Tuesday, March 16, 1999POSTER SESSION I

7:00 p.m. UHCL

Impacts I

Takayama H. Tada R. Matsui T. Iturralde-Vinent M. A. Oji T. Tajika E. Kiyokawa S. Garciaanmd D.Okada H. Hasegawa T. Toyoda K.

Origin of a Giant Event Deposit in Northwestern Cuba and Its Relation to K/T Boundary Impact [#1534]We investigated the Penalver Formation in northwestern Cuba, which is a <180 m thick, normal-gradedcalcareous clastic deposit. This formation must have been formed by a grain flow and huge tsunami wavescaused by the K/T boundary impact.

Kiyokawa S. Tada R. Matsui T. Tajika E. Takayama H. Iturralde-Vinent M. R.Extraordinary Thick K/T Boundary Sequence; Cacarajicara Formation, Western Cuba [#1577]Cacarajicara Formation of western Cuba is the thickest K/T boundary sequence in the world. It is an upward-fining carbonate clastics, at least 300m in thickness. It might be a distal part of the ejector blanket of Chicxulubimpact crater or a giant tsunami deposit.

Ward S. N. Asphaug E.Impact Tsunami: A Probabilistic Hazard Assessment [#1475]We apply linear tsunami theory to the NEO flux to investigate the generation, propagation, and hazard oftsunami spawned by oceanic asteroid impacts.

Matsui T. Imamura F. Tajika E. Nakano Y. Fujisawa Y.K/T Impact Tsunami [#1527]Numerical simulation of generation and propagation of the tsunami caused by the impact at the K/T boundarypredicts that unusually gigantic tsunami attacked the coast of the Gulf of Mexico and the tsunami furtherpropagated worldwide.

Nakamura Y. Christeson G. L. Buffler R. T. Morgan J. Warner M. Chicxulub Working Group Structure of the Chicxulub Impact Crater as Determined from Large-Offset Onshore-Offshore Seismic Data[#1288]Onshore-offshore seismic data over the Chicxulub impact crater revealed for the first time its deep structure.Major features include a central basement uplift and inward-dipping low-angle faults well outside the collapsedtransient cavity.

Poag C. W.Secondary Craters from the Chesapeake Bay Impact [#1047]I document 23 secondary craters on two seismic reflection profiles north of the Chesapeake Bay primary crater(85-km-diameter). Seismostratigraphic analysis is calibrated with lithostratigraphy and biostratigraphy fromnearby outcrops and bore holes.

Plescia J. B.Mulkarra Impact Structure, South Australia: A Complex Impact Structure [#1889]The Mulkarra impact structure in South Australia is interpreted as a complex crater having a diameter of ~20 kmwith a 9 km central pit or peak ring. This interpretation differs from that of Flynn (1988) who interpreted it as asimple 9 km bowl shaped crater.

Henkel H. Reimold W. U.Magnetic Model of the Central Uplift of the Vredefort Impact Structure [#1336]The negative magnetic anomalies occurring within the area of the central uplift of the Vredefort impact structurehave been analyzed. The final model shows the extent of post-impact thermal re-magnetization and revealsstructures related to the collapse of the central rise.

Popov Y. Pohl J. Romushkevich R.Geothermal Investigations of the Ries Impact Structure [#1623]Thermal conductivity measurements of the 1206 m Ries 1973 drill core characterize and differentiate impactformations. From temperature gradients and the new conductivity data, inferences about heat flow and fluidmovements can be drawn.

Pohl J. Geiss E.Investigations of the Ries Crater Ejecta Using a Digital Geological Map, DEM and GIS [#1531]A digital edition of the Ries crater geological map allows, together with digital elevation data, a newinvestigation of the statistical geological and morphological characteristics of the ejecta and their radial andazimuthal distribution.

Herkenhoff K. E. Giegengack R. Kriens B. J. Louie J. N. Omar G. I. Plescia J. B. Shoemaker E. M.Geological and Geophysical Studies of the Upheaval Dome Impact Structure, Utah [#1932]Detailed geologic mapping and geophysical data indicate that Upheaval Dome in Canyonlands National Park(southeastern Utah) originated by collapse of a transient cavity formed by impact, not by salt diapirism aspreviously proposed.

Kenkmann T. Ivanov B. A.Low-Angle Faulting in the Basement of Complex Impact Craters: Numerical Modelling and Field Observationsin the Rochechouart Structure, France [#1544]Low-angle normal faults generate during crater modification in acoustically-fluidized rocks. Faulting is a resultof an impact-induced rheological stratification of the crater floor and a passive rotation due to the uplift of thecentral peak.

Buchanan P. C. Koeberl C. Reimold W. U.Petrogenetic Modeling of the Dullstroom Formation, South Africa [#1290]The Dullstroom Formation was the first phase of a magmatic episode culminating in the formation of theBushveld Complex. This study models available geochemical data to determine whether impact or endogenicterrestrial processes were responsible.

Hough R. M. Vishnevsky S. Abbott J. I. Pal’chik N. Raitala J. Gilmour I.New Data on the Nature of Impact Diamonds from the Lappajärvi Impact Structure, Finland [#1571]New data is presented on the characteristics of impact diamonds from the Lappajärvi impact structure, Finland.

Wolbach W. S. Widicus S. French B. M.Carbon-bearing Impactites from the Gardnos Impact Structure, Norway: No Evidence for Soot [#1043]To test the idea that combustion of the impactor or carbon-bearing rocks could have occurred during the impactthat produced the Gardnos crater, we searched for soot in Gardnos impactites and related rocks. No detectablesoot is observed in any of these Gardnos samples.

King D. T. Jr. Neathery T. L. Petruny L. W.Impactite Facies Within the Wetumpka Impact-Crater Fill, Alabama [#1634]This paper summarizes initial findings from drilling two 190-m deep holes within the Late CretaceousWetumpka impact crater, Alabama. Five intercalated impactite facies from the crater fill unit are described:sands; sandy and cataclastic diamictites; breccias; and blocks.

Komatsu G. Olsen J. W. Baker V. R.Field Observation of a Possible Impact Structure (Tsenkher Structure) in Southern Mongolia [#1041]In 1998 summer, we visited the Tsenkher Structure, a possible impact structure in southern Mongolia. Thestructure’s rim to rim diameter is 3.6 km, and a raised outer rim was observed. Prehistoric stone artifacts werefound near the structure.

Master S. Diallo D. P. Kande S. Wade S.The Velingara Ring Structure in Haute Casamance, Senegal: A Possible Large Buried Meteorite Impact Crater[#1926]A new 48 km-diameter postulated impact crater, the Velingara structure, discovered on satellite images (centredon 14°7'40" W, 13°02'13.2" N, in Senegal), is developed in Mid-Eocene sediments, has a central uplift ofmetamorphic basement, and is buried by Neocene rocks.

Arday A(t). Bérczi Sz. Don Gy. Lukács B.Preliminary Report of Szilvágy-Patkó (Horseshoe): A New (Possible) Impact Crater Remnant in Hungary[#1384]In the last autumn during aerial photographing a central symmetric form was found 250 km SW from Budapest,in Hungary. We consider it to be a crater remnant, it is 500 m diameter and 25 m deep, and its southern wall ismissing. Yet no transformed matter proves its origin.

Origin of a giant event deposit in northwestern Cuba and its relation to K/T boundary impact. H. Taka-yama1, R. Tada1, T. Matsui2, M. A. Iturralde-Vinent3, T. Oji1, E. Tajika1, S. Kiyokawa4, D. Garciaanmd3, H.Okada5, T, Hasegawa6 and K. Toyoda5, 1Geological Inst., Univ. of Tokyo (Tokyo, 133-0033, Japan, taka-yama@geol.s.u-tokyo.ac.jp), 2Department of Earth and Planetary Physics, Univ. of Tokyo, 3Museo Nacional deHistorica Natural, Cuba, 4National Science Museum, Japan, 5Hokkaido Univ., Japan, 6Kanazawa Univ., Japan.

Introduction: Since late 1980's, characteristic

sandstone layers probably related to the Cretaceous-Tertiary (K/T) boundary impact have been found in theGulf of Mexico region and several studies advocatethey were formed by huge tsunami waves caused by theimpact [e.g.,1]. However their tsunami origin is stillcontroversial and one of criticism is the lack of ho-mogenite, a thick, normal graded, structureless depositformed by tsunami in deep sea environment. On theother hand, the Peñalver Formation and its equivalentsin the northwestern Cuba, which contain Late Maas-trichtian fossils, are characterized with a < 180 mthick, normal-graded calcareous clastic deposit withdistinct basal conglomerate, and have been suspectedas a single event deposit possibly related to the K/Tboundary impact [2]. However, convincing evidence fortheir association with the impact has never been pre-sented and its depositional mechanism has not beeninvestigated in detail. In this sturdy, we conducteddetailed field survey of the Peñalver Formation in orderto prove its relation with the K/T impact and to clarifyits sedimentary process. If the 180 m thick PeñalverFormation has genetic relation with the K/T boundaryimpact, it becomes the thickest event deposit formedby the impact, and understanding its sources andsedimentary processes is essential to understand theconsequences of the impact.

Geological Setting: Present Cuban island con-sists of several allochtonous tectonic units which wereassembled by a collision between the North Americanpassive margin and the Proto-Cuban Arc during Paleo-cene to latest Eocene [3]. The Peñalver Formation isconsidered to have been deposited in a northern forearcbasin on the Proto-Cuban Arc which was probablylocated approximately 500 km to the south of its pre-sent position during the late Maastrichtian. The Peñal-ver Formation overlies the Campanian to MaastrichtianVia Blanca Formation with disconformity and is dis-conformably overlain by the Paleocene Apolo Forma-tion [4]. The Via Blanca and Apolo Formations aremainly composed of hemipelagic calcareous sedimentswhich were occasionally interrupted by turbidites fromthe Proto-Cuban Arc. The depositional depth of theunderlying Via Blanca Formation is estimated as 600to 2000+ m based on the ratio of planktonic to benthicforaminifers [4].

The Peñalver Formation: The Peñalver Forma-tion in Havana is approximately 180 m thick and issubdivided into the Basal, Lower, Middle, Upper andUppermost Members. One of the conspicuous charac-teristics of this formation is lack of any evidence of

bioturbation. The Basal Member overlies the underly-ing Via Blanca Formation with irregular erosional sur-face, and consists of massive, poorly-sorted calciruditewith grain-supported fabric. The calcirudite containsabundant macrofossils of shallow-marine origin andoccasional large intraclasts derived from the Via BlancaFormation. These features of the Basal Member areconsistent with those of a grain flow deposit. TheLower, Middle, and Upper Members mainly consistsof massive, calcarenite which shows a single normalgrading. In addition, calcarenite of the Lower Memberis characterized by intercalations of thin conglomeratelayers which are composed of greenish gray, well-sorted, well-rounded mud clasts with a small amountshallow marine fossils such as rudist and repeat at least14 times. The mud clasts are generally 1 - 2 cm indiameter except a few out size intraclasts of over severaltens centimeters in diameter and show a preferred orien-tation with their major axes parallel to the bedding.The calcarenite of these members is characterized bydominance of serpentine fragments and occasionalspinels within non-carbonate fraction and the near ab-sence of shallow water macrofossils, which is distinctlydifferent in composition from the calcirudite of the Ba-sal Member. Abundant large-scale water escape struc-tures in the Lower and Middle Members which suggestrapid sedimentation are another conspicuous features ofthe Peñalver Formation. On the other hand, the UpperMember consists of weakly bedded well-sorted finecalcarenite, and water escape structures are absent. Theresult of grain-size analysis of > 32 µm fraction of in-soluble residues of the calcarenite of the Lower to Up-per Members shows coarse-tail normal grading whichalso suggest settling from high density suspension [5].The Uppermost Member consists of massive homoge-neous calcilutite which contains abundant planktonicforaminifera and cocolith, and shows matrix supportedfabric. This member can be regarded as representingcontinuous upward fining from the underlying UpperMember. The contact between the Peñalver Formationand the Apolo Formation is not exposed at the studiedsites. The chemical composition of the detrital spinelfrom the Lower to Uppermost Members suggests thatit was derived from upper mantle peridotite of islandarc or forearc setting, most likely the ultramafic body ofProto-Cuban Arc. Age-diagnostic planktonic foramini-fers and calcareous nannofossils found through the for-mation indicate various ages ranging from Aptian toMaastrichtian with the assemblage of Campanian toMaastrichtian age, which is similar to that of the un-derlying Via Blanca Formation being most abundant.

Origin of a giant event deposit in northwestern Cuba and its relation to K/T boundary impact: H. Takayama et al.

A calcareous nannofossil, Micula prinsii whose firstoccurrence is estimated as 65.4 Ma [6] is found from alarge intraclast in the Lower Member. No Tertiary spe-cies are found. These microfossil assemblage of thePeñalver Formation suggests that the Peñalver Forma-tion is a mixture of reworked materials and its deposi-tional age is constrained as between 65.4 Ma and 65.0Ma, the first occurrence of Tertiary species. These fea-tures suggest the Lower to Uppermost Members wereformed by settling from a high density suspensionwhich has a different source from the Basal Member.The cause of the suspension could have been explainedby a huge tsunami because the features are similar tothose of a deep-sea tsunami induced deposit, ho-mogenite [7], in following respects; 1) the massivelithology with no sedimentary structure suggestive ofturbidite, 2) the fossil assemblage suggestive of mainlyreworking from underlying strata. Furthermore, it ispossible that thin conglomerate layers in the LowerMember may have been induced by sires of large waveactions during the early stage of settling from dispersedsuspension.

Identification of Ejecta Materials: Altered ve-sicular glass grains are found in the Basal Member andthe lower part of the Lower Member. These vesicularglass grains reaches 2 mm in diameter and show bub-bly texture with bubble diameter of several tens µm.The bubbles generally have smectite rim-cement andcemented either with sparry calcite, smectite or heulan-dite. Heulandite sometimes replaces part of the glass.The concentration of the vesicular glass is highest inthe basal part of the Basal Member and decrease up-ward. Although most of the glass grains are angular inshape, they are similar in texture to bubbly spherule ofprobable ejecta origin described from the lower part ofthe K/T boundary layers in the Gulf of Mexico andCaribbean regions [1].

> 60 µm size quartz grains are isolated and exam-ined to find shocked quartz. Small number of quartzgrains with at least 4 sets of planar lamellas of < 1 µmin width and 5 µm in spacing are found from theLower, Middle and Upper Members. These featuresagree with planar deformation features of shocked quartzwhich is produced by a impact [e.g., 8]. The Maxi-mum size of shocked quartz grains reaches ~ 380 µmin the Lower Member and their grain size decreasesupward in the similar way as other grains. Togetherwith biostratigraphically constrained age, existence ofthese ejecta material suggests that the Peñalver Forma-tion has genetic relation to the K/T impact.

Trigger of the grain flow and tsunamis: TheBasal Member is considered as a grain flow depositderived from a carbonate platform on the Proto-CubanArc to the south of the depositional site [9]. Whereas,the Lower to Uppermost Member is considered to havebeen formed by rapid settling of grains from a highdensity suspension possibly produced by Tsunami.Assuming 600 to 2000 m water depth and using set-

tling velocity for the maximum grain size of 500 µm atthe top of the Lower Member, the time required fordeposition of the Lower Member is estimated as ~ 2 to8 hours after generation of the suspension. In similarmanner, deposition of the Upper Member requires ~ 3to 12 days.

Distribution of altered vesicular glass and theirhigher concentration in the basal part suggest deposi-tion of vesicular glass on the sea floor before arrival ofthe grain flow. Whereas absence of shocked quartzgrains in the Basal Member and their distributionthrough the Lower to Upper Member suggest thatshocked quartz reached the sea floor after deposition ofthe grain flow and the high density suspension wasformed while shocked quartz were still within the watercolumn. Assuming the water depth of the depositionalsite in a forearc basin of the Proto-Cuban Arc as be-tween 600 and 2000 m, the distance from the shallowcarbonate platform to the site as 100 km and the veloc-ity of the grain flow as 30 to 100 km/h, travel timecalculation based on settling velocity of altered vesicu-lar glass and shocked quartz grains within the forma-tion suggests that the initial grain flow started between2 hours before and 11 hours after the impact whereasthe high density suspension occurred within 4 and 12hours after the impact. Consequently, seismic shockwave generated by the impact is most likely for thetrigger of the initial grain flow, on the basis of goodagreement of its estimated arrival time and difference inits composition from overlying calcarenite. Since thefollowing high density suspension is considered tohave produced by tsunami, the first tsunami waveshould have arrived at the depositional site within lessthan 4 and 12 hours after the impact. Furthermore, theconglomerate layers which possibly represent series offollowing large waves which hit the coastal area sug-gest repeated tsunami waves passed intermittentlywithin 6 to 20 hours after the impact. This estimationis consistent with the result of numerical simulation fortsunami waves generated by the K/T boundary impact[10].

References: [1] Smit J. et al. (1996) Geol. Soc.Am. Spec. Pap., 307, 151-182. [2] Bohor B. F. andSeitz R. (1990) Nature, 344, 593. [3] Bralower T. J.and Iturralde-Vinent, M. A. (1997) Palaios, 12, 133-150. [4] Brönnimann P. and Rigassi D. (1963) Ec-logae Geol. Helv., 56 , 193_480. [5] Allen, J. R. L.(1982) Develop in Sedimentology, 30B, 405. [6]Bralower T. J. et al. (1995) Spec. Publ. -Soc. Econ.Paleont. Min., 54, 65-79. [7] Cita M. B. et al. (1996)Sediment. Geol., 104, 155-173. [8] Koeberl C. andAnderson R. R. (1996) Geol. Soc. Am. Spec. Pap.,302, 1-29. [9] Rojas R. et al. (1995) Rev. Mex.Cien.Geol., 12, 272-291. [10] Imamura F. et al.(1999) submitted to Science.

