Martian Rampart Crater Ejecta: Experiments and Analysis of Melt-Water Interaction ~

Home , Bingham (crater), Clark (Martian crater), Deutsch (crater), Ejecta, Greeley (Martian crater), Jones (Martian crater)

nCARUS 56, 15-37 (1983)

K. H. WOHLETZ* AND M. F. SHERIDANt *Earth and Space Science Division, Los Alamos National Lahoratot3', Los Alamos, New Mexico 87545 and ~fDepartment of Geology, Arizona State University, Tempe, Arizona 85276

Received October 18, 1982: revised May 31. 1983

Viking images of Martian craters with rampart-bordered ejecta deposits reveal distinct impact ejecta morphology when compared to that associated with similar-sized craters on the Moon and Mercury. Topographic control of distribution, lobate and terraced margins, cross-cutting relationships, and multiple stratigraphic units are evidence for ejccta emplacement by surface flowage. It is suggested that target water explosively vaporized during impact alters initial ballistic trajectories of ejecta and produces surging flow emplacement. The dispersal of particulates during a series of controlled steam explosions generated by interaction of a thermite melt with water has been experimentally modeled. Preliminary results indicate that the mass ratio of water to melt and confining pressure control the degree of melt fragmentation (ejecta particle size) and the energy and mode of melt-ejecta dispersal. Study of terrestrial, lobate, volcanic ejecta produced by steam-blast explosions reveals that particle size and vapor to clast volume ratio are primary parameters charac- terizing the emplacement mechanism and deposit morphology. Martian crater ramparts are formed when ejecta surges lose fluidizing vapors and transported particles are deposited cn masse. This deposition results from flow yield strength increasing above shear stress duc to interparticlc friction.

INTRODUCTION Carr and Schaber, 1977). Therefore, water is an important factor for consideration Studies by Head and Roth (1976) and when comparing the lobate ejecta patterns Cart et al. (1977) of Viking imagery point of Mars to the simple ballistic patterns on out that ejecta deposits surrounding large, the Moon and Mercury, where the environ- Martian impact craters, up to 50 km in ment has been essentially water free (Cin- diameter, are different from those around tala and Mouginis-Mark, 1980). To better similar-sized craters on the Moon and Mer- understand the relationship of target water cury. The Martian craters are surrounded to ejecta morphology, Johansen (1979), by lobate or multilobate ejecta deposits and Mouginis-Mark (1979a), and Blasius and have been termed "spiosh," "fluidized," Cutts (1980) attempt to correlate morphol- "rampart," or "pedestal" craters (Match ogy variation with latitude, elevation, and and Worronow, 1980; Mouginis-Mark, target lithology. However, the interpreta- 1979a). Their deposits appear to have been tion of morphology statistics has been lim- emplaced by ejecta flow over the surface ited by a lack of both information concern- away from the crater. The presence of ing distribution of water on the surface of abundant surface or near-surface water Mars as well as a generally accepted model early in Martian history is indicated by con- explaining formation of lobate ejecta de- siderable evidence (Toulmin et al., 1977; posits. A few recent papers relate water content t This work was performed under the auspices of the of target material to crater formation, dis- U.S. Department of Energy. The U.S. Government's right to retain a nonexclusive royalty-free license in persal of products, and types of deposits. and to the copyright covering this paper, for govern- Gault and Greeley (1978) and Greeley et al. mental purposes, is acknowledged. (1980) impacted targets of varying water 15 001% 1035/83 $3.00 Copyright ~ 1983 by Academic Press, Inc. All right~ of reproduction in any form reserved. 16 WOHLETZ AND SHERIDAN content using the NASA-Ames Research volcanoes where magma interacts with Center vertical gun. Preliminary results in- near-surface water (Colgate and Sigurgeirs- dicate a strong effect on crater and ejecta son, 1973; Peckover et al., 1973; Wohlctz, deposit morphology by yield strength and 1980; Sheridan and Wohletz, 1981). Numer- viscosity, which are dependent on water ous papers describe the phcnomenology content. Kieffer and Simonds (1980), in and products of such hydromagmatic erup- their review of the literature on terrestrial tions as reviewed by Wohletz and Sheridan impact structures, have shown that al- (1983). Usually flash vaporation leads to though impact melt is more readily formed strong fragmentation of the melt; dispersal in lithologies rich in volatiles, melts associ- of ejecta from the explosion center is by ated with such targets are widely dispersed ground-hugging, partly fluidized density from their craters by rapid volatile expan- currents termed surges (Moore, 1966: Wa- sion. Hence dry terrestrial targets contain ters and Fisher, 1971; Wohletz and Sheri- thick melt sheets, whereas wet targets lack dan, 1979). melt sheets but are surrounded by suevite Also of importance to ejecta emplace- blankets. ment is the affect of shock waves. The af- This paper describes possible effects that fect of shock waves on the atmospheric explosive water vaporization might have on flow field produced by the projectile bow ejecta emplacement after impact into a wet shock, vapor explosion, or other mecha- target. A general model (Fig. !) is formu- nism is complex and beyond the scope of lated from analysis of Viking imagery and this paper (see Jones and Kodis, 1982: experimental vapor explosions as well as Schultz and Gault, 1982). It is worthwhile consideration of fluidized particulate trans- to note. however, that the experimental port and lobate volcanic deposits. The work to bc discussed indicates that shock model contends that as target water content phenomena trigger and enhance vapor ex- increases, the effects of vapor expansion plosion processes. due to impact increasingly modify thc ballistic flow field during crater excavation. VIKING IMAGERY DESCRIPTION This modification results in transport by Observations of Martian rampart crater gravity driven surface flowage, and is simi- deposits based on interpretation of Viking lar to that of atmospheric drag effects on imagery tbrm the principal constraints on ejecta modeled by Schuitz and Gault our models of ejecta emplacement. Twelve (1979). Martian rampart craters were studied to Critical to this problem is the mechanism quantitatively characterize their deposit by which water and impact melt become morphology. Maps of three of the craters intimately intermixed so that efficient en- (Figs. 2-4) are presented to show represen- ergy transfer is possible. We consider sev- tative deposit characteristics, aereal extent eral other situations of melt-water interac- of depositional units, and spatial relation- tion, albeit quite different physically. ships of rampart lobes. The following map however, instructive in the dynamics of va- units are based in part on interpretations of por explosions. The dynamic interaction of Mouginis-Mark (1978, 1979b). molten materials and water has been stud- Wall material (14/). This material com- ied by numerous workers concerned with prises deposits that form discontinuous, vapor explosions at foundaries (Lipsctt, concentric scarp ridges parallel to the cir- 1966). The literature on such industrial cumferencc rim on the inner slopes of cra- melt-coolant explosions is summarized by ters. Some craters have multiple rings of Sandia Laboratories (1975). Another wall material that form concentric tilted ter- source of data comes from natural steam races. The inward decrease in elevation of explosions which are common in terrestrial multiple ridges suggests an origin by col- MARTIAN RAMPART CRATER EJECTA 17

