Chapter 19

New Insights on the Formation of the Caribbean Basalt Province Revealed by Multichannel Seismic Images of Volcanic Structures in the Venezuelan Basin

J O H N D I E B O L D , N E A L D R I S C O L L and EW-9501 S C I E N C E T E A M 1

A regional multichannel seismic survey (EW-9501) was carried out in the Venezuelan Basin aboard R/V Ewing. The resulting reflection profiles image previously unseen structures within the entire thickness of the Caribbean oceanic plateau basalts east of Beata Ridge. Identifiable volcanic products include large amounts of extrusive material, which form two morphologically distinct, vertically stacked sequences. The lower sequence consists of local highs and ridges, flanked by wedges made up of dipping flows. The upper sequence is more homogeneous, featuring widespread flows which fill morphological and extensional lows in the lower sequence. Dipping sequences of volcanic flows are seen with length scales varying from 20 km to over 100 km. In contrast to seaward-dipping reflectors described elsewhere, these flows appear to be of submarine origin, and to have maintained their primary sense of dip. The morphology and scales of the flows are controlled by the volume of source material. The top of the upper 'high-volume' sequence forms the smooth B" horizon sampled by DSDP and ODP drilling. A more detailed survey was made of the southeastern edge of the plateau which includes a 100-kin-wide sequence of flows forming gently dipping wedges overlying thinned oceanic crust. The resulting images show clearly that the present edge is constructional, and that buried beneath it is an earlier, possibly rifted edge. This two-phase development is characteristic of the plateau everywhere in the Venezuelan Basin, and interpretations of the Caribbean's history must take this into account. It is not clear how much time elapsed between the end of the first, and the onset of the later phases, but our results indicate that the later phase is the only one so far sampled by Caribbean drilling.

INTRODUCTION Two-ship refraction profiles carried out in the 1950s yielded velocity functions of a Caribbean crust that was anomalously thick (up to 20 km), with an unusual two-layer structure. Based on early analog single-channel seismic reflection data, two distinctive horizons were designated A" and B", according to their resemblance to A and B in the Atlantic, and A', B' in the Pacific. Sediment cores, carefully located on the basis of those single-channel profiles, showed A" to correlate with an Eocene chert layer, a result later confirmed by D S D P drilling, which indicated that layers of limestone also contributed to the reflective character of A". The makeup of B" (stands for acoustic ' b a s e m e n t ' ) remained unknown for a while longer. Magnetic anomalies in 1Lewis Abrams, Peter Buhl, Thomas Donnelly, Edward Laine, Sylvie Leroy, Adrienne Toy

the Venezuelan Basin seem to form linear N E S W trends, which correspond with those of minor faults and grabens i m a g e d in reflection profiles and m a p p e d by Case and H o l c o m b e (1980). A n u m b e r of efforts (Christofferson, 1973; Ghosh et al., 1984) have been m a d e to fit the magnetic 'lineations' in the Venezuelan and C o l o m b i a n Basins to known seafloor spreading sequences, but the results have been unpersuasive. Drilling, during D S D P Leg 15, sampled B" for the first time, and found it to be the top of a Cretaceous volcanic sequence. The igneous rocks of the Caribbean basalt province, as sampled by drilling at five sites during D S D P Leg 15, Site 1001 during O D P Leg 165, and in on-land occurrences, are virtually all basalt. Their character has been s u m m a r i z e d by Donnelly et al. (1990; Fig. 1). The bulk of the basaltic material occurs in three petrochemical varieties: transitional tholeiitic basalt, highly magnesian basalt, and mildly alkalic basalts. The transitional

Caribbean Basins. Sedimentary Basins of the World, 4 edited by P. Mann (Series Editor: K.J. Hsti), pp. 561-589. 9 1999 Elsevier Science B.V., Amsterdam. All rights reserved.