EXTRAORDINARY THICK K/T BOUNDARY SEQUENCE; CACARAJICARA FORMATION,WESTERN CUBA. S. Kiyokawa1, R. Tada2, T. Matsui2, E. Tajika2, H. Takayama2 and M.A.Iturralde-Vinent4;1National Science Museum, Tokyo, kiyokawa@kahaku.go.jp, 2Univ. of Tokyo, 3National Science Museum, Cuba.

Introduction: Western Cuba was situated about 300km distant from the Chicxulub impact crater in theYucatan peninsula at the late Cretaceous [1]. Accordingto Palmer in 1934 and 1945 [2,3], Cretaceous toPaleogene sedimentary units with an unusual upperMaastrichtian sequence are widely distributed in theCuba Island. Pazczolkowski [4] pointed out that thisunusual sequence might be mega-turbidites possiblycaused by the Cretaceous-Tertiary (K/T) boundaryimpact event. Based on thses references [2,3,4], Bohorand Seitz [5] considered that there could be an ejectablanket sequence produced by the K/T impact in theCuba Island. However, for the political reason, theCuban geology was veiled for long time.

Since 1997, we have conducted the geologicalresearch project to clarify the unknown K/T boundarysequences in Cuba. We have studied the sequences fromthree formations; Moncada, Cacarajicara and PenalverFormations, in different tectonic belts. Detailed researchhave revealed that these three formations are the K/Tboundary sequences formed directly by the Chixulubimpact event [6,7]. Here we report the unusual K/Tsequence with thickness of more than 300 m, called theCacarajicara Formation, which is identified as thethickest K/T formation in the world. The followings arethe interim reports of the Cacarajicara Formation.

Tectonic setting of the western Cuba: In recentstudy, the western Cuba is tectonically divided intothree belts; Los Organos, Rosario and Habama (BahiaHonda to Matanzas) belts [e.g., 4]. Each belt is formedby the fold and thrust with serpentinised shear zone andis redeformed by the strike-slip deformations. All ofthese sequences were truncated by left-lateral PinaerFault [8]. The Los Organos belt consists mainly ofJurassic terrestrial shallow water sequence, Cretaceousopen water limestone, and Paleocene micritic limestone.The Rosario belt forms Jurassic shallow watersequence, Cretaceous deep marine carbonate-clasticfacies and Paleocene micritic limestone to the top.Based on the stratigraphic relationship and fossil faunarelationship, these belts were situated as a continentalfragment of the North American Craton at Jurassic[9,10]. The Cacarajicara Formation unconformablyoverlies the Cretaceous deep marine sandstone-mudstone alternation. In Cretaceous, the Rosario belt ismore distal facies than that of the Los Organos belt[10,11]. On the other hand the Habana belt ischaracterized by thin-skinned fold-and-thrust whichbounded to the Rosario belt by ophiolitic andCretaceous volcanic sequences. The Cretaceous toPaleocene sedimentary sequences of the Habana belt areidentified as overthrust allochthon on the Rosario belts[10,11]. In the thrust piles, carbonate rich mega-turbidite sequences, called Penalver Formation, are well

preserved between the upper Cretaceous andPaleocene turbidite sequences. More detaileddescription of the Penalver Formation is reported byTakayama et al. in this volume [6].

Overview of the Cacarajicara Formation: TheCacarajicara Formation was formed in top-to-the northnorthwest fold-thrust belts of the Rosario belt [8,9,10].It consists of a mega upward-fining carbonate clasticssequence. Continuos section with at least 300 m inthickness is observed near the Soroa area. Bedding dipis changed from 80ª to 30ª the northwest. Our researcharea is situated southeast-rim of mapscale synclinal-syncline.

Stratigraphy: The more than 300 m thickCacarajicara Formation is subdivided into fivemembers, that is, Basal, Lower, Middle, Upper andUppermost members. The grain size of clasts in thesequence is gradually decreased to the upwardmembers.

The Basal member disconformably overlies on themiddle Cretaceous Santa Teresa Formation. Theformation is composed of carbonated sandstone-shalealternation formed by distal sequence of the gravityflow. This member is approximately more than 100m inthickness and is characterized by the grain-supportedboulder breccia. Rock-species of the breccia are micriticlimestone, foraminiferal limestone, banded radiolarianchert and radiolarian-bearing black chert. Averagebreccia size is approximately 1m in diameter at basalunit and 20-50 cm at the top in this member. Thebiggest breccia is more than 3 m bedded chert. Theradiolarian chert and black chert boulders may bederived from the underlying Middle Cretaceoussequence (Santa Teresa Formation). Petrographicobservation shows that matrix of the breccia, very fewin amount, contains carbonate sandstone with flowtexture, siltstone with a few foraminiferas and carbonateclast. There are some spherule-like grains in the matrix.Outcrop scale intrusions of sand-siltstone are preservedin this breccia zone. These intrusions contain quartz,foraminifera and fine carbonate abundantly. Sand-siltstone intrusions of microscopic size are alsoobserved in some carbonate breccia. These evidencesuggest that the Basal member was formed in the highpore-pressure situation. Each breccia preserves steep-dipping fracture joints with northwest-trend. Quartzgrains in the sand-siltstone intrusion also contain thesame direction dislocation creek as fracture jointmentioned above. These structres were related by thetectonic deformations in this formations. We have toexclude this fabric to find the shocked quartz.

Lower member contains the clast-support pebblebreccia sequence which is more than 50m in thickness.This member consists of pebbles of micritic limestone,

Extraordinary Thick K/T Boundary Sequence: S. Kiyokawa et al.

foraminifera limestone, rudist fragment and radiolarianblack chert and fragments of quartz, greenish volcanicsand serpentinite. There are some highly deformedquartz grains similar to shocked quartz. The grainboundaries of each calcareous clasts were recrystalised.The outermost rims, a few mm, of the black chertpebble was replaced into carbonate zone. There aremany stylolite zones in this member, especiallydeveloping in grain boundary.

Middle to Upper members contain coarse to finecarbonate sandstone which contains mainly micriticlimestone, foraminifera, rudist fragment with minoramounts of quartz, serpentinite and volcanics grains.Each unit is very homogenous and well sorted facies.Dewatering pipe or vein structure during sedimentationwas partly preserved in medium-grained sandstonefacies. Sedimentary structure such as cross-bed, cross-to parallel-laminations and bioturbation is not wellpreserved. The lithology of this member, however, issimilar to that of the middle part of the PenalverFormation. According to the microscopic observations,grain boundaries are partly recrystalised and lowgrademetamorphic minerals are observed. This sequence wasaffected by low-grade metamorphism which may havedestroyed the original sedimentary structure.

Upper most members consist mainly of silt whichcontains fine-grained carbonate, foraminifera and quartzgrain. The Paleocene micritic limestone (AnconFormations) occurs above this member. However, theupper boundary between them is not well exposed.

Sedimentary environment and conclusion:Characteristics of the Cacarajicarra Formation aresummarized as follows; 1) upward-fining sequence withmore than 300 m in thickness, 2) very thick boulderzone, 3) well sorted stratigraphic sequence, 4) highwater-pressure at the base of the sequence, 5) presenceof the shocked quartz and spherule (analyses of thesegrains are now in progress ), 6) similar lithology to thePenalver Formation. We recognized that basal boulderbreccia zone with high pore-pressure matrix may beformed by the hyperconcentrated flow. An extremelyhuge sequence with exotic blocks and impact relatedmaterials in the Cacarajicara Formation is probably adistal facies impact ejecta sequence or an impact relatedmega-Tsunami deposit.

0 1000 km

Chicxulub

Yucatan

LosOrganos

Rosario

Habana belt

25 km

CacarajicaraFormation

Cuba

References: [1] Pindell, J. L. and Barrett, S.F. (1990)Geology of North America, Geological Society ofAmerica. 405-432. [2] Plamer R. H. (1934) J. Geol.,43, 123-145. [3] Plamer R. H. (1945) J. Geol., 53, 1-34.[4] Pazczolkowski (1986) A. Bull. Pol. Acad. Sci.(Earth Sci.) 34, 81-94. [5] Bohor, B. F. and Seitz, R.(1990), Nature, vol 344, 593. [6] Takayama et al.,Sedimentary Geology (submitted), LPS XXX (in thisvolume), [7] Tada et al. (in preparation), [8] Gordon etal. (1997) J. Geophys. Res. 102, 10055-10082. [9]Piotrowska K. (1978) Acta Geol. Pol., 28, 97-170. [10]Iturralde-Vinent, M. A. (1994) J. Petrolumme Geology,vol 17 (1). 39-70. [11] Iturralde-Vinent, M. A. (1994)Tectonophysics, 234, 345-348.

IMPACT TSUNAMI: A PROBABILISTIC HAZARD ASSESSMENT S. N. Ward and E. Asphaug, Instituteof Tectonics, University of California, Santa Cruz CA 95064

We investigate the generation, propagation, andhazard of tsunami spawned by oceanic asteroid im-pacts. Linear tsunami theory dictates that radiallysymmetric, initial impact cavities u z

impact(r0 ) , evolveinto vertical sea surface waveforms at position r andtime t as

u zsurf (r,t ) = k dk J0 (kr )cos(ω(k )t)

0

∞

∫ r0dr0 uzimpact(r0 ) J 0(kr0 )

r0

∫ (1)

with r=|r|; ω(k)=kc(k)=kvt[tanh(kh)/kh]1/2; vt=(gh)1/2; h,a constant ocean depth; and J0, a cylindrical Besselfunction. By being valid in both shallow and deepwater, and by properly accounting for losses due togeometrical spreading and frequency dispersion1, for-mula (1) can determine maximum tsunami amplitudeexpected at distance r from cavities created by anydiameter impactor. Coupling this information with thestatistics of such falls, we assess the probabilistic haz-ard of impact tsunami shoaling upon global coastlines.

Our research is motivated by the fact that 2/3 of allobjects striking Earth impact the ocean2. Geologicalevidence3 for oceanic impacts include mesosideritefragments from the Eltanin meteorite that struck theSouthern Ocean of the Pliocene (~2.15Ma), and tsu-nami deposits discovered from Texas to Haiti4 thatdate to the K/T impact (~66 Ma). Moderate size im-pactors 30 to 300 m diameter are thought to strikeEarth’s oceans every 100 to 10,000 years. Impactors ofmoderate scale draw our attention because they mayproduce a perceptible safety hazard within historicalcontexts, or within the life span of a human being. The initial stage of cratering by moderate size im-pactors may be characterized5 by the excavation of aparaboloid transient cavity with a depth(dC)/diameterratio ~1:3. Accordingly, we model 1:3 initial cavitiesthat include an outer lip such that there is no net waterloss:

u zimpact(r) = dC(1− 4 r2 9dC

2 );r ≤ R D = 3 2dC / 2

This initial cavity is transient, transforming to propa-gating tsunami waves

u zsurf (r, t) = 4dC dk

F(k, dC )J0 (kr) cos(ω(k)t)k0

∞

∫ (2)

1 Ward, S.N. (1982), Phys. Earth Planetary Interiors27, 273-285.2 e.g. Toon, O.B. et al. (1994), in Hazards due toComets and Asteroids (T. Gehrels, Ed.), U. of ArizonaPress, pp. 791-826; Hills, J.G. and C. Mader (1995),in Proc. Planetary Defense Workshop, LLNL.3 Gersonde, R. et al. (1997), Nature 390, 357-3634 Maurrasse, F.J.-M. and G. Sen (1991), Science 252,1690-1693; Bourgeois, J. et al. (1988), Science 241,567-570.5 Grieve, R.A.F. et al. (1989), Meteoritics 24, 83-88.

where F(k,dC ) = J2 (kR D) − kR DJ1(kR D) / 4 . Figure 1shows a cross section of the birth and development ofan impact tsunami as computed from (2).

By equating tsunami energy to some fraction ε, ofthe impactor kinetic energy, crater depth dc is tied toimpactor radius RI, velocity VI, and density ρI by

dC= (8ερIR3

I V2

I /9ρwg)1/4 (3)

For ε=0.135, equation (3) reproduces traditional craterdiameter predictions6 to within 5% for impactor radii25m<RI<500m. Repeated evaluation of (2) with (3),exposed an empirical law for maximum tsunamiamplitude versus distance r, and impactor radiusas:

u zmax(r,R I) = dC 1+ 2r/3dC[ ]−(1.07 +0.03dC / h)

(4)

Figure 2 plots (4) for VI =20 km/s and ρI =3 gm/cm3.Note that attenuation losses in maximum tsunami am-plitude are nearly (1/r) for RI<500m and ocean depthsh>1000m. Given a particular tsunami hazard thresh-old u z

crit , and impact-site/coast-site distance r, (4) can

be used to find R Icrit(r,uz

crit ) , the critical impactor ra-dius. Any impact at distance r from a bodyRI>R I

crit(r,uzcrit ) will produce tsunami heights ex-

ceeding u zcrit . Given the annual impact flux n(RI) for

all RI, the annual rate of falls exceeding the thresholdis

N(r,uzcrit ) = n(

R Icrit (r,u z

crit )

∞

∫ RI )dRI (5)

For impact rate density we consider

n(RI)= a RI-2

where a is fixed to generate one Earth-striking im-pactor RI>1km per 100,000 years – a conservativeestimate of the flux of small Near Earth Objects 7.

Work in progress includes the production of aglobal coastline map of impact tsunami hazard from(4) and (5) based on the fluxes of stony, iron, andcometary bolides, and the incorporation of atmos-pheric filtering and tsunami shoaling effects.

6 Housen, K.R., R.M. Schmidt and K.A. Holsapple(1983), J. Geophys Res. 88, 2485-2499.7 Rabinowitz, D. et al. (1994), in Hazards due toComets and Asteroids (T. Gehrels, Ed.), U. of ArizonaPress, 285-312.

IMPACT TSUNAMI: S. N. Ward and E. Asphaug

Figure 1. Tsunami inducedby the impact of 200 m di-ameter asteroid at 20 km/s ascomputed by equations (2)and (3). Waveforms areshown at 3 minute intervals.Maximum amplitude islisted to the left. Note thestrong effects of frequencydispersion in “pulling apart”the initial impact cavity.Peak amplitude is found atthe wavelength correspond-ing to the cavity diameter.Maximum tsunami ampli-tude versus distance readfrom many plots like this arecapsulized in Figure 2.

Figure 2. Tsunami attenuationfrom equation (4). The curvestrace maximum tsunami heightversus distance for asteroid radiibetween 1 and 500 meters. Notethat there is little dependence ofattenuation on ocean depth forimpactors of this size range.Asteroids smaller than ~15 mradius typically airburst ratherthan impact.

K/T IMPACT TSUNAMI. T. Matsui1 , F. Imamura2, E. Tajika3 , Y. Nakano1 , Y. Fujisawa4; 1Department of Earthand Planetary Physics, University of Tokyo, Tokyo 113-0033, 2Disaster Control Research Institute, Tohoku University,3Geological Institute, University of Tokyo, 4Technical Research Institute, Obayashi Cooperation.

Introduction: Tsunami may have been generated byasteroid impact which formed the Chicxulub crater inthe Yucatan Peninsula, Mexico, at the K/T boundary(about 65 million years ago). This is because the Yuca-tan Peninsula was covered with shallow water at thetime of the impact [1] and, in fact, the K/T sandstonecomplexes which are interpreted as possible tsunamideposits have been found around the Gulf of Mexico[2,3,4]. Although there are several attempts to study theimpact-induced-tsunami [5,6], nature of the gigantictsunami caused by the impact at the K/T boundary hasnot been investigated so far. Here we present the resultsof the numerical simulation of generation and propaga-tion of the tsunami over the oceans, and show the con-sequences of the tsunami generated by the impact at 65million years ago.

Models: Hypervelocity impact of a large asteroid onthe Earth results in release of significant energy, va-porizing and melting crusts at the impact point, ejectingcrustal materials outward, and creating shock waveswhich propagate in the atmosphere as well as in thecrust. When asteroid impacts the area covered withseawater, impact-induced-tsunami will be generated.We can consider three stages of tsunami generations:(1) the wave coupled with high air pressure and windgenerated by the entry of the asteroid into the atmos-phere, (2) the rim wave formed at the front of the ejectacurtain as well as by the shock waves and (3) the wavecaused by movement of water to fill and flow out of thecrater cavity after the crater formation (hereafter calledreceding and rushing waves, respectively). The wavegeneration at the first stage might be negligible becausewaves generated by strong winds are dispersive andrapidly damped during propagation over a long dis-tance. The wave generated by the second stage is mod-eled here as the initial condition of the rim wave withwave height and length estimated from the experimen-tal results [5]. The wave generated by the third stagemight be the most devastating, and is simulated basedon the nonlinear long wave theory.

In this study, we consider two models for oceanicpropagation of tsunami based on the linearized long-wave theory in the spherical coordinates; one is for tsu-nami generation covering the Gulf of Mexico with adetailed coastal run-up condition. The second is foroceanic propagation over the entire globe. For thespherical coordinate system, the 20 min. spatial gridsize was used together with the time interval adjusted to

satisfy the Courant-Friedrichs-Lewy stability conditionat all times. Details of the model such as numericalscheme and stability condition are referred to Imamuraand Shuto [7].

The asteroid struck the Earth at the K/T boundaryhas been estimated to be about 10 km in diameter withimpact velocity of several tens of km/s, resulting in theformation of the Chicxulub crater approximately 200km in diameter [8]. Although the crater floor is consid-ered to have rebounded rapidly, details of the craterformation process are still unclear. Hence, we simplyapproximate the shape of the crater based on the presentshape of the Chicxulub crater [8]. It is noted that a de-tailed crater shape is not important for the tsunami gen-eration, as shown in the results.

We need to reconstruct the continental distribution,paleobathymetry and sea level of 65 million years ago tosimulate tsunami propagation across the oceans. Thepaleobathymetry was estimated from the plate motionsand the cooling model of the plate. We used theETOPO-5 as the topographic data of the present Earthand a digital age map of the ocean floor. For the recon-struction of the paleobathymetry of the Gulf of Mexicoand Caribbean region, we adopted the tectonic model byRoss and Scotesse [9] with slight modifications basedon our fieldwork in Cuba [10].