INTERACTION OF IMPACT / MELT AND WATER Hz0 ~ H~0 $H~ 0/MELT RATIO ,CONFINEMENT $MIXING AND FRAGMENTATION

BALLISTIC EJECTION FOLLOWED BY VAPOR EXPLOSION *VAPOR TO PAIITICI,E ,,,--7e--7--..,z.~ RATIO oVAPOB ENTROPY f-.~,, k'~k~.~:U-,.~/. BALLISTIC (SUPERllEATING) 3. ~(. t' " "-~/.~ff CURTAIN INITIAL SURGE FOLLOWED BY REPEATED VAPOR EXPLOSIONS OF DECREASING ENERGY • AERODYNAMIC EFFEC'I ,'S -~'~ VISCOUS-- / INERTIAL-- UPON BAI,I.ISTIC MATFBIAL % .. • . -", .'."" ..- : tRANSPORT-tRANSItION 4. FROM VAPOR-SUPPORTED VISCOUS TO INERTIAL FLOW eE]FCTA SIZE oVOID SPACE C R H La L~ *"WETNESS"

5. MULTI-LOBED EJECTA FLOW DEPOSIT

Fn(;. I. Schematic model of rampart ejecta emplacement. lapse of slump blocks. Such terraces are Hummocky material (H). The outer common in large impact craters regardless slopes of rampart crater rims are character- of whether the target was dry, such as the ized by hummocky ridged material that ex- Moon or Mercury, or volatile rich, as ap- tends out to the surrounding plain. This ma- pears to be the case for many Martian cra- terial resembles short and stubby stacked ters. We suggest that in a volatile-rich envi- platcs or "shingles" of ejecta layers that ronment these features could also be may be related to an overturned flap of rim explained by rapid, local withdrawal of sup- materials. Multiple layers are likely the port related to vapor explosions in the cra- result of excavation through a crustal stratig- ter during and shortly after initial excava- raphy composed of several different litbo- tion. Terrestrial analogs of such features logic units. are common within maar craters produced Lobate ejecta (L). These relatively thin by hydrovolcanic steam-blast explosions ejecta sheets extend beyond the hummocky (Lorenz, 1970). material to the limits of continuous ejecta. 18 WOHLETZ AND SHERIDAN

25 km ...... "..... It

"" ::', |,z

RI ""

0 b

FIG. 2. Paired photo (a) and map (b) of Yuty crater, C, showing multiple lobes of rampart ejecta sheets. Note the topographic effect of crater, C~, on the dispersal pattern of 1.~ and H. Unit terminol- ogy and descriptions are given in the text. Viking image 003A07. \ PLATEAU

•.r- 100 km

:! I., .,j

PI,AINS R

1.2

..t. 0

tlIGHLANDS b

FIG. 3. Paired photo (a) and map (b) showing the constrainl of rampart ejecta by highland topography. The plateau to the top was insufficient to block flowage of the lobe. Viking image 545A30. 19 20 WOHLETZ AND SHERIDAN

In plan view, individual lobe terminations rampart deposits from pedestal deposits have a tongue-like shape with the apex that resemble an overturned saucer. pointing away from the crater. Intercon- The observed stratigraphic relationships nected distal lobes define a scalloped but of ejecta deposits outline some constraints continuous outer margin of ejecta flows. for a ejecta emplacement model. These re- Concentric lobes are here designated by nu- lationships are discussed below. merical subscripts; subscript 1 applies to Ejecta lobes (L~ and L2) from Yuty crater the lobe set of greatest runout distance and (C) shown in Figs. 2a and b have overrun a successive lobe sets of lesser runout dis- nearby older crater (C~). However. a later tance are designated by subscripts 2 and 3. flow lobe (L3) has deflected around the Ramparts (R). Some lobes have a termi- raised rim of Ci and is locally covered by nal scarp which is marked by either a flat the hummocky deposits (H). Cross-cutting ridge or a gradual increase in deposit thick- relationships of ramparts indicate that the ness. These ramparts have steep outer ejecta lobes were emplaced in a general se- walls that mark step-like boundaries for quence: L~, L_,, and 1-3 followed by H. The flow lobes. The raised rim distinguishes fact that L~ and L2 have passed over the

FIG. 4. Paired photo (a} and map (bJ showing the effect of small hills on the distribution of rampart ciecta, l.obe l.f was blocked by some hills, passed through some saddles, and was channelized into a long runout stream where flow was confined. Viking image 220A20. MARTIAN RAMPART CRATER EJECTA 21 \,

T 25 km

FIG. 4~Continued. topographic rim of C~ leaving a thin man- along the margins of Lj. The proximity of tling deposit but that L3 was deflected wind streaks to L~ suggests that this ram- around C~ suggests that the earlier flows part lobe consists of unconsolidated parti- were of higher energy. Hummocky material cles that were readily transported by Mar- (H) from Yuty also appears to have de- tian winds. flected around the topographic rim of C~, A 17-km-diameter crater (Figs. 4a and b) with some of this material spilling over the has an extensive apron with distinct termi- rim of Ct to cover the crater floor. These nal terraces but poorly defined interior features demonstrate the strong effect of to- lobes. The small group of hills in the upper pography on the distribution of ejected ma- part of the photo were a strong topographic terial. control on the distribution of flow lobes LI. A 28-km-diameter crater (Figs. 3a and b) High ridges have blocked the flow, saddles has two distinct lobe sets, each with a well- have allowed passage without deposition, developed outer rampart. L~ has spread and plains beyond gaps in the hills have onto a low plateau region, but abuts against provided runout chutes for flow lobes. the highlands near the bottom of the map. Relationships recorded on these maps This observation demonstrates the ability constructed from Viking images include: (1) of ejecta lobes to overtop low topographic Emplacement of outer lobe sets occurs be- barriers but not high mountain ranges. fore inner lobe sets. (2) Outer lobes may Ramparts did not form where the highlands pass over higher topographic barriers than caused the flows to pond before reaching inner lobes. (3) Distribution of hummocky their maximum runout distance. Wind material is also topographically controlled. streaks, apparently adjacent to small fresh (4) Lobes that pond where they abut against craters in the plains, are also observable existing topography lack ramparts. (5) Flow 22 WOHLETZ AND SHERIDAN ejecta deposits are composed of unconsoli- of thermite was ignited to form a melt of dated, fine-grained particulate matter that similar physical properties as those of ba- can be retransported by Martian winds. (6) salt. This melt is brought into contact with Flows that are channelized through gaps in water within a steel confinement chamber. topography may attain increased runout. The Series I chamber (Fig. 5) is equipped (7) Flow lobes may run into and out of topo- with a vent designed to burst at predeter- graphic depressions leaving only a thin de- mined pressures. The confinement chamber posit without complete ponding and spill- is monitored by continuously recording over. pressure and temperature gauges. The mass The details of rampart lobe distribution of water contacting the melt. burst pres- and morphology provide additional insight sure, and contact geometry have been var- into their emplacement. Lobe set perime- ied in this series of experiments. High- ters seem to be continuous, albeit undulat- speed cinematography is used to record the ing, which suggests that each lobe set (L~, phenomenology of the vapor explosion dur- L2, L~, etc.) originates from an individual ing venting. ejecta pulse of a characteristic transport A representative experiment with a wa- energy. Scalloping is greater for the outer ter-melt mass ratio of about 0.20 was per- ring lobes (L0 than for inner lobes (L 2 and formed in December, 1980 (Figs. 6 and 7). L3). The rampart width of the outer lobes is Our main purpose was to analyze the explo- likewise greater than that for inner lobes sion characteristics and the interaction of indicating that their flow continued for a ejected melt fragments with the external longer distance after the development of flow field produced by the expanding vola- yield strength in the flows. tile cloud. Larger fragments follow unmodi- fied parabolic paths, whereas small-sized EXPERIMENTATION ejecta experience significant aerodynamic Vapor explosions resulting from the in- drag due to their interaction with the atmo- teraction of a molten fluid with water are sphere and superheated steam. The drag usually termed "fuel-coolant interactions" turbulence causes separation of fine parti- (Colgate and Sigurgeirsson, 1973; Peckover cles from the ballistic plume. This sepa- et al., 1973; Corradini, 1981). As mentioned rated material collapses as a density cloud earlier, this process has attracted (surge/ and flows horizontally away from much study due to its devastating affects the crater (Fig. 6). in steel foundaries and its potential haz- Although the cratering mechanism for ard in the event of nuclear reactor core this experiment is quite different from that meltdown (Sandia Laboratories, 1975). The of an impact event, the deposit morphology rapid conversion of thermal to mechanical (Fig. 7) resembles that associated with energy and the remarkable efficiency of the small chemical explosions and small-scale process are still poorly understood even af- Martian craters observed on Viking images. ter numerous laboratory studies (Sandia Data on the resulting deposit (Table I) show References). This phenomenon is similar to that 65% of the total displaced mass lies phreatomagmatic explosions which pro- within two crater radii. This mass distribu- duce maar volcanoes (Wohletz. 1980; tion is similar to that observed by Stoffler et Heiken, 1971; Lorenz, 1973; Oilier, 1967). al. (1975) for deposits formed by ballistic Our first series of experiments at Los A1- transport due to impact of projectiles into amos National Laboratory were designed dry sand. This result shows that there is to document the phenomenology of this little difference in morphology between de- process on a larger scale than past investi- posits formed by small-scale melt-water gations in order to assess its affects upon explosions and those related to impact and planetary cratering. Approximately 100 kg explosion cratering in dry material. How'- MARTIAN RAMPART CRATER EJECTA 23

