Dynamic modeling suggests terrace zone asymmetry in the Chicxulub crater is caused by target heterogeneity

Earth and Planetary Science Letters 270 (2008) 221–230

Contents lists available at ScienceDirect

Earth and Planetary Science Letters

j ourna l homepage: www.e lsev ie r.com/ locate /eps l

Dynamic modeling suggests terrace zone asymmetry in the Chicxulub crater iscaused by target heterogeneity

Gareth S. Collins a,⁎, Joanna Morgan a, Penny Barton b, Gail L. Christeson c, Sean Gulick c, Jaime Urrutia d,Michael Warner a, Kai Wünnemann e

a Earth Science and Engineering, Imperial College London, UKb Earth Sciences, University of Cambridge, UKc Institute for Geophysics, Jackson School of Geosciences, Texas, USAd Institute of Geophysics, UNAM, Mexicoe Museum für Naturkunde, Humboldt-Universität Berlin, Germany

⁎ Corresponding author.E-mail address: [email protected] (G.S. Collins

0012-821X/$ – see front matter © 2008 Elsevier B.V. Aldoi:10.1016/j.epsl.2008.03.032

A B S T R A C T

Article history:

Editor: T. Spohn

Keywords:Chicxulubimpact angledynamic modelingcrater asymmetry

terrace zone asymmetry in the Chicxulub impact crater using dynamic models ofcrater formation. Marine seismic data acquired across the crater show that the geometry of the crater'sterrace zone, a series of sedimentary megablocks that slumped into the crater from the crater rim, variessignificantly around the offshore half of the crater. The seismic data also reveal that, at the time of impact,both the water depth and sediment thickness varied with azimuth around the impact site. To test whetherthe observed heterogeneity in the pre-impact target might have affected terrace zone geometry weconstructed two end-member models of upper-target structure at Chicxulub, based on the seismic data atdifferent azimuths. One model, representing the northwest sector, had no water layer and a 3-km thicksediment layer; the other model, representing the northeast sector, had a 2-km water layer above a 4-kmsediment layer. Numerical models of vertical impacts into these two targets produced final craters that differsubstantially in terrace zone geometry, suggesting that the initial water depth and sediment thicknessvariations affected the structure of the terrace zone at Chicxulub. Moreover, the differences in terrace zonegeometry between the two numerical models are consistent with the observed differences in the geometryof the terrace zone at different azimuths around the Chicxulub crater. We conclude that asymmetry in thepre-impact target rocks at Chicxulub is likely to be the primary cause of asymmetry in the terrace zone.

© 2008 Elsevier B.V. All rights reserved.

1. Introduction

Meteorite impact is a fundamental geologic process that affects allplanetary bodies. In terms of energy deposition, mass andmomentumtransfer, and environmental consequences, the largest yet mostinfrequent impacts are by far the most important. Our understandingof impacts has advanced through observational studies of impactcraters on Earth (e.g. Grieve et al., 1977) and other planets (e.g. Pike,1974); through small-scale impact experiments in the laboratory (e.g.Gault et al., 1968), and high-energy explosions (e.g. Roddy, 1977); andthrough numerical modelling (e.g. Pierazzo and Collins, 2003).However, little is known about large impact crater formation becausethere are few pristine large craters on Earth, and it is difficult toextrapolate from laboratory experiments to planetary-scale impacts.The Earth has only three large (N150 km diameter) impact craters:Chicxulub, Vredefort and Sudbury (Grieve and Therriault, 2000). Of

).

l rights reserved.

these three structures, Chicxulub is the best preserved, and henceprovides the greatest insight into large impact crater formation.

The Chicxulub impact structure is buried beneath several hundredmeters of sediments, and is located half offshore and half onshore onthe Yucatán peninsula, Mexico (Hildebrand et al., 1991). Seismicreflection, refraction and gravity data were acquired across the offshorehalf of the crater in 1996 and 2005 (Fig.1). The 1996 seismic data consistof 4 reflection profiles and a refraction dataset. These data revealedclear images of the target rocks and impact basin (Morgan et al., 1997)and strong changes in velocity across the crater (Morgan et al., 2000;Christeson et al., 2001). In 2005 we acquired a larger dataset (Fig. 1),that included a high-resolution grid of reflection profiles within theoffshore impact basin, as well as 3 new radial profiles, one concentricprofile, and refraction data (Morgan et al., 2005). These 2005 data haveprovided us with a more detailed picture of both the crater structureand initial target stratigraphy (Gulick et al., in press).

The geophysical data at Chicxulub show that the crater is notcylindrically symmetric about an axis through the crater center interms of its gravity and magnetic anomalies (Schultz and D'Hondt,1996; Hildebrand et al., 1998b) or its seismic reflection signature

Fig. 1. Experimental geometry of Chicxulub seismic surveys. Solid black lines are marine reflection profiles acquired in 1996 and 2005. Color is depth to the base Tertiary.

222 G.S. Collins et al. / Earth and Planetary Science Letters 270 (2008) 221–230

(Gulick et al., in press). Aspects of crater asymmetry may be due toasymmetry in the pre-impact target (Gulick et al., in press) or obliqueimpact (Schultz and D'Hondt, 1996; Hildebrand et al., 1998a).According to this latter idea, the growth and collapse of the cratervaries depending on azimuth relative to the direction of impact and,hence, crater asymmetry is diagnostic of impact direction. Moreover,the amount of asymmetry might depend on how oblique the impactwas (i.e. the departure of the impact angle from the vertical). In thiscase, the crater asymmetry may also be diagnostic of impact angle.

Knowledge of impact direction and impact angle is important,particularly in the case of Chicxulub. Regional environmentalconsequences of impacts are much more severe in the downrangedirection, and there is evidence that impact vapor production – likelyone of the major contributing factors to the K–T mass extinction – is afew to many times more efficient in oblique impacts than in verticalimpacts (Schultz, 1996; Schultz and D'Hondt, 1996; Pierazzo andMelosh, 1999). Currently, the only unequivocal method for ascertain-ing impact direction and angle for a given crater is by analyzing thegeometry of proximal continuous ejecta deposits (Gault and Wede-kind,1978). Unfortunately, ejecta deposits are often buried, re-workedor eroded at terrestrial crater localities. Hence, if crater asymmetry canbe used to infer impact angle and/or direction, understanding howwould constitute a major advance in impact cratering.