Numerical Results: We summarize the numericalresults of tsunami generation processes, tsunami propa-gation within the Gulf of Mexico, and tsunami propa-gation over the oceans.

Tsunami generation: We found that the seawaterflow rate into the crater cavity is independent of thedepth of the crater because the surrounding water depthof the Yucatan platform may have been much shallower(200 m) than the crater depth (3,000 m) [1,8]. We alsofound that the rushing wave heights and periods aredetermined by the horizontal scale of the crater because,the larger the crater, the longer the circumference,which results in a greater inflow rate, hence more timeto fill the cavity and a longer time to rebound to createthe rushing wave. The results of numerical simulationdemonstrate the water movement that generates thereceding and rushing waves. The water flowed into thecrater cavity accumulates and the crater cavity over-filled, thus, generating the rushing wave outward. Wefound that the waves going out of the crater are con-trolled by the depth of the shallow water region sur-rounding the crater. The results indicate that the

K/T IMPACT TSUNAMI: T. Matsui et al.

smaller the flow velocity in the shallow water region,the smaller the wave height of the rushing wave withthe longer oscillation period. When we assume the wa-ter depth of the Yucatan platform to have been 200 m atthe time of impact [1], the wave height and period areestimated to be about 50 m and 10 hours, respectively,at the rim of the crater.

Tsunami propagation within the Gulf of Mexico:There are two types of tsunamis that attacked the coastnear the impact crater; one is the receding wave, origi-nally generated by the water flowing into the crater, andthe other is the rushing wave, generated by water flow-ing out of the crater. Because we cannot find any wavesahead of the receding wave, the rim wave must havequickly dispersed during the propagation to the shore.The receding wave covers the entire Gulf over 10 hoursafter the impact. The positive wave follows with aheight of more than 200 m, reaching the coastal area ofNorth America. It runs up over the Plains and passesthe Mississippi embayment, a distance of over 300 km.The time history of the simulated tsunami shows multi-ple waves (approximately 1-2 hour wave periods) ridingon the main slow water-level fluctuations at Mimbral(Mexico), Brazos River (USA), and Mussel creek(USA). The simulated wave heights at Brazos Rivercorrespond well with the results of Bourgeois et al. [2]who estimated the runup from the geological inferenceof the tsunami deposits. The maximum run-up height of300 m is found near the Rio Grande embayment (USA)with an average value of more than 150 m. Smit et al.[4] estimated that more than 9 tsunamis were generatedbased on the analysis of paleocurrent directions re-corded in the K/T sandstone complex at La Lajilla(Mexico) near Mimbral. Such an observation is inagreement with simulated multiple wave-runup charac-teristics. The period of waves observed at DSDP sites inthe Gulf of Mexico is very different from those at othersites. This indicates that, short period waves of less than20 minutes are dominant in deep water, whereas thewaves with a one-hour period are dominant near thecoast. Along the coast, multiple and significant tsunamirunup with the periods of 1-2 hours resulted because ofthe reflections and energy entrapment of the tsunami.For waves of less than a 20-minute period, the majorityof the wave energy is dissipated via wave breaking atthe shore. For the tsunamis with periods longer than 10hours, the energy is simply reflected out to the oceanswithout entrapment.

Tsunami propagation over the oceans: The numeri-cal results of oceanic propagation show that some of thewave energy is trapped in the Gulf of Mexico and thenpropagated mostly towards Europe and North Africa,but less to the Pacific Ocean. This is because the straitto the Pacific Ocean was narrow 65 million years ago

whereas the Caribbean Sea was wide open to the Atlan-tic Ocean. The simulation suggests that, besides theNorth American coasts, significant tsunami runup musthave occurred in South America: the results showedapproximately 200-250 m high runup in what is nowVenezuela. Furthermore, the simulation shows therunup height of approximately 100-150 m in NorthwestAfrican coasts and 100 m along Northern Europe coasts(note that the distance to Africa and Europe was muchshorter at 65 million years ago than it is today). Oursimulation also shows two separate waves, one propa-gated across the Atlantic ocean and the other across thePacific ocean, met at the eastern part of the IndianOcean at about 28 hours after the impact. The simulatedwave height in the Indian Ocean was found to be ap-proximately 10 m with the 10 hours wave period.

References: [1] Sohl, N. F. et al. (1991) In The Ge-ology of North America, Vol. J, The Gulf of MexicoBasin, The Geological Society of America. [2] Bour-geois, J. et al. (1988) Science, 241, 567-570. [3] Smit,J. et al. (1992) Geology, 20, 99-103. [4] Smit, J. et al.(1996) Geological Society of America Special Paper307, 151-182. [5] Gault, D. E. and Sonett, C. P. (1982)Geological Society of America Special Paper 190, 69-92. [6] Hills, J. G. et al. (1994) In The Hazards due toComets and Asteroids, 779-789. [7] Imamura, F. andShuto, N. (1990) Proc. of Int. Sym. Comp. Fluid Dy-namics, Nagoya, 390, 15. [8] Morgan, J. et al. (1997)Nature, 390, 472. [9] Ross, M. I. and Scotese, C. R.(1988) Tectonophysics, 155, 139-168. [10] Tada, R. etal. (submitted).

STRUCTURE OF THE CHICXULUB IMPACT CRATER AS DETERMINED FROM LARGE-OFFSETONSHORE-OFFSHORE SEISMIC DATA. Yosio Nakamura1, G. L. Christeson1, R. T. Buffler1, J. Morgan 2,M. Warner2 and the Chicxulub Working Group, 1Institute for Geophysics, University of Texas at Austin (4412Spicewood Springs Road, Austin, TX 78759-8500, U.S.A., yosio@uitg.ig.utexas.edu), 2T. H. Huxley School,Imperial College (Prince Consort Road., London SW7 2BP, U.K.).

Introduction: In the fall of 1996, an internationalgroup of scientists from the U.K. (Imperial College,University of Cambridge and University of Leicester),U.S.A. (University of Texas at Austin and Lunar andPlanetary Institute), Canada (Geological Survey ofCanada) and Mexico (Universidad Nacional Autonomade Mexico and Petroleos Mexicanos) conducted a geo-physical survey over the Chicxulub impact crater, nowwidely believed to represent a large Cretaceous-Tertiary(K/T) boundary impact that led to the mass extinctionof species. Located along the northern coast of YucatánPeninsula, Mexico, under a thick (~1 km) sedimentarycover, it is the best preserved large impact crater knowon Earth. It thus is expected to provide highly valu-able information on the deep structure of large impactcraters, which is difficult to obtain on other planetarybodies.

In this survey, we combined techniques of offshoremulti-channel seismic reflection, wide-angle seismicrecordings on both ocean-bottom seismographs (OBSs)offshore and land stations onshore, and onshore andoffshore gravity measurements. An initial report of thisexperiment, mainly dealing with interpretation of theoffshore seismic reflection data, shows reflection signa-tures that indicate a multi-ring structure with a 100-kmdiameter transient cavity [1]. An earlier analysis of theOBS data from one of the offshore seismic lines showsshallow structures near the crater center, which includea low-velocity Tertiary basin, high-velocity reefalbuildups at the margins of the basin, and negative ve-locity anomalies at about 1 to 2 km depth interpretedto be melt rocks [2].

In this paper, we will report on our recent analysisof combined OBS and land-station data for deeperstructures.

Data Sets: The data sets used in this analysis con-sist of the large-offset seismic signals from air-gunshots as recorded on 10 OBS stations along a line ex-tending 180 km to northwest from the approximatecenter of the crater (Fig. 1, Line B), 23 land seismicstations along a line extending 95 km southeast fromthe approximate center of the crater (Line F), 23 OBSstations along a 345-km line parallel to the northerncoast of the peninsula and offset about 26 km from theapproximate center of the crater (Line A), and 46 landseismic stations on a 143-km line along the northerncoast of the peninsula (Line D). The OBS data werecollected by the group from the University of Texasusing their 4-channel (3-component geophones and ahydrophone) instruments and the land station data werecollected by the groups from Imperial College and

Universidad Nacional Autonoma de Mexico usingPASSCAL single-component instruments.

Fig. 1. Seismic lines and recording stations of the Chicxu-lub experiment. Large and small dots give the locations ofOBS and land stations, respectively.

Upper Crustal Structure: We used 2D and 3Dtomographic codes of Hole and Zelt [3] and Zelt andBarton [4] to invert the seismic first arrival times forupper crustal structure. In one case, the Line-B (OBS)and Line-F (land) data from sources on Line B wereanalyzed for a 2D structure along the line crossing theapproximate center of the crater. In another case, Line-A (OBS) and Line-D (land) data from sources alongLine A were analyzed for a 3D structure between thesetwo lines encompassing a 42-km wide strip across thecrater center.

The most prominent feature seen on these to-mographic results, in addition to the shallow, low-velocity Tertiary basin referred to above, is a basementhigh clearly seen near the center of the crater on bothsections. It represents a P-wave velocity contrast ofabout 0.5 km/s and an uplift of about 5 km relative tothe surrounding areas with its top reaching a depthbelow the surface of about 2.5-3 km. The upliftedblock is 30 to 50 km wide and slightly offset to southand west from the apparent geometrical center of thecrater. A similar offset is also seen on the contours ofpositive surface gravity anomaly, supporting the base-ment uplift as the cause of the observed central positivegravity anomaly.

A second basement high is observed along Line Bbeyond 90 km from the crater center. This coincideswith the prominent gravity high observed only in thisnorthwest quadrant, and thus may not be related to the

DEEP SEISMIC STRUCTURE OF CHICXULUB IMPACT CRATER: Y. Nakamura et al.

crater formation, although one may argue that it is amanifestation of an oblique impact.

Mid-Crustal Reflections: Tomographic inver-sions of seismic first-arrival times do not show anyappreciable lateral velocity variations below about 15km depth. This, however, does not necessarily meanthat the velocity structure below this level is laterallyuniform. It simply means that relatively high veloci-ties of the shallow Cretaceous carbonate and basementrocks prevent deeper seismic refractions from appearingas first arrivals. In fact, several prominent seismic re-flections are observed coming from mid-crustal depths.

Some distinct bands of mid-crustal reflections werealso seen on vertical reflection data outside the col-lapsed transient cavity, generally dipping towards thecrater center at between 30 and 40° [1]. A prominentreflector consistently seen on Line B OBS record sec-tions, also located outside the collapsed transient cav-ity and dipping towards the crater center, appears tohave a smaller dip of between 10 and 15°. This reflec-tor extends from about 90 to 150 km from the cratercenter. Similar low-angle reflections are observed onOBSs on the western half of Line A, and also on theeastern half of Line A with somewhat lesser consis-tency. The apparent lack of observable velocity con-trast between the layers above and below the reflectorsis consistent with faulting caused by impact, wherein alarge volume of melt within the fault zone producedextensive pseudotachlites [1, 5].

Moho Refractions and Reflections: Clear refrac-tions from the upper mantle (Pn phase) were observedas first arrivals on most seismograms beyond about160 km in distance along Line B/Line F. To-mographic inversion of these arrivals, however, failedto provide clear variation in depth to Moho along theline.

In contrast, wide-angle reflections from Moho(PmP phase), prominently observed on most OBS andland seismograms, give some constraints on depthvariations along the seismic lines. CharacteristicMoho reflections, consisting of a band of lower-crustreflections terminating at Moho, are also observed onthe vertical reflection data [1]. These data indicate thatMoho in this area generally lies at a depth of 35±3 kmwith some interesting topography. There is a stepdown of about 3 km from west to east along Line Anear the place where the line passes just north of thecrater center. The Moho is deepest under most of LineB, but shallows somewhat at its southeastern end to-wards the crater center. Some of the long-wavelengthvariations may represent regional trends and thus maynot be related to the formation of the impact crater.However, short-wavelength variations seen within 50-60 km of the crater center may well be of impact ori-gin. We need more detailed studies, possibly withadditional data, to further elucidate the likely impacteffect on Moho.

Acknowledgments: Tomographic codes used in thisstudy were kindly provided to us by John Hole andColin Zelt.

References: [1] Morgan J. et al. (1997) Nature,390, 472-476. [2] Christeson G. L. et al. (1999) GSASp. Paper, (in press). [3] Hole J. A. and Zelt B. C.(1995) Geophys. J. Int., 121., 427–434. [4] Zelt C.A. and Barton P. C. (1998) JGR, 103., 7187–7210.[5] Spray J. G. (1997) Geology, 25., 579–582.

SECONDARY CRATERS FROM THE CHESAPEAKE BAY IMPACT. C. Wylie Poag, U.S. Geological Sur-vey, 384 Woods Hole Road, Woods Hole, MA 02543-1598 USA.

Despite the near ubiquity of secondary cra-ters on other planetary bodies, secondary craters havenot been documented on Earth. The numerous simpleterrestrial craters (<5 km in diameter) identified sofar are interpreted to be primary craters. Nor havesecondaries been previously reported to be associatedwith known complex or multiringed structures.

The Chesapeake Bay primary crater (85-km-diameter) is unusually well preserved, because it isrelatively young, it formed under water, and it occu-pies a basin characterized by relatively rapid postim-pact marine sedimentation [1]. This advantageoussetting appears also to account for the presencenearby of at least 23 smaller fault-bounded excava-tions, which I interpret to be secondary craters. The23 secondaries can be identified north and northwestof the primary crater on two of the multichannelseismic reflection profiles that help define the Chesa-peake Bay primary crater. A seismostratigraphicanalysis of these profiles, calibrated withlithostratigraphy and biostratigraphy from nearbyoutcrops and bore holes, highlights the structural andstratigraphic contrasts between the normal successionof sedimentary coastal plain rocks and those of theinferred secondary craters.

The closest secondary crater is 20 km fromthe primary crater, whereas the farthest is 85 kmaway; all secondaries are outside the seismicallyidentifiable periphery of the continuous ejecta blan-ket. Relatively undisturbed, flat-lying, coastal plainformations separate the secondaries from each otherand from the primary crater. Five secondary craters(C-1 through C-5) are imaged by north-south seismicprofile T-1-CB, where it crosses Chesapeake Baynear the mouth of the Potomac River. Secondaries C-1 through C-5 have similar general characteristics,except for varying apparent diameters and depths.Each crater is marked by clearly expressed rim es-carpments constructed by en echelon (presumablyconcentric) normal faults, which dip into the craters.The rim faults truncate horizontal, parallel, continu-ous to subcontinuous reflections, which represent thesame sedimentary target rocks disrupted by the pri-mary impact (Lower Cretaceous to lower Eocenesiliciclastic sediments and middle Eocene bioclasticlimestone). Inside each secondary crater, seismicreflections are chaotic or incoherent. I interpret theseto represent impact breccia, equivalent to the Exmorebreccia, which fills the primary crater. The postim-pact formations (mainly middle Miocene to Quater-nary sediments) thicken and sag into all five secon-dary craters on this profile.

Eighteen similar small craters (P-1 through

P-18) are distributed along a 110-km segment ofprofile T-11-PR, which extends from the northernrim of the primary crater up the Potomac River to alocation near the town of Colonial Beach. Thoughmany features of the Potomac (P) secondaries aresimilar or identical to those of the CB (C) seconda-ries, some differences can be noted. For example,eight of the P secondaries have well-developed,raised, sedimentary rims, in contrast to the lack ofraised rims on the primary crater. Perhaps the mostimportant difference, however, is that some of thenormal faults of the P secondaries disrupt the surfaceof the crystalline basement (P-5, 7-10, 13, 14, 17, and18). Furthermore, the basement surface along thePotomac River profile is cut by eight reverse faults of100 m or more vertical displacement (lateral dis-placement is unknown), seven of which display un-derthrusting to the west. The basement, as well asthe entire preimpact sedimentary section at P-18, hasbeen thrust into two anticlinal folds. This dates thethrusting as late Eocene, coincident with the primaryimpact. I interpret the reverse faults to be early prod-ucts of compressive shock radiating from the primaryChesapeake Bay impact. Most of the reverse faults,however, appear to have been reactivated as normalfaults during later stages of ejecta bombardment anddeformation of the secondary craters.

Apparent diameters of the secondary cratersrange from 0.4 km to 4.7 km, and average 1.9 km;only four have apparent diameters greater than 3 km.Apparent depth of the secondaries (measured fromsedimentary lip to crater floor) ranges from 50-710m, averaging 370 m; in six of the secondaries, theentire preimpact sedimentary section has been exca-vated and replaced by impact breccia. Breccia fillranges from 30 m to 680 m in apparent thickness, andaverages 266 m.

Stratigraphic and structural characteristics ofthe 23 secondary craters coincide with the generalfeatures of simple primary craters (as opposed tocomplex craters). The principal difference is the ap-parent lack of overturned flaps (though they may betoo small to be resolved on our profiles). This is notsurprising, however, because the impacts took placein the late Eocene ocean, and there is ample evidencethat oceanic impact craters lack these features,probably as a result of more extensive slumping oftheir water-saturated walls and of hydraulic erosionresulting from collapse of the oceanic water column[2,3,4,5].

The characteristics of the Chesapeake Baysecondary craters also are in general agreement withthe features of secondary craters observed on other

SECONDARY CRATERS: C. W. Poag

planetary bodies [6]. They occur in distinct clusters,or perhaps chains, and their apparent diameters fallwithin the expected ranges relative to the diameter ofthe primary (less than 10% of the diameter of theChesapeake Bay primary crater).

Acknowledgments: I am grateful to Texaco,Inc., for providing the multichannel seismic reflec-tion profiles upon which this analysis is based.

References: [1] Poag, C.W., Sed. Geol. 108,45-90, 1997. [2] Kieffer, S.W., and Simonds, C.H.,Rev. Geophys. Space Phys., 18, 143-181, 1980. [3]Roddy, D.J., Schuster, S.H., Rosenblatt, M., Grant,L.B., Hassig, P.J., and Kreyenhagen, K.N., Int. J.Impact Eng., 5, 525-541, 1987. [4] Mckinnon, W.B.,Geol. Soc. Am. Sp. Pap. 190, pp. 129-142, 1982. [5]Poag, C.W., and Poppe, L.J., Mar. Geol., 145, 23-60,1998. [6] Melosh, H.J., Impact Cratering - A Geo-logic Process, pp.1-245, 1989.