FIG. 5. Steel vessel used in the thermite-water experiments. Note the connectors for P-T readout. A Kodak film box is shown for scale.

ever, an increase in the mass ratio above cus centers on the properties of the 0.20 drastically increases vapor expansion vapor-particulate cloud that collapses after and ejecta runout as observed for large, impact crater formation. Important param- Martian rampart craters (Mouginis-Mark, eters are: size, velocity, and trajectories of 1979a). particles, as well as the physical state of the Since the excavation dynamics of the ex- vapor (i.e., condensing, saturated steam vs perimental vapor explosions are quite dif- expanding, superheated steam). These pa- ferent from those of impact origin, our fo- rameters have been shown by experiments 24 WOHLETZ AND SHERIDAN

5 :3 l- y-

z .A

E l- Y_ 09 o

E E

/- MARTIAN RAMPART CRATER EJECTA 25

.~-2

.-a E .~ r~ C .

=~-~

-],8~

a_E~ • ~ ~ ~'

"~sE ~>,

•~ .~ -

~-E .-

",~ to 26 WOHLETZ AND SHERIDAN

TABI.I! 1

MORPtlOI OGY ANI) DISPI.ACI-~I) Mass ot ~XPI!IUMhN IAI. CRAIF, R

Radius (m) Area (m:) Thickness" (cm) Mass/boar& (kg) Total mass (kg)

0.00-0.66 1.36 0.00 -- 0 0.66-0.81 O. 70 15.00 -- I 10 0.81-0.97 0.89 30.0(I -- 356 0.97- I. 12 0.99 29.00 -- 426 1.12- 1.27 I. 12 26.00 -- 489 1.27- 1.42 1.27 20.00 -- 484 1.42-1.58 1.51 16.(X1 -- 547 1.58-1.73 1.56 I 1.00 -- 402 1.73- 1.88 1.70 7.00 -- 303 1.88-2.03 1.85 3.00 -- 154 2.03-2.19 2. I I 0.50 -- 34 2.3-3.3 17.61 2.55 13.78 674 3.3-4.3 23.87 I. 14 6.14 407 4.3-5.3 30.86 0.67 3.64 305 5.3-6.3 36.46 0.38 2.(15 208 6.3-7.3 42.70 0.08 0.45 53 7.3-8.3 49.00 0.02 0.12 16 8.3-9.3 55.80 0.01 0.3 5

" Thickness is calculated from mass collected on boards (2.3-9.3 m). b Collection boards are (I.36 m-'.

(Wohletz and McQueen, 1981) to be namic drag effects are important. Size data strongly influenced by the mass interaction from samples produced by naturally occur- of water and melt (Fig. 8). ring steam explosions at terrestrial volca- Ejection phenomenology can be com- nocs (Sheridan and Wohletz, 1983) relate pared with mechanical energy efficiency three main groupings: (I) ballistic, (2) dry calculations (Table II) based on pressure rec- surge, and (3) wet surge to the dominant ords from four experiments. A Redlich- eruption mode (Fig. 10). These groups, Kwong equation of state is used instead of a which are a function of water-melt interac- Murnaghan equation (Kieffer and Simonds, tion, have a direct bearing on rampart cra- 1980) in order to better approximate the ter ejecta morphology. P-V relationships of a water-rich system. The systematics of magma-water inter- More detailed descriptions of this experi- action has only recently been investigated. mentation are given by Wohletz and Studies in progress (Sheridan and Wohletz, McQueen (1981, 1983) and Wohletz (1980). 1981; Wohletz and Sheridan, 1983)of small The efficiency of conversion of melt ther- terrestrial volcanoes show a strong correla- mal energy to mechanical energy of ejecta- tion among deposit morphology, particle menta is a function of water-melt ratio (Fig. size, and eruption mode related to various 9). An increase in water interaction causcs degrees of water interaction. Products greater melt fragmentation; associated va- range from cinder cone (ballistic), through porization results in high-energy surges. tuff rings (dry surge), to tuff cones (wet The maximum degree of superheating and surge). Interaction of water with basaltic melt fragmentation for thermite occurs for magma at small mass ratios (H20/melt less mass ratios between 0.3 and 0.6. than 0.10) produces cinder cones composed The mode of particle transport is directly of centimeter-sized, ballistic clasts that fol- related to particle size, especially if aerody- lowed parabolic trajectories. Highly explo- MARTIAN RAMPART CRATER EJECTA 27

{']

BALLISTIC FLUIDIZED FLUIDIZED SUPERHEATED CONDENSING STEAM STEAM AND BALLISTIC ~MALL AMOUNT OF 1"120 ~ EQUAl. MASSES 1-120 AND MELT

FIG. 8. Phenomenology of water-thermite experiments. As the ratio of water to melt increases the ejection changes from ballistic to clouds of fine-surge clasts to sputtering emission of coarse and fine clasts in weak surge-like explosions. sire interactions with mass ratios from 0. I to teractions with mass ratios above 1.00 pro- 1.00 create tuff rings composed of micron- duce tuff cones of millimeter-sized particles sized particles transported to extreme run- that are deposited from dense surges with out distances by low-density surges with wet, saturated steam. We speculate that dry, superheated steam. Less explosive in- ballistic, dry surge, and wet surge deposits