Schultz and D'Hondt (1996) and Hildebrand et al. (1998a)interpreted Chicxulub crater asymmetry to be related to impactobliquity. Schultz and D'Hondt (1996) postulated that the gravity highin the northwestern section of the crater represented an offset of thecentral uplift in that direction. In an alternative interpretation,Hildebrand et al. (1998a) modeled the central gravity high, which isoffset to the southwest of the crater center, as representing offset ofthe central uplift in the SW direction. In both studies, central upliftoffset is interpreted as being produced by an oblique impact, althoughSchultz and D'Hondt (1996) proposed that the offset would occurdownrange of the impact direction, while Hildebrand et al. (1998a)suggested it would be uprange. However, statistical studies ofextraterrestrial craters suggest no connection between the offset ofthe central uplift from the crater center and impact direction or angle(Ekholm and Melosh, 2001; Herrick and Forsberg-Taylor, 2003;McDonald et al., in press).

In this paper we investigate whether pre-impact target hetero-geneity, instead of obliquity of impact, could have produced aspects ofthe observed crater asymmetry at Chicxulub, using dynamicmodels ofcrater formation. Specifically, we test the hypothesis that hetero-geneity in the target rocks might be the cause of terrace zoneasymmetry at Chicxulub, inspired by seismic data interpretation(Gulick et al., in press). The gradual rise in computing power, as well asrecent improvements to dynamic modeling codes, means that largevertical impacts can now be modeled at sufficient resolution insensible computing times using two-dimensional models (e.g. Ivanov,2005). By combining and comparing such models with the extensivegeophysical data at Chicxulub we provide insight into large craterformation.

2. Geophysical observations

Surrounding the central zone of uplifted rocks in large impactcraters is a “terrace-” or “megablock-zone” (Fig. 2), formed from near-surface target rocks that have been faulted and down-dropped duringcrater formation (Melosh, 1989). The head scarp of the terrace zoneforms the crater rim (Morgan and Warner, 1999; Turtle et al., 2005),and target rocks outside the crater rim area, although often faulted, lieclose to their original stratigraphic position. At Chicxulub, the terracezone is best observed on the new seismic data at radial distances ofbetween 40 and 77 km (Gulick et al., in press). In this section weinvestigate the nature of the pre-impact target using seismic dataacquired outside the crater's terrace zone, and examine the crater'sterrace zone along two radial profiles along which we observe twoextremes in terrace zone geometry.

2.1. Pre-impact target

Fig. 1 shows the depth to the base Tertiary, mapped using thereflection data, and converted to depth using a 3D tomographicvelocity model (Gulick et al., in press). The resolution of this depthmeasurement is dependent on both the choice of the base Tertiaryreflector and the velocity model, which is difficult to quantify, but isprobably accurate, on average, to around ±200 m. Even allowing forsome error, it is clear that there are significant variations in depth to

Fig. 2. Simplified sketch of the structure of the Chicxulub crater, redrawn from Vermeesch and Morgan (in press). During crater formation, material close to the impact site isexcavated and melted, and rocks rise up to form a central zone of structural uplift covered by allocthonous impact melt rocks and impact breccias. Outside the excavation zone, thenear-surface target rocks collapse inwards and downwards to form a terrace zone.

223G.S. Collins et al. / Earth and Planetary Science Letters 270 (2008) 221–230

the base Tertiary around the crater. Fig. 3 shows the concentricreflection profile that tracks around the crater at ~85 km radialdistance from the crater center (see Fig. 1 for location). This profile isoutside the crater's terrace zone, and thus provides us with data onpre-impact target structure at different azimuths around the crater.The depth to the base Tertiary increases from a few hundred meters inthe northwest direction to ~2.5 km in the northeast. Onshore theTertiary is between 250 and 450 m thick in boreholes outside theterrace zone (Ward et al., 1995; Urrutia-Fucugauchi et al., 1996). Thus,the current base of the Tertiary rocks in the north and northeastdirection is up to 2 km deeper than elsewhere. Although some post-impact modification by subsidence and erosion is expected, particu-larly in the deepest part of the basin, it does appear that there was apre-existing deep basin in the north and northeast quadrant at thetime of impact (Gulick et al., in press). It is likely that this basin is thecause of the broad gravity low over the northeast sector of the crater.

Fig. 3 also illustrates the variation in depth around the crater ofanother package of bright reflectors (labeled Lower K), which weinterpret as originating from a Lower Cretaceous anhydrite/carbonatesequence in accordance with the interpretation by the Mexicannational oil company, Petróleos Mexicanos (Camargo-Zanoguera andSuárez-Reynoso,1994). If this interpretation is correct, these reflectorsprovide an estimate of the variation in the thickness of the pre-impactCretaceous sediment sequence with azimuth around the crater. Fig. 3shows that Lower Cretaceous reflectors deepen in correlationwith thedeepening of the base Tertiary, and that the thickness of theCretaceous sediments varies from a minimum of ~3 km in thenorthwest, to 4–4.5 km in the northeast. Data from onshoreexploration wells outside the crater rim indicate that the Cretaceoussediments were thinner onshore. The Cretaceous section thickenstowards the crater from E–W, from ~1.6 km in Y4 to N2 km in Y5A, and~2.4 km in Y1 and Y2 (Hildebrand et al., 1991; Ward et al., 1995;Urrutia-Fucugauchi et al., 1996).

Fig. 3. Concentric reflection profile (see Fig. 1 for location) acquired at a constant radial distanthe crater to N2 km in the northeast, while the Cretaceous sequence deepens, and thickens

2.2. Crater structure

Seismic reflection data map the position of the Cretaceous targetrocks, which are relatively flat outside the crater, and are down-thrown and moved inwards during crater formation to form a terracezone (Figs. 2 and 4). The start of the terrace zone (head scarp) occurs atdifferent radial distances (between 58 and 77 km) on differentprofiles, whereas the innermost observable terrace zone is consis-tently around 40–43 km from the crater center (Fig. 4). Fig. 4 showstwo radial reflection profiles: a) Chicx-B and b) Chicx-C. These profilesrun though the center of the gravity high and low in the northwestand northeast quadrant of the crater, respectively (see Fig. 1), andillustrate two extremes in offshore terrace geometry. There are severaldistinct differences in the terrace zone on the two profiles in Fig. 4. OnChicx-B the terrace zone starts much further out from the cratercenter than on Chicx-C. Also, on Chicx-B the terrace zone displays thelargest total offset (~6 km) over the widest distance (25–35 km), andthe innermost visible Lower Cretaceous target rocks are the deepestobserved at ~9 km beneath the base Tertiary sediments. The terracezone on Chicx-C displays the smallest total offset (~2 km) over theshortest distance (~15 km), and the innermost visible lower Cretac-eous target rocks are relatively shallow at ~6.5 km below the baseTertiary. On the other five profiles that cross the terrace zone in radialor sub-radial directions (see Fig. 1), the geometry of the terrace zone isbetween these two extremes.