MULKARRA IMPACT STRUCTURE, SOUTH AUSTRALIA: A COMPLEX IMPACTSTRUCTURE. J. B. Plescia, U. S. Geological Survey, 2255 N. Gemini Drive, Flagstaff, AZ 86001

Introduction: Flynn (1) identified a struc-ture in the subsurface of the western EromangaBasin that he interpreted as a buried impact.The structure, based on seismic reflection andgravity, was considered a simple bowl shapedcrater 9 km in diameter. Residual gravity datashowed a circular anomaly 6-7 km in diameterhaving a central low and a surrounding high;total gravity relief IS ~1 mGal (residual gravityover the remainder of the region was not illus-trated). The nature of a disturbed zone sur-rounding the suggested 9 km structure was notaddressed.

The question of the diameter remains. (1)concluded the diameter was 9 km and thestructure was that of a simple crater. However,the seismic deformation encompasses a region17 km in diameter. Thus, two models may besuggested: a 9 km diameter structure whoseformation disturbed the surrounding sediment,or a larger feature with a central pit or peakring.

Geologic Context: The stratigraphic sec-tion consists of fluvio-lacustrine sediments(Hutton Sandstone, Birkhead Fm., Mooga Fm.)overlain by a shallow marine and marginal ma-rine sequence (Cadna-Owie Fm., Bulldog Shale,Coorikiana Sandstone, Oodnadatta Fm., andWinton Fm.). Unconformities separate thesedimentary section from underlying basementand from overlying Tertiary and Recent sedi-ments. The only post-Jurassic regional defor-mation in the area is the Tertiary age northeast-trending Birdsville Track Ridge Anticline ofTertiary age to the northwest.

Seismic reflection data indicate reflectorsare flat lying and undeformed outside thestructure. Within the interior, reflectors are bro-ken up and somewhat chaotic. A series ofhorsts and grabens (~ 500 m wide) and normalfaults with perhaps 50-100 m of offset occur.The deformed zone is ~ 17 km across. At thecenter is a bowl shaped depression 9 km acrossoverlying incoherent reflectors. Deformationoccurs below the Coorikiana Sandstone (750 mdeep), but extends through the remaining sec-tion and possibly into the basement (1400 mdeep).

Gravity Survey: To better constrain thecrustal structure, data were collected in Sep-

tember 1993 using a Lacoste Romberg meter;elevation control was established using a lasertheodolite. Data were collected along four di-agonal profiles centered on the gravity anomalydefined by (1). The new data were combinedwith data presented by (1) that were collected in1988 (2). The combined set consists of >600stations.

The uppermost part of the stratigraphic sec-tion at Mulkarra is characterized by a seismicvelocity of 2.4 km sec-1, thus a density of 2.1 gcm-3 was used in the reduction. The more de-tailed gravity survey reported here defines aBouguer gravity field decreasing to the west-southwest, consistent with the broader gravityfield (3). Removal of 2nd and 3rd order surfacesproduces residual maps that isolate the anomalydirectly associated with the Mulkarra structure.

The third order residual gravity map (Figure1) shows a central low surrounded by a high ~8km in diameter having ~1 mGal relief. Thisfeature is again surrounded by a low having adiameter of 15-16 km (measured to the axis ofthe outer low) or 20 km (measured to the outeredge of the low).

Interpretation: Gravity and seismic datasuggest that the Mulkarra structure is not a 9 kmsimple crater. Rather, it is a 20 km complexcrater with a 9 km central pit or peak ring. Thestratigraphic context of the structure suggests itwas formed in a shallow marine environment inunconsolidated sediments. A 20 km diameterstructure is within the complex size range, al-though the diameter is smaller than the transi-tion diameters for central pits (22 km) and peakrings (25 km) (4). However, as the final form ofthe crater is controlled by the strength of thematerial (5), a low cohesion and low viscositytarget (i.e., unconsolidated marine sediments)could result in the onset of a central pit or peakring at smaller diameters.

The Marquez structure in Texas showssimilar limited deformation in seismic profilesacross the feature. Marquez is a 13 - 20 km di-ameter complex impact crater (6). Seismic re-flection data (7) show a central peak (charac-terized by chaotic and incoherent reflectors) anda surrounding annulus where the strata is onlymildly deformed. Concentric inward dippingnormal faults with <100 m of displacement sur-

MULKARRA IMPACT STRUCTURE: J. B. Plescia

round the structure. A well defined rim does notoccur. The suggestion has been made (8) that asthe Marquez structure formed in a marine envi-ronment in unconsolidated sediments, the clas-sic bowl shaped structure with a central peakand well defined rim did not form. Rather theenergy was expended in a more chaotic distur-bance of the stratigraphy.

A similar situation may have occurred atMulkarra. Impact into a marine section oflargely unconsolidated sediments may haveproduced a crater lacking the well definedstructural elements of impacts into more com-petent rock. As a result, the gravity signature issimilarly less well developed

150000 155000 160000 165000 170000 175000

Easting (M)

1420000

1415000

1410000

1405000

1400000

No

rt

hi

ng

(

M)

MULKARRA STRUCTURETHIRD-ORDER RESIDUAL GRAVITY (DENSITY =2.1)

-1.50

-1.25

-1.00

-0.75

-0.50

-0.25

0.00

0.25

0.50

0.75

1.00

1.25

1.50

1.75

2.00

2.25

2.50

Figure 1. Color coded contour map of the third order residual gravity. Reduction density was 2.1 gcm-3. A 50 x 50 grid was calculated and contoured. Contour interval is 0.25 mGal.

References: (1) Flynn, M., 1989, The Coo-per and Eromanga Basins Australia, B. O’Neiled., Australian Society of Exploration Geo-physicists, p. 431-439. (2) Busuttil, S., 1988,SANTOS Ltd. Unpublished Report. (3) Wil-liams, A., 1975, Geological Survey South Aus-tralia. (4) Pike, R., 1983, J. Geophys. Res., 88,

2500-2504. (5) Melosh, H., 1989, Oxford Uni-versity Press, 245 pp. (6) Sharpton, V., and Gib-son, J., 1990, LPSC 21, 1136-1137. (7) Gibson,J., 1990, M. Sc. thesis, University of Houston,65 pp. (8) Buchanan, P., et al., 1998, Mete-oritics, 33, 1053-1064.

MAGNETIC MODEL OF THE CENTRAL UPLIFT OF THE VREDEFORT IMPACTSTRUCTURE. H. Henkel1 and W.U. Reimold2, 1Dept. of Geodesy & Photogrammetry, Royal Institute ofTechnology, S-100 44 Stockholm, Sweden (herbert@geomatics.kth.se); 2Department of Geology,University of the Witwatersrand, Private Bag 3, P.O. Wits 2050, Johannesburg, South Africa(065wur@cosmos.wits.ac.za).

Summary. The very characteristic negative magneticanomalies occurring within the area of the centraluplift (the Vredefort Dome) of the Vredefort impactstructure have been analyzed in detail by interactivemagnetic modelling. A final model was derived thatshows the extent of post-impact thermal re-magnetization to about 2.5-5 km depth and 30 kmradial distance from the center of the structure(Figure 1). Details of the model reveal structuresrelated to the collapse of the central rise. Introduction. The Vredefort impact structure is,with an estimated original diameter of 250 km [1, andrefs. therein], the largest astrobleme so far detectedon Earth. It comprises the entire extent of theWitwatersrand basin and is characterised by a 40 kmwide central rise comprising Archean crystallinebasement rocks and Archean-Proterozoic supracrustalsequences [2]. Modeling based on gravity and seismicdata has shown that this rise is an antiformalstructure with a vertical displacement of upper crustalmarker strata by about 12 km. This differential upliftdecreases to about 4 km at the base of the crust [1].At the present erosion surface, the rise is surroundedby a 20 km wide up- and overturned sequence of pre-impact cover rocks ranging from the ArchaenDominion Group (at the contact to the crystallinebasement) and Witwatersrand Supergroup to the earlyProterozoic Transvaal Supergroup. The central upliftis also characterised by an anomalous magnetic fieldwith strong negative anomalies indicative of reversedmagnetisation (compare Figure 1C of [1]). The mostprominent anomalies are located in the annular areaaround the contact between supracrustal collar andcrystalline core, where magnetic shales of the WestRand Group (Witwatersrand Supergroup) occur andwithin the crystalline core, where a major semi-annular anomaly straddles the area around thetransition from Outer Granite Gneiss to InlandseeLeucogranofels [1,2]. Modeling Method. Aeromagnetic measurementstogether with measurements of the magneticproperties of rocks sampled at the surface [1] providea basis for modeling of the magnetic structure of thecentral uplift down to considerable depth. Asadditional constraint, the gravity model of the centralrise is used, which describes the general shape of thecrystalline core [1]. For the modeling , the versatile2.5-d-software by Geovista AB [3] was used, whichallows the modeling of individual strike direction andlength, and off-set structures with polygonal cross-sections. The calculated magnetic field from suchsource structures is compared with the measuredmagnetic anomaly and the modeling is iterated until

the discrepancies between measured and calculatedmagnetic fields are minimised. The resultingconfiguration of magnetic source structures representsa possible magnetic structure of the central rise of theVredefort astrobleme. The Magnetic Models. Three different modelshave been derived, each with increasing complexityof the source structures, progressively accounting formore aspects of the cratering process: [A] - Auniform layer of remanent magnetised lithologies, [B]- individual magnetic sources in the collar and core,respectively, and [C] - an assemblage of sourcestructures in both the collar and core regions. Each ofthe models reproduces the measured magneticanomalies. They all represent structures extendingfrom the surface to restricted depth and they allinvolve the occurrence of remanent magnetisedsource structures with properties similar to thosemeasured at exposures or on rock samples. The layermodel [A] demonstrates that the variation of thenegative anomaly can be reproduced by variation ofthe thickness of one layer. This gives a general clueabout the extent of reversed magnetisation in thehorizontal and vertical directions. Such remanence,thus, extends radially for about 30 km from the centerof the structure and extends to about 2 km depth. It ishowever not likely that the remanent source structureis that uniform. Therefore, the model involvingseparate core and collar magnetisations [B] wouldbe more realistic, allowing different magneticproperties to be applied in the calculations. Thismodel demonstrated that the magnetic structure of thecrystalline core follows its antiformal shape. Aremanent envelope is seen around the top of thecentral rise. The depth extent of remanentmagnetisation is slightly deeper in this envelope,whereas it is less in the central part of the rise. As theantiformal shape contrasts with the overturnedgeometry of the collar, this could imply that thecollapse of the central rise may be responsible of theobserved pattern. The final model [C, Figure 1]takes into account the findings of the earlier modelingattempts [A,B], but has an added structuralcomponent related to the outward and downwardsliding of the upper parts of the central rise. Thisoverprint, thus, disrupted the conformal arrangementof lithologies that characterized the pre-impact targetand the geometry during the rise of the central uplift.The collapse phase involved the final overturning ofthe upturned collar sequence. Conclusions. A number of conclusions can bedrawn from the modeling results. They consider thesource for the anomalous magnetisation and the

VREDEFORT IMPACT: H. Henkel and W. U. Reimold

structure of the central rise as seen from thearrangement of magnetic source structures:Magnetisation. The reversed remanence reflects theambient geomagnetic field shortly after the impactevent. It was imprinted into highly coercive magneticgrains probably formed, to a large extent, by shockdissociation of Fe-Mg-silicates, by the excesstemperature introduced by a now eroded impact meltbody, and the rise of high temperature crustal rocksfrom originally deeper crustal levels. It typicallycovers the upper central region of the astrobleme andhas remagnetised both collar and core lithologies. Theregion of shock dissociation, which originallyfollowed a concentric (to the explosion centre)structure, now forms an envelope around originallydeeper and, thus, less shocked lithologies.Sedimentary rocks that lack Fe-Mg-silicates can notbe remagnetised. The remanence seen is athermoremenance and was acquired long after theimpact event when ambient temperatures fell belowthe blocking temperatures of ferrimagnetic minerals.Some of these may have existed before the impact,but the highly coercive nature of the remanence in thecrystalline core rocks indicates that they probablyformed due to the impact event. Crystalline rocksoutside of the central rise are low-coercive and unableto preserve remanence over geological time periods.Structure. Two structural patterns are revealed thathave bearing on the impact process. First, thedistribution of a remanent magnetic envelop aroundthe central part of the crystalline core mirrors thezoned structure of the spatial distribution of shockdeformation degrees, which originally werespherically arranged around the explosion centre, butthat after uplift had evolved into an anticlinalstructure. Second, the magnetic boundaries withinthis updomed envelope indicate the effect of thecollapse of the central rise by outward flow of itsouter upper segments along relatively low angle shearzones.

Refs: [1] Henkel, H. and Reimold, W.U., 1998.Tectonophysics 287, 1-20. [2] Reimold, W.U. andGibson, R.L., 1996. J. African earth Sci., 23, 125-163. [3] Geovista, 1994. GMM gravity and magneticmodelling. Users Manual. Geovista AB, Luleå,Sweden, 41 pp.

Figure 1: The final magnetic model derived for theVredefort central uplift shows a 60 km wide regionwith reversed (remanent) magnetized sourcestructures both within the crystalline core (insideheavy line) and in the overturned sedimentary collar(outside heavy line). Details of the structure indicatecollapse of the central rise along low angle outwarddipping thrust surfaces. Vertical lines indicate‘dominant remanent magnetization’ (continuouslines: relatively high r.m.; dashed lines: relativelylow), horizontal lines indicate ‘dominantly inducedmagnetization.

GEOTHERMAL INVESTIGATIONS OF THE RIES IMPACT STRUCTURE. Y. Popov1, J. Pohl2 and R.Romushkevich1,. 1Moscow State Geological Prospecting Academy, Miklukho-Maklai-Str. 39-1-191, Moscow 117485,Russia, yupopov@mgga-sec.msk.ru, 2Institut für Allgemeine und Angewandte Geophysik, Theresienstr. 41, D-80333Munich, Germany, pohl@alice.geophysik.uni-muenchen.de.

Introduction: In 1973 a 1206 m deep drill core wasobtained from the Ries crater, about 3 km off the cratercenter (FBN73, e.g. [1]). The drill hole penetrated ca.315 m post-impact lake sediments, ca. 16 m graded,aerial fall-back, ca. 275 m of high-shock, high-temperature and low-temperature suevite, containingvarious amounts of crystalline blocks, then ca. 600 m ofcrystalline basement with variable degree of fracturingand content of dike breccias. The drill core can beconsidered as one of the important and characteristicdrill cores in impact structures. The drill cores from 300to 1206 m are stored since 1998 in a new storage placein Nördlingen and are available for new investigations.In 1998 we had the opportunity to use a modern methodto measure the thermal conductivity of the FBN73 drillcores with close spacing. The thermal conductivity logprovides an excellent means for characterizing anddifferentiating impact formations and target rocksaffected by the impact process. In addition the thermalconductivity, together with measured temperatures andtemperature gradients, can be used for an interpretationof the geothermal anomalies characteristic of impactstructures.

Measurements: The new optical scanningtechnology [2] was used for the rock thermalconductivity measurements on 530 dry andwater-saturated cores from the FBN 73. Anisotropycoefficients and thermal inhomogeneity factors of therocks were also determined. Open porosity values andthe thermal conductivity of the rock matrix werecalculated using these experimental data and thegeometric theoretical model for effective thermalconductivity of porous rocks [3]. The measured thermalconductivity L (for dry rock samples) and the thermalinhomogeneity factor B logs (Fig.) reflect the structureof the studied formations and indicate a three-layerthermophysical model of the Ries structure. The smallestmean values of L (0.79 W/m·K) and B (0.25)characterize the sedimentary part of the strata in thedepth interval of 56-331 m. The intermediate values ofL (1.46 W/m·K) and the largest values of B (0.56) arecharacteristic for suevites with intercalations ofcrystalline rocks. The largest values of L (1.99 W/m·K)and intermediate values of B are characteristic forcrystalline rocks and breccia dikes. High- and low-temperature suevites have different average L (1.20 and1.05 W/m·K) and B values (0.71 and 0.54correspondingly). Thermal conductivity is different forgrey and brown breccia dikes too (1.11 and 1.77W/m·K). Similar average values of L of the rock matrixof grey and brown breccia dikes (2.45 and 2.58 W/m·K)indicate a similar mineralogical composition. Thereforetheir difference in L can be explained by a significant

difference (more than two times) in calculated porosity.The thermal conductivity (L) log (for dry cores)correlates well with the sonic log (Fig.) except for ashort depth interval of 825-840 m with alternation ofamphibolites and gneisses. The regression equation isL=0.876+exp(8.67+1.54Vp) and the correlationcoefficient was found to be 0.87. It allows to predictsonic log data for the depth interval 1030-1200 m. Thecalculated open porosity values give an additionalinformation to other porosity logs because they reflectthe geometry of porous space. The thermal anisotropy ofgneisses is found to be negligible although significantanisotropy for these rocks (1.2-1.7) is usually observed.A similar pecularity in thermal anisotropy coefficientvalues was observed earlier for the rocks of thePuchezh-Katunk impact structure [4]. Significantdifferences in temporal variations of temperaturegradient determined from the temperature logs measuredat different times after termination of drilling (from 1day up to 10 months) indicate significant variations instrata permeability and fluid activity. Determinedaverage heat flow density values (calculated using thethermal conductivity of water-saturated rocks andtemperature gradients from the final temperature logafter 10 month) vary from 117 mW/m2 in the depthinterval of 70-250 m to 68 mW/m2 in the depth intervalof 250-500 m and to 72 mW/m2 in the depth interval of500-830 m and are predicted to be 77 mW/m2 in thedepth interval of 870-1200 m, i.e. 33-45 % higher forthese depths than in previously published data. A similarsituation is observed in the Chicxulub crater [5]. Thevery significant increase of the heat flow density in theupper part of the strata could be explained by fluidmovement from deeper parts of the strata throughfractured and permeable rocks to its upper part withhorizontal deflection of the fluid under the unpermeableclay layer. A similar explanation of a significant increaseof the heat flow density in the upper parts of strata wasproposed by [6].