TABI.E 11

THF.RMITE-WATER EXPERIMENTAL SUMMARY

Water-melt Approximate Ejecta Pressure ~ Transport ratio efficiency" size (cm) (% of maximum)

A. 0-0.2 <10 10° 10°-I01 sec, <100 bar Ballistic B. 0.2-1.0 10 to 100 10 -4 10 ~ sec, >220 bar Fluidized in superheated steam C. 1.0-10.0 10 to 20 10-: 10°-101 sec, 0-220 bar Fluidized in condensing steam and ballistic D. > 10.0 < 1 l0 t 102 sec, 0-40 bar Fluid flow

Efficiencies--ratio of mechanical to thermal energy calculated as percent of the thermodynamic maximum for fuel-coolant explosions. h Duration and magnitude of pressure pulse. 28 WOHLETZ AND SHERIDAN

I.{~ F" t ---" ?~y." r ..... I i force between the fluid and particles bal- ances the weight of the grains and the J L -T" "I" "X a: 6 A ~ i whole mixture behaves like a fluid." This '," 0.30 r- ~, \ definition has been adapted to current anal- . 1 \\ I , \ yses of pyroclastic flow emplacement (for "'" /B "" "-- 0.20 " -. / example: Sparks, 1976: Wilson, 1980). However, it does not include target fluid- ~, o.10 ization which results from acoustic and "." ,, seismic energy present in the material sur- i..... I 1 .3_...... ~ ~ a . ~.L~ 0 20 0.40 0.60 0.80 I.O0 rounding the crater after impact (Schttltz and Gault, 1975: Melosh, 1979: 1982). As MASS RATIO !1~0 /MEI.T natural fluidized systems move laterally, FI(;. 9. Efficiency vs mass ratio for water-melt ex- the upward fluid flux generally decreases periments. Dashed line show.s the inferred projection with time and distance. Concurrently, the of the efficiency curves. Curve A represents super- frictional resistance among grains and the heating and curve B is for nonsuperheating. substrate increases. This friction slows lateral movement and eventually brings the could correspond to lunar-type, lobate- flow to rest. type, and pedestal-type ejecta blankets ob- l,aboratory experiments have been con- served on other planetary surfaces. ducted to determine the upward velocity of expanding gases required to fluidize pyro- FI.UII)IZATI(.)N clasts in the size range produced by vapor Fluidization has been used in the litera- explosions (Sheridan, 1979: Sparks. 1979: ture in reference to a wide spectrum of pro- Wilson, 1980). As expected, the experimen- cesses involving the mass flowagc of partic- tal minimum fluidizing velocity on Earth in- ulate materials. For our purpose we define creases with grain size (Fig. 11). The re- fluidization as "a mixture of particles (solid quired velocities would be much less on or liquid) suspended by an upward moving Mars than on Earth because of the reduced fluid (liquid or gas) such that the frictional gravitational field. Particles in the size

lO.O I I i I I ]0.0 Ballist c ,~ ...... Mass Flow or Field ....!!:iii~iii::i::iii::i::i::i:iiiii::. Surge Field I :'::: Martian i!!iii :::::::::!:::~: Tuff : i:::::::-...,..:.::::i:i:!:i:i:i:i:i:i:i:i"':~:: z :c ~r B ~ Cone i::M i::iiiii::i:::;ii~::~::~i!ii:::i::!::H i::ii i::ii: eedestat :~i~i~: 1.0 'i!!!!:i:.:c :<<; Type i)!:!:!ii::::::::ii!!>-.- .: :--. ~::~:~:~:i:i:!:i~!~:.:.:~i:i:i:i:~:~:~:~:~:~:~:!:i:?:i:!:i:i:i:i:i:i:i:i~i:i:i:~:i::i::: ~ i i~::i~! P ::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::.... 0 , .:i:iiiii~i~iiii:::::::iii~ Ring ~i~iiiii'~'iill :ii:iiii!!/!::~/:::~::~ima.ian :: ::: :::::::: ~: :~:I: :::::::::::::::::::::::::::::::::::::: ~ :~:~ /~" O.I ,,::::::::::::::: ~::::: ::::::::::::: :::::::::::::::::::::::::::::::: 0.1 ...... ii::::i::i!ii::i::::i:ii!i::::i:/iiii Rampart ::ii ...... ::zgsf!i:f:)i=ili::iiif=:ii::::ii:::/::/:=:iiif:.iii:=::iii!::5!~ss: Type ii+SSi:: •:::::':':':: -: ::::::::::::::::~'~" '::'::~:::iii:.:.:.:.:::ii~i~ '"' ""'"<:i:i:::!::':" : :::.,: .: :.:: ::.: ::i:: ~::~i!::iC i n d e t '::':::::::::::::::~ :':::'~ iL.... ray :~...... Type

,z:!:!:!:!i!!~:!:::::. / ======0.01 , F" ~ , A 0.01 ' 'i" t I I -2 -1 0 I 2 COARSE FINE Median Diameter ~ Grain Size

El(;. 10. Plots of water-melt mass ratio versus grain size. Data are from interred terrestrial volcanic analogs and show grain size regions fl)r ballistic emplacement and vapor supported surge transport, where B ~- ballistic emplacement typified by grow, th of cinder cones (MeGetchin et./., 1974): P planar surge: M : massive surge: S = sandwave surge: H - hummocky material. Tuffcones consist of wet surge products characterized by planar and massive deposits and tuff rings consist of dry surge products which are sandwave and massive surge. These volcanic ejecta morphologies are related to impact ejecta morphologies in the right-hand diagram. MARTIAN RAMPART CRATER EJECTA 29

F MedianS'zeq levees). In column C particles are separated Range -- by a high degree of fluidization where void 16 space (~) is greater than 90%. In this example the viscous effects of the interparticle gas dominate over inertial effects of the particles. Saltating viscous flow of particles 2 bouncing over a substrate results in devel- -° ! l opment of dunes similar to desert dunes studied by Bagnold (1954a,b). An initial 0.50 highly fluidized ejecta cloud (column C) will Z 0.25 lose fluidizing gases (defluidize) as it flows

~ 0.13 away from the crater and pass through an intermediate state (column B) where vis- 0.06 cous and inertial forces are nearly equal.