In summary, the geometry of the terrace zone varies around theoffshore half of the crater, and the variations in the terrace zoneappear to correlate with changes in the initial target stratigraphy. Thewater depth and sediment thickness were greatest in the northeastand least in the northwest—the directions in which the terrace zonesare most different. There are no reflection data onshore, hence thegeometry of the terrace zone is not known. Onshore, there is lessradial variation in the gravity field, and this may suggest a more

ce of ~85 km. The base Tertiary deepens from a few hundred meters to the northwest offrom 3 to 4–4.5 km.

Fig. 4. Two radial reflection profiles: a) Chicx-B and b) Chicx-C. These profiles are located in northwest and northeast quadrants of the crater (see Fig. 1 for location) and illustrate thetwo extremes in terrace geometry. The terrace zone along Chicx-B (a) displays the largest offset (~6 km) over the widest distance (25–35 km), and the innermost visible LowerCretaceous (Lower K) target rocks lie ~9 km beneath the middle of the topographic peak ring. The terrace zone along Chicx-C (b) displays the smallest offset (~2 km) over the shortestdistance (~15 km), and the innermost visible Lower Cretaceous target rocks are only 6.5 km below the outermost edge of the topographic peak ring.

224 G.S. Collins et al. / Earth and Planetary Science Letters 270 (2008) 221–230

regular terrace zone geometry onshore than offshore. In the followingsection we present numerical models to test whether asymmetry inthe pre-impact target structure at Chicxulub could be responsible forthe observed terrace zone asymmetry in the offshore crater.

3. Numerical modeling

3.1. Rationale

Formation of the Chicxulub impact crater has been the subject ofmany recent two-dimensional (2D) numerical modeling studies(O'Keefe and Ahrens, 1999; Collins et al., 2002; Ivanov and Artemieva,2002; Ivanov, 2005). In particular, the most recent 2D model Ivanov(2005) showed very good agreement with geophysical modelscompiled by Morgan et al. (1997, 2000). To test rigorously whetherlateral variations present in the pre-impact target at Chicxulub couldcause the observed crater asymmetry requires a three-dimensional(3D) numerical model. However, despite the constant improvement incomputer power, 3D calculations at a resolution comparable withcurrent 2D models are not yet possible. An alternative modelingstrategy, which allows the use of 2D models, is to approximate thelateral variations in target structure using a number of differenthorizontal layered targets, with layer thicknesses that represent end-member target structures. The results of separate vertical impactmodels into these end-member targets may then be compared toexplore the range of possible effects of target variations. Following thisrationale, and using the model of Ivanov (2005) as a starting point, wepresent two numerical models with different initial target structuresthat reflect theminimum andmaximum likely water and sedimentary

thicknesses at Chicxulub, which the seismic data suggest occur in thenorthwest and northeast of the crater, respectively.

The extent to which the approach of comparing two end-membervertical impact models provides meaningful insight into craterasymmetry depends on how much of an effect the evolution of thecrater in one direction has on crater development in other directions.In some circumstances this approach is not appropriate; for example,where the impact trajectory is at a moderately or highly-oblique angleto the target surface, or where large, short-wavelength azimuthalvariations, or radial variations in target structure are present. In suchcases, 3D numerical modelling is the only recourse to quantify theeffect of variations in target structure on crater formation. Here, weassume that the Chicxulub impact was vertical, to investigate theeffect of pre-impact target asymmetry on crater asymmetry indepen-dently. Moreover, at Chicxulub, the major asymmetry in pre-impacttarget structure is large-scale; the sediment thickness and waterdepth increase over a range in azimuth of ~30 degrees (in the NNW),and away from this direction, in the WNW and NNE sectors, there islittle variation in pre-impact sediment thickness and water depth(Fig. 3). Our approach does not provide reliable insight into the effectof lateral target variations on all aspects of Chicxulub crater formation;however, it should provide a robust estimate of themaximumpossibleeffect of lateral target variation on terrace zone geometry at Chicxulub.

3.2. Method

For this study we used the iSALE hydrocode (Collins andWünnemann, 2005; Wünnemann et al., 2006), a multi-rheology,multi-material extension to SALE (Amsden et al., 1980). iSALE is very

Table 1Numerical model parameters

Symbol Definition Value

L Impactor diameter (km) 14vi Impact velocity (km/s) 12ρi Impactor density (kg/m3) 2680Tdec Decay time of acoustic vibrations (s) 160υlim Kinematic viscosity of acoustically fluidized region

(m2/s)280,000

Crust Sedimentsρ Reference density (kg/m3) 2680 2700Y0 Cohesion (Yield strength at zero pressure; MPa) 50 5Ym von Mises plastic limit (theoretical yield strength at

infinite pressure; GPa)2.5 0.5

μi Coefficient of internal friction 1.5 1.0μd Coefficient of friction (damaged material) 0.6 0.4Tm Melt temperature (°K) 1500 1500ξ Thermal softening parameter 1.2 1.2pbd Brittle–ductile transition pressure (GPa) 2.59 0.76pbp Brittle–plastic transition pressure (GPa) 3.41 0.92

225G.S. Collins et al. / Earth and Planetary Science Letters 270 (2008) 221–230

similar to SALEB developed by Boris Ivanov (Ivanov et al., 1997; Ivanovand Artemieva, 2002; Ivanov, 2005). The code is well tested againstlaboratory experiments at low and high strain-rates (Wünnemannet al., 2006) and other hydrocodes (Pierazzo et al., 2007). It has alsobeen used in the previous numerical simulation of several terrestrialimpacts (Chicxulub, Collins et al., 2005; Chesapeake Bay, Collins andWünnemann, 2005; Ries, Wünnemann et al., 2005; Haughton, Collinsand Wünnemann, 2006; Sierra Madera, Goldin et al., 2006). Thethermodynamic behavior of each material in the model is describedby an equation of state (EoS). We used tables generated using theAnalytic EoS (ANEOS, Thompson and Lauson, 1972) for dunite torepresent the mantle, granite to represent the crust, calcite torepresent the sedimentary sequence, and water to represent the sea.