562

J. DIEBOLD et al.

Fig. 1. EW-9501 track, with line numbers and sonobuoylocations on bathymetry; 500 m contours. DSDP and ODP sites where volcanic basement was sampled by drilling are indicated. tholeiitic basalts are the most widely distributed and the most voluminous of the basalt types sampled in the Caribbean region. The highly magnesian basalts have been termed 'komatiite' on Isla Gorgona by Echeverria (1980) and in Colombia by Spadea et al. (1989), 'picrite' on Curaqao by Beets et al. (1982) and Klaver (1987), and 'low-Ti-basalts' in the Santa Elena Peninsula of Costa Rica by Wildberg (1984). The mildly alkalic basalts are best known from the Beata Ridge (DSDP Site 151) and from the Dumisseau Formation of southern Haiti. They are also represented in the younger (latest Cretaceous or Paleogene) basaltic centers from Costa Rica (Wildberg, 1983; Frisch et al., 1992). These basalts are enriched in K, Th, and several minor elements. They resemble well-known, mildly alkalic basalts of Hawaii, and are characteristic of ocean island basalts. Where there is some local basis for stratigraphic positioning, they seem to occur among the later eruptive products. One of the most unsatisfactory aspects of the previous studies of the Cretaceous flood basalts has been establishing ages of magmatic activity. Conventional K-Ar age determinations that are very numerous in the literature (Donnelly et al., 1990) can be shown in several instances to contradict clear stratigraphic constraints. In general, the ages ascertained from K-Ar are too young because of Ar loss or K uptake during seafloor weathering or during burial metamorphism. Problems with biostratigraphic ages

also occur because it is difficult to assess the duration of the hiatus between basalt emplacement and pelagic accumulation. Even worse, if the basalts are sills then the biostratigraphic age only yields a maximum age of emplacement. Despite the numerous dating problems, there are several firm ages for the termination of magmatic activity, indicated by the locally youngest ages of the Cretaceous basalts. As reviewed in Donnelly et al. (1990) these dates are mainly derived from dating sedimentary deposition on the basalt complex, including DSDP Leg 15 sites. These ages cluster tightly at the latest Turonian, but in a few cases could be as young as Santonian, and even early Campanian (DSDP Site 152 and ODP Leg 165). Recently, a detailed radiometric study of basalts from Gorgona, Costa Rica, Haiti, and Curaqao has been undertaken by Sinton et al. (1998) using 4~ incremental heating methods, which yield dates consistent with the sedimentary ages for the termination of the event. The number and geographic extent of consistent age determinations at 88-90 Ma suggest that a truly vast igneous event occurred at this time. Nevertheless, there remain several circumstantial fossil occurrences of Albian and Aptian age associated with the Cretaceous basalt province. For example, intercalated within the Curaqao basalt complex is a sedimentary succession with six genera of ammonites that indicate a latemiddle Albian age (Weidmann, 1978), which is significantly older than the biostratigraphic and ra-

MULTICHANNEL SEISMIC IMAGES OF VOLCANIC STRUCTURES IN THE VENEZUELAN BASIN

diometric ages obtained elsewhere. There are two interpretations: (1) the ammonite ages are incorrect, or (2) there was an earlier age of magmatic activity of unknown extent, and the 88-90 Ma event was a terminal event for the main Caribbean Cretaceous basalt province, but not the initial event. Early multichannel seismic (MCS) data acquired by IFP (France), UTIG (Texas) and L-DEO (Lamont) occasionally imaged internal layering beneath B", and showed that to the southeast, the volcanic sequence had a distinct edge, beyond which lay a triangular area of deep, rough acoustic basement (called 'rough B"') whose reflective character resembled oceanic crust. This area was found, however, to be practically devoid of magnetic signature (Donnelly, 1973), suggesting that the crust was either created by seafloor spreading during a magnetically quiet period, or that the original magnetic signature was degraded by thermal or tectonic modification. Early MCS surveys, carried out during the 1970s, defined the approximate area of rough basement. The distinction between the rough and smooth varieties of B" was first described by Talwani et al. (1977) and a series of normal faults was shown to coincide with the rough-smooth boundary by BijuDuval et al. (1978). The acquisition of additional crossings of the boundary showed that it had a variety of manifestations, ranging from imperceptibly gradual to abrupt (Diebold et al., 1981). Lamont data (described by Diebold et al., 1981) provided five crossings of the rough-smooth B" boundary, which were not enough to characterize the exact nature of the transition. Additional crossings by French (IFP) and US (UTIG) investigators all converged at the boundary in the same area as Lamont line 108. The resulting coverage showed that the basement faults controlling the rough-smooth boundary had strikes oblique to the general trend of the transition, suggesting that the basement fabric pre-dated the volcanic emplacement. MCS line 108 was unique in that it showed clearly how depth to Moho increases, in a step-wise fashion, northward across the roughsmooth transition, leading Diebold et al. (1981) to interpret smooth B" as the top of Cretaceous flows which had overrun and depressed rough B" crust. Sonobuoy refraction lines shot in 1974 in conjunction with MCS data (Talwani et al., 1977) followed by two-ship multichannel seismic expanding spread profiles (ESPs) carried out in 1976 showed that the rough basement horizon B" was the top of what appeared to be anomalously thin oceanic crust, and that the crust underlying the only ESP located over smooth B" was, as expected, unusually thick, and comprising (in a gross sense) two layers (Diebold et al., 1981). Though it was the first to be discovered, the area of smooth B" has been the most difficult to pene-