0.03 Resulting deposits are hummocky and non- ,,p, continuous due to clots of particles segre- qb l J I II gating in the flow and falling out of the 2 5 I0 20 50 cloud to form mounds [see "'clot slugging" MINIMUM FLUIDIZING VELOCITY (CM/S) (Leva, 1959)]. FiG. I I. Relationship of grain size to minimum fluidizing velocity of pyroclasts on Earth (from Sheridan. PARTICUI.ATE DISPERSAL 1979). Grain size has a critical effect on the mechanism of transport and deposition in range produced by maximum-efficiency fluidized systems. True fluidization affects steam-blast explosions (I to 50/,tm) can be only a restricted range of particle sizes fluidized by a modest upward expansion ve- which approximate three size classes for locity. These velocities of steam relative to terrestrial ash flows (Sparks. 1976). Elutria- that of particles probably exist near the cra- tion is a common feature of fluidized sys- ter by superheated steam expansion. tems when the upward velocity of the fluid The effects of fluidization upon transport is greater than the downward terminal fall of particulate materials have been studied velocity of particles in that medium (Kunii extensively in industrial processes (for ex- and Levenspiel, 1969). The smallest parti- ample, Leva, 1959; Kunii and Levenspiel, cles, elutriated from the top of a flow by the 1969). These basic principles have been ap- escaping fluid, may be carried away as a plied to geologic phenomena (Reynolds, 1954: Sparks, 1976; Wohletz and Sheridan, 1979). The columns in Fig. 12 show particle A B C separation for three degrees of fluidization. In column A, particles are barely separated at the incipient level of fluidization where Oo void space (~) is 60%. Grains are suffi- :,',,;'.¶ ciently close so that flow energy is transmitted by grain collisions (inertial effects) and nonfluidized, laminar flow results. Result- ~-60 ing deposition is related to traction load or ~=.75 ¢-95 avalanching flow of particles and produces FIG. 12. (;rain dispersal in flows of various types. planar deposits with inverse grading. Such (A) Grain flow during rampart tbrmation. (B} Quasi- fluidized flow prior to rampart lbrmation. (C) Dis- flows have a finite yield strength that pro- persed viscous flow in surge clouds proximal to the duces flow front scarps (ramparts and crater. 30 WOHLETZ AND SHERIDAN dilute suspension in an overriding cloud. ing expansion of the heated vapor. The role Particles that are too large to be fluidized of fluidization diminishes outward from the may sink, float, or be transported depend- crater as the fluid (steam) is lost or con- ing on their density and the Reynolds num- densed so that the main part of transport is ber of the flow. incipiently fluidized. By the terminal stagc Fluidization enhances viscous transport of emplacement the system is bcst de- and grain-flow (inertial flow), although the scribed as laminar grain flow or viscous latter can occur in a vacuum. Grain-flow transport, depending on the content of en- (Bagnold, 1954, 1956: Hsu, 1978) refers to trained fluid. movement of cohesionless particles under the influence of gravity in which the flow SURGE MECHANISM energy is transmitted through the collisions Several recent papers on volcanic base of grains. The range of particle concentra- surges, pyroclastic surges, or gravity- tion in grain flow may be sufficient to dupli- driven density flows have a bearing on our cate the bed forms exhibited by fluidized model for emplacement of vapor-modified transport. For this reason an atmosphere is ejecta sheets on Mars (Waters and Fisher, not necessary for the formation of pat- 1971; Wohletz and Sheridan, 1979: Fisher. terned ejecta deposits produced by surge 1979; Reimers and Komar, 1979). Surge flowage although it is necessary to produce transport has been defined (Wohlctz and the massive flow of Schultz and Gault Sheridan, 1979) as a "'time-transient, un- (1979). In viscous flow, common to terres- steady horizontal flow of particles and gas trial fluviatile systems, grains arc trans- that occurs as a pulse or series of pulses in ported due to shear and buoyancy stresses which the kinetic energy rapidly decays.'" exerted by the moving fluid medium. If sufficient kinetic energy is transferred by From the preceding discussions, ejecta grain-to-grain collisions an interparticulatc dispersal may be classified as ballistic, fluid phase is not necessary for flowagc of a grain-flow, or fluidized. Schultz and Gault cohesionless mass of solid fragments, such (1979) describe ballistic emplacement in as a landslide (Helm, 1932; Hsu, 1978). As which the largest fragments follow para- in the case of industrial grain flows, the bolic paths with little atmospheric effect. spacing of particles necessary to allow mo- Intermediate-sized particles follow ballistic bility is caused by the net dispersive pres- paths strongly affected by atmospheric drag sure due to the kinetic energy of particle due to a greater surface area to mass ratio. collisions (Bagnoid, 1954: 1956). Small particles, however, may become flu- Viscous, drag effects of an interparticu- idized by rapidly expanding gasses or en- late fluid on grains is important in fluidized trained in grain flow. During fluidized trans- systems (Kunii and Lcvcnspicl, 1969) and port, the finest-sized population of particles in natural fluid transport of grains as slurrys may be removed from thc cjecta flow by or dilute grain dispersions. In high-concen- elutriation and transported by suspension tration flows (Fisher, 19711 the void frac- in a cloud. Particles traveling at high veloci- tion or particle concentration per unit vol- ties parallel to the ground surface may ume can strongly affect such important move by grain flow if there is little or no fluid properties as viscosity and yield fluidization. Viscous transport of these par- strength. Thus, a surge pulse having a high ticles occurs if an intergranular fluid trans- velocity and low viscosity is turbulent at mits sufficient lateral shear stress to cause high void fractions and will approximate particle motion. Thus, during and after cra- Newtonian behavior. Frictional forccs after excavation of wet targets several inter- fecting particle motion increase in signifi- gradational processes may occur. Initially, cance with decreasing void fraction. This ballistic and fluidized transport occurs dur- produces a higher viscosity with a finite MARTIAN RAMPART CRATER EJECTA 31