The most important aspect of impact models for properly simulatingcrater formation is the constitutive model. iSALE uses a constitutivemodel that accounts for changes in material shear strength that resultfrom changes in pressure, temperature, and both shear and tensiledamage (Melosh et al., 1992; Ivanov et al., 1997; Collins et al., 2004). Forlarge impact crater formation this must be supplemented by some formof transient target weakening model that facilitates deep-seatedgravitational collapse of the initial bowl-shaped cavity (Melosh, 1989;Melosh and Ivanov, 1999). The physical explanation for this apparenttargetweakening is still amatterof debate; inourmodels, and themodelsof Ivanov (2005), the assumed explanation is acoustic fluidization(Melosh, 1979). The effects of acoustic fluidization are incorporated intoourmodel using the “block-model” (Melosh and Ivanov,1999; IvanovandArtemieva, 2002; Wünnemann and Ivanov, 2003). Our choice of block-model and other input parameters was based on previous successfulmodels of the Chicxulub impact (Collins et al., 2005; Ivanov, 2005). All theimportant model parameters used in the simulations presented here areincluded inTable1; the interested reader is referred toCollins et al. (2004)for more detailed parameter definitions.

4. Results

Fig. 5 shows the development of the crater during twomodels withdifferent initial target structures. The image sequence on the leftillustrates results from the NW Model, representing the approximatetarget structure in the northwest; nowater layer was included and thesediment thickness was 3 km. The image sequence on the rightillustrates results from the NE Model, representing the approximatetarget structure in the northeast; the water layer was 2-km deep andthe sediment thickness was 4 km. In addition to illustrating theposition of the different material layers (blue=water; brown=sedi-ments; gray=crust), Fig. 5 also shows the deformation of a grid of

Lagrangian tracer particles with an original separation of 2 km (10computational cells).

Below the sediment layer in both the NE and NW models was agranite half-space. A different mantle material was not considered inthese simulations because iSALE can model a maximum of threedifferent materials plus vacuum (i.e. crust/sediment/water). In themodel of Ivanov (2005) the Chicxulub target was approximated usingthree horizontal layers representing sediments, crust and mantle. Weperformed test models using mantle/crust/sediment and comparedthe results with two-layer crust/sediment models. An example 3-layer(mantle/crust/sediment) model is shown in Fig. 5f. The 3-layer modelssuccessfully reproduced the results of Ivanov (2005) and demonstratethat inclusion of mantle beneath the crust does not substantially affectthe model results away from the crust/mantle boundary. Thus,although the deformation at depth may not be accurately reproducedby our models that omit the mantle, the formation of the terrace zone,which is the focus of this study, is well modeled by a simple sediment-over-crust target.

Our model results illustrate that differences in near-surface targetstructure at Chicxulub do not substantially alter crater formation. Theprogression of crater development in both the NE model and the NWmodel is very similar to that observed in earlier models (O'Keefe andAhrens,1999; Collins et al., 2002; Ivanov and Artemieva, 2002). A deepbowl-shaped cavity is formed in the crust, excavating all sediments,water, and some upper crustal material within a radius of ~40–45 km(Fig. 5b). This cavity collapses through uplift of the crater floor, to forma large central uplift (Fig. 5c). The central uplift then collapsesdownward and outward as the crater rim collapses inward anddownward—the collision and interaction of these two flow regimesgives rise to the topographic peak ring (Figs. 5d and e). Importantly,the evolution of the central zone of the crater is very similar in bothmodels as can be seen by comparing deformation near the cratercenter in the NW and NE models. (Note that, in both models,deformation in the very center of the crater is unreliable as it isaffected by numerical artifacts along the symmetry axis). This suggeststhat, in a vertical Chicxulub impact, crater growth and collapse in oneazimuthal direction should not have a strong influence on craterdevelopment in another direction. Hence, we expect that ourapproach of comparing two end-member models of water andsediment thickness does provide reliable insight into the effect oflateral target variations at Chicxulub on crater formation in a verticalimpact.

The two models shown in Fig. 5 do result in important differencesin the late-stage behavior of the near-surface target material, and theevolution of the zone of inwardly collapsed sediments—equivalent tothe terrace zone imaged by the seismic data. The fact that the base ofthe sediment layer is 3-km deeper, initially, in the NE model than inthe NWmodel implies that its final position after collapsewill be quitedifferent to the final position of the base of the sediment layer in theNW model; in fact, its final position is similar to the final position ofthe line of tracer particles at the equivalent depth (but in the crust) inthe NW model. This further illustrates that the difference in sedimentthickness between the NW and NE models does not affect cratercollapse strongly; rather, the principal effect of increased sedimentthickness and a water layer is to change the stratigraphic position ofthe sediment/crust interface in the final crater. Also evident from Fig. 5is that the crater is smaller in the NEmodel than in the NWmodel. Theinnermost position of the base of the sediment layer is closer to thecrater center in the NE model than in the NW model. Moreover, theradius at which the base of the sediments starts to drop down towardsthe crater center is substantially smaller in the NE model than in theNW model.

The presence of a 2-km deep water layer in the NE model,compared with no water layer in the NW model, also affects the finalcrater morphology and structure. In the NW model, sediment andcrustal material is ejected from the crater and lands close to the crater

Fig. 5. Results from the numerical modeling. Gray is crystalline basement, brown is sediments and blue is water. The grid connecting Lagrangian tracer particles shows the materialdeformation. The arrows indicate directions of major motions. The NW model (left) used no water layer and a 3-km thick sediment layer, and represents the target structure to thenorthwest of the crater center. The NE model (right) used a 2-km deep water layer and a 4-km thick sediment layer, and represents the target structure to the northeast of the cratercenter (a–e). Also shown is the final frame from a 3-layer NW model (f), including mantle (dark gray), for comparison with the 2-layer NW model.

226 G.S. Collins et al. / Earth and Planetary Science Letters 270 (2008) 221–230

to form a proximal ejecta deposit and a raised rim at ~70–80 kmradius. In the NE model, less sediment and crustal material is ejecteddue to the presence of the deep water layer above. In this case, theejected water, sediment and crust that lands just outside the cratergenerates a huge wave of water as it collides with the ocean (Fig. 5c).

This wave propagates outward at hundreds of meters per second,before a resurge flow returns water back towards and into the crater.The rapid movement of the ejecta and water layer precludes theformation of a sizable ejecta deposit. As a result, there is little-to-noraised rim observed in the final crater in the NE model (Fig. 5e).

Fig. 6. Close-up of themodeled terrace zone in Fig. 5 and the observed location of the terrace zone from the seismic reflection profiles shown in Fig. 4. The first-order differences in theobserved terrace zone (start position, total offset, width and maximum burial depth) are all replicated by the numerical models.