563

trate by seismic reflection. This is principally due to the high reflection coefficients of the sedimentbasement interface there, though in early work, large water depths and weak profiling systems played a part. Reflections beneath the smooth B" horizon were detected from time to time in early MCS data (Hopkins, 1973; Ladd and Watkins, 1980; Stoffa et al., 1981; Diebold et al., 1981). In some cases, these reflectors were observed to dip in relation to the B" surface, though their identification was often open to doubt. The strong basement reflection coefficient gives rise to strong intrabed ('peg-leg') multiples that reverberate efficiently within the sedimentary section. The fact that water depths were quite large, compared to the length of receiving arrays used in the early work, made it difficult to discriminate between those multiples and primary arrivals. Intrabed multiples are still a problem, even with the 4 km receiving array of R/V Ewing.

EW-9501

In February and March, 1995, 5200 km of MCS data, shot with a 20-airgun (8415 cubic inch) source array, and a 4-km, 160-channel hydrophone streamer, were acquired aboard R/V Ewing. Underway geophysics included measurements of gravity, magnetics, and swath bathymetry. 104 successful deployments of expendable sonobuoys were made. A track map, identifying the seismic reflection lines by number, with sonobuoy deployment locations, is shown as Fig. 1. The purpose of this NSF-funded cruise was to better define the boundary between smooth and rough B", to image structures within and beneath the Caribbean volcanic complex, and to provide new data on the structure and development of the Beata Ridge. The new data meet these goals, and also reveal several unexpected features of the sedimentary column which provide important new constraints on the tectonic and oceanographic development of the Caribbean region. The motivation behind the cruise was to obtain deep reflection seismic data in the Venezuelan Basin and on the Beata Ridge crest and its eastern flank. In the Venezuelan Basin, the survey track was laid out to complement MCS data previously acquired by R/V CONRAD in the 1970s (Talwani et al., 1977; Stoffa et al., 1981; Diebold et al., 1981) specifically where these lines crossed the rough-smooth B" boundary. The general NNW-SSE orientation of the lines, and their northwestward extent, reflected the desire to image deep structure (if any) related to previously mapped basement topography (Case and Holcombe, 1980) and magnetic lineations (Donnelly, 1973) which trend ESE-WSW. In retrospect, the Venezuelan Basin lines should have extended farther