References: [1] Pohl J. et al. (1977) in Roddy et al.(eds.): Impact and explosion Cratering, 343-404. [2]Popov Yu.A. (1997) in Proceedings of the 4-th WorldConference on Experimental Heat Transfer, FluidMechanics and Thermodynamics. Brussels, Belgium, 1,109-117. [3] Pribnow D. and Sass J.H. (1995). J.Geophys. Res. 100, 9981-9994. [4] Popov Yu.A. et al.(1998) Tectonophysics 291, 205-223. [5] Matsui T. et al.(1998) LPS 29, Abstract no. 1255. [6] Kutas R.I. andBevzyuk M.I. (1989) Geophysicheskiy sbornik 87, 68-71 (in Russian)

GEOTHERMAL STRUCTURE OF THE RIES CRATER. Yu. Popov et al.

Legend:

1-2 Lake sediments3-Graded suevitic layer4-Suevites with intercalations of crystalline rocks5-Amphibolites6-Gneisses7-Granites8-Ultrabasites9-Breccia dikesLSC: Lithologic-structural complex

INVESTIGATIONS OF THE RIES CRATER EJECTA USING A DIGITAL GEOLOGICAL MAP, DEM andGIS. J. Pohl1 and E. Geiss2, 1Institut für Allgemeine und Angewandte Geophysik, Theresienstr. 41, D-80333 Munich,Germany, pohl@alice.geophysik.uni-muenchen.de, 2 Bayerisches Geologisches Landesamt, Hess-Str. 128, D-80797München, Germany, erwin.geiss@gla.bayern.de

Introduction: For a new and updated edition of theGeological map of the Ries crater at the scale of 1:50000the map was digitized at the Bayerisches GeologischesLandesamt and incorporated into a geographicalinformation system. This gives us the possibility to makenew investigations of the impact formations such asallochthonous ejecta and parautochthonous blocks withinthe megablock zone and fall-out ejecta outside of thestructural boundary of the crater. In the map blockslarger than 25 m are treated as separate units. Massescontaining parts smaller than 25 m are included in theso-called Bunte Breccia. The complete geologicalmapping of impact formations is of course hampered bynumerous biases due to erosion, difficulties to recognizeunits in the field, coverage by later sediments etc.

In addition to the digital geological map detaileddigital terrain models are also available. Both data setscan be used for combined statistical geological andmorphological investigations of the radial and azimuthaldistribution of the ejecta, of their block sizes as afunction of radius, of their elevation position relative tothe elevation of the autochthonous pre-impact formationsetc..

Target stratigraphy: The pre-impact stratigraphy inthe Ries crater consisted from top to bottom of a thincover of 0 - 50 m of unconsolidated Tertiary sediments(0 - 50 m), of 100 - 150 m of Malmian, predominantlylimestone, of ca. 150 m of Dogger, mainly sandstonesand marls, of ca. 30 m of Liassic, mainly marl andlimestone, of about 230 m of Keuper, mainly sandstonesand shale, of possibly 30 m of Muschelkalk and may besome additional tens of meters of lower Triassic andupper Permian deposits above the crystalline basement,predominantly granites, gneisses and amphibolites.

Ejecta distribution: In a first step we began toseparate different stratigraphic units in the allochthonousejecta formations, differentiating the Jurassic (Liassic,Dogger and Malmian), the Triassic and the crystallinebasement, including the highly shocked suevite. Thefigure shows the distribution of the mapped blocks largerthan ca. 25 m. The distribution of the Bunte Breccia,which contains variable amounts of the sedimentarycover formations, but almost no crystalline rocks, isshown on the lower right.

Strongly asymmetrical distributions are evident. Thelack of ejecta north of the structural crater rim has beeninterpreted as the effect important erosion since thecrater formation. Other asymmetries are not easilyunderstandable assuming a centro-symmetric ejection.The small amount of Liassic ejecta may be due to thesmall thickness of Liassic target, but from the ratherthick Triassic a more imprtant contribution to the ejectacould be expected. In order to estimate the possible

amounts of ejecta of the different target rock formationswe use in a first step a z-model calculation for evaluatingthe ejecta distribution of the different pre-impactformations, taking into account their dipping in NS andEW directions and the probable variations of thicknessin the target area. Other causes of asymmetries should ofcourse also be considered, especially the possible effectsof oblique impact, which have been discussed recently.Asymmetries in the Ries crater are not only visible in theejecta distribution. There are also pronouncedasymmetries in the sub-surface structure of the crater,especially in the morphology of the inner ringsurrounding the central crater. A structural asymmetrymay also be reflected in the gravity anomalies of theRies crater with an offset of the minimum to the north ofthe crater center. This offset has been interpreted untilnow as an effect of the regional gravity field.

References: Bayerisches Geologisches Landesamt(1998), Geologische Karte des Rieses 1: 50 000, 2nd

edition, in press. Engelhardt, W. v. (1990)Tectonophysics 171, 259-273, and literature therein.Pohl, J. et al. (1977) in Roddy et al. (eds.): Impact andexplosion Cratering, 343-404, and literature therein.

Figure: Distribution of ejecta in the Ries crater(blocks larger than ca. 25 m). In the upper left thedistribution of crystalline ejecta is shown. Granites andsuevite are plotted together to give an idea of the overalldistribution of crystalline ejecta, although suevite is abreccia with clasts of a maximum size of a few dm. Inthe lower right the distribution of the Bunte Breccia isshown, which may contain single clasts up to about 25m.

EJECTA OF THE RIES IMPACT CRATER: J. Pohl and E. Geiss

GEOLOGICAL AND GEOPHYSICAL STUDIES OF THE UPHEAVAL DOME IMPACT STRUCTURE,UTAH. K. E. Herkenhoff1, R. Giegengack2, B. J. Kriens, J. N. Louie3, G. I. Omar2, J. B. Plescia1, and E. MShoemaker1, 1United States Geological Survey, 2255 N. Gemini Drive, Flagstaff, AZ 86001-1698 (kherken-hoff@flagmail.wr.usgs.gov), 2Department of Earth and Environmental Science, University of Pennsylvania, Phila-delphia, PA 19104-6316 (gieg@sas.upenn.edu), 3Seismological Laboratory 174, University of Nevada, Reno NV89557-0141 (louie@seismo.unr.edu).

Introduction: Two vastly different phenomena,impact [1] and salt diapirism [2], have been proposedfor the origin of Upheaval Dome, a spectacular scenicfeature in southeast Utah. Detailed geologic mappingand seismic refraction data indicate that the domeoriginated by collapse of a transient cavity formed byimpact. Evidence that Upheaval Dome is an erodedimpact structure includes: 1) sedimentary strata in thecenter of the structure are complexly folded and per-vasively imbricated by top-toward-the-center thrustfaulting, 2) top-toward-the-center normal faults arefound at the perimeter of the structure, 3) clastic dikesare widespread, 4) the top of the underlying salt hori-zon is relatively flat, at least 500 meters below thesurface at the center of the dome, and there are noexposures of salt or associated rocks of the ParadoxFormation in the dome to support the possibility that asalt diapir has ascended through it, 5) the lack of agravity anomaly over the structure is consistent withthe shallow deformation and flat salt horizon inferredfrom geologic mapping and seismic studies, 6) fan-tailed fracture surfaces (shatter surfaces) and rareshatter cones are present near the center of the struc-ture, and 7) planar microstructures have been found insamples from the clastic dikes in the center of thedome.

Geologic Mapping: Detailed geologic mappingindicates that the dome formed mainly by centerwardmotion of rock units along listric faults. Outcrop-scale folding and upturning of beds, especially com-mon in the center, are largely a consequence of thismotion. We have also detected some centerward mo-tion of fault-bounded wedges resulting from displace-ments on subhorizontal faults that conjoin and die outwithin horizontal bedding near the perimeter of thestructure. The observed deformation corresponds tothe central uplift and the encircling ring structuraldepression seen in complex impact craters [3]. Theapparent depth of erosion of the structure (between 0.1and 2 km) suggests that the impact occurred eitherduring the late Jurassic/early Cretaceous or during thelate Triassic.

Recent study of the rounded cobbles found at Up-heaval Dome that were previously interpreted as “im-pactites” [1] suggests that they may be a lag deposit ofchert nodules commonly found at the top of the Na-vajo Sandstone in the Canyonlands region [4]. Planarmicrostructures have been recognized in quartz grains

in thin sections of some samples, some of which re-semble planar deformation features. Results of ourcontinuing studies of these samples will be reported atthe conference.

Seismic Reflection Results: We obtained a 5 kmseismic section extending radially from the Dome'scentral depression using a 320 kg weight-drop sourceand a 48-channel off-end receiver spread 0.5 km long.The data show clear reflections as deep as 1.5 km.Imaging of the reflection section with velocity filteringand 3-D prestack Kirchhoff migration techniques re-veals the geometries of deformed stratigraphy from thesurface to the top of the Paradox Formation at 1.2 kmdepth. Stratigraphic terminations and fault-plane im-ages show the paths of listric faults. We tied our sec-tions to two well logs, one in the ring syncline and oneoutside the zone of deformation. Listric faults flattenand sole into the clastic formations above the calcare-ous layers of the Hermosa Formation at 1.0 km depth.At the base of the Hermosa, on the axis of the ringsyncline, the Paradox has forced the Hermosa 0.1 kmup and broken it with thrust faults. Post-impact re-laxation of the crater form may have driven thisdeeper uplift.

Seismic Refraction Results: Refraction rayspassing within 500 m below the center of UpheavalDome show no evidence of early arrivals. Rays pass-ing below Buck Mesa and Syncline Valley have veryearly arrivals. There is no evidence of any salt diapirbelow the center of Upheaval Dome [5]. High veloci-ties at depth ringing Upheaval Dome may be due to:1) an asymmetric bulge of the top of the Paradox; or2) to the presence of a relatively low-velocity shatteredzone at the center of the structure. A central shatteredzone is more consistent with the minor deformation ofstratigraphy observed on Buck Mesa. A more com-plete description of our seismic results can be foundat:http://www.seismo.unr.edu/ftp/pub/louie/dome/index.html

Gravity Survey Results: Joesting and Plouff’sgravity data [7] indicated a positive anomaly associ-ated with the structure. Their figure 4 shows aBouguer map (assuming a density of 2.5 g cm-3) thatindicates a positive anomaly of about 5 mGal; in thetext they describe the anomaly as +3 mGal. However,they further state that the anomaly is only about +1mGal when errors associated with the station eleva-tions are considered.

STUDIES OF THE UPHEAVAL DOME IMPACT STRUCTURE, UTAH: K. E. Herkenhoff et al.

In light of these uncertainties and to constrain thecrustal structure, a gravity survey was conducted.Data were collected at each of the seismic refractionstations, along the access road on the southeast side ofthe structure, and at scattered locations around andwithin the structure. The Canyonlands region is oneof rugged topography; terrain corrections are requiredand are significant. Near-station corrections are madeby hand, hence there is significant uncertainty in es-timating the correction. We estimate that the uncer-tainty in estimating the near-station topography couldcorrespond to perhaps a milliGal. Thus, in order toattach geologic significance to an anomaly, it musthave an amplitude of several milliGals.

Our results suggest that no gravity anomaly existsassociated with Upheaval Dome. If a reduction den-sity of 2.67 g cm-3 is used (a standard reduction den-sity), a positive anomaly of about 5 mGal is indicated.However, this density is significantly greater than thatof the sedimentary rocks exposed at Upheaval Dome.If the reductions are made using a density of 2.3 g cm-

3 (± 0.1 g cm-3), a value more consistent with the den-sity of the rocks, no Bouguer gravity anomaly is indi-cated.

The absence of a gravity anomaly is consistentwith the seismic data that suggest that the Paradoxsalt formation is essentially flat beneath the structureand the geologic mapping which suggests deformationis limited to shallow crustal levels. Structural defor-mation of the clastic rocks above a decollement willnot significantly affect the density of those rocks,hence a gravity anomaly would not be expected. Theabsence of a gravity anomaly at Upheaval Dome, vis-à-vis other impact structures, is consistent with thegeology. Gravity anomalies at impact structures typi-cally result from the impact breccia layer within thestructure and/or higher density rocks exposed in thecentral peak, both of which are absent at UpheavalDome.

Fission Track Analysis: In June 1990 we col-lected samples of all rock types exposed in the centraluplift. We collected the same rock types from thewalls of the canyon of the Colorado River along theShafer Trail, 11 km NE of the center of UpheavalDome. That locality is far enough from the center ofUpheaval Dome to have escaped the effects of shockmetamorphism from the presumed impact event, butclose enough to offer some assurance that we weresampling the same rock units that we had collectedwithin Upheaval Dome. Of the 10 samples acquired(5 from within Upheaval Dome; 5 from Shafer Trail),only two (one from each locality) yielded enough apa-tite grains to enable us to determine fission-track ageswithin acceptable statistical limits. Both of thosesamples were taken from the Moss Back conglomeratenear the base of the Chinle Formation. We returned

to those two sites in 1994 and 1995 to acquire moresamples.

Samples of Moss Back conglomerate from both lo-calities contain abundant woody material; in samplesfrom within Upheaval Dome much of that wood isfully carbonized. A vitrinite reflectance measurement(undertaken by Gareth Mitchell and Alan Davis, ofthe Coal and Organic Petrology Laboratory of PennState University) shows no increase in reflectance overnon-carbonized wood, suggesting that the elevatedtemperature that affected the wood in samples fromwithin Upheaval Dome was of short duration. Pet-rographic study of samples of Moss Back conglomer-ate from within Upheaval Dome revealed abundantinterstitial glass, partially devitrified, and an abun-dance of a well crystallized mineral phase that weidentified as [Ba(0.75), Sr(0.25)]SO4, a member of aseries between Barite (BaSO4) and Strontianite(SrSO4). This mineral may be synthesized in thelaboratory at ~1,000°C [6]. Neither the interstitialglass, the sulfate mineral, nor the carbonized woodwas observed in the sample of Moss Back conglomer-ate collected from Shafer Trail Road.

Even in these two samples, apatite is not abundant,and must be hand-picked from the heavy-mineralfraction. To improve the statistics of the track-lengthmeasurements, we will irradiate the samples withneutrons from 252Cf to expose horizontal confinedtracks in the interior of the apatite grains to theetchant. We expect to be able to announce a fission-track age, a track-length histogram, and a modeledthermal history for each sample at the meeting.

Summary: Stratigraphic uplift observed in thecenter of Upheaval Dome is the result of convergentdisplacement of the wall of a transient cavity formedby hypervelocity impact, not Paradox salt diapirism.Geophysical studies indicate that the top of the Para-dox is deformed up to 100 m vertically, but the top ofthe Hermosa appears undeformed. At least 5 km indiameter, Upheaval Dome is the largest recognizedimpact structure on the Colorado Plateau.

References: [1] Kriens, B. J. et al. (1997) BYUGeol. Studies, 42, Part II, 19-31. [2] Jackson, M. P.A. et al. (1998) GSA Bull., 110, 1547-1573. [3] Wil-shire, H. G. et al. (1972) USGS Prof. Paper 599-H, 42pp. [4] Pipiringos, G. N. and R. B. O’Sullivan (1975)in Canyonlands Country: Four Corners Geol. Soc.Guidebook, 8th Field Conference, edited by J. E. Fas-sett, pp. 149-156; Peterson, F., and G. N. Pipiringos(1979) USGS Prof. Paper 1035-B, 44 pp. [5] Louie, J.N. et al. (1995) Eos, 76, 337. [6] Mitchell and Davis(1995) pers. comm. [7] Joesting, H. R. and D. Plouff(1958) Utah Intermountain Assoc. Petrol. Geol., 9thAnnual Field Conf., 86-92.

LOW-ANGLE FAULTING IN THE BASEMENT OF COMPLEX IMPACT CRATERS:NUMERICAL MODELLING AND FIELD OBSERVATIONS IN THE ROCHECHOUARTSTRUCTURE, FRANCE. T. Kenkmann1 and B. A. Ivanov2, 1Museum für Naturkunde, Institut fürMineralogie, Humboldt-Universität Berlin, Invalidenstraße 43, D-10115 Berlin, Germany, 2Institute forDynamics of Geospheres, Russian Academy of Science, Moscow, Russia 117939.

Introduction. The formation of complexcraters on the earth and other terrestrial planets isinduced by a collapse of a parabolically shapedtransient crater cavity that formed after thehypervelocity impact of an asteroid or a comet.Interplanetary comparison of the crater diametersfor the transition from simple to complex cratersprove that gravity is the principal force drivingthe collapse. Failure of the transient cavity occursin acoustically fluidized rocks with low cohesionand reduced friction coefficients. Deformation isstrongly localized and occurs along discrete faultzones. The distribution, geometry and kinematichistory of fault zones exposed in the crater floorof complex impact craters has not beenunderstood very well so far and only a fewinvestigations have focussed on this problemrecently [1]. Morphological criterions of extra-terrestrial craters (headscarps, terrace-likefeatures) are helpful to deduce the mechanisms ofnear-surface faulting. However, the investigationof deeper structures rests upon observations andinterpretations of terrestrial craters sincegeological cross-sections are not available for anyextraterrestrial crater.

Structural investigations. The Rochechouartimpact crater is located about 40 km SW ofLimoges, France (latitude N 45°50´, longitude EO°56´). The structure has a diameter ofapproximately 25 km [2] and an age of 214 ± 8Mio. a [3]. Morphologically, the crater has beendestroyed by erosion of more than 1 km depth.Impact-induced faults have been studiedpredominantly in a quarry near the village ofChampagnac north of Rochechouart (6 km NE ofthe impact centre). In order to distinguish faultsinduced by the impact event from pre-impacttarget faults our main criterion was the presenceof pseudotachylites.