227G.S. Collins et al. / Earth and Planetary Science Letters 270 (2008) 221–230

Although no water layer is present in the NW model to erode thecrater rim, it is likely that resurging water from the NE sector wouldflood the whole crater andmaymodify the crater rim in the NW. Here,insight provided by our 2D modelling approach is limited; high-resolution 3D impact models are required to simulate accurately theeffect of asymmetric water resurge on crater morphology.

Fig. 6 shows a close-up of themodeled terrace zone comparedwiththe inferred location of the terrace zone from the seismic reflectionprofiles shown in Fig. 4. The innermost observed location of theterrace zone in the reflection data is between 40 and 43 km radialdistance (Morgan et al., 1997), whereas the numerical models showinwardly collapsed sediments as far inward as 30 km. The NW and NEmodels both show amarked increase in deformation of the Lagrangiantracer grid as the crater center is approached. The total displacement

Fig. 7. Close-up of themodeled terrace zone in Fig. 5, showing total displacement of targetmadeformation; in both models deformation in the terrace zone increases toward the crater cenradius of ~40 km (see Fig. 6).

of each Lagrangian tracer particle is shown graphically in Fig. 7. Totaldisplacement is a rough proxy for the amount of deformation thematerial represented by the tracer particle experiences during theimpact; contour plots of total plastic strain show similar, but morediffuse, patterns of deformation. The increase in deformation withdistance toward the crater center is similar in both models and, thus,does not depend strongly on the thickness or depth of thesedimentary layer. We suggest that the reflective terrace zonedisappears at ~40 km radial distance in the seismic data, at allazimuths, due to the increase in deformation with decreasing radialdistance.

Assuming that the terrace zone is not imaged by the seismic datainward of 40-km radius, in any direction, the agreement between thenumerical and geophysical model is good (Fig. 6). Although the

terial during impact crater growth and collapse. Total displacement (in km) is a proxy forter. This may explain the loss of coherent seismic reflectivity in the terrace zone inside a

1 Calculated cratering efficiency uses scaling laws of Holsapple (1993) and assumesvertical impact at a velocity of 18 km/s. Target and projectile density are both assumedto be 3 g/cc.

228 G.S. Collins et al. / Earth and Planetary Science Letters 270 (2008) 221–230

observed Lower Cretaceous rocks do not lie at exactly the location ofthe lowermost sedimentary section in the models, the first-orderdifferences in terrace zone geometry are replicated in the modeling.The start of the terrace zone occurs at a greater radial distance in theNW model than in the NE model. The total offset, width, and finaldepth of the terrace zone is greater in the NW model than in the NEmodel. In addition, there is little-to-no raised rim in the NE model,compared with the NW model. Each of these differences between themodels is consistent with the differences between the interpretedseismic data profiles in the NW and NE radial directions.

5. Discussion

Our numerical model results show that changing the water depthand sediment thickness at Chicxulub affects the geometry of theterrace zone. Moreover, the qualitative observational differences interrace zone geometry between the NWand NE radial seismic profilesare reproduced by models that use a pre-impact water depth andsediment thickness appropriate for those sectors of the crater. Thissuggests that target heterogeneity may be the principal cause ofasymmetry in the terrace zone of the Chicxulub crater. As discussedearlier, an alternative explanation for this and other asymmetries ofthe Chicxulub crater is oblique impact (Schultz and D'Hondt, 1996;Hildebrand et al., 1998a). Our 2D axis-symmetric models enforce anassumption of vertical impact; hence, the effect of impact angle oncrater formation was not addressed in our study. In this section wediscuss the possibility that the asymmetry in the terrace zone atChicxulub may be due to oblique impact instead of, or in addition to,target asymmetry.

Laboratory-scale oblique impact experiments show important,quantitative effects of impact angle and direction on crater size. Forexample, Gault and Wedekind (1978) observed that the diameter ofcraters formed in sand decreased with decreasing impact angle (to thetarget plane). Laboratory experiments, 3D numerical modeling, andextraterrestrial crater observations also show that the direction ofmomentum in an oblique impact has a strong effect on shock wavedecay (Pierazzo and Melosh, 2000a; Dahl and Schultz, 2001), and theamount and velocity of material ejected from the crater at differentazimuths (e.g. Gault and Wedekind, 1978; Stöffler et al., 2002; Herrickand Forsberg-Taylor, 2003; Artemieva and Ivanov, 2004). However, theeffect of impact angle on crater morphology and sub-surface structureis less well understood (Pierazzo and Melosh, 2000b).

Laboratory-scale oblique impact experiments and 3D numericalmodeling studies suggest that in oblique impacts the transient cavityis very asymmetric in the earliest stages of formation and progres-sively loses its asymmetry as it grows, due to geometric spreading(Schultz and D'Hondt, 1996; Anderson et al., 2003; Shuvalov andDypvik, 2004; Elbeshausen et al., 2007). The asymmetry of the finalcrater, therefore, depends not only on the angle of impact, but also onthe size of the crater in relation to the size of the projectile—commonly referred to as the cratering efficiency. Observable asym-metry may only occur for moderately oblique (30–60°) impacts if thecratering efficiency is low. For example, Gault and Wedekind (1978)observed that craters formed in sand only became noticeably elliptical(length/widthN1.1) at angles below 5–10° (to the target plane). In thiscase the ratio of crater diameter to projectile diameter was 30–40:1.Impact experiments in metals, on the other hand, produced cratersthat exhibited a strong dependence of morphology on impact angle(Burchell and Mackay, 1998), becoming elliptical at an impact angle ashigh as 40°. However, in this case crater formation was controlled bythe high cohesive strength of the target and crater growth wasarrested much earlier than in a granular target, like sand. The ratio ofcrater diameter to projectile diameter in these experimentswas 2–4:1.

The applicability of these experimental results to planetary-scalecratering is uncertain. In impacts of all scales a deep bowl-shapedcavity, referred to as the transient crater, is formed by the excavation

and outward displacement of target material (Dence et al., 1977;Melosh, 1989). However, an important difference between km-scalecratering and cm-scale cratering (even in granular targets) is that inlarge-scale impacts this cavity subsequently undergoes substantialgravitational collapse, leading to a considerably broader and shallowerfinal crater (see Fig. 5, for example). Thus, for impact direction relatedasymmetry to be present in a planetary-scale crater, the asymmetry ofthe transient crater must persist through subsequent collapse. Inlaboratory-scale impacts little or no collapse of the transient crater isobserved. Hence, experiments demonstrate only the effect of impactangle and direction on transient crater asymmetry.

