Eddy-Suspended Flow
'-.D-1/2
•,•
Pe>>PA
/ D
"-....
outer
eject.•..//•
MARS
inner
ejecta
•
MOON
MERCURY
I
I
i
I I i i J
1
I
10
CRATER
DIAMETER
i
i
I
, , !
100
(km•
Fig.20.Predicted
andobserved
run-out
of rampart-bordered
ejecta
controlled
by smaller
than20kmin diameter
duetomuch
finerconstituent
ejecta.
(c) Summary
windandeddy-suspended
eraplacement
mechanisms.
(a)Summary
oftheexpectedofthecontrasting
run-out
distances
expected
fordiff•t styles
oferaplacement.
relation
between
thedistance
fromcrater
rimto rampart
asa function
of crater Run-out
distance
represents
thedistance
between
therimandtheedge
ofthecjecta
diameter
based
ondiscussion
in text.If themoving
ejecta
curtain
creates
strongfacies
andis scaled
tothecrater
diameter.
Theouter
limitofthecontinuous
ejecta
windsthatremobilizc
cjecta
without
extensive
entrainment,
thentherim-rampart
faciesontheMoonandMercury
representing
ballistic
eraplacement
undertwo
distance
should
beproportional
to crater
diameter
(i.e.,a constant
valuewhen different
gravitational
fields[fromGaultet al., 1975].Innercontinuous
ejecta
distance
is scaled
to craterdiameter).
Theonsetof rampart
formation
shouldaround
Martiancraters
[fromHomerandGreeley,
1982]extend
to greater
depend
oncjecta
density
8 andsizea fora givenatmospheric
density
(p, i.e., distances
thancontinuous
cjecta
around
Mercurian
craters
withidentical
gravity,
ambient
pressure).
As aerodynamic
dragincreases
dueto increases
in ejectionperhaps
asa resultof continued
outward
flow.Outercjectafaciesdrivenby
velocity
(givensubstrate),
e_n_trainment
in turbulent
eddies
develop
witha run-outatmospheric
interactions
maytakeontwodifferent
relations.
Turbulent
eddy-
distance
proportional
toD1.5(D0.$
when
scaled
todiameter).
Atany
diameter,
thedriven
eraplacement
with
ancjecta
flow
density
(Pe)
much
greater
than
the
ambient
autosuspension
process
canenhance
run-out.
(b)DataforLunac
Planurn.
The atmospheric
density
(pA)should
exhibit
crater-scaled
run-out
distances
increasing
rampart
run-out
distances
aregenerally
consistent
with
eddy-driven
run-out
flows.asD1J2.
Ejocta
flows
with
maximum
entrainment
(asonVenus),
however,
should
Between
3and
5km,
however,
variations
insubstrate
and
crater
widening
processes
decrease
asD'1/'2
(equation
(6a))
provided
that
additional
turbulent
energy
(e.g.,
(rim/wall
slumping)
willresult
inconsiderable
scatter.
Open
circles
represent
datathermal
effects)
isnot
introduced.
AsPeapproaches
PA'
aconstant
run-out
distance
for cratersin northernlowlandplains(35ø-60
ø) as in Figurel Sb (right).Their is expected.
longerrun-outdistances
areconsistent
with an autosuspension
process
for craters
stylesareproposed
hereforMars,asillustrated
in Figure21: [Schultz,1988]. Much of the impact-generated
vaporis
curtain-wind
flow,eddy-suspended
flow, andautosuspended
contained
andredirected
upward
bythegrowing
transient
cavity
flow. All stylesbeginwith ballisticejection,emplacement
of behindthe curtain,therebybeingpartly isolatedfrom the
near-rim
ejecta,
andthedevelopment
of recovery
winds(Figure atmosphere
except
foroblique
impacts
asdiscussed
bySchultz
21a). Impact-generated
vaporexpanding
in advanceof the and Gault[1979,1982,1990].Laterstages
of cratergrowth,
ejectacurtainwouldhavescouredthe surfacewith supersonicthen,are believedto takeplaceundernear-ambient
conditions.
winds,andatmospheric
densities
wouldhavereturnedto near Theadvancing
ejectacurtainresembles
a movingsolidwalland
ambientvalueswell beforethe craterhad finishedforming createsa pressuredifferentialthat inducesstrongrecovery
11,656
SCHULTZ:ATMOSPIIERI½EFFE•S OF FJEL•A F_.MPLACEMENT
shock
front
/.
ballistic
trajectories
withincurtain
....
i!:" '
L:...' ß
I'
.....
ß '
|
post-emplacement
-.'•.. • •
ii•=--shear-driven
vortices
ß .[':ivpor
winnowingand
ejecta / drag-disruption
• ( debrisslide
•;'..:'
'J .'•:"/•'•'••
•.•._
CURTAIN-WIND
eiec,,on
'
'•?-
v.,.,:,o,
coarse
rampart
on ejectafacies
rimuplift
•.....
supersonic
winds
over-ridden
ß
o.::
- .
...• .. :...
- ..:
....'
flow
..
ba Illstic
ejecta
separation
."• recovery
winds
•
air-entrained
ß' ß
:.
•.--• •".'--- deceleratedfines
• ........