The present outcrop situation exposes aprominent low-angle normal fault that dips gentlywith 5-35° towards the crater centre [4]. This faultcrosscuts all lithologies. Its azimuth of inclinationis SSW. Striations on the undulating fault planehave partly oblique dip-slip trajectories andindicate normal faulting with an average slipvector of 210° (SSW). The major fault partlyconsists of an anastomosing network of sub-faults. Extensional duplexes of deca-metre sizedeveloped along this detachment. They haveformed in order to compensate local ramps and

flats of the normal fault. In addition to this,conjugate faults with antithetic dip have formed.More steeply-inclined subsidiary faults thatbranch off from the main low-angle fault form anextensional imbricate fan. These faults join thebasal low-angle detachment in a depth of 30-100m below the transition from the parautochthonousbreccias to the autochthonous country rock. Theyare synthetic to the basal shear plane – that is theydip in the same direction. Some of the subsidaryfaults are connected by splay faults. Steepersubsidiary faults partly transsect parautochto-nuous monomict breccias which lie on the top ofthe basement. These faults may have undergoneunconstrained (free surface) dip-slip. They mayrepresent analogues to headscarpes at the surfaceof extraterrestrial craters, which separate terrace-like features.

Main and subsidiary faults of the Champagnacquarry are covered with grey-black colored rockswhich resemble pseudotachylites. These dyke-likemasses of centimetres to several decimetres widthand tenth of metres length are strongly alignedalong the discrete fault zones. Pseudotachylitescontain considerable amounts of hydrothermalphases. Thinner pseudotachylite veins (< 1mm –cm) also cover subordinary faults, veinlets andfracture joints of the footwall block below thedetachment. They formed more pervasively alongan irregular or anastomozing fracture network ofveins and feather joints. A few subordinary faultspreserve evidence for multiple slip, which ispossibly related to an oscillation mechanism ofblocks. This is proposed to occur during shock-induced acoustic fluidization of the target [5].

Numerical modelling of crater modification.In an attempt to understand the occurrence oflow-angle normal faults in the crater basement ofcomplex craters a numerical model of cratermodification was designed. The modified SALE-2D hydrocode [6] was used in the Lagrangianmode. We used the Tillotson equation of state forbasalt to describe the thermodynamics of thetarget rocks. Computation was started when thetransient cavity stops to grow. In order to take theeffects of fluidization into account we artificiallydecreased internal friction and took a cohesion of0.1 MPa. In our model we assumed that theeffects of acoustic fluidization are not equallydistributed throughout the whole crater floor.However, a function of decay of oscillation in

LOW-ANGLE FAULTING IN THE BASEMENT OF COMPLEX IMPACT CRATERS: T. Kenkmann and B. A. Ivanov

time and space is not available up to date. Weapplied a very simple model and assumed a lineardecrease of the friction coefficient, µ, with depthstarting with normal values of 0.43 at the freesurface and reaching µ = 0.058 at the level of thedeepest point of the transient cavity. The frictioncoefficient remains constant towards the bottomof the disrupted crater basement. We assumed thatthe transient cavity was excavated to a depth of 4km and that the depth/diameter ratio was 0.33.

Model results. Sliding at the crater rim startsat the crest where several slip lines form after 20sec. The inclination of slip zones decreases withdepth. Shear zones in the periphery remainsteeply inclined, but in the central part they bendinto a subhorizontal orientation. Uplift of thecentral area reverses the dip orientation of the slipzone by passive rotation. The central uplift startsto collapse after 60 sec and slip trajectoriesindicate reverse motion (outward flow). Sinceinward flow continues in the more distal parts,collision occurs in a transition zone. Thisultimately leads to the squeezing-out of rockmaterial to the surface. After 90 sec themovement stops. A main characteristic of thecollapsed crater model are three weakly-inclinedshear zones occurring in a distance ofapproximately 3 to 8 km away from the cratercentre (Fig. 1). They have a lateral extent ofapproximately 5 km at 0.5 - 2 km depth (Fig. 1).Because of the circular shape of impact cratersthese faults should be arranged in a ring-likefashion. Less-localized shear zones also developwith opposite dip orientation. A very complexstyle of extensional and compressionaldeformation dominates in the uppermost region ofthe central peak.

Discussion. The significance of low-anglenormal faulting during crater modification can bededuced in combining field observations withnumerical computations. The low-angle normalfaults are interpreted as being the base ofdownward and inward moving slides and slumps

of rock units located primarily in the walls andnear the rim of the transient crater. The degree offault inclination during crater modificationchanges due to the uplift and collapse of thecentral uplift (passive rotation). Detachmentfaulting is also caused by the impact-inducedrheological stratification of the crater basement. Itis suggested that the degree of fluidizationchanges with depth and distance from the impactcentre. The effects of acoustic fluidization arestrongest at the base of the transient cavity. Thisresults in a mechanically weak zone which issandwiched between regions of relatively higherstrength. In this mechanically-weak zone simpleshear may dominate along horizontal flow linesand leads to the formation of low-angle faults.

The displacements along the low-angle faultscalculated with the model vary strongly andlocally exceed 1 km. Based on the width ofpseudotachylites in Champagnac we estimatedisplacements of about 1.2 km (width: 0.3 m,coefficient of friction µ: 0.1, latent heat fusion:4E+5 Jkg-1, depth of burial: 100 m). Spray [1]proposed the term ”superfault” for faults charac-terized by unconstrained, single-slip movements.Effects of thermal softening which may be quiteeffective in narrow shear zones, are under-estimated in our model because the resolution ofthe mesh does not allow any localization ofdeformation below the grid size.

References [1] Spray J. G. (1997). Geology,25, 579. [2] Stöffler D. et al. (1988). In Boden A.and Eriksson K. G. (eds.), Deep Drilling inCrystalline Bedrock, 1, 277. [3] Kelley S. P. andSpray J. G. (1997). MPS, 32, 629. [4] Kenkmann,T. et al. (1998) ESF-workshop. Impacts and theearly earth, Cambridge, 13. [5] Ivanov B. A. andKostuchenko V. N. (1997). LPSC, 28, 631. [6]Amsden A. A., et al. (1980). Los Alamos NationalLaboratory LA-8095, Los Alamos, NM, 101 pp.

Fig. 1 Fault distribution and fault kinematics inferred from the numerical model after movement hasstopped. Some of the faults in the central area are activated in a multiple fashion. For explanation of thekinematic history, see text.

PETROGENETIC MODELING OF THE DULLSTROOM FORMATION, SOUTH AFRICA. P. C. Bu-chanan1, C. Koeberl2, and W. U. Reimold3, 1Mail Code: SN2, NASA Johnson Space Center, Houston, Texas, USA77058, pbuchana@ems.jsc.nasa.gov, 2University of Vienna, Althanstrasse 14, A-1090 Vienna, Austria,3Department of Geology, University of the Witwatersrand, Private Bag 3, Wits 2050, Johannesburg, South Africa.

Introduction: The intrusive Bushveld Complexand the surrounding volcanic units of the RooibergGroup and the Dullstroom Formation are products ofan extended period of magmatic activity that affectedsouthern Africa. Some previous workers [e.g., 1] haveinterpreted these units as the result of the impact ofseveral comets or asteroids. Other workers have at-tributed these magmatic rocks to the effects of a man-tle plume [e.g., 2] or a subduction zone [3]. TheDullstroom Formation, which is predominantly com-posed of mafic to intermediate volcanic strata inter-bedded with rare sedimentary strata and felsic vol-canic units [4], makes up the floor of the complex andrepresents the first phase of this magmatic episode.This study is an attempt to model available geochemi-cal data for the Dullstroom Formation in order to un-derstand the processes responsible for the initiation ofthis magmatic episode.

Discussion: Harmer and von Gruenewaldt [5] andSchweitzer et al. [4] determined that volcanic units ofthe Dullstroom Formation could be subdivided into atleast three interbedded compositional groups: low-Timafic to intermediate strata, high-Ti mafic to inter-mediate strata, and rare felsic strata. Our geochemicalanalyses of these volcanic rocks support this subdivi-sion and indicate that these strata are tholeiitic totransitional in character. Low-Ti and high-Ti unitsrange from basalts to andesites, whereas felsic vol-canic rocks range from andesites to dacites. Low-Tiunits are distinguished by compositions that are in-compatible trace element-rich and are similar to manyother volcanic rocks from continental environments(Fig. 1). Pyrolite-normalized concentrations of Nb,Sr, Sm, and Ti for these rocks are low relative to otherincompatible elements.

A variety of models were calculated in an attemptto reproduce compositions of the low-Ti units of theDullstroom Formation. The model that best matchedthe compositional data for these strata involved bulkassimilation into a picritic melt of 30% upper conti-nental crust accompanied by 30% fractional crystalli-zation of olivine of composition Fo85 (model 1, Fig. 1).The composition of the picritic melt used in these cal-culations is that of a very MgO-rich picrite from west-ern Greenland [6], which presumably represents ahigh degree of partial melting of a mantle source.Pyrolite-normalized trace element abundances of this

picrite are low in Ba, Th, Nb, Sr, and Zr relative toother incompatible elements (Fig. 1). The averagecomposition of the upper continental crust used inthese calculations was taken from the estimates ofTaylor and McLennan [7] (Fig. 1).

Fig. 1. Pyrolite-normalized spider diagrams com-paring the average composition of low-Ti volcanicrocks of the Dullstroom Formation with model 1 (thisstudy). Also included in this diagram are composi-tions of the high-MgO picrite [6] and the estimatedcomposition of the upper continental crust [7] used inthese calculations. Composition of pyrolite is takenfrom McDonough and Sun [8].

Rare felsic volcanic units of the Dullstroom For-mation are more enriched in incompatible trace ele-ments than low-Ti units. Pyrolite-normalized compo-sitions also display minima in Nb, Sr, Sm, and Tirelative to other trace elements. These felsic volcanicrocks are similar in composition to a melt derived bysubjecting the average composition of low-Ti volcanicunits to an additional assimilation of 25% upper con-tinental crust and 25% crystallization of a mixture of60% plagioclase and 40% low-Ca pyroxene (model 2,Fig. 2).

High-Ti volcanic units are more incompatible traceelement-rich than low-Ti strata and generally havesmaller depletions in pyrolite-normalized abundancesof Nb, Sr, Sm, and Ti relative to other incompatibleelements (Fig. 3). For several elements (e.g., La),abundances of high-Ti volcanic units are similar to orgreater than those estimated for the upper continental

1

10

100

1000

Ba

Rb Th

K2O N

b La Sr

Nd

Sm Zr

Hf

TiO

2 Y Yb

AB

UN

DA

NC

E/P

YR

OL

ITE

ave. low-Ti

model 1

up. crust

picrite

DULLSTROOM FORMATION: P. C. Buchanan, C. Koeberl, and W. U. Reimold

crust by Taylor and McLennan [7]. Such abundancesmake it difficult to derive these high-Ti melts by proc-esses similar to those suggested for low-Ti volcanicunits, except by unrealistically large proportions ofassimilation of upper continental crust into the samepicritic melt. It is possible, however, that these high-Ti volcanic units are derived from parental melts thatare much richer in incompatible trace elements thanthe picritic melt used to calculate model 1. This pa-rental melt could have formed by lower proportions ofpartial melting of the same source area as the parentalmelts of the low-Ti volcanic units or by partial melt-ing of a different, more incompatible element-richsource region.

Fig. 2. Pyrolite-normalized spider diagrams com-paring the compositions of two felsic units from theDullstroom Formation with model 2. Composition ofpyrolite is taken from McDonough and Sun [8].

Fig. 3 Pyrolite-normalized spider diagrams com-paring the average composition of high-Ti volcanicrocks of the Dullstroom Formation with compositionsof the high-MgO picrite [6] and the estimated compo-sition of the upper continental crust [7] used in calcu-

lating model 1 and model 2. Composition of pyroliteis taken from McDonough and Sun [8].

Conclusions: The present study suggests that low-Ti and felsic volcanic units of the Dullstroom Forma-tion can best be modeled as a sequence of fractionatedliquids that have been affected by significant amountsof assimilation of upper continental crust and crystal-lization of olivine, pyroxene, and plagioclase. Themost reasonable interpretation for such large amountsof assimilation and fractional crystallization is thatthese melts resided for significant periods of time inone or more shallow magma chambers before beingextruded. In contrast, high-Ti units may be derived bylower degrees of partial melting of similar source ar-eas or by partial melting of source areas that are moreincompatible trace element-rich. These characteristicsand the volcanic stratigraphy of the formation [seealso 4 and 9], which is distinguished by interbeddingof volcanic units with rare sedimentary strata, suggestthat rocks of the Dullstroom Formation are the resultof endogenic terrestrial processes and are difficult toreconcile with a direct origin by bolide impact. Thesedata may be more consistent with an origin associatedwith a mantle plume [2], since low-Ti and felsic unitsmay be derived from a plume-generated, high propor-tion partial melt by assimilation and fractional crys-tallization in one or more shallow magma chambers,whereas high-Ti units may represent lower proportionpartial melts from the periphery of the plume.

Acknowledgments: Part of this work was sup-ported by the Austrian FWF, project Y58-GEO.Funding was also provided by the Foundation for Re-search Development of the Republic of South Africa.

References: [1] Rhodes R. C. (1975) Geology, 3,549-554. [2] Hatton C. J. (1995) J. Afr. Earth Sci., 21,571-577. [3] Hatton C. J. (1988) Abstracts of theCongress of the Geological Society of South Africa,22, 251-254. [4] Schweitzer J. K. et al. (1995) S Afr JGeol, 98, 245-255. [5] Harmer R. E. and von Grue-newaldt G. (1991) S. Afr. J. of Geol., 94, 104-122. [6]Thompson R. N. et al. (1983) in Hawkesworth C. J.and Norry M. J. (eds.) Continental basalts and mantlexenoliths, Shiva Publishing, Nantwich, U.K., pp 158-185. [7] Taylor S. R. and McLennan S. M. (1985) Thecontinental crust: Its composition and evolution.Blackwell Scientific Publications, Oxford, U.K., 312p. [8] McDonough W. F. and Sun S.-s. (1995) Chemi-cal Geology, 120, 223-253. [9] Buchanan P. C. andReimold W. U. (1998) EPSL, 155, 149-165.

1

10

100

1000

Ba

Rb

Th

K2O N

b La Sr

Nd

Sm Zr

Hf

TiO

2 Y

Yb

AB

UN

DA

NC

E/P

YR

OL

ITE

ave. high-Ti

up. crust

picrite

1

10

100

1000

Ba

Rb

Th

K2O N

b La Sr

Nd

Sm Zr

Hf

TiO

2 Y

Yb

AB

UN

DA

NC

E/P

YR

OL

ITE

felsic kk22

felsic kk23

model 2

NEW DATA ON THE NATURE OF IMPACT DIAMONDS FROM THE LAPPAJÄRVI IMPACTSTRUCTURE, FINLAND. R. M. Hough1, S. Vishnevsky2, J. I. Abbott1, N. Pal’chik2, J. Raitala3 and I. Gil-mour1, 1P.S.R.I, The Open University, Walton Hall, Milton Keynes, MK76AA, UK, 2Institute of Mineralogy and Petrol-ogy, Novosibirsk, Russia, 3Astronomy Space Institute, University of Oulu, 90570 Oulu, Finland. (email:r.m.hough@open.ac.uk )

Impact derived diamonds are an increasingly com-mon phenomenon associated with a number of impactcraters including Popigai, Ries, Chicxulub, Kara, Za-padnaya, Terny and Obolon. New discoveries of dia-monds at Sudbury [1] and Lappajärvi [2] further sup-port the minerals’ use as an indicator of impact proc-esses. Impact diamonds commonly occur as pseudo-hexagonal plates after their crystalline parental graphitegrains. They do however, also occur as skeletal laths,as minor nano-diamonds and as much more anhedralforms lacking any platey or hexagonal structure pre-sumably as a remnant of a poorly crystalline carbonsource. Recognition of the different diamond forms isrequired to fully understand their origins.

In order to further characterise impact diamonds wehave undertaken a detailed study of samples from theLappajärvi structure. Samples of suevites, a large bombof suevite glass and a sample of Kärnäite from the 77Ma old Lappajärvi impact structure in Finland wereinvestigated. These were decomposed using the alkalifusion technique at 550°C as in [1, 2]. The residueshave been investigated using optical, scanning andtransmission electron microscopy, X-ray analyses andcarbon isotopic analyses. A total of 54 diamond grainswere extracted from the suevite samples, the other twosamples were found to be devoid of diamonds. Thesuevite samples also contained many graphite grains.Optically, the cubic (as confirmed by X-ray analyses)diamond grains appear yellow to white and often milkydue to surface textures. The diamond grains range from0.1 to 0.4 mm in size with platey, rounded and angu-lar grains as well as the occasional so-called volume-xenomorphic grain. The surface textures viewed bySEM indicate corrosion possibly by the exothermicnature of the extraction technique but more likely fromlate stage oxidation and reaction with other gases whilethe grains are still hot after the impact. Deeper traces ofetching are exhibited by some grains with complextunneling and cavities. Individual plates/flakes showthe impact diamond characteristic of surface hatchingand lineations as described in [2, 3]. The X-ray analy-ses indicate a preferred orientation of the crystalliteswithin the polycrystalline grains and this is confirmedby TEM analyses as is the cubic structure (by SAED).The presence of lineations on a very fine scale is alsoobserved using TEM which is again considered to be awider characteristic of impact diamonds [3, 4]. Pre-liminary carbon isotopic analyses have been performedon graphite and diamond from the residues. Staticmass spectrometry combined with stepped heating

provides a very sensitive tool for the analysis of smallsamples [5]. The grains extracted in this study are onaverage <0.1mm in size. The grains were placed inpre-cleaned platinum buckets (by baking overnight at1,000°C). The buckets plus sample are loaded into agas extraction glass line and into a section heated by afurnace. For these analyses the samples were heated to1,200°C after a pre-combustion step at 200°C to re-move organic contamination. The heating in the pres-ence of oxygen forms CO2 which is extracted usingcryogenic techniques before entering the mass spec-trometer. Initial results indicate that the diamond andgraphite grains are within the typical carbon isotopicrange for other impact diamonds analysed from boththe Popigai [6] and Ries craters [7], with a δ13C for thediamond of –19‰ and of –25‰ for the graphite grain.