Cratering efficiency for typical terrestrial impacts is not wellunderstood, but is controlled by gravity and friction (as is the case ingranular targets like sand), rather than cohesive strength, and isthought to range from approximately 23 for a 1-km diameter transientcrater to about 6 for a 100-km diameter transient crater.1 Thus, theefficiency of small impacts on Earth is closer to that of experimentalimpacts in granular targets, like sand; whereas, the efficiency of verylarge impacts on Earth is closer to that of experimental impacts intometal targets, but still several times larger. The stronger effect ofimpact angle on crater shape in low-efficiency, metal-target impactssuggests that transient crater asymmetry related to impact direction ismost likely to be observed in the largest impact structures, such asChicxulub. Three-dimensional (3D) numerical impact models thatsimulate the complete impact process, including collapse, are requiredto ascertain whether or not any transient crater asymmetry ispreserved in the final crater.

Contrary to the idea that impact direction and angle affect cratermorphology, there are several observations that suggest complexcrater formation, apart from ejection, is a predominantly axis-symmetric process for even highly-oblique impacts. Most signifi-cantly, the vast majority of impact craters are almost circular in planview, even though the average, and statistically most frequent, impactangle is 45°. Less than ~5% of craters onMars, Venus and the Moon aresignificantly elliptical (length/widthN1.1), suggesting that craterformation is only strongly asymmetric when the impact angle islower than ~12° (Bottke et al., 2000). This angle is not very different tothe 5–10° threshold observed in small-scale impacts into sand (Gaultand Wedekind, 1978). A statistical study of central peak craters onVenus showed that offset of the central uplift was not related toimpact direction (Ekholm and Melosh, 2001). Herrick and Forsberg-Taylor (2003) observed no relationship between impact angle andeither location or shape of the central structure on Venus or themoon.In addition, McDonald et al. (in press) showed that in larger Venusiancraters offset in the crater's peak ring is also randomly oriented withrespect to impact direction. Together these observations suggest thatcrater morphology cannot be used to infer impact direction or angle,even in large craters.

Although offset of the topographically exposed central craterstructures on Venus show no relationship to impact angle, this doesnot necessarily imply that terrestrial craters show no features that arediagnostic of impact angle. Scherler et al. (2006) recently argued that,although the Upheaval Dome crater in Utah is roughly circular, faultsand folds within the central uplift show a preferred transportdirection that might be indicative of the direction of impact. There isalso no extraterrestrial data on the geometry of terrace zones as theseare largely buried beneath impact breccias and melt rocks. Hence itremains unproven as to whether an oblique impact could produceasymmetries in either the crater's terrace zone, or the root of thecentral uplift.

In summary, despite observational evidence that for impact anglesgreater than 10–15° to the horizontal crater formation is predominantly

229G.S. Collins et al. / Earth and Planetary Science Letters 270 (2008) 221–230

an axis-symmetric process, there still remains the likelihood thatmoderately oblique impacts might also produce some observable craterasymmetry. The reduction in crater efficiency with increasing impactenergy implies that if crater asymmetry related to impact directionexists it is most likely to be apparent in the very largest impactstructures, such as Chicxulub. Hence 3Dmodels of the Chicxulub impactare clearly required to assess whether changing the direction and angleof impact can also produce some of the changes in terrace zonegeometry observed in the seismic data, and cause other asymmetries incrater structure. However, the fact that the observed target asymmetryat Chicxulub can explain the crater's terrace zone asymmetry, and thatthe asymmetry in target and terrace zone correlate, suggests that pre-impact target asymmetry is the primary cause of the crater's terracezone asymmetry. If this interpretation is correct, it implies thatasymmetry in the terrace zone at Chicxulub cannot be used to inferimpact angle or direction.

6. Conclusions

Marine seismic data acquired across the Chicxulub crater areinterpreted to show that the geometry of the crater's terrace zonevaries significantly around the offshore half of the crater. The seismicdata also reveal that, at the time of impact, both the water depth andsediment thickness varied with azimuth around the impact site. Theterrace zone asymmetry and the pre-impact target asymmetry arestrongly correlated, suggesting that heterogeneity in the initial targetmight have affected final crater geometry.

We constructed two end-member models of upper-target structureat Chicxulub from the seismic data at different azimuths. The modelrepresenting the NW had no water layer and a 3-km thick sedimentlayer; the model representing the NE had a 2-kmwater layer above a 4-km sediment layer. Numerical models of vertical impacts into these twotargets produced final craters that differ substantially in the geometryof the inwardly slumped sediment layers. Moreover, the differences inthe geometry of the sediment layers between the two numericalmodels are consistent with the differences in the geometry of theterrace zone at different azimuths, as interpreted from the seismic data.This demonstrates that target heterogeneity can cause the observedterrace zone asymmetry at Chicxulub. To definitively address whetherimpact angle could also be responsible for some of the observedasymmetries in the Chicxulub crater requires 3D modeling. However,although impact direction and angle may play a role in craterasymmetry, pre-impact asymmetry in the target is sufficient to explainthe asymmetric terrace zone observations at Chicxulub.

Acknowledgments

Both the 1996 and 2005 seismic experiments were jointly fundedby the Natural Environment Research Council (NERC) and the NationalScience Foundation (NSF). We thank Boris Ivanov and Jay Melosh fortheir support in developing iSALE. GSC was funded by NERC grant NE/B501871/1. We are grateful to N. Artemieva and an anonymousreviewer for their comments that improved the quality of this paper.This is IARC contribution no. 2007-1252 and UTIG contribution no.1934.

References

Amsden, A.A., Ruppel, H.M., Hirt, C.W., 1980. SALE: Simplified ALE Computer Programfor Fluid Flow at all Speeds, LA-8095. Los Alamos National Laboratory, Los Alamos,NM, p. 101.

Anderson, J.L.B., Schultz, P.H., Heineck, J.T., 2003. Asymmetry of ejecta flow duringoblique impacts using three-dimensional particle image velocimetry. J. Geophys.Res.-Planets 108, 5094. doi:10.1029/2003JE002075.

Artemieva, N., Ivanov, B., 2004. Launch of martian meteorites in oblique impacts. Icarus171, 84–101.

Bottke, W.F., Love, S.G., Tytell, D., Glotch, T., 2000. Interpreting the elliptical craterpopulations on Mars, Venus, and the Moon. Icarus 145, 108–121.

Burchell, M.J., Mackay, N.G., 1998. Crater ellipticity in hypervelocity impacts on metals.J. Geophys. Res. 103, 22761–22774.