•
•
inner ejecta
'=••"•\ [bw
EDDY-SUSPENDED
•. ';:•.•
dmiuS•t'all•
ie •,,sduesP•)esl•løe•
•foa
rsefract
ion•
GENERATION
lobes
(erosion-resistant
rampart)
scoured
,•
ballistic •_'•"
•
'•,;"
fraction
•;.'..
finesentrained
•i'-':•'• inwinds
andvortices
,•'.?l'.'..'•l
•'•"-•'•
"'-"'
ejecta-entrained
turbidity flow
•l •
scoured
_ /
\ /
moat
\.,......• AUTO-SUSPEND
ENTRAINMENT
,•///••/'•///
'•FLOW
[ enhanced run-out
•, fromsecondaries.
'entrained fines.
wind/turbulence-driven
inner
facies
volatile
release
deposition of
suspended fines
(erosion-sensitive lobes)
turbidity flow
Fig. 21. Generalscenariodetailingprocesses
that may controlran-outof ejecta suspended
clastsamdeposited
in anejectaridge,thatis, a rampaxt.
Withincreasing
through
atmospheric
interactions.
(,2)Starting
conditions
for alternative
modesof utrbulence
duetogreater
induced
windvelocities
orenhanced
entrainment
of fines,
eraplacement
depicted
in Figure2lb. At thefinal stagesof cratergrowth,ejecta turbulent
eddiesoverrun
andscourtheinnerejectasurfaces.
Thisprocess
canform
followballistictrajectories
withinanejectacurtain.
Theprincipal
contrast
with multiple
lobeswithtcnninal
ramparts.
Flowseparation
maycreate
additional,
thin
vacuumconditions
is thedevelopment
of a largeejccta-curtain
angleandvorticesdistalejectalobes.Lithelogics
composed
of veryfine grainedparticles
resultin
createdalongthe curtaindueto redirected
air flow. Experiments
andtheoreticalgreaterentrainment
and increased
turbulentpower.The turbulentpowermay
considerations
suggest
thatanyimpact-generated
vaporcloudwouldhaveexpandedincrease
astheflow travelsdownandoff theinnerejecta,asit picksdebrispe•aps
to largedistances
by thistimeandmayhavecreated
a strong
precursor
scouring
of alreadymobili•.edby precursorblast or secondaries,
or as entraineddebris
thesurface[see$cAultz,1988].As the wall of ejectamovesoutward,it cream a dovolatize
dueto collisions.
Theresulting
autosuspended
flow mayextendto great
largedifferentialpressure,
therebyinducingstrongrecoverywindsthatentrain distances,
leavingbehinda verythinveneerof ejecta.The roleof volatilesin the
decelerated
finesintoa turbulent
flow. (b) Threecmplacement
processes
discussedproposed
scenarios
wouldenhance
ejectarun-out,butlongejectaflowsarepossible
in the text. Curtain-windsuspended
flow with minimal entrainmentresultsin fromlithelogicslargelycomposed
of loessor silt-sizedeposits.
temporarysuspension
of ballisticejecta.As the wind-drivingcurtaindissipates,
winds. Sufficiently small ballistic ejecta deceleratedto near
The startingconditionsdepictedin Figure 21a can evolve
terminalvelocitieswill be entrainedin the resultingturbulence into the three later stageemplacementstylesas depictedin
and winds.The size of the entrainedejectadependson crater Figure 2lb. Curtain-wind flow is proposedto characterize
size (Table 8). To first order, a 10-km-diameter crater smaller craters where the ballistically eraplaced continuous
(observed) would result in entrainmentof ejecta fractions ejecta is reworked and entrainedin recovery winds trailing
smaller than about 200 [tin. As the ballistic curtain thins, it behindthe ejectacurtain.As the windscease,largeclasts(>10
resembles a porous wall, and recovery winds abate. cm) are depositedto form a rampartsuperimposed
on the bulk
Ballistically eraplacedejecta near the rim of smaller craters ejecta deposits. In the laboratory experiments, differential
(<10 km) shouldbe characterized
by the invertedstratigraphy pressurecreated by the vortex appeared to control onset of
observedunder vacuumconditions,but this orderedsequence rampart formation. At the scalesof Martian craters,however,
will evolve into a more chaoticejectadebrisflow with distance both retainedvelocities(momentum)in impactingejecta and
from the rim. Becausestructuraluplift of the rim createsan their larger sizes result in viscous forces controlling
outwardslope,the eraplacedejectamay form an en massedebris entrainment,if laboratorystudiesof vortex entrainment(see
slide that preservesthe invertedstratigraphy
while creatingthe Greeley and Iversen, 1985) can be applied by analogy.
appearanceof nonballisticeraplacement.
Consequently, the occurrence and onset diameter of this
SCHULTZ:ATMOSPHERICEFFE•S OF Ea-ECr^EMPLACE.MENT
emplacementstyle depend on the average ejecta size. If the
ejecta exhibits a bimodal size distribution,entrainmentof the
finer fraction in the turbulent atmosphericresponsecreates a
two-phase flow (atmosphere and fine ejecta) that further
enhancestransport of coarser fraction. As a result, bimodal
ejecta size distributions (e.g., created by layered lithologies
with contrastingstrengths)will enhancethe developmentof
the contiguousrampart. The net outward momentum of the
ejectaand structuralrim uplift may resultin basalshearingand
a postemplacementdebris slide within the inner ejecta,
particularlyif a fluidizing agentis present.