The carbon isotopic compositions of impact dia-monds are thought to be related to carbon sources,combining these initial results with structural and tex-tural information confirms a probable carbon source ofgraphite within the crystalline target rocks as describedby [2]. Such a process of transformation of graphite todiamond has been proposed for the impact diamondsfrom Popigai and Ries and has been confirmed as plau-sible by experimental production of shock diamonds.Pyrolitic graphite experimentally shocked at modestshock temperatures (circa 1,000°C) and at pressureswithin the diamond stability field has been shown toproduce vitreous diamond flakes preserving many ofthe characteristics observed in the impact diamondsfrom Popigai, Ries and now Lappajärvi (Paul De Carlipers. comm.). The experimentally produced flakes haveboth surface features and internal lineations presumablyproduced by the passage of shock waves through thehighly crystalline grains.

The Lappajärvi diamonds have enabled us to fur-ther constrain the form, structure and composition ofimpact diamonds and to support the need for graphiterich target rocks in order to produce these “characteris-tic” shock produced diamonds.

[1] Masaitis V. L. et al. (1997) LPI contributionNo 922, p. 33. [2] Masaitis V. L. et al. (1998) LPSXXIX,. [3] Langenhorst. F. et al. (1998) Meteoritics &Planet. Sci., 33, A90. [4] Hough R. M. et al. (1998)Meteoritics & Planet. Sci., 33, A70. [5] Wright I. P.& Pillinger C. T. (1989) USGS Bull, 1890, 9-34. [6]Shelkov D. A. et al. (1998) Meteoritics & Planet. Sci.33. No. 5. 985-992. [7] Abbott J. I. et al. (1998) Me-teoritics & Planet. Sci., 33, A7.

CARBON-BEARING IMPACTITES FROM THE GARDNOS IMPACT STRUCTURE,NORWAY: NO EVIDENCE FOR SOOT Wendy S. Wolbach1, Susanna Widicus1, and BevanM. French2, 1Department of Chemistry, Illinois Wesleyan University, P.O. Box 2900, Blooming-ton, IL 61702-2900, 2Department of Mineral Sciences, Smithsonian Institution, NHB MRC 119,Washington, DC 20560.

The 5-km-diameter Gardnos, Norway im-pact structure [1] is one of only two impactstructures whose shocked basement rocks andimpact-produced breccias contain significantcarbon contents, typically 0.1 – 1 wt% totalcarbon. (Sudbury, Canada is the other.)These values are 5 – 10 times greater thanthose of the exposed basement target rocks atGardnos. The source(s) of the additional car-bon is still undetermined. Possible sourcesinclude a carbonaceous impactor, carbona-ceous target sediments no longer preserved atthe impact site, and fluids introduced duringsubsequent metamorphism. δ13C values (–29to –32 permil) for Gardnos carbon are consis-tent with terrestrial values, indicating that nomajor meteoritic component is present [1].

To test the possibility that combustion ofthe impactor or carbon-bearing rocks couldhave occurred during the impact, we searchedfor soot in a wide variety of Gardnos impac-tites and related rocks. Samples of Gardnosimpactites included: shocked autochthonousbasement rocks beneath the crater (quartziteand granitic “Gardnos breccia”); possiblebreccia dikes (“black-matrix breccia”) in thebasement rocks; melt-bearing allochthonousbreccias (suevite and impact melt breccias).We also analyzed post-impact crater-fillingsediments (sandstone and black shales) fromGardnos, as well as black shale units (Pre-cambrian Biri Shale and Cambrian AlumShale) from distant (>50 km) localities thatcould have been present at the impact site atthe time of impact.

Dissolution and analysis procedures were

based on those used successfully for detectingsoot from impact-produced wildfires at theCretaceous-Tertiary (K/T) boundary [2–5].Reduced carbon was isolated from sedimentsusing HCl and HCl/HF. Elemental carbonwas separated from organic carbon by acidicdichromate oxidation under controlled condi-tions. The elemental carbon of interest (soot,charcoal) was identified and characterized us-ing SEM imaging and quantified by weighingand particle size analysis [4].

Unlike rocks which have been analyzedacross the K/T boundary, many of thesesamples contained significant quantities ofHF-resistant (or slow to dissolve) silicateminerals. In some cases, significant quantitiesof these undissolved silicate and other miner-als were left along with the desired elementalcarbon following demineralization and oxida-tion. We corrected our post-oxidation carbonweights using SEM imaging and particle sizeanalysis of both carbon and silicate fractions(we were able to distinguish between thesefractions using backscattered electron imagingon the SEM). Most required corrections werelarge, and soot contents were not consideredsignificant unless the amount exceeded theestimated error. After corrections were ap-plied, none of the Gardnos impactites showedsignificant soot contents (>1 ppm). Thenegative results do establish that surface con-tamination by soot was not a problem, eventhough many sample locations were close toautomobile traffic, mechanical operations, andpossible vegetation fires. No detectable com-bustion products are observed in any of the

CARBON-BEARING IMPACTITES FROM THE GARDNOS IMPACT STRUCTURE, NORWAY: W. S. Wolbach et al.

Gardnos samples directly associated with theimpact event.

Surprisingly, significant soot contentswere found in three black shale samples notdirectly related to the Gardnos impact event,one (sample 96-161a, with 1100 ± 300 ppm)from post-impact crater-filling sedimentswithin the Gardnos crater, and two (sample96-170, with 2000 ± 1500 ppm and sample96-171, with 6200 ± 4700 ppm) from thepre-impact Precambrian Biri Shale over 100km east of the crater. The origin of the sootin these samples is currently not understood,and more research is in progress. Because theGardnos impact event occurred at least 400Ma ago and possibly 650-700 Ma ago [1], itis also not clear what source could have fur-nished the carbon for combustion to soot at atime when land vegetation was sparse or evenabsent. The occurrence of soot in black shalesediments of varying age may be an importantgeological problem for future study, even ifthe soot observed in these samples is not re-lated to the Gardnos impact itself.

References: [1] French B. M., KoeberlC., Gilmour I., Shirey S. B., Dons J.A., andNaterstad J. (1997) Geochimica et Cosmo-chimica Acta 61, 873-904. [2] Wolbach W.S., Lewis R. S., and Anders E. (1985) Science230, 167-170. [3] Wolbach W.S., Gilmour I.,Anders E., Orth C.J., and Brooks R.R. (1988)Nature 334, 665-669. [4] Wolbach W.S. andAnders E. (1989) Geochimica et Cosmo-chimica Acta 53, 1637-1647. [5] WolbachW.S., Gilmour I., and Anders E. (1990) In:Global Catastrophes in Earth History (eds.V.L. Sharpton and P. Ward). Geological Soci-ety of America Special Paper 247, 391-400.

IMPACTITE FACIES WITHIN THE WETUMPKA IMPACT-CRATER FILL, ALABAMA. D. T. King,Jr.1, T. L. Neathery2, and L. W. Petruny3, 1Dept. Geology, Auburn University, Auburn, AL 36849-5305 [king-dat@mail.auburn.edu], 2Neathery and Associates, 1212-H 15th Street East, Tuscaloosa, AL 35404, 3WesleyanUniversity, Dept. Earth and Environmental Sciences, Middletown, CT 06457.

Introduction: Alabama’s Wetumpka impact cra-ter is a 6.5-km diameter, Late Cretaceous structurethat formed in a two-layer target (i.e., soft sedimentoverlying crystalline rock). At the time of cratering,estimated at 80 to 83 m.y. ago, this target was overlainby up to 100m of seawater covering a shallow conti-nental shelf [1], [2]. Target stratigraphy: Target stratigraphy atWetumpka includes three essentially unconsolidatedUpper Cretaceous units which disconformably overliepre-Cretaceous crystalline basement of the Appala-chian piedmont. These units, in stratigraphic order,are: Tuscaloosa Group (including the Coker Forma-tion and overlying Gordo Formation; total 60 mthick); Eutaw Formation (30 m); and MoorevilleChalk (30 m) [3]. All three stratigraphic units pluscrystalline basement rocks were involved in the im-pact event and contributed clastic material to the cra-ter-filling impactite unit [4]. Surficial crater geology: Surficial crater geologyconsists mainly of two distinctive terraines: 1) a semi-circular rim (having up to 87 m relief) composed ofAppalachian piedmont micacous schist and gneiss and2) a relatively low-relief crater floor composed of con-tiguous tracts of highly disturbed, clastic dike-injectedtarget strata (mostly Tuscaloosa Group, but significanttracts of Eutaw Formation and Mooreville Chalk alsooccur in the crater floor [2], [3]). Directly adjacent tothe crater rim, slump-derived megablocks of Tusca-loosa and Eutaw strata crop out as minor crater-floordeposits [1]. Only small outcrops of impactite facies,like those cored in this study, crop out on the craterfloor, and these impactite outcrops are located atWetumpka’s central rebound peak [3]. Drilling: During July-August 1998, two coreholes were drilled on the eastern flank of Wetumpka’scentral rebound peak. Core holes were sited approxi-mately 200 m apart and at approximately the sameelevation, and both reached depths of approximately190 m. Both core holes penetrated steeply dippingUpper Cretaceous target sediments (specifically, Tus-caloosa Group) to a depth of approximately 64 m.Below that level, both core holes penetrated interca-lated impactite deposits, which encompass five mainfacies.

Impactite facies: Wetumpka impactite facies typesinclude: (1) lignitic, clast-bearing, clay-rich sands;(2) sandy, matrix-supported, monomict crystallineclast diamictites; (3) monomict, clast-supported, im-pact breccias; (4) cataclastic, matrix-supporteddiamictites; and (5) target-rock blocks. Impactitesands have a green clay matrix, are rich in very finelignite, and contain minor crystalline and sedimentaryclasts, including rounded quartz pebbles from theTuscaloosa Group. Monomict diamictites and brec-cias contain only crystalline (i.e., basement-derived)clasts, which are generally less than 0.2 m across.Cataclastic diamictites display crystalline clasts,which are generally less than 0.1 m across, within amatrix of finely comminuted crystalline and sedi-mentary rock. Target-rock blocks are internally de-formed megaclasts (<1 m) of clayey sand, schist orgneiss. Origin of facies: All impactite facies are inter-preted to have formed from catastrophic disintegrationof both sedimentary and crystalline target layers. Im-pactite sands are preliminarily interpreted to representimpact fluidization of the upper sedimentary targetlayer. Monomict diamictites and breccias represent,respectively, matrix-rich and matrix-poor end-members in a continuum of amalgamation betweenbasement-derived clasts and fluidized impactite sands.Cataclastic diamictites, which superficially resemblesuevites but apparently lack significant melted com-ponents, are interpreted as proximal ejecta fallout de-posits. Sedimentary target-rock blocks are megaclaststhat apparently escaped impact-induced fluidization,whereas crystalline target-rock blocks were brought upfrom depth at the crater’s rebound peak.

References: [1] King D. T., Jr. (1997) Ala. Geol.Soc. Guidebk., 34c, 25-56. [2] King D. T., Jr., andNeathery T. L. (1998) Amer. Assoc. Petrol. Geol. Ann.Conv. Abst., 7, 358a-f. [3] Neathery T. L. et al.(1976) Geol. Soc. Amer. Bull., 87, 567-573. [4] KingD. T., Jr. et al. (1998) Geol. Soc. Amer. Ann. Mtg.Abst., 30(7), Add. 4.

Acknowledgments: Drilling at Wetumpka was anin-kind gift to Auburn University by Vulcan MaterialsCompany of Birmingham, Alabama.

FIELD OBSERVATION OF A POSSIBLE IMPACT STRUCTURE (TSENKHER STRUCTURE)IN SOUTHERN MONGOLIA. G. Komatsu1,3, J. W. Olsen2, and V. R. Baker3, 1International Re-search School of Planetary Sciences, Universita’ d’Annunzio, Viale Pindaro 42, 65127 Pescara, Italy(goro@sci.unich.it); 2Department of Anthropology, University of Arizona, Tucson, AZ 85721,U.S.A.; 3Lunar and Planetary Laboratory, University of Arizona, AZ 85721, U.S.A.

Introduction: Compared with NorthAmerica and Europe, little is known about im-pact structures in Asia despite its areal extentand a large span of geological ages. In Mongo-lia, only one impact has been identified to date[1]. Here we report results of our 1998 expedi-tion to a hypothesized impact crater namedTsenkher Structure (43°37’N, 98°22’E, GPScoordinates 43°38'41.1"N, 98°22' 08.5"E) [2]in southern Mongolia. We visited the site as apart of the Joint Mongolian-Russian-AmericanArchaeological Expedition conducted in Mayand June of 1998. Geological setting: The Tsenkher Struc-ture was identified in the transition zone be-tween the Gobi Desert and the Altai MountainRange. This area is characterized by ranges andbasins. The ranges originated as Paleozoic islandarc assemblages [3] and were later subjected totranspressional deformation during late Ceno-zoic [4]. The area also experienced Carbonifer-ous and Permian intrusive events. The basinfills are Mesozoic, mostly Cretaceous in age, butthin Quaternary alluvial materials overlie theMesozoic sequence. The Tsenkher Structure islocated in the middle of an east-west trendingbasin about 10-20 km wide. The basin is tiltedslightly toward the south. The Tsenkher Struc-ture must be younger than Cretaceous, and itspreservation indicates an age as young as Qua-ternary. Structure: The Tsenkher Structure is rela-tively well-preserved with a raised rim outliningits circular shape (Figure 1). The rim to rim dis-tance is about 3.6 km (previously reported 3.2km [2]). The rim lacks its northwestern seg-ment, and it is fluvially dissected radially. Thearea inside the rim is filled with fluvial sedi-ments, which were deposited from north tosouth. There is a major spillway connecting theinner part of the structure with a radar-darkfan-shaped area outside of the structure to thesouth. The radar-dark area is actually a channelincised in the gravel-covered valley floor. Thechannel floor is covered with fine-grained flu-vial sediments making the area dark in the radar

images. The maximum rim height may reach 50meters. The rim surfaces are covered with rockfragments ranging in size from 1 to 50 centime-ters, with occasional boulders exceeding onemeter in size (Figure 2). These rock fragmentscause the radar bright signature of the rim (Fig-ure 1). Because of the heavily-fractured natureof the rim and the desert varnish apparent onrock surfaces, it is very difficult to observe anystratigraphic sequence. On the eastern side ofthe structure, we observed a raised outer rim(Figure 3), about 1 to 2 km outside the innerrim. This outer rim is lower than the inner rimin height. The area between the inner and outerrims is covered with desert pavement, and thisarea gently dips down and outward. We are in the process of analyzing samplesfrom Tsenkher in order to confirm or reject theimpact hypothesis. If this structure is an impactcrater, it would be the second one identified inMongolia. The outer rim is possibly a rampart,which is commonly observed with Martian cra-ters. The rampart implies that the impact oc-curred in a ground water layer or permafrostzone. Late Pleistocene and early Holocene stoneartifacts were found around the margins of acurrently desiccated playa a few kilometerssouth of the Tsenkher Structure. These arti-facts, fashioned by direct percussion and, insome cases, subsequent pressure flaking, aremade on a variety of raw materials includingjasper, chert, and an as yet unidentified fine-grained metamorphic rock. We are investigat-ing a link between the structure, particularly thepossible ponding inside the rim, and human oc-cupation in the area.

References:[1] McHone, J.F., and Dietz, R.S. (1976) Me-teoritics, 11, 332-334 [2] Komatsu, G. et al.(1998) LPSC [3] Sengor, A.M.C., and Natal’in,B.A. (1996) In The tectonic evolution of Asia,Cambridge Univ. Press [4] Cunningham, W.D.et al. (1997) Tectonophysics, 277, 285-306

TSENKHER STRUCTURE: G. Komatsu, J.W. Olsen and V. R. Baker

Figure 1. RADARSAT imagery of the possibleimpact structure in southern Mongolia. Fig. 2and Fig. 3 were taken from point A and B tothe west respectively. The image was providedby Canadian Space Agency/Agence Spatiale Ca-nadienne (1997)

Figure 2. Inside the inner rim.

Figure 3. Outer rim.

THE VELINGARA RING STRUCTURE IN HAUTE CASAMANCE, SENEGAL: A POSSIBLE LARGEBURIED METEORITE IMPACT CRATER. S.Master1*, D.P.Diallo2,3, S.Kande4 & S.Wade2,5. 1Dept. of Geology,Univ. of the Witwatersrand, P. Bag 3, WITS 2050, Johannesburg, South Africa, 065sha@cosmos.wits.ac.za. 2Institutdes Sciences de la Terre, Univ. Cheikh Anta Diop, Dakar-Fann, Senegal. 3dpdiallo@ucad.refer.sn. 4MIFERSO, BP6082, Dakar, Senegal. 5wsouleye@ucad.sn. *Corresponding Author.

A 48 km-diameter multi-ring feature, the Velingarastructure, has been discovered in Landsat and NOAA-AVHRR images of southern Senegal (centred on 14°7’40”W, 13°02’13.2” N)[1]. The structure is developed in Mid-Eocene marine sediments of the coastal Senegal Basin, andhas been buried by up to 90 m of post-Eoceneunfossiliferous continental sediments. Drilling andresistivity surveys have revealed the presence ofsubcropping Neoproterozoic or Palaeozoic metamorphicbasement rocks in the centre of the structure. There is also apositive Bouguer gravity anomaly associated with thestructure. We interpret these features as representing apossible shallowly-buried complex impact crater with acentral uplift, which may be linked with other impactevents of the Late Eocene, or the terminal Eocene massextinction.

The feature, which appears in published Landsat MSS& TM images (Ref. 2, p. 370; Ref. 3, p. 100), as well as inNOAA-AVHRR images, is a circular multiple ringstructure with an overall diameter of 48 km. It is centred onan area about 12 km SSW of the town of Velingara inHaute Casamance Province, southern Senegal. Thenorthern rim of the structure just touches the border withthe Gambia, while the southern rim is about 20 km N of theborder with Guinea-Bissau. The region is one of relativelyflat topography and low elevations (<70 m above sealevel), and is characterised by thick ferruginous lateriteswith no outcrops, and an open tree savannah vegetation.