Camargo-Zanoguera, A., Suárez-Reynoso, G., 1994. Evidencia sísmica del cráter deimpacto de Chicxulub. Bol. Asoc. Mex. GeofiÂs. Explor. 34, 1–28.

Christeson, G.L., Nakamura, Y., Buffler, R.T., Morgan, J., Warner, M., 2001. Deep crustalstructure of the Chicxulub impact crater. J. Geophys. Res. 106, 21751–21770.

Collins, G.S., Wünnemann, K., 2005. How big was the Chesapeake Bay impact? Insightfrom numerical modeling. Geology 33, 925–928.

Collins, G.S., Wünnemann, K., 2006. Impact crater formation in mixed crystalline andsedimentary targets: Insight from numerical modelling. Geophys. Res. Abstr. 8,06901.

Collins, G.S., Melosh, H.J., Morgan, J.V., Warner, M.R., 2002. Hydrocode simulations ofChicxulub crater collapse and peak-ring formation. Icarus 157, 24–33.

Collins, G.S., Melosh, H.J., Ivanov, B.A., 2004. Modeling damage and deformation inimpact simulations. Meteorit. Planet. Sci. 39, 217–231.

Collins, G.S., Ivanov, B.A., Melosh, H.J., Morgan, J.V., Wünnemann, K., 2005. NumericalModeling of the Chicxulub Impact Structure. Spring Meeting. American Geophy-sical Union. Abstract #P21A-02.

Dahl, J.M., Schultz, P.H., 2001. Measurement of stress wave asymmetries inhypervelocity projectile impact experiments. Int. J. Impact. Eng. 26, 145–155.

Dence, M.R., Grieve, R.A.F., Robertson, P.B., 1977. Terrestrial impact structures: principalcharacteristics and energy considerations. In: Roddy, D.J., Pepin, R.O., Merrill, R.B.(Eds.), Impact and Explosion Cratering. Pergamon Press, New York, pp. 247–275.

Ekholm, A.G., Melosh, H.J., 2001. Crater features diagnostic of oblique impacts: the sizeand position of the central peak. Geophys. Res. Lett. 28, 623–626.

Elbeshausen, D., Wunnemann, K., Collins, G.S., 2007. Three-dimensional numericalmodeling of oblique impact processes: scaling of cratering efficiency. 38th Lunarand Planetary Science Conference, Houston, TX. Abstract #1952.

Gault, D.E., Wedekind, J., 1978. Experimental studies of oblique impact. Proc. Lunar andPlanet. Sci. Conf. 9th, pp. 3843–3875.

Gault, D.E., Quaide, W.L., Oberbeck, V.R., 1968. Impact cratering mechanics andstructures. In: French, B.M., Short, N.M. (Eds.), Shock Metamorphism of NaturalMaterials. Mono Book Corp., Maryland, Baltimore, pp. 87–99.

Goldin, T.J., Wünnemann, K., Melosh, H.J., Collins, G.S., 2006. Hydrocodemodeling of theSierra Madera impact structure. Meteorit. Planet. Sci. 41, 1947–1958.

Grieve, R.A.F., Therriault, A., 2000. Vredefort, Sudbury, Chicxulub: three of a kind? Annu.Rev. Earth Planet. Sci. 28, 305–338.

Grieve, R.A.F., Dence, M.R., Robertson, P.B., 1977. Cratering processes: as interpretedfrom the occurrence of impact melts. In: Roddy, D.J., Pepin, R.O., Merrill, R.B. (Eds.),Impact and Explosion Cratering. Pergamon Press, New York, pp. 791–814.

Gulick, S., Barton, P., Christeson, G., Morgan, J., McDonald, M., Mendoza, K., Pearson, Z.,Surendra, A., Urrutia, J., Vermeesch, P.M., Warner, M., 2008. Importance of pre-impact crustal structure for the asymmetry of the Chicxulub impact crater. NatureGeoscience 1, 131–135.

Herrick, R.R., Forsberg-Taylor, N.K., 2003. The shape and appearance of craters formedby oblique impact on the Moon and Venus. Meteorit. Planet. Sci. 38, 1551–1578.

Hildebrand, A.R., Penfield, G.T., Kring, D.A., Pilkington, M., Camargo, A., Jacobsen, S.B.,Boyton, W.V., 1991. Chicxulub crater: a possible Cretaceous/Tertiary boundaryimpact crater on the Yucatán peninsula, Mexico. Geology 19, 867–871.

Hildebrand, A.R., Pilkington, M., Halpenny, J.F., Cooper, R., Connors, M., Ortiz-Aleman, C.,Chavez, R.E., Urrutia-Fucugauchi, J., Graniel-Castro, E., Camara-Zi, A., Buffler, R.T.,1998a. Mapping Chicxulub crater structure with overlapping gravity and seismicsurveys. Lunar and Planetary Science Conference XXIX, Houston, TX. Abstract#1821.

Hildebrand, A.R., Pilkington, M., Ortiz-Aleman, C., Cavez, R.E., Urrutia-Fucugauchi, J.,Connors, M., Graniel-Castro, E., Camara-Zi, A., Halpenny, J.F., Niehaus, D., 1998b.Mapping Chicxulub crater structure with gravity and seismic reflection data. In:Grady, M.M., Hutchison, R., McCall, G.J.H., Rothery, D.A. (Eds.), Meteorites: Flux withTime and Impact Effects. Geol. Soc. Spec. Pub., vol. 140. The Geol. Soc., pp. 155–176.

Holsapple, K.A., 1993. The scaling of impact processes in planetary sciences. Annu. Rev.Earth Planet. Sci. 21, 333–373.

Ivanov, B.A., 2005. Numerical modeling of the largest terrestrial meteorite craters. Sol.Syst. Res. 39, 381–409.

Ivanov, B.A., Artemieva, N.A., 2002. Numerical modeling of the formation of largeimpact craters, catastrophic events and mass extinctions. Impact and BeyondSpecial Paper, vol. 356. Geological Society of America, pp. 619–630.

Ivanov, B.A., Deniem, D., Neukum, G., 1997. Implementation of dynamic strengthmodelsinto 2D hydrocodes: applications for atmospheric breakup and impact cratering.Int. J. Impact Eng. 20, 411–430.

McDonald, M.A., Melosh, H.J., Gulick, S.P.S., 2008. Oblique impacts and peak ringposition: Venus and Chicxulub. Geophys. Res. Lett. 35, L07203. doi:10.1029/2008GL033346.

Melosh, H.J., 1979. Acoustic fluidization: a new geologic process? J. Geophys. Res. 84,7513–7520.