564 to the north-northwest, and more cross-lines should have been shot, but to do this extra work would have required an impossibly long cruise. 104 Navy-supplied expendable sonobuoys were deployed during the survey. Data from most of these were recorded to maximum offsets of between 30 and 40 km, providing refraction and wide-angle reflection data for crustal velocity analysis. Onedimensional velocity analysis, using interactive ray trace modeling has been attempted for all of the sonobuoys. Velocity functions were obtained for 85 buoys. Results from this analysis, combined with earlier wide-angle reflection and refraction work, provide several important constraints on the structure and mode of emplacement of the Caribbean LIP (Large Igneous Province). The 4-km-long hydrophone array used in acquiring the reflection profiles, along with the powerful seismic source array of R/V Ewing, has produced the best MCS data ever shot in the deep-water basins of the Caribbean. The streamer length, in particular, allows high resolution of hyperbolic stacking velocity, which, if accurately determined, results in amazingly detailed images of what now can be seen as intricate and convoluted sub-basement structures in the Venezuelan Basin. The other side of this coin is that stacking velocity analysis is a painstaking and time-consuming process. The high velocity contrasts at the sediment-basement interface, and the large range of source-receiver offsets present in the data (and required, to ameliorate the effects of intrabed multiples), cause unusually extreme lateral changes in the stacking velocities, despite the large water depths of the Caribbean, which normally could be expected to minimize these changes. As a result, most of the reflection data we present here are in the form of 'brute' stacks. In processing, normal moveout corrections were made based on a velocity model created by extrapolation from sonobuoy results, and the stacking velocity analyses manually carried out for line 1293, the only line whose stacking velocities have so far been completely determined. Therefore, the foldout section for line 1293 (Fig. 2) is very important in evaluating the discussions presented below. While the other, brute-stack sections show features never before seen, the 1293 profile gives a feeling for the level of fine detail that will eventually be imaged by these data. Excepting velocity analysis, all of the profiles were processed identically. The field data were filtered and decimated to 4-ms sample size. Water depth-dependent outer and inner mutes were applied (the effect of outer muting is to make imaging of sedimentary structures less sensitive to errors in stacking velocities; the inner mute reduces the effects of peg-leg and water column multiples). After stacking, the data were filtered again and AGC was

J. DIEBOLD et al. applied. The reflection data plots presented here have been horizontally compressed so that they can be shown in their entirety. This has resulted in a vertical exaggeration of about 20.8 in the water column. A special dip-adaptive trace mixing process was employed, to enhance the appearance of coherent reflectors, so that when only every fifth trace is plotted, the result is to retain most of the structural information in the data. All of the reflection profiles are plotted with north and west to the left, south and east to the right. CDP numbers are in chronological sequence in each line. CDP spacing is 12.5 m, so 800 CDPs equals 10 km horizontal distance.

CRUSTAL ELEMENTS OF THE COLOMBIAN AND VENEZUELAN BASINS Thin crust

Despite 20 years accumulation of evidence to the contrary, the concept that the Colombian and Venezuelan Basins are uniformly capped by a Cretaceous igneous body persists. In fact, the crustal structure of these basins is much more complex, and is made up of at least four categories of lithological elements: (1) thin ocean-like crust; (2) volcanic extrusive sequences with dipping reflectors; (3) massive bodies of volcanic extrusives and intrusives; and (4) extrusive volcanic mounds. While it might be argued that the last three categories listed above are simply different manifestations of the same crust-thickening Cretaceous volcanic event, the Caribbean Sea contains a significant amount of crust whose thickness varies from normal (6-8 km) to abnormally thin (3-5 km). Although analysis of many of the early two-ship refraction profiles indicated the presence of anomalously thick crust, several, particularly west of Beata Ridge, detected crust of normal oceanic thickness (Ewing et al., 1960; Edgar et al., 1971). Ludwig et al. (1975) described the basement reflector in the Colombian Basin as 'rough' in comparison to that in the Venezuelan Basin. In 1977, Houtz and Ludwig mapped a number of areas in the Colombian Basin where crust was deep, rough and thin. Bowland and Rosencrantz (1988) re-mapped a roughsmooth boundary in one of the areas outlined by Houtz and Ludwig (1977) and mapped several others. To the east, Talwani et al. (1977) provided MCS profiles showing rough basement in the Venezuelan Basin. Simultaneously recorded sonobuoys showed that crust in those areas was thin, compared to normal oceanic crust. Biju-Duval et al. (1978) also observed rough, deep basement in the southeastern comer of the Venezuelan Basin and suggested that it corresponded to oceanic crust, upon which was su-