Photogeologically, the Velingara structure has a darkcore, c. 20 km in diameter, which is surrounded by a light-coloured ring (defined by land use patterns), which is about7 km wide. This is followed by an outer dark ring, and onthe perimeter, a discontinuous light ring, from which anumber of streams emerge in a radial centrifugal pattern.Four of the most prominent of these streams, to the W ofthe structure, form the source of the Casamance River.Other streams issuing to the NW, N and E flow into theGambia River or its tributaries. Streams from the SE, S andSW parts of the outer ring drain into the Kayanga River,which flows SW into the Geba estuary of Guinea-Bissau.There is a superficial similarity in terms of size andappearance to the Araguinha ring structure in Brazil [2], inwhich a basement core is exposed. The central part of ofthe Velingara structure is a broad topographic depression,the Anambé basin, in which centripetal drainageconcentrates in a central swampy area at 22m elevation,which has an outlet in the SE-flowing Anambé river, atributary of the Kayanga. The soils of the central Anambébasin are hydromorphic sandy clay loams with a heaviersubsurface horizon that thickens towards the centre of thedepression, where the whole profile may have a clayeytexture [4]. These wet soils, which are regularly inundatedduring the rainy season as well as part of the dry season,are responsible for the dark colour of the core, which isaccentuated in Landsat Band 5. The light and dark rings

surrounding the Anambé depression define a broad flattorus, with elevations between 40 and 70 m above sea level,which forms a watershed between the centripetal drainageof the Anambé basin and the centrifugal drainage arisingfrom the outer light ring. The inner light ring is producedby deforestation of the tree savannah related to farmingactivities in a series of small villages along main roadswhich follow the more elevated area surrounding thecentral depression.

The country rocks are Paleogene (Paleocene and Earlyto Middle Eocene) marine carbonate sediments of theSenegal Basin, which are overlain by an average of 40 m ofunfossiliferous continental sandstones and shales of post-Eocene (Neogene) age (locally called “Continentalterminal”) [5]. The Paleogene sediments overlieNeoproterozoic to Paleozoic rocks of the Mauritanide belt,and were deposited in a passive margin following riftingand the opening of the Atlantic Ocean [6]. Themetamorphic Mauritanide basement rocks are exposed inthe core of the Velingara structure, which although it hasthe form of a structural basin (the Anambé depression), isclearly a zone of basement uplift.

The central basement uplift was discovered duringdrilling for an agricultural project in the Anambé basin in1979 [4], as well during a major phosphate explorationprogram in 1980-1982, when about 20 vertical drillholes,up to 90 m in depth, were drilled in the region of theVelingara structure [7]. The drilling showed that thebasement uplift, which consists of biotite and muscoviteschists with abundant quartz veins, as well as talc schistsand quartzites, is present under 3m of sand cover in the S ofthe Anambé depression, while 4 km N it is found at a depthof 114 m, and is at a depth greater than 60m 10 km E andW [7].

During our initial field investigation, from 20-23December 1998, we found no outcrops except thickferruginous laterites developed over the Neogene andQuaternary continental sand cover that blankets the region.The subcropping basement rocks were located in a trenchdug for an irrigation canal near Soutouré. The abundantquartz veins, which cut through quartz-muscovite schists,have a shattered appearance, with a “gries” texturecharacterised by the presence of several closely-spaced (2-3mm) fracture sets, yielding lozenge-shaped fragments ofquartz. A preliminary petrographic examination of 3basement rocks from this locality has shown that the quartzveins are strongly deformed and have undulose extinction,as well as 2 intersecting sets of non-planar deformationtwin lamellae, which intersect at an angle of 70°. Agarnetiferous quartzite shows completely annealed granularrecrystallized textures. Quartz-muscovite schists havecrenulation cleavages. A sample of lateritic hematite-cemented sandstone from the post-Eocene cover showswell rounded quartz grains displaying triangular etch pitson their surfaces, with some grains having

Velingara Ring Structure in Haute Casamance, Sengal: S. Master et al.

crystallographically oriented rutile needles. No PDFs havebeen identified as yet in the basement rocks, and no Eocene“target” rocks have yet been seen or examined.

In one borehole close to the basement uplift, there is,within the post-Eocene continental sequence, a 24 m-thicksedimentary breccia consisting of angular fragments ofmica schists and quartzites, in a medium-grained sandymatrix with a clayey cement. Surrounding the basementuplift, there is a trough in which the post-Eocene sedimentsare more than 68 to 90 m in thickness, compared to anaverage thickness of 30-50 m within the region [7].

A geophysical resistivity survey of the Anambé basinin 1979 indicated the presence of an arcuate belt of highresistivity (>200 Ohm.m) basement rocks underlying thesouthern part of the depression [8]. A regional gravitysurvey [9] shows that there is an oval-shaped Bouguergravity high of +30 mgal over the N part of the Velingarastructure, which rises above a regional plateau of c. 20mgal in southern Senegal and the Gambia.

Because of the regular concentric arrangement of lightand dark bands in the Landsat image, which reflectschanges in soil, vegetation and hydrology, together with theannular elevated ring surrounding a central topographicdepression, in which there is evidence of a major uplift ofmetamorphic basement, it is postulated that the Velingararing structure is due to a complex impact crater, with apeak-ring-type central uplift. The presence of ametamorphic basement core uplift in passive marginsediments rules out an origin of the ring structure bymechanisms such as salt diapirism, igneous intrusion,interference folding or karst dissolution. Evidence fromabout 20 boreholes indicates that the basement uplift, aswell as the Mid-Eocene and older sediments, are buried byNeogene continental sediments, which thicken in a troughsurrounding the central uplift. The age of the postulatedimpact crater is thus constrained to be Late Eocene orOligocene. There may be a link between this structure andthe multiple tektite layers and iridium anomalies of the LateEocene, which have been associated with the 35.35±1.5 MaChesapeake Bay structure [10] and the 35.7±0.2 MaPopigai structure (Russia) [11], or with the terminal Eocene(33.7±0.2 Ma) mass extinction [13].

The economic importance of the Velingara structure isevidenced by the major SODAGRI agricultural project [4]being carried out in the Anambé basin, which owes itspeculiar hydrology and waterlogged soils to the presence ofan impermeable basement close to surface. As a buriedpostulated impact crater, the Velingara structure has thepotential for hosting hydrocarbons (oil or gas), as is thecase with several buried impact structures in NorthAmerica [14, 15]. Further field and petrographic studies(including the search for microscopic shock deformationfeatures) are in progress. Attempts are being made to locatethe original drillcores of boreholes drilled in the Velingarastructure in 1979-1982, and to sample Eocene sedimentaryrocks which would have formed the target rocks during thepostulated impact event.

Acknowledgements: We thank Prof. Abdoulaye Dia(Director IST, Dakar) and El Hadji Abdoul Aziz Fall(SODAGRI, Soutouré) for their logistical and practicalsupport, and Aminata Kande and Boudy Mballo(Velingara) for their hospitality during the fieldinvestigation.

References: [1] Master, S. & Woldai, T. (1998). Abstr.,AfricaGIS Conference, Abidjan, Côte d’Ivoire, 4-10October 1998. [2] Short, N.M. et al. (1975). Mission toEarth: Landsat Views the World. NASA, Washington,D.C., 459 pp. [3] Stancioff, A. et al. (1986). Mapping andRemote Sensing of the Resources of the Republic ofSenegal. The Remote Sensing Institute, South Dakota StateUniv., Brookings, S.D., 655 pp. [4] Electrowatt S.A.(1980). Aménagement du bassin de l’Anambé, Vol. 2,Hydrogéologie. SODAGRI, Dakar, 75 pp. [5] Monciardini,C. (1966). La sédimentation éocène au Sénégal. Mém.BRGM No. 43, Paris, 65 pp. [6] Van der Linden, W.J.M.(1981). Geol. Mijnbouw, 60, 257-266. [7] Pascal,M.(1983). Recherche de phosphates au Sénégal, SecondeCampagne (1981-1982). Rapp. BRGM 83DAK003, Dakar,42 pp. [8] Ogier, M. (1979). Etude géophysique parsondages éléctriques dans la région de Kounkane (Dépt. deVélingara). Rapp. BRGM 79 GPH021, Paris, 11 pp. [9]Roussel, J. et al. (1984). Tectonophysics, 109, 41-59. [10]Koeberl, C. et al. (1996). Science, 271, 1263-1266. [11]Grieve, R.A.F. (1996). Meteoritics, 31(2), 166-167. [13]Stoeffler, D & Claeys, P. (1997). Nature, 388, 331-332.[14] Donofrio, R.R. (1997). Oklahoma Geol. Surv. Circ.100, 17-29. [15] Dumestre, M.A. (1985). Oil & Gas J.,83(43), 146-148, 151-152.

PRELIMINARY REPORT OF SZILVÁGY-PATKÓ (HORSESHOE): A NEW (POSSIBLE) IMPACT CRATERREMNANT IN HUNGARY. A(t). Arday 1, Sz. Bérczi 2, Gy. Don 3, B. Lukács 4, 1Eötvös University, Dept. Appl. Engi-neering Geology, H-1088, Budapest, Múzeum krt. 4/a, Hungary, 2Eötvös University, Faculty of Sci., Cosmic Mat. Sp.Res. Gr. Dept. G. Technology, H-1117, Budapest, Pázmány Péter sétány 1/a, Hungary, (bercziszani@ludens.elte.hu),3Hungarian Geological Institute, H-1146, Budapest, Stefánia út 14. Hungary, 4Central Research Inst. Physics, RMKI,H-1525 Budapest, 114. P.O.Box. 49. Hungary,

ABSTRACT Photographing from his airplane, A(n).Arday recognized a very expressive central symmetricform in Zala County, West Hungary. It can be foundin the vicinity (3 kms from) Szilvágy, in SE directionfrom the village. Its diameter is 500 meter, depth fromthe top of the rim is cca. 25 m. Earlier drilling re-search data has shown that limonitic grains form alayer at shallow depths (between 0-5 meters) andshows a peak population at about 2 meters deep.10,000:1 mapping expressively shows the horseshoelike crater rim. The horseshoe like crater-remnant isopened toward the SSE direction from which we im-ply that the impacting body may arrive from this di-rection. Because of the Pannonian layers on the sur-face, the age of the possible crater remnant may be afew 10,000 years.INTRODUCTION In autumn of 1998 A(n). Arday,pilot at aerial photography survey in Hungary, duringhis work recognized a well developed and exclusivecircular form in Zala County, West Hungary. Thisregion is about 250 kilometers WSW from Budapest,in Transdanubia province, near to the Hungarian-Slovenian boundary. The place is in agricultural use,wine-yards are in the upper inner slopes of Szilvágy-Patkó (Horseshoe was its popular name). The geo-graphical coordinates are as follows: 46°43’18.5” N,16°38’17” E.

FIELD WORKS On 18th November we visited thesite. It was a nice early winter day with sunshinetherefore we could take good morphological observa-

tions. We could imagine, standing on the bottom ofthe crater, as if we were on the Moon, so expressivewas the central symmetry, looking northern direction.All this northern part of the depression (the slopesfacing southward) have been used for grape agricul-ture since the known times. From the aerial view itcould be recognized that boundaries of the small-holder's wine-yards were arranged along the radials ofthe circle, the fences also emphasized the centralsymmetric arrangement. We could see the same fromthe center of the horseshoe crater. The wine-yardsfrom the upper small rim plato came down toward thecentral depression till the inflexion point (which is theplace where the upper rim region's curvature radius -focus in deep - changes to the depression's curvatureradius up above the crater). Therefore the last indi-vidual grape blocks downslope in all yards showed acircular edge (we called it inflexion circle), lookingfrom the center of the depression. With a hand drillerwe took soil samples till 1.5 meter deep. On the sitewe could observe that some red clay grains can befound at about 1 meter depth.DATA FROM EARLIER DRILLING One of us(Gy.D.) worked earlier on the fields and could re-member that in 1986 they took drilling samples fromthe section NW rim of the Szilvágy structure. Lookingfor these data we could identify important layers asfollows:Between 0.0 - 4.8 m : after the 20 cm thick upper soila yellow-yellow-brown silt, its has micro-layeredstructure, and contains limonithic spherules in greatquantity (the largest is max. 1 cm, ferrous concre-tions). Their amount has maximum between 2.0-2.1meter. Its color, from 2.5 meter is more intensive yel-low, but does not contain the spherules. Between 4.8-4.9 meters brick-red microlayered clay can be found.Between 4.9-9.8 meters the brown-yellowish-brownsilt with fine grained sand continues downward. Be-tween 9.8-10.0 meters (the bottom) : Greenish-blue-gray microlayered clay. On the whole region the lay-ers are from the Upper Pannonian age, and these lay-ers are covered by thin Quaternary sediments, whichis missing in some patches. On the geophysical mapsno anomaly can be seen in the deep on this region.THE RECONSTRUCTION OF THEPROJECTILE. We know the sizes of the crater: ra-dius 250 m, depth cca. 30 m. To estimate the impac-tor, there is the thumb rule for terrestrial impacts

SZILVÁGY-PATKÓ: NEW, PROBABLE IMPACT FORM IN HUNGARY: A. Arday, Sz. Bérczi, Gy. Don, B. Lukács

when cca. 97 % of the material is ejecta; the case ofthe Ries crater where the ejected volume is 124-200km3 (Pohl & al., 1977) while the impactor is esti-mated to r=0.5-1 km (Hörz, 1982); and rough ener-getic considerations. All of them suggest cca. 30-100times of projectile mass to be ejected. Newton's esti-mate is that the projectile traverses cca. its own lengthin a target of similar density. Now we choose the fac-tor 30. Then we get a diameter/depth ratio cca. 6,roughly valid for the Arizona meteor crater and inaccordance with the empirical Baldwin rule in thesize range considered. Then cca. R=40 m, i.e. for astony body M~8*105 t, and the double for iron. Thisis an order of magnitude higher than the estimate forthe Tunguz impact (Menzel, 1975). So the Szilvágyevent must have been mechanically at least as devas-tating as the Tunguz one. The latter uprooted thewoods of dense forest at 50 km distance and buildingswere told to be damaged at 200 km.ON THE AGE OF THE IMPACT (HISTORICALRETROSPECTIVE). The Pannonian layers are notdisturbed, the event cannot have happened beforePliocene. The geological details are still being inves-tigated, but standard geologic maps of the Szilvágyarea suggest Quaternary.

A very strong constraint is that up to now noconjecture exists for myths, tales, &c. reflecting aTunguz-type event in the Carpathian Basin W. of theDanube. The general area is populated for at least350,000 y, but cultural continuity is expected sinceHomo Sapiens. The area was Neanderthal-roameduntil BC 35000. If later geologic surveys permit ear-lier event, then no problem arises.

In latter ages it would have been difficult notto observe something devastating at 50 km and verytroublesome at 200 km. It is worthwhile to note, how-ever, that the general Szilvágy area is devoid of defi-nite sites up to the XIIIth century, AD. There is nospecific explanation for this.

From Neolithic we know the connections ofthe peoples of the area. Changing uncalibrated data ofMakkay (1982) into calibrated ones following Suess(1970), (an educated guess before 5300 BC) we getthe following story of the Szilvágy area.

Before 6100 BC a Mesolithic hunter-gathererpeople. 6100-5900: connection with the Körös-Starcevo culture. Körös-Starcevo had strong connec-tion with the Neolithic population of Balkan, AsiaMinor and even of Mesopotamia (Renfrew, 1979),having later the variety of myths recorded. 5900-4800: the Körös-Starcevo culture, Central EuropeanCorded Ware. 4800-4400: the Lengyeli culture, withsouthern connections. 4400 BC: Early Copper Age. Inthe Copper Age the area belongs to the Baden cultureextending to Austria. The 2nd millennium BC is there

Bronze Age. First half: People of Pots with Lime In-set. Second half: People of Urnfields, also in Austria.1st millennium: Hallstadt Iron Age.

In the IIIrd c. BC Celts reach the territorywhich becomes a part of a linguistic community ex-tending to half of Europe. Henceforth news anystrange event can propagate without problems. BC 8:the Roman Empire annexes the area which so be-comes part of a worldwide administration. During thewhole period it is impossible for such an event re-maining unobserved by neighbors.

However there remains an interesting chancewhose possibility can be determined by later geologicsurvey. In 433 AD West Rome ceded Pannonia to theHuns. They evacuated the territory in 454, when Em-peror Avitus reoccupied but after some months histroops were utterly defeated by the incoming EasternGoths (Sági, 1978). In the following turmoil starts aflight of the remainder of literate Roman populationof the Western cities nearby Szilvágy. There is nomore Imperial administration. Now, on Sep. 7, 456,according to late Roman records and also detectedfrom findings, an earthquake demolishes Savaria, 55km from Szilvágy, maybe the nearest city still in ex-istence. In the vicinity of the epicenter such an event,of course, could erase the archaeologic sites.

The cause and details of that earthquake arestill obscure. The inner Carpathian Basin lack globaltectonic events and the few devastating earthquakesalways have local mechanisms. The distance is not toolarge for a Tunguz event. So, while data before 35000BC would be less problematic, if later investigationssuggested more recent times, Sep. 7, 456 AD wouldstill be at hand.REFERENCES Menzel D.H.: Astronomy. Chanti-cleer, NY. 1975; Pohl & al., in Impact and ExplosionCratering, ed. Roddy D.J., p. 342 Pergamon, NY1977; Hörz F.: Geol. Soc. Amer. Sel. Pap. 190, 39(1982); Makkay J.: A magyarországi neolítikum ku-tatásának új eredményai. Akadémiai, Budapest, 1982;Suess H.E. in Radiocarbon Variations and AbsoluteChronology, ed. by. I.U. Olsson, Wiley, p. 303 (1970);Renfrew C.: Before Civilization. Penguin, Harmond-sworth, 1979; Sági K.: in Régészeti barangolásokMagyarországon, p. 101, Panoráma, Budapest, 1978