Bingham yield strength, so that, with in- tance, the resulting increase in their particle creasing particle concentration, flows be- concentration causes a change to inertial come laminar and nonNewtonian (Sheridan grain flow with a finite yield strength. Dur- and Updike, 1975; Wilson, 1980). ing this transition from viscous to grain- Our model of rampart ejecta emplace- flow motion, deposition occurs en masse ment (Fig. 1) assumes that the initial para- producing thick deposits that pond and are bolic trajectories of ejected particles may strongly affected by topography. Flow be strongly modified by aerodynamic drag. lobes terminate in ramparts at this stage. Because this drag effect dramatically in- Laboratory fluidization experiments creases as the grain size decreases (Schuitz (Wilson, 1980) and field studies (Sparks, and Gault, 1979), ballistic fragments less 1976; Wright, 1979; Wright and Walker, than a few millimeters in diameter are lim- 1981) support the above conclusion. Par- ited to low-altitude paths during excavation tially fuidized particulate flows, such as of Martian craters. If sufficient fine material pyroclastic fows, do not slow down gradu- is produced, it would collapse downward as ally, but suddenly decelerate to a halt (Wolf a density flow. As this turbulent mixture of and Wright, 1981), producing prominent fine particles and vapor reaches the ground, flow scarps. it spreads outward as a horizontally surging cloud driven by the force of gravity ENERGY LINE (Young, 1965). Explosive melt-water inter- A simple heuristic model of surge ejecta action causes a series of pulses as demon- emplacement can be formulated based on strated by drop size and large (100 kg) ex- gravity-driven dispersal controlled by an periments (Nelson and Duda, 1981; energy line (Sheridan, 1982). A detailed de- Wohletz and McQueen, 1981). Larger ex- scription of the energy line model is given plosions produced by the interaction of by Hsu (1975, 1978). Hsu explains mass magma and sea water likewise are pulsing. transport of landslide material by grain The frequency of explosive pulses is depen- flow. His model also is useful in analyzing dent upon system size. Experimental fre- emplacement of pyroclastic flows (Sheri- quencies are on the order of micro- to milli- dan, 1979) and steam-blast explosions seconds, whereas volcanic systems display (Sheridan, 1980). The principal assumption pulse frequencies of minutes. For this rea- is that an energy line can be drawn connect- son, vapor explosion during Martian crater ing the highest elevation of the material be- excavation would probably produce multi- fore flowage to the toe of the resulting de- ple surges that decrease in energy with time posit. The Heim (1882) constant (tangent of and distance from the crater. As these tur- the energy line slope, tz), gravitational ac- bulent mixtures of fine particles and vapor celeration (g), and the configuration of the collapse to the ground, they develop into underlying topography are sufficient con- horizontally surging clouds that are essen- straints to permit calculation of the main tially driven by the force of gravity. Contin- flow parameters. The elevation of the en- ued vapor explosions and collapses during ergy line above the ground surface (Ah) de- crater excavation would produce multiple fines the velocity potential (vp), surges that decrease in energy with time Vp = (2gAh) le. (I) and distance from the crater. The surges initially move as dilute, vis- The horizontal acceleration (a~) is a func- cous flows from which the suspended and tion of the topographic slope (/3) and the saltating grains are deposited, particle by Heim constant (~) (Hsu, 1978), particle, in thin beds that mantle the exist- an = g(sin fl - /z cos/3). (2) ing topography. As fluidization levels in the surges drop below a critical value with dis- The velocity and runout times for any part 32 WOHLETZ AND SHERIDAN of the flow field can be calculated by equa- constructing an appropriate energy surface tions given by Hsu (1978). instead of an energy line for the three-di- For example, apply the model to a 20- mensional field. From the potential veloc- km-diameter Martian crater with a rampart ity, a first-order flow field can be calculated lobe extending 30 km from the rim. The assuming steady, nonviscous flow. Because bulk of the particles comprising the rampart of the sensitivity of flows to topography, ejecta deposit are probably less than 250 the limiting constraints for optimum appli- /xm in diameter, assuming high fragmenta- cation of the model to an analysis of ram- tion caused by the presence of the water. part ejecta is the availability of adequate The Martian atmospheric drag on grains topographic base maps. less than 1 cm diameter would limit thc vertical ballistic elevation to less than one-half DISCUSSION the crater radius which in this case would Consideration of the rampart ejecta be 5 km (Schultz and Gault, 1979). A grav- maps, the experimental work, and observa- ity collapse of a density cloud from this tions of terrestrial deposits of vapor explo- height yields a potential velocity of about sion origin allows the following illustrations 200 m/see. Assuming a level runout sur- of the model to be put forth (Fig. 13). This face, the deceleration of the cloud would be figure incorporates the aspects of fluidized -0.6 m/see 2 and the minimum runout time transport and maintains that ramparts and would be about 300 sec. terraces are deposited as a result of the This model can be applied to more com- transition from quasi-fluidized viscous plicated three-dimensional situations by ejecta to inertial fixed-bed emplacement.

"~ ~.~ ~1 - VOID SPACE

¢ i-- 2.<- • "~ ~ 0 "o.9 viscous. SALTA11NG • . ' "\ "~ /¢ TRANSPORT II. : " ~ DUNE DEPOSITION

..1--/.'- ." "-" ~ / VISCOUS INERTIAl

b. .~.S i ." ~ " • ~ ..: .-. ~ MASSIVE-HUMMOCK "~: " " ::; :':,"~ ~: "': " "" '/"" -- DEPOSITION

~0.6 INERTIAl. FLOW PLANE DEP~

=0.4 "FROZEN FLOW" FIXED BED RAMPART d.

DEPOSIT MORPHOLOGY

INCREASING DEPOSITION- Dun~ Hummocka Rampart

FiG. 13. Detailed model of rampart ejecta emplacement. MARTIAN RAMPART CRATER EJECTA 33

Figure 13 (a-d) illustrates the develop- emplacement depends upon the degree of ment of deposit morphology by fluidized segregation of volatiles from particles dur- transport of ejecta (stages 3 and 4 shown in ing emplacement. If steam is highly super- Fig. 1). A facies distribution develops with heated during transport, the tendency of dunes or ridges of ejecta near the crater, segregation of the steam from the particles thin hummocky deposits at intermediate is increased and deposits will be dry. On distances from the crater, and gradually the other hand, if steam is saturated during thickening planar deposits with distal ram- transport, much of it will condense on parts (Fig. 13e) at the limits of continuous ejecta particles resulting in wet deposits ejecta deposits. Repeated vapor explosion subject to postemplacement hydration. The pulses during and after crater excavation experimental curve of Fig. 9 is shown in produce multiple lobes that obscure the Fig. 14 with a dry and wet division. Increas- proximal ejecta morphology and result in ing water-melt ratios in the dry region overlapping ramparts. Consolidation of the results in higher melt fragmentation, higher deposits is primarily dependent upon the rates of thermal energy release from melt to degree of post emplacement ejecta altera- the water, and increased vapor superheat- tion (Wohletz and Sheridan, 1981). ing. In the wet region, however, increasing Findings of the Viking lander (Toulmin et water-melt ratios will quench the melt be- al., 1977) suggest that much of the Martian cause of the excess total heat capacity of surface material is composed of hydrated the water. Steam produced becomes in- silicate or clays. Hydration of ejecta from creasingly cooler and water droplets larger. vapor explosions in impact events (Kieffer The division between wet and dry is of and Simonds, 1980)is expected. The degree course gradational, and dependent upon the of hydration depends upon the wetness of melt enthalpy and the degree of confine- ejecta during and after emplacement. The ment in the area of interaction. amount of water trapped in a deposit after The result of wet and dry vapor expio-

MECHANICAL 100

ul LOBATe E.IEcTA ~ PEDESTAL EaECTA

< ,cE~ / a_ \ \ \

WET THERMAL 0 I I I .01 .1 1 10 100 MASS MIXING RATIO (WATER/MELT) FIG. 14. Partitioning of thermal and mechanical (0-100%) emplacement energy to the water-melt ratio. The wet (water, steam, c!ast) field and the dry (steam, clast) field are shown as well as curves for water and ice. The partition values are approximated and mixing ratios are calculated from thermite- melt experiments. 34 WOHLETZ AND SHERIDAN sions upon ejecta morphology is summa- relatively short runout distances should be rized in Fig. 15. Impacts into a dry target expected where target water is in the form generally result in dominantly ballistic em- of ice. This is because of the increased heat placement of ejecta with subsequent atmo- needed to vaporize ice over that required spheric effects that may limit the range of for liquid water. Hence, maximum, explo- continuous ejecta deposits. An increase in sive heat exchange occurs at lower water- the mass ratio of water to impact melt in the melt mass ratios in an icy terrain. For a zone of excavation increases the effect of water-rich Martian crust of constant water fluidized transport. Flow emplacement oc- content (near the explosive maximum), ter- curs where target water abundance pro- raced, stuby ejecta flow deposits are ex- duces high-energy vapor explosions. These pected to form at higher latitudes and alti- target areas are characterized by the forma- tudes where water is ice. This result is tion of ramparts and maximum ejecta run- supported by Mouginis-Mark (1979a) who out. Water contents in excess of that shows preliminary data indicating that pan- needed for high-energy vapor explosion cake (type 6) craters are more common in result in wetter, less-fluidized flows with in- high latitude and altitude areas and that cra- creasing cohesion. Still wetter conditions ters with ejecta runout distances (ER) of 0 tend toward ejecta terrace formation and a to 4 crater radii are favored in lower lati- diminished runout distance. The tendency tudes. However, Mouginis-Mark (1979a) for formation of terraced ejecta deposits of also demonstrates that high latitudes are fa-