MULTICHANNEL SEISMIC IMAGES OF VOLCANIC STRUCTURES IN THE VENEZUELAN BASIN perimposed the volcanics to the northwest. Diebold et al. (1981) interpreted additional two-ship velocity profiles and obtained velocity-depth functions showing that the rough B" crust resembled unusually thin oceanic crust. They also extended mapping of the rough-smooth B" transition around the Venezuelan Basin and imaged Moho in a few places beneath both rough and smooth crustal types. A compendium map of crustal thickness, throughout the Caribbean, was presented by Leroy (1995). The Moho reflection can be seen beneath rough B" on all of the EW-9501 profiles, and on a few of the older RC- 1904 and RC-2103 lines (Diebold et al., 1981). Interval two-way time of the rough crust is typically 1.2 s, with a corresponding sonobuoyderived thickness of 3800 m. In a few places, crustal thickness varies, thinning to as little as 0.9 s, and thickening to as much as 1.6 s. Velocity seems to vary also, though no systematic changes have yet been detected. Our results agree with those of Talwani et al. (1977) and Diebold et al. (1981). Upper crustal velocities are commonly around 5.5 kin/s, though occasionally, a thin upper layer is detected with velocities from 4.2 to 4.9 km/s. Mid-crustal velocities of 6.6 km/s are typical, and often velocities around 7.1-7.2 km are seen below these. 'Normal' Moho velocities are never seen, but instead, critically refracted energy with velocities between 7.4 and 7.8 km/s appears, corresponding to rays turning at depths below reflection Moho. These crustal thicknesses are well below the normal range, as defined by White et al. (1992), who also demonstrated a historical bias towards the underestimation of oceanic crustal thickness. White et al. (1992) compared inversion results based on forward synthetic seismogram modeling to those obtained by traditional slope-intercept methods and found that they differed (typically) by 20%. Even if such a bias is present in our methods (neither slope-intercept, nor synthetics-based modeling) the thicknesses we (and Talwani et al., 1977) determine for rough B" crust are abnormally thin. EW-9501 MCS profiles in the southeastern Venezuelan Basin indicate that the seismically fast (3.5-4 km/s) 'transparent' sediments identified by Biju-Duval et al. (1978) are, in our view, laminated turbidite sequences directly overlying rough basement there. The stratal relationships that make this obvious (filling bathymetric lows and onlapping highs) are particularly well represented in line 1298 (Fig. 3). The interval two-way time between the easily identifiable A" and underlying 'basement' increases dramatically, from 0.2 s at CDP 14,500 to about 0.5 s at CDP 12,500. The number of reflectors within the A"-B" interval increases dramatically across the rough-smooth B" transition, and the reflections become more distinct. Similar sediments

565

have been imaged (though not always recognized for what they are) in the Colombian Basin (Lu and McMillen, 1982; Kolla et al., 1984; Bowland and Rosencrantz, 1988; Bowland, 1993). The presence of these sediments appears to be diagnostic of deep, rough, thin crust, and identifying them in older data can further extend the mapped areas of thin crust. Several EW-9501 MCS profiles show that in a few places the anomalously thin rough B" crust of the SE Venezuelan Basin produces low-angle, upward-concave reflectors which appear to be the surfaces of listric faults extending from the seafloor to the Moho; see for example line 1293 (Fig. 2) CDPs 24,000-25,000. As a rule, these do not correlate with crustal offsets, even where the reflections are seen to intersect the top of crust. Similar reflections have been imaged in several areas, including marginal NW Atlantic crust of Jurassic age (McCarthy et al., 1988; Mutter and Karson, 1992; Rosendahl et al., 1992; Zehnder Mutter, 1992) and are, apparently, the products of some aspect of seafloor spreading. A recent study by Kent et al. (1997) has demonstrated, however, that the same pattern of reflections may arise due to sideswipe from a sub-parallel basement ridge, and it is certainly possible that some of the 'faults' we see are artifacts of this kind. Although velocity analysis (not yet complete) indicates that these are intra-crustal events, Kent et al. (1997) have shown that streamer feathering can be the cause of high stacking velocities for sideswipe events, which would otherwise be better imaged with lower velocities. The question remains: was Venezuelan Basin rough crust formed thin, or was it thinned by some tectonic process, presumably, extension. Unusually thin oceanic crust has been associated with the development of passive margins, e.g. the Iberian margin (Whitmarsh et al., 1990) and in the Labrador Sea (Hinz et al., 1979; Srivastava and Keen, 1995) and occasionally in ocean basins (Jackson et al., 1982; Muller et al., 1997) presumably created at slowspreading ridges (Reid and Jackson, 1981; Chen, 1992; White et al., 1992). Mantle material beneath this thin crust often exhibits abnormally low velocity, presumably from serpentinization, resulting from invasion of seawater along fault planes. Similar velocities are observed in the southeastern Venezuelan Basin (Talwani et al., 1977). We note that the 'originally thin' oceanic crust of the Iberian margin forms a narrow strip (Whitmarsh et al., 1990), while the rough Venezuelan Basin crust is much wider in all directions. The southeastern extent of thin, rough B" crust is entirely unknown; it can be seen (for example in EW-9501 line 1319, Fig. 4) being subducted beneath the accretionary wedge of the Curacao Ridge. The destruction of the crust by normal faulting, apparent in this profile, is typical of