Eddy-suspended flow develops as ejecta entrainment
increases.For small craters(<1 km) in fine-grained(<100 gm)
substrates,this processcan extendthe continuousdepositsand
ejectarampart.For largercraters(>5-10 km), it is characterized
by overrunningand partial scouringthe inner continuousejecta
depositsin its courseover the surroundingplains.As the eddy
loses energy, the suspendedload is suddenlydeposited.In a
well-mixed debrisflow of fine-grainedejecta,the depositforms
a lobate deposit.The distal lobes of ejecta recently recognized
at Meteor Crater [Schultz and Gram, 1989; Grant and Schultz,
19 8 9a] most likely reflect this process. If a significant
quantity of coarse clasts (>500 gm) is present or becomes
entrainedduring run-out, however, the lobe will form a rampart
terminus.The inner ejecta depositsalso may continueoutward
movement,due to either a lubricatingagent or simply residual
lateral momentum. In general, however, this emplacement
style shouldcharacterizea dry lithology. Both wind-suspended
and eddy-suspendedflows should create erosion-resistant
ramparts, since subsequentambient winds will not approach
the intensityof crater-generatedwinds.
As turbulent eddies run-out beyond the continuousejecta,
they may entrain surface materials with the appropriatesize
distribution. Material can be set in motion by a precursor
atmospheric blast, by the pressure differential created in
vortices, by preceding impacting ballistic ejecta, and by size
fractions susceptible to saltation/suspension. On Mars,
particles less than about 200 gm appear to be susceptibleto
suspension[Greeley and Iversen, 1985], and curtain-induced
wind speedsshould greatly exceed the necessarythreshold
speeds.Consequently,regions on Mars that have been longterm sinks for wind-blown debris (e.g., high latitudes) may
contribute a significant local componentto the ejecta run-out,
thereby enhancing the appearanceof both excessive ejecta
fluidity and a mass-balance paradox (ejecta versus crater
11,657
flows by obstacles indicated a slurry-like mode of
emplacement, not a gas-supported ground surge. This
conclusion does not challenge the interpretations here.
Observationscited by Carr et al. [1977] referred to the inner
facies where ballistic deposition(with possibledevelopment
of a debris slide) is expectedas depictedin Figure 21. They
noted,however,that ejecta flow associatedwith the outer facies
commonlyoverrun obstaclessimilar in height to the deposit
and thin inward from the flow termini. The emplacement
sequenceproposedby Schultzand Gault [ 1979], observedin the
laboratory, and generalized in Figure 21, therefore is
completely consistent with the conclusion of Carr et al.
[1977].
High-energy chemical and nuclear explosionsprovide a
possibletest for laboratory-scalephenomenaat a much larger
scale.Parallel aspectsof aerodynamiceffectson ballisticejecta
(including aerodynamic sorting) can be documented [see
Wisotski, 1977]. Nevertheless,the processof energytransfer
and the degree of atmosphericcoupling are significantly
different [Gault et al., 1968; Cooper, 1977; Knowles and
Brode, 1977; Schultz and Gault, 1982], and this difference can
significantly modify the processes affecting ejecta
emplacementobservedin the laboratory. Comparisonof the
evolution of ejecta cloud from a 0.5 ton (TNT) half-buried
explosionfrom the Pre-Mine Throw IV seriesin desertalluvium
(left sideof Plate 3) with a laboratoryimpactexperimentunder
full atmosphericpressureillustratesthis point (right side of
Plate3). A half-buriedexplosivechargetransfersonly about10
to 20% of its energyto the target,in contrastwith an impact
which transfers about 80% [Cooper and Sauer, 1977].
Consequently80-90% of the released explosive energy is
directly coupledwith the atmospherein contrastwith lessthan
1% for the lower velocity(1.5 km/s) impactshownin the right
sideof Plate 3 and the 20% possiblefor muchhighervelocity
(20 km/s) impactson the planets.Althoughan outwardmoving
conical ejecta curtain developsfrom an explosion(Plate 3,
left), the rising fireball carriesdecelerateddebrisupwardand
creates inward flow at the base. The laboratory impact
experimentshowsdeceleratedfractionssuspendedabove,but
volume).
Autosuspended
flow can result from extremelyfine grained
ejecta entrained by turbulence created in the ambient
atmosphere. Impact-generated water vapor [Wohletz and
Sheridan, 1983] or vapor released by mechanical friction
created by impacting ejecta could significantly enhance this
process as indicated by equation (9) and illustrated by
experimentsinvolving powdered tempera. Additional studies
are needed, however, to assesssuch processesquantitatively.
Diagnosticfeaturesassociatedwith this style include a scoured
inner rim creating a moatlike depression,radial grooving of
the continuousdeposits,a bulbousroundedprofile of the inner
ejecta deposits,absenceof secondarycraters (except for very
poroustargets),and a thin, fluid-appearingouter ejecta with
long run-out. Lithologies characterizedby a narrow size
distributionof fine material (loessor silt) shouldproduceliule,
if any, rampart-borderedejectafacies.
The proposed scenario emphasizes the response of the
atmosphereto the ejecta, thereby complementingthe scenario
by Schultz and Gault [1979] that focusedon the responseof
ejecta to the atmosphere. Nevertheless, both approaches
distinguishbetween ballistic emplacementof the inner ejecta
deposits but nonballistic emplacement of the outer deposits.
Carr et al. [1977] concluded that deflection of martian ejecta
Plate3. Time-resolved
comparison
of a largehalf-buried
explosion
of 0.5 tonTNT
in dry alluvium (from the Pre Mine Throw IV seriesin Nevada)shownat left and a
laboratoryimpact experimentshownat right (0.635 cm aluminumprojectile
impactingat about1.5 'kin/s).As in previousfigures,time stepsprogress
from
coolerto warmercolors.Direct contactbetweenthe explosivechargeand the
atmosphere
resultsin more than 80% of energycoupledwith the atmosphere,
whereasthe impacttransfersits energyto the atmosphere
indirectlythrough
aerodynamic
deceleration
of ejectaandturbulence
createdby the outwardmoving
ejecta curtain.This contrastin couplingproducesa contrastin atmospheric
circulationwherethe explosiongenerates
a risingfireballthat lofts (arrowat top)
anddraws(arrowat bottom)aerodynamically
decelerated
fineswhilethe impact
createsa basalturbidity current.