Melosh, H.J., 1989. Impact Cratering: A Geological Process. Oxford University Press.245 pp.

Melosh, H.J., Ivanov, B.A., 1999. Impact crater collapse. Annu. Rev. Earth Planet. Sci. 27,385–415.

Melosh, H.J., Ryan, E.V., Asphaug, E., 1992. Dynamic fragmentation in impacts:hydrocode simulation of laboratory impacts. J. Geophys. Res. 97, 14735–14759.

Morgan, J.V., Warner, M.R., 1999. Chicxulub: the third dimension of a multi-ring impactbasin. Geology 27, 407–410.

Morgan, J.V., Warner, M.R., Brittan, J., Buffler, R., Camargo, A., Christeson, G., Denton, P.,Hildebrand, A., Hobbs, R., Macintyre, H., Mackenzie, G., Maguire, P., Marin, L.,Nakamura, Y., Pilkington, M., Sharpton, V., Snyder, D., Suarez, G., Trejo, A., 1997. Sizeand morphology of the Chicxulub impact crater. Nature 390, 472–476.

230 G.S. Collins et al. / Earth and Planetary Science Letters 270 (2008) 221–230

Morgan, J.V., Warner, M.R., Collins, G.S., Melosh, H.J., Christeson, G.L., 2000. Peak ringformation in large impact craters. Earth Planet. Sci. Lett. 183, 347–354.

Morgan, J., Warner, M., Urrutia-Fucugauchi, J., Gulick, S., Christeson, G., Barton, P.,Rebolledo-Vierya, M., Melosh, J., 2005. Chicxulub Crater Seismic Survey PreparesWay for Future Drilling. . EOS Transactions, vol. 86. American Geophysical Union,pp. 325–328.

O'Keefe, J.D., Ahrens, T.J., 1999. Complex craters: relationship of stratigraphy and rings toimpact conditions. J. Geophys. Res. 104, 27091–27104.

Pierazzo, E., Melosh, H.J., 1999. Hydrocode modeling of Chicxulub as an oblique impactevent. Earth Planet. Sci. Lett. 165, 163–176.

Pierazzo, E., Melosh, H.J., 2000a. Melt production in oblique impacts. Icarus 145,252–261.

Pierazzo, E., Melosh, H.J., 2000b. Understanding oblique impacts from experiments,observations, and modeling. Annu. Rev. Earth Planet. Sci. 98, 10–96.

Pierazzo, E., Collins, G.S., 2003. A brief introduction to hydrocode modelling of impactcratering. In: Dypvik, H., Burchell, M., Claeys, P. (Eds.), Cratering in MarineEnvironments and on Ice. Springer, Berlin, p. 340.

Pierazzo, E., Artemieva, N., Asphaug, E., Cazamias, J., Coker, R., Collins, G.S., Gisler, G.,Holsapple, K.A., Housen, K.R., Ivanov, B., 2007. The impact hydrocode benchmarkand validation project: initial results. Lunar and Planetary Science Conference,Houston, Texas.

Pike, R.J., 1974. Depth/diameter relations of fresh lunar craters: revision from spacecraftdata. Geophys. Res. Lett. 1, 291–294.

Roddy, D.J., 1977. Large-scale impact and explosion craters: comparisons of morpho-logical and structural analogs. In: Roddy, D.J., Pepin, R.O., Merrill, R.B. (Eds.), Impactand Explosion Cratering. Pergamon Press, New York, pp. 185–246.

Scherler, D., Kenkmann, T., Jahn, A., 2006. Structural record of an oblique impact. EarthPlanet. Sci. Lett. 248, 43–53.

Schultz, P.H.,1996.Effectof impact angle onvaporization. J.Geophys.Res.100, 21117–21135.Schultz, P.H., D'Hondt, S., 1996. Cretaceous–Tertiary (Chicxulub) impact angle and its

consequences. Geology 24, 963–967.

Shuvalov, V., Dypvik, H., 2004. Ejecta formation and crater development of the Mjolnirimpact. Meteorit. Planet. Sci. 39, 467–479.

Stöffler, D., Artemieva, N.A., Pierazzo, E., 2002. Modeling the Ries–Steinheim impactevent and the formation of the moldavite strewn field. Meteorit. Planet. Sci. 37,1893–1908.

Thompson, S.L., Lauson, H.S., 1972. Improvements in the Chart D radiation-hydro-dynamic CODE III: revised analytic equation of state. Sandia Laboratories, ReportSC-RR-71-0714, Albuquerque, N. Mex., USA, vol. 119.

Turtle, E.P., Pierazzo, E., Collins, G.S., Osinski, G.R., Melosh, H.J., Morgan, J.V., Reimold, W.U.,2005. Impact structures: what does crater diameter mean? In: Kenkmann, T., Hörz, F.,Deutsch, A. (Eds.), Large Meteorite Impacts III. GSA Special Paper, vol. 384. Geol. Soc.Am., Boulder, CO, pp. 1–24.

Urrutia-Fucugauchi, J., Marin, L., Trejo Garcia, A., 1996. UNAM scientific drilling programof the Chicxulub impact structure—evidence for a 300-kilometre crater diameter.Geophys. Res. Lett. 23, 1565–1568.

Vermeesch, P., Morgan, J., in press. Structural uplift beneath the Chicxulub impactstructure. J. Geophys. Res. doi:10.1029/2007JB005393, (accepted 24 March 2008)http://www.agu.org/contents/journals/ViewPapersInPress.do?journalCode=JB#id2007JB005393.

Ward, W.C., Keller, G., Stinnesbeck, W., Adatte, T., 1995. Yucatán stratigraphy:implications and constraints for the Chicxulub impact. Geology 23, 873–876.

Wünnemann, K., Ivanov, B.A., 2003. Numerical modelling of the impact crater depth-diameter dependence in an acoustically fluidized target. Sol. Syst. Res. 51, 831–845.

Wünnemann, K., Morgan, J.V., Jödicke, H., 2005. Is Ries crater typical for its size? Ananalysis based upon old and new geophysical data and numerical modeling. In:Kenkmann, T., Hörz, F., Deutsch, A. (Eds.), Large Meteorite Impacts III. GSA SpecialPaper, vol. 384. Geol. Soc. Am., Boulder, CO, pp. 67–83. 384.

Wünnemann, K., Collins, G.S., Melosh, H.J., 2006. A strain-based porosity model for usein hydrocode simulations of impacts and implications for transient crater growth inporous targets. Icarus 180, 514–527.