DRY o BAI,LISTIC

O.S

FLOW

1.0

/ ~'~

WET

FIG. 15. Schematic profile of ejecta blankets showing the relationship of deposit morphology to water content of the target material. Water contents are shown on the vertical axis as mass ratio of water to melt in the zone of target melting. The runout distances of ejccta shown are variable and scaled approximately to crater radius. MARTIAN RAMPART CRATER EJECTA 35 vorable for ejecta runout distances greater determined the degree of vapor expansion than 4. This suggests that strongest ejecta and melt fragment size. The movement of fluidization may occur in areas where the the resulting vapor-fluidized ejecta clouds optimum water-melt ratio is controlled by followed a predictable path where initially water in the form of ice. In other words, the viscous flows decrease in energy and de- presence of ice shifts the efficiency curve gree of fluidization as they move away from peak nearer to the water-melt ratio of wa- the crater. At the distance from the crater ter-poor targets. where flows were deflated below the critical fluidizing level, deposition of ramparts be- CONCLUSIONS gan en masse as the shear stress of the flow Much of our present understanding of the decreased below the yield strength. Ter- evolution of planetary bodies is based upon raced deposits resulted from poorly fluid- studies of meteorite impacts. Comparative ized ejecta moving in laminar flows away analysis of impact structures on Moon and from craters. The morphology of Martian Mercury with those of the Earth and Mars impact craters is therefore strongly related strongly supports the hypothesis that sur- to the physical state and amount of water face water can greatly influence the impact present in the crust at the time of impact. process. Existence of ejecta showing evidence of surface flow such as lobate mar- ACKNOWI.EDGMENTS gins, ramparts, and terraces, cross-cutting relationships, topographic effects, and Robert McQueen directed and provided technical support for the experimental work. Ronald Greeley multistratigraphic deposits around craters encouraged our research in impact ejecta emplace- suggests emplacement as fluidized flows. ment and provided Viking imagery as well as technical Studies of terrestrial fluidized flows of par- advice on cratering mechanisms. Financial support ticles as well as industrial fluidization from NASA Grants NSG-7642 and NAWG-245 is processes provide adequate analogs to gratefully acknowledged. interpret the emplacement process. Experiments dealing with the explosive in- REFERENCES teraction of thermite melt and water show BAGNOLD, R. A. (1954a). The Physics of Blown Sand that the mass ratio of water to melt and the and Desert Dunes. Methuen, London. degree of confinement in the zone of inter- BAGNOLD, R. A. (1954b). Experiments on a gravity- action strongly affect the emplacement of free dispersion of large-solid spheres in a Newtonian fluid under shear. Proc. R. Soc. London 225, 49-63. ejecta. Important factors of emplacement BAGNOLD. R. A. (1956). The flow of cohesionless of ejecta fluidized by vaporized water are grains in fluids. Phil. Trans. R. Soc. London 249, the size population of particles, the physi- 291-293. cal state of the volatile substance (condens- BLASIUS, K. R., AND J. A. CUTTS (1980). Global patterns of primary crater ejecta morphology on Mars. ing-saturated or superheated-expanding NASA Tech. Memo. 82385, 147-149. steam) and the level of ejecta cloud fluidiza- CARR, M. H., L. S. CRUMPLER, J. A. CUTTS, R. tion (the volume ratio of gases to particles). GREELEY, J. E. GUEST, AND H. MASURSKY (1977). We conclude that water either on or be- Martian impact craters and emplacement of ejecta neath the surface of Mars interacted with by surface flow. J. Geophys. Res. 82, 4055-4065. impact melt during crater excavation to CARR, M. H., AND G. S. SCHABER (1977). Martian permafrost features. J. Geophys. Res. 82, 4039- produce vapor explosions. These explo- 4054. sions altered initial parabolic trajectories of CINTALA, M. J., AND P. J. MOUGINIS-MARK (1980). ejecta to form partly fluidized flows that de- Martian fresh crater depth: More evidence for sub- posited the rampart and terraced ejecta, a surface volatiles? Geophys. Res. Letters, 7, 329- 332. conclusion in agreement with Schuitz and COI.GATE, S. A., AND R. SIGURGEIRSSON (1973). Dy- Gault (1979). Furthermore, the water-to- namic mixing of water and lava. Nature 244, 552- melt ratio in the zone of crater excavation 555. 36 WOHLETZ AND SHERIDAN