11,658
SCHULTZ:
ATMOSPHERIC
EFFE•SOFEJECTA
EMPLACEMENT
' .:':"
: ::-'
..
.:.
ß
..
...
...
ß .: • .:;½.
,%: •- ..
.:.:
•:..-::::•........::::
:,:;:,::::•:
.............
....:....,.:?.....:•:..::.•.:::-,:•:
.......
'½::::i:;-::;::::
:::.•!•
.....
::..::
,::
•. ....
.
.....:::::::::::::::::::::::::::::::::::
.....
..j..:::
...........
...•
......
...
,.
,...:
Fig.22. Contrasting
ejectaeraplacement
stylesfor identicalsize,nearbycraters.
(a) by morefluidizedejectabelow,all withinabouta •0 km instance
in the Tempo
Two 5-kin-diametercratersin the ridgedplainsof Hesperia(28.6øS,241.9øW; region (45øN, 60øW; Viking frames704B36 and 704B37.). Such contrastsin
Vikingframe418S37).The exampleat bottomexhibitsa well-developed
contiguous eraplacement
stylescouldrepresent
a changein aunospheric
pressure,
a changein a
ejectarampart,whereas
theexampleat topexhibitsrounded
innercontinuous
facies volatilereservoirat depth,and/ora subtlecontrast
in in situparticlesizeoccurring
with thin distalejectalobes.Scalebar corresponds
to 10 kin. (b) Comparison
of a oversmalldistances.
Scalebar corresponds
to 20 kin.
craterwith rampart-bordered
multiplelobesabovewith smallercraterssurrounded
this debris fraction was ballistically delivered. Moreover, the
flow field. As a result, some of the systematicsof emplacement
hook-shapedprofile (yellow time slice) above the impact observed in the laboratory should be destroyed around an
reflects a clockwise circulation, while the outward bulge at the
base results
from
a counterclockwise
circulation.
Other
explosioncrater.The comparison
between2 x 1016erg
explosion
and3 x 109 erg impactnevertheless
demonstrates
experimentsinvolving higher velocity impactsinto easily the importanceof both the atmosphericdecelerationof finevolatizeddry ice and carbonates
resultin a rising fireball, but grainedejecta and the atmosphericcirculatoryresponseto the
such fireballs are restrictedto very early times (1% of crater event. It should be expected that the exaggerated thermal
growth).Plate3 (left) revealsthatstrongatmospheric
coupling effects shown in Plate 3 will occur to a lesser degree on the
andthe risingfireballfrom an explosionalterthe atmospheric Earth (e.g., the suevitearoundthe Ries Crater [sdeSchultz and
SCI-I•TZ: ATMOSPHERIC
E•S
OF•A
EMPLACEMENT
11,659
::: ..:::3.....•.•
ß.,r,:'-:½,'
...
.•'4:.
X'..
.;;'.,::::."-i
"::::.::;;!'!-'
'-'--'½':':
•5½ •.::;::.'
-25.;;::;!&' .
.?"-•:"'
•':';•::::'*
"**""":'
'*;;;;i:•:
......
-•:%•...
......
.•:...
; .•
,
•:::..,:..
%:
:'*:;:*:, ,:,.:•."-"-.a
'*
:.:
•:::;.:
.....
.,
....",.-,,•..
:.:
- ....
,.,.
....
Fig. 22. (continued)
Gault, 1979; Newsom et al., 1991] and Venus where high planetswith atmospheres,
increasingdistances
from the crater
atmospheric
pressurescan constrainexpansionof the rising rim also must reflect increasing levels of the depositional
energy field, but the mode of emplacementcan range from
In general,different modesof ejecta emplacementwith ballistic kinetic energy to potential energy controlling a
fireball.
distance from a crater reflects the level of energy and ground-hugging
debrissurge.The latter stylereflectsthe degree
momentum associatedwith deposition. On planets without of entrainmentand the power containedin inducedatmospheric
atmospheres,
the deposifionalenergyfield (energyper unit turbulence,whether mechanically transferredby the wall of
areaat a givendistance)is controlledby the ballisticrangeas ejecta or augmentedby storedenergy (e.g., releasedvolafiles
underscoredby progressivechange from continuousejecta during entrainment).
An important implication of the proposed scenario in
facies to secondarycrater fields on the Moon [Shoemaker,
1962; Oberbeck et al., 1975; Oberbeck, 1975] and Mercury Figure 21 is that ejecta emplacementstyles could be used to
[Gault et al., 1975]. As a result, ejecta facies should be characterize lithologies in the absence of other clues.
separated
into annular
zones[e.g.,seeSchultz,1976].On Moreover, contrasting styles in a given lithology could
11,660
SCHULTZ:ATMOSPlt]•tlC EFFECTSOFEIECrA F3{r,LACEMENT
Crater
Atmospheric Volatiles
Size
Pressure
size and size distribution) far exceed variations due to
atmospheric
pressurewith elevation.
5. Systematic
trendsin ejectaemplacement
stylesandejecta
run-out distanceswith latitude could reflect latitude-dependent
Ballistic
lithelogicscharacterized
by aeoliandepositionof fine-grained
Contiguousramparts
materialratherthanvolatilesburied at depth.Surfacematerials
with adhered frozen volatiles (frost) or bound volatiles
(clathrates)could significantlyenhanceejectam-out through
the autosuspension
processas these volatiles are released
Lobed rampart
within a turbulent run-out flow.
(outer radial or thin
6. Unboundwaterexcavatedat impactmay be bestidentified
by its role in reducingbasal shear within the inner ejecta
deposits,therebycausingenhanced
run-outof the innerfacies
[seeHorner and Greeley, 1982], althoughgasentrainmentalso
couldplay a role. Aerodynamicdisruptionof excavatedwater
dropletswill producefiner sizesthat areevenmoresusceptible
distal flows)
to entrainment.
Multilobed
Inner bulbous facies
Fig. 23. Schematic
representation
of the effectsof cratersize, ejectasize
(lithology),atmospheric
pressure,and presenceof volatileson different
emplacement
stylesof ejectaon Mars.Arrowsreflectsystematic
changein
emplacement
styleasa function
of increasing
thevaluesof theselected
variable
whileholdingothervariables
constant.
7. Variationsin atmospheric
pressureexceedinga factorof
5 could changethe style of emplacementfrom rampartto runout flow for a given lithology at a given craterdiameterover
the last 3 billion years.
8. The largenumberof asymmetricejectafaciesindicating
obliqueimpacttrajectories(45ø-15ø from the horizontal)is an
expectedconsequence
of atmospheric
modificationof ejecta
emplacement.
indicatepastchanges
•n atmospheric
pressure
in additionto (or
even instead of) the presenceof buried water. Figure 22
The extremeatmosphericpressureon Venus will result in
flows entrainingejecta of nearly all sizes [see Schultz and
Gault, 1979]. Extrapolationof laboratoryresultsindicatesthat
aerodynamicdrag should create a near-verticalcurtain of
ballisticejectathroughoutcratergrowthwith ejectaexceeding
providestwo of manyexamples
of thelatterpossibility
where
a factor of 5 changein ambientpressurecould induce the
observed
changes
in ejectaemplacement
style.Suchvariations 100 m across. As the ballistic component impacts the
arepossiblefromorbitalforcingandthe globalredistribution structurally uplifted outer rim slope, it should resemble a
of polar-trapped
volatiles[e.g.,Fanaleet al., 1986].
massive air-entrainedrock avalanche.The high atmospheric
7. SUMMARY AND CONCLUDING REMARKS
A wide range of fluidlike ejecta emplacement styles is
possibleon planets with atmospheres,even in the absenceof
water or volatiles in the target. This conclusion does not
precludethe possiblerole of volatilesin enhancingejectaflow
run-out but raisesthe possibilitythat the morphologyof ejecta
facies on Mars could reflect local lithology and past
atmosphericconditionsas schematicallysummarizedin Figure
23. Mars, in fact, provides ideal conditions for recognizing
atmosphericeffects since the present atmosphericpressureis
sufficient to decelerate and entrain particles sensitive to
aeolian reworking for craters larger than about 0.5 km in
diameter. With this perspective, the following general
conclusionsand implicationsemergefor Mars.
1. Ejecta flows around Martian craters may reflect
entrainment of ejecta in winds and turbulencegeneratedin
responseto the outwardmoving ejectacurtain.
2. Different emplacementstylesmay reflect the degreeof
ejecta entrainmentin the dynamic atmosphericresponsewhich
in turn depends on crater size (controlling ejecta curtain
velocity) and ejecta size (controlling drag and sensitivity to
winds). Consequently,increasedejecta run-out with increasing
cratersize neednot indicategreateramountsof crater-excavated
densityresultsin near-rimejecta flows that shoulddecreasein
run-out distance(scaled to crater radius) with an increasein
crater size, therebycontrastingwith ejecta flows on Mars as
illustratedin Figure20c. Althoughdecelerated"finer"fractions
for conditionson Venus could include meter-sizedblocks,drag
forces and mechanical interactions should produce much
smaller debris that will be entrained in turbulentautosuspended
flows with much greaterrun-out distancesextendingover and
beyond the coarse inner facies. The distal eddy-drivenand
autosuspended
flows shouldbe highly channelizedwith fanshapedtermini analogousto ejecta flows developedin targets
of low-densitymicrospheres.
Becausethe saltation-suspension
boundaryon Venusoccursfor particlesonly 30 Ilm in diameter
[Greeleyand lversen, 1985], entrainmentof local materialsby
residualeddies(e.g., after flow separation)shouldcreatedistant
but thin "secondary"flows around the most pristinecraters.
Justas the coarserampart termini on Mars are more erosion
resistantthan the outer facies, the extremely coarseinner ejecta
around craters on Venus should persist over long times,
whereas
finer
fractions
in the distal
facies would be more
susceptible.
From the experiments,
subtledifferences
in ejecta
asymmetrydue to impactangleshouldbe accentuated.