CORRADINI, M. L. (1981). Phenomenological modeling MCGETCHIN, T. R.. M. S~t'rLE, AND B. A. CHOULT of the triggering phase of small-scale steam explo- (1974). Cinder cone growth modeled after Northeast sion experiments. Nucl. Sci. l:.ng. 78, 154-170. Crater, Mount Etna, Sicily. J. Geophys. Res. 79, FISHER, R. V. (1971). Features of coarse-grained, high 3257-3272. concentration fluids and their deposits. J. Sediment. MELOSH. H. J. (1979). Acoustic fluidization: A new Petrol. 41, 916-927. geologic process'? J. Geophys. Res. 84, 7513-7520. FtS~IF.R, R. V. (1979). Models for pyroclastic surges MELOSH, H. J. (1982). A schematic model of crater and pyroclastic flows. J. Voh'anol. Geotherm. Res. modification by gravity. J. Geophys. Re.s. 87, 371- 6, 305-318. 380. GAUl.T, D. E.. AND R. GREELEY (1978). Exploratory MOORE. J. G. (1967). Base surge in recent volcanic experiments of impact craters formed in viscous- eruptions. Bull. Volcanol. 30, 337-363. liquid targets: Analogs for Martian rampart craters? MOtJGINIS-MARK, P. J. (1978). Morphology of Martian Icarus 34, 486-495. rampart craters. Nature 272, 691-694. GREELEY, R., J. FINK, D. E. GAULT, D. B. SNYDF.R, MOUGINIS-MARK. P. J. (1979a). Martian fluidized cra- J. E. GUEST, AND P. H. SC:HULTZ (1980). Impact ter morphology: Variations with crater size. lati- cratering in viscous targets: Laboratory experi- tude, altitude, and target material. J. Geophys. Res. ments. Proe. Lunar Plan. Sci. ConJ: llth, 2075- 84, 8011-8022. 2097. MOUGtNtS-MARK, P. J. (1979b). Ejecta emplacemcnt HEAD, J. W., AND R. ROTH (1976). Mars pedestal cra- of Martian impact crater Bamburg. Proc. Lunar ter escarpments: Evidence for ejecta-related em- Planet. Sci. Cot~f. IOth. 1651 - 1668. placement. In papers presented to the Symposium ~]" MUrCH. P., AND A. WORONOW (1980). Martian ram- Planetary Cratering Mechanics, pp. 50-52. The Lu- part and pedestal craters" ejecta emplacement: Co- nar Science Institute. Houston, Tex. prates Quadrangle. Icarus 41, 259-268. HEIKEN, G. H. ( 1971 ). Tuff rings: Examples from Fort NF.LSON. L. S., AND P. M. DUDA (1981). Steam explo- Rock-Christmas Lake Valley, south-central Ore- sion experiments with single drops of CO: laser- gon. J. Geophys. Res. 76, 5615-5626. melted iron oxide. Trans. Amer. Nucl. Soc. 38, 453- HEtM, A. (1882). Der Bergsturz von Elm. Z. Deutsch 454. Geol. Ges. 34, 74-115. OLLIER, C. C. (1967). Maars, their characteristics. HELM, A. (1932). Bertsturz und Menchenleben. l-retz varieties, and definition. Bull. Volcanol. 30, 45-73. and Wasmuth. Zurich. PECKOVER, R. S.. D. J. BUCIIANAN. AND D. E. ASHBY Hsu, K. J. (1975). On sturzstroms--Catastrophic de- (1973). Fuel-coolant interactions in submarine vol- bris streams generated by rockfalls. Geol. Soc. canism. Nature 245, 307-308. Amer. Bull. 86, 129-140. REYNOtDS, D. L. 11954). Fluidization as a geologic Hsu, K. J. (1978). Albert Helm: Observations on land- process and its bearing on the problems of intrusive slides and relevance to modern interpretations. In granites. Amer. J. Set. 252, 577-613. Rockslides and Avalanches, Vol. I. Natural Phe- REnMERS, C. E.. AND P. D. KOMAR (1979). Evidence nomena (B. Voight, ed.), pp. 71-93. Elsevier, New for explosive volcanic density currents on certain York. Martian volcanoes, h'arus 39, 88-110. JOHANSEN, L. A. (1979). The latitude dependence of SANDIA I.ABORATORIES (1975). ('ore-Meltdown Ex- Martian splosh cratering and its relationship to wa- perimental Review, SAND-74-0382. Sandia l,abora- ter. NASA Tech. Memo. 80339, 123-125. tories, New Mexico. JONES, E. M., AND J. W. KODIS (1982). Atmospheric S(:HULTZ, P. H., AND 1). E. GAUI.T (1975). Seismi- effects of large-body impacts: the first few minutes. cally induced modification of lunar surface features. Proe. Snowbird Conf.. Geol. Soc. Amer. Spec. Pup. Proe. Lunar Sci. Conf. 6th, 2845-2862. 190, 175-186. SCHUL'IZ, P. H., AND D. E. GAUIT 11979). Atmo- KIEFFER, S. W., ANn C. H. SIMONDS (1980). The role spheric effects on Martian ejecta emplacement. J. of volatiles and lithology in the impact cratering pro- Geophys. Res. 84, 7669-7687. cess. Rev. Geophys. Space Phys. 18, 143-181. SCHUI.TZ, P. H., AND D. E. GAt'I.T (1982). Impact KUNII. D., AND O. I.EVENSPIEI. (1969). Fluidization ejecta dynamics in an atmosphere: Experimental Engineering. Wiley, New York. results and extrapolations. Proc. Snowbird Con/:. I.EVA, M. (1959). Fhddization. McGraw-Hill, New Geol. Soc. Amer. Spec. Pap. 190, 153-174. York. SHERUDAN, M. F. (1979). Emplacement of pyroclastic LtPSETT, S. G. (1966). Explosions from molten materi- flows--A review. Geol. So('. Amer. Spe(. Paper als in water. Fire Teehnol. May, 118-126. 180, 125-136. LORENZ, V. (1970). Some aspects of the eruption SHERIDAN. M. F. (1980). Pyroclastic block flow from mechanism of the Big Hole Maar, Central Oregon. the September, 1976 eruption of l.a Soufriere Vol- Geol. Soe. Amer. Bull. 81, 1823-1830. cano, Guadeloupe. Bull. Voh'anol. 43, 397-402. LORENZ, V. (1973). On the formation of maars. Bull. SHERtDAr,,', M. F. (1982). The energy line: A heuristic Volcanol. 37, 183-204. model for Martian rampart ejecta sheets. Reports of MARTIAN RAMPART CRATER EJECTA 37

the Planetary Geology Program, 1981-1982. NASA WOHI.ETZ, K. H., AND R. McQUEEN (1981). Experi- Tech. Memo. 84211. mental hydromagmatic volcanism. Trans. Amer. SHERIDAN, M. F., AND R. G. Ur'DIKE (1975). Sugar- Geophys. Union, Los, 62, 1085. loaf Mountain tephra--A Pleistocene rhyolitic de- WOHLETZ, K. H., AND R. McQUEEN (1983). Experi- posit of base-surge origin. Geol. Soc. Amer. Bull. mental studies of hydromagmatic volcanism, In Ex- 86, 571-58 I. plosive Volcanism: Inception. Evolution, and Haz- SHERIDAN, M. F., AND K. H. WOHLETZ (1981). Hy- ards (F. R. Boyd, ed.), Studies in Geophysics, drovolcanic explosions: The systematics of water- National Academy of Sciences, in press. pyroclast equilibration. Science 212, 1387-1389. WOHI.ETZ, K. H., AND M. F. SHERIDAN (1983). Hy- SHERIDAN, M. F., AND K. H. WOHLETZ (1983). Hy- drovolcanic explosions 11: the evolution of tuff drovolcanism: basic considerations and review. J. cones and tuff rings. Amer. J. Sci. 283, 385-413. Volcanol. Geotherm. Res. 17, 1-29. SPARKS, R. S. J. (1976). Grain size variations in ignim- WOHLETZ, K. H. (1980). Explosive hydromagmatic brites and implications for the transport of pyroclas- volcanism. Ph.D. dissertation, Arizona State Uni- tic flows. Sedimentology 23, 147-188. versity, Tempe, Ariz. STOFFLER, D., D. E. GAULT, J. WEDEKIND, AND G. WOHI.ETZ, K. H., AND M. F. SHERIDAN (1979). A POI.KOWSKI (1975). Experimental hypervelocity im- model of pyroclastic surge. Geol. Soc. Amer. Spec. pact into quartz sand: distribution and shock meta- Paper 180, 177-193. morphism of ejecta. J. Geophys. Res. 80, 4062- WOLFF. J. A., AND J. V. WRIGHT (1981). Rhe- 4077. morphism of welded tuffs. J. Volcanol. Geotherm. TOULMIN I11, P., H. K. BAIRD, B. C. CLARK, K. Res. 10, 13-34. KEIL. H. J. ROSE JR., R. P. CHRISTIAN, P. H. EVANS, AND W. C. KELLIHER (1977). Geochemical WRIGHT. J. V. (1979). Formation, Transport and Dep- and mineralogical interpretation of Viking in- osition of lgnimbrites and Welded Tuff~'. Ph.D. the- organic chemical results. J. Geophys. Res. 82, 5625- sis, Imperial College. London. 5634. WRIGHT, J. V., ANDG. P. L. WALKER(1981). Eruption, WATERS, A. C., AND R. V. FISHER (1971). Base surges transport, and deposition of ignimbrites: A case and their deposits: Capelinhos and Taal Volcanoes. study from Mexico. J. Volcanol. Geotherm. Res. 9, J. Geophys. Res. 76, 5596-5614. 11-131. WILSON, C. J. N. (1980). The role of fluidization in the YOUNG, G. A. (1965). The Physics of the Base Surge. emplacement of pyroclastic flows. J. Volcanol. Geo- U.S. Naval Ordnance Lab., NOLTR 64-103. White therm. Res. 8, 231-249. Oak, Md.