Experiments with easily vaporized targets (dry ice and
carbonates)at different impact angles [Schultz, 1988; Schultz
and Gault, 1990] provide additional clues for atmospheric
water-rich substrates.
effects on early-time processesperhapsuniquely revealed on
3. Onset diameter for nonballistic ejecta emplacement
Venus.At higherimpactangles(>60ø),impact-generated
vapor
is partly contained within the crater cavity and early-time
styles may indicate critical conditionsrequired for drag
decelerationand entrainmentof ejecta rather than presenceof
ejecta plume. In contrast with the later stage mechanically
driven ballistic ejection, this early-time cloud of hot vapor
volatiles. Under present atmospheric conditions on Mars,
should resemble a short-lived gas-driven plinian eruption
onset diameter may prove useful to characterize impacted
column. At lower impact angles (<45ø), however, increasing
lithologies.
fractionsof the early-timevapor cloud and entrainedprojectile
4. Rampart-bordered ejecta facies are expected to
characterizebasaltic bedrock lithologies with diameter-scaled become decoupledfrom the later stage processesand travel
run-outdistances
increasing
asD1/2,whereas
multilobed
ejecta downrange[see Schultz and Gault, 1990]. Under the extreme
patternsare expectedfor well-sorted, fine-grainedsedimentary atmospheric pressures on Venus, lateral expansion of this
lithologies. Variations in lithology (controlling averagegrain
early-time cloud will be prevented, while the downrange
SCHULTZ: ATMOSPFIERICEFFECTSOF EJECFA•CEMEWr
11,661
quartz sand,II, Effectsof gravitationalacceleration,in Impact and
Explosion Cratering, editedby D.J. Roddyet al.,pp. 1231-1244,
Pergamon,New York, 1977.
atmosphere
cannotrestrainthis expandingcloud [Schultz, Gault,
D.E. and J.Wedekind,Experimentalstudiesof oblique impact,
1988], whereason Eaxththe atmosphere
servesto localizesuch
Proc. Lunar Planet. Sci. Conf., 9th, 3843-3875, 1978.
effects into a broad "fireline" [Schultzand Gault, 1990]. On Gault, D.E., J.E. Guest, J.B. Murray, D. Dzurisin, and M.C. Malin,
Venus, however, the downrange complex of vapor and
Some comparisons
of impactcraterson Mercury and the Moon, J.
projectilericochetshouldcreatea turbidityflow extending
to
Geophys.Res., 80, 2444-2460, 1975.
greatdistances.
As in laboratory
experiments,
laterstageejecta Gault, D.E., W.L. Quaide, and V.R. Oberbeck, Impact crateting
mechanics and structures,in Shock Metamorphism of Natural
will be eraplaced
on top of thiscomponent
but maybe drawn
Materials, edited by B.M. Frenchand N.M. Short,Mono,.pp.87downrange
by strongrecoverywindstrailingthe flow of hot
distancewill be arrestedby aerodynamicdrag acting on the
ensemble(as observedin the laboratory).On Mars, the tenuous
vapor and projectiledebris.Suchlaboratoryobservations 100, Baltimore, Md., 1968,
J., and P.H. Schultz, The erosional state and style of Meteor
providequalitativeyet specificpredictions
for the styleand Grant,
Crater, Lunar Planet. Sci., XX, 355-356, 1989a.
sequence
of eraplacement
thatmaybe uniquelytestedto the Grant, J. and P.H. Schultz, Gradation on Mars: From west and
extremeatmospheric
conditionson Venus.
The intentof thisstudywasnot to simulatedirectlyall the
northwestof Isidis basin and the Electris region, Icarus, 84, 166195, 1989b.
conditions
for ejectaeraplacement
on thevariousplanetswith Greeley,R. and J.D. Iversen,Wind as a GeologicalProcess,321 pp,
atmospheres.
Rather,the intentwasto demonstrate
that the
CambridgeUniversityPress,New York, 1985.
atmosphere
canplay a criticalrole in modifyingeraplacement,Greeley, R., J. Fink, D.E. Gault, J. Guest, and P.H. Schultz, P.H,
Impact crateringin viscoustargets:Laboratoryexperiments.Proc.
and to explorecontrollingprocesses
revealedin laboratory
Lunar Planet. Sci. Conf. X/, 2075-2097, 1980.
experiments.
Thisperspective
challenges
thewidelyheldview
debris flow in generating
that buried water controlsejecta morphologieson Mars and Hampton,M.A., The role of subaqueous
turbiditycurrents,J. Sediment.Petrol.,42, 775-793, 1972.
proposesinsteadthat both lithology(grain size) and near- Herr, R.W., Effects of the atmospheric-lithostatic
pressureratio on
surfacevolatiles(e.g., frost)play equal(if not the dominant)
explosivecratersin dry soil,NASA, Tech.Rep.,TRR-366, 1971.
roles,particularly
for distalfacies.Theobserved
processes
also Hess, S.L., R.M. Henry, C.B. Leovy, J.A. Ryan, and J.E. Tillman,
allow formulationof simplifiedtheoretical
modelsandtestsof
Meteorologicalresultsfrom the surfaceof Mars: Viking I and 2, J.
theverycomplexinteractions
actingfrommicronto kilometer Geophys.Res., 82, 4559-4574, 1977
scales.
Holsapple, K.A., Impact experimentswith ambient atmospheric
pressure
(abstract),
EOS,Trans.Am.Geophys.
Union,60, 1979.
Acknowledgments. I wish to acknowledge
the assistance
and Holsapple, K.A. and R.M. Schmidt, On the scaling of crater
dimensions,2, Impact processes,J. Geophys.Res., 87, 1949skills of the crew of the NASA Ames Vertical Gun Range who have
1970, 1982.
participated
in various
stages
of thisstudy:
WayneLogsdon,
JohnVon
grey,Ben Langedyk,
RogerKraus,andEarl Brooks.I alsowishto Homer, V.M., and R. Greeley, Pedestalcraterson Ganymede,Icarus,
51, 544-562, 1982.
Housen,K.R., R.M. Schmidt,and K.A. Holsapple,Crater ejectascaling
laws: Fundamental forms based on dimensional analysis, J.
enablingthe early stagesof this work. Lastly, this paper was
thank Don Gault for sharinghis insight,Ann Meloy for helpingto
producePlatesI and2, andthe LunarandPlanetar]Institutefor
significantly
improved
by theconscientious
reviewsby J.D. O'Keefe,
an anonymous
reviewer,andClarkChapman.
Geophys.Res., 88, 2485-2499, 1983.
Iranov, B.A., A.T. Basilevsky,V.P.Kryuchkov, and I.M. Chernaya,
Impact cratersof Venus:Analysisof Venera 15 and 16 data,J.
Geophys.Res., 91, D413-D430, 1986.
REFERENCES
Jones,E. and J.W. Kodis, Atmosphericeffectsof large body impacts.
The first few minutes, in Geological Implications of Impacts of
Alvarez,L.W., W. Alvarez,F. Asaro,andtI.V. Michel. Extraterrestrial Large Asteroidsand Cometson the Earth, editedby L. Silver and
P.H. Schultz,Geol.Soc.
Am. Spec.Pap. 190, 175-186, 1882.
causefor the Cretaceous-Tertiary
extinction,
Science,208, 1095Knowles, C.P., and H.L. Brode, H.L., The theory of crateting
1108, 1980.
Arvidson,R.E, E.A. Guinness,
and S.W. Lee. Differentialaeolian
redistributionrateson Mars, Nature, 278, 533-535, 1979.
Bagnold,
R.A., Auto-suspension
of transported
sediment:
turbidity
phenomena,an overview,in Impact and ExplosionCratering,
editedby D.J. Roddy et al., pp. 869-895, Pergamon,New York,
1977.
Langhaar,H.L., DimensionalAnalysisand Theoryof Models, 166pp,
JohnWiley, New York, 1951..
Bagnold,
R.A.,Mechanics
of marine
sedimentation,
in TheSea,edited
McGetchin, T.R., M. Settle, and J.W. Head, Radial thicknessvariation
by M.N. Hill, pp.507-523,JohnWiley,NewYork,1963.
in impactcraterejecta:Implicationsfor lunar basindeposits,Earth
Cart, M.H., L.S. Crumpier,
J.A. Cutts,R.Greeley,
J.E. Guest,andH.
currents.Proc. R. Soc.Lond. (A), 265, 315-319, 1962.
Planet. Sci. Lett., 20, 226-236, 1973.
Masursky.
Martianimpactcratersandemplacement
of ejectaby
Mellor, M., Dynamics of snow avalanches,in Rockslides and
surfaceflow, J. Geophys.
Res.,82, 4055-4065,1977.
Avalanches,1, Natural Phenomena,editedby B. Voight, pp. 752Chabai, A.J., Influence of gravitationalfields and atmospheric
pressures
onscaling
of explosion
craters,
in ImpactandExplosion
792, Elsevier Science,New York, 1978.
New York, 1977.
Ring Basins,editedby P.H. SchultzandR.B. Merrill, pp. 29-35,
Cratering,editedby D.J.Roddyet al.,pp.1191-1214,
Pergamon, Melosh,H.J., Atmosphericbreakupof terrestrialimpactors,in Multi-
Pergamon,New York, 1981.
Cooper,H.F.,A summary
of explosion
cratering
phenomena
relevant
to meteorimpactevents,
in ImpactandExplosion
Cratering,
edited Melosh, H.J., The mechanicsof large meteoroidimpactsin the Earth's
oceans,in GeologicalImplicationsof Impacts of Large Asteroids
by D.J.Roddyet al.,pp.11-44,Pergamon,
NewYork,1977.
and Comets on the Earth, Geol.Soc.Am.Spec.Pap., 190, edited by
Cooper,H.F., andF.M. Sauer.Crater-related
groundmotionsand
L. Silver and P.H. Schultz, 121-127, 1982.
implications
forcraterscaling,
in ImpactandExplosion
Cratering,
editedby D.J.Roddyet al.,pp.1133-1163,
Pergamon,
NewYork, Mouginis-Mark,P., Martian fluidized cratermorphology:Variations
1977.
with crater size, latitude, altitude, and target material, J. Geophys.
Res., 84, 8011-8022, 1979.
Curran,D.R., D.A. Shoekey,
L. Seaman,
andM. Austin,Mechanisms
andmodelsof crateringin earthmedia,in Impactand Explosion Newsom,H.E., G. Graup,D.A. Iseri, J.W. Geissman,and K. Keil, The
formation of the Ries Crater, West Germany; Evidence of
Cratering,editedby D.J.Roddyet al.,pp.1057-1087,
Pergamon,
New York, 1977.
Fanale, F., J.R.Salvail,A.P. Zent, and S.E. Postawko,Global
distribution
andmigration
of sub-surface
iceonMars,Icarus,67, 118, 1986.
atmosphericinteractionsduringa large crateringevent, in Global
Catastrophesin Earth History, edited by V.L. Sharptonand P.S.
Ward, Geol. Soc.Am. Spec.Pap., 247, 195-206, 1990.
Oberbeck,V.R., F. Horz, R.H. Morrison, W,L. Quaide, and D.E. Gault,
On the origin of the lunar smoothplains,Moon, 12, 19-54, 1975.
Gault,D.E., andR. Greeley,Exploratory
experiments
of impactcraters
formed in viscous-liquidtargets'Analogsfor martian rampart Oberbeck, V.R., The role of ballistic erosion and sedimentation h•
craters?,Icarus,34, 386-495, 1978.
Gault,D.E., andJ. Wedekind,Experimental
hypervelocity
impactinto
lunar stratigraphy, Rev. Geophys. Space Phys., 13, 337-362,
1975.
11,662
O'Keefe,
SCHULTZ:ATMOSPHERIC
•S
J.D.,
and T.A.
Ahrens, The
interaction
of the
OFE.IF.L'TA
F.MIn_.ACEMENT
Schultz, P.H. and D.E. Gault, Clustered impacts: Experimentsand
implications,J. Geophys.Res., 90, 3701-3732, 1985.
Cretaceous/Tertiary
extinctionbolidewith the atmosphere,
ocean,
and solid Earth, in GeologicalImplicationsof Impactsof Large Schultz, P.H., and D.E, Gault, Prolongedglobal catastrophesfrom
Asteroidsand Comets on the Earth, edited by L. Silver and P.H.
Schultz,Geol.Soc.Am.Spec.Pap.190,103-120, 1982.
Pantin, H.M., Interaction between velocity and effective density in
turbidity flow: phase plane analysis with criteria for
autosuspension,
Mar. Geol., 31, 59-99, 1979.
Post,R.L., Ejectsdistributions
fromnear-surface
nuclearandHE bursts.
Rep. Ab'WL-TR-74-51,
Air ForceWeaponsLab.,KirfiandAir Force
Base,Albuquerque,
NM, 1974.
Schultz, P.H., Moon Morphology, Univiversity of Texas Press,
Austin, 1976.
Schultz,P.H., Crater ejectsmorphologyand the presenceof water on
Mars, in MECA Symposium
on Mars: Evolutionof Its Climateand
Atmosphere,pp. 95-97, Lunar and PlanetaryInstitute,Houston,
oblique impacts,Spec.Pap. Geol. Soc.Am., 247, 239-261, 1990,
Schultz, P.H. and J.A. Grant, Styles of ejects emplacement:Meteor
Crater, Lunar Planet. Sci. XX, 972-973, 1989,
Schultz,P.H., and A.B. Lutz, Polar wanderingon Mars, Icarus, 73, 91141, 1988.
Schultz, P.H., and W. Mendell, Orbital infrared observationsof lunar
cratersand possibleimplicationsfor impact ejecta emplacement,
Proc. Lunar.Planet.Sci. Conf./X, 2857-2883, 1978.
Schultz, P.H., and J. Singer, Secondaryimpact cratersaroundlunar,
mercurian, and martian craters,Lunar Planet. Sci. X/, 2309-2259,
1982.
Sherwood,A.E., Effect of air dragon particlesejectedduringexplosive
cratering,J. Geophys.Res., 72, 1783-1791, 1967.
Shoemaker, E.M., Interpretation of lunar craters, in Physics and
Tex., 1986.
Astronomy of the Moon, edited by Z. Kopal, pp. 283-359,
Schultz, P.H., Impact vaporization of volatile-rich targets:
Academic,San Diego, Calif., 1962.
Experimentalresultsand implications,Lunar Planet. Sci., XIX,
Taylor, G.T., The formation of a blast wave by a very intense
1039-1040, 1988.
explosion; Part I. The atomic explosion of 1946, Proc. R.
Schultz, P.H., Factors controlling impact ejects eraplacementon
Soc.LondonSer. A, 201, 175-177, 1950.
Mars(abstract),in Fourth InternationalConferenceon Mars, pp.
Vickery, A., Effect of an impact-generated
gascloudon the acceleration
181-182,Universityof ArizonaPress,Tucson,1989.
of solidejects,J. Geophys.Res., 91, 14,139-14,160, 1986.
Schultz,P.H., Atmospheric
effectson crateringefficiency,Lunar
Wisotski, J., Dynamic ejects parameters from high explosive
Planet. Sci. XXI, 1095-1096, 1990a.
detonations,in Impact and Explosion Cratering, edited by D.J.
Schultz, P.H., Evidence for atmosphericeffects on martian crater
Roddy et al.,pp. l101-1121, Pergamon,New York, 1977.
shape,LunarPlanetSci.,XXI, 1097-1098,1990b.
Schultz,P.H., Atmosphericeffectson cratetingefficiency,J. Geophys. Wohletz, K.H., and M.F. Sheridan, Martian rampart crater ejects:
Experimentsand analysisof melt-waterinteraction,Icar•, 56, 15Res. 97, 975-1005, 1992.
37, 1983.
Schultz, P.H., and D.E. Gault, Atmosphericeffects on martian ejects
eraplacement,J. Geophys.Res., 84, 7669-7687, 1979..
Peter H. Schultz, Department of Geological Sciences, Brown
Schultz,P.H. and D.E. Gault, Impact ejectsdynamicsin an atmosphere: University,Providence,RI 02912.
Experimental results and extrapolations, in Geological
Implications of Impacts of Large Asteroids and Comets on the
Earth, editedby L. Silver and P.H. Schultz,Geol. Soc. Am. Spec.
Pap. 190, 153-174, 1982.
(ReceivedJanuary31, 1991;
Schultz,P.H. andD.E. Gault,Impact-induced
vaporization:Effect of
revisedFebruary21, 1992;
impactangle,Lunar Planet.Sci.XV, 732-733, 1984.
acceptedMarch 13, 1992.)