566 that seen in other MCS profiles crossing this zone. The amount of thin crust consumed by subduction and obduction is difficult to estimate, and its original southern boundary is difficult to reconstruct. The Iberian margin also features a narrow zone (25-30 km wide) of greatly thinned continental crust whose abnormally high basal velocities are thought to result from underplating (Whitmarsh et al., 1990). Both of these Iberian scenarios (thinned continental crust, thin transitional oceanic crust) produce crust having a thickness similar to that of the southeast Venezuelan Basin, and also having little or no magnetic signature. The variation of apparent velocities in rough Venezuelan Basin crust is such that seismic velocity alone cannot be used to make the distinction between crustal types described by Whitmarsh et al. (1990). As in the case of its thin, early-stage oceanic crust, the radically thinned continental crust of the Iberian margin is found only in a narrow band. Rosendahl et al. (1992) have determined a wider (not specified, but shown as at least 65 km) zone of attenuated continental crust off the African margin in the Gulf of Guinea. This crust has also, apparently, been thinned by faulting, which causes a rough 'pseudo-oceanic' appearancepseudo-oceanic appearance in reflection profiles. This crust is of comparable thinness to that of Caribbean rough B", but this thinness is defined by an elevated 'equilibrium' Moho whose smooth reflection character is quite dissimilar to that of Moho imaged anywhere in the Caribbean. Next, we present evidence that crustal extension was taking place during the emplacement of Venezuelan Basin smooth B" volcanics. It is likely that such extension may have reactivated preexisting faults in rough B" crust.

Rough-smooth B" transition It is apparent from the EW-9501 data that at least some of the area of thin rough crust pre-dates the emplacement of smooth B" volcanic material, and that the area of 'originally' rough crust is, therefore, significantly larger than previously realized. Both the reflection profiles and the sonobuoy data show that the thin crust underlies the Cretaceous volcanics, and is present at least 50 km northwest of the roughsmooth boundary along some profiles. Ewing profiles 1300-1316 provide 3-D coverage dense enough to show that the model proposed by Diebold et al. (1981) for the rough-smooth boundary as an overrunning of en-echelon normal fault blocks by lava flows encroaching from the northwest is basically correct. The uppermost, and presumably youngest, sequence of extrusives included highly mobile flows, which were able to spread in thin fingers, along the valleys separating gently tilted fault blocks in the

J. DIEBOLD et al. rough B" crust. The strike of those faults which can be accurately mapped is ENE-WSW, and none seems to extend for more than 100 km m most are shorter. Their en-echelon pattern may be indicative of wrench faulting. Some of the extrusive fingers can be seen as bright, smooth spots as far as 20-30 km from the principal rough-smooth boundary (for example, line 1298, CDP 11,000; line 1293, CDP 20,350). When successive flows filled and finally buried the valleys, overlying flows were able to spill over into the next valley, thus spreading quickly to the east-northeast, more slowly south-southeast. In several places, the tilting of the normally faulted blocks continued after a thin veneer of volcanic material had been emplaced, creating the smoothtopped tilted blocks ('ski-jumps') characteristic of the Central Venezuelan Fault Zone (so named by Biju-Duval et al., 1978), and seen here in Ewing lines 1293 (Fig. 2, CDP 19,100), 1298 (Fig. 3, CDP 14,000), and 1300 (Fig. 5, CDP 15,000). Lines 1300 and 1302, though only 35 km apart, show different manifestations of the rough-smooth B" transition. On line 1300 the distal edge of the smooth B" volcanics is tilted back into a 'skijump' form, probably caused by post-emplacement extensional rotation of underlying blocks of rough basement. The fault system that controls the geometry and structure of the 'ski-jump' must be relatively surficial and antithetical to a larger fault system that dips north-northwest and controls the location and geometry of the large divergent wedge observed beneath B" (CDPs 9000-13,000; in line 1293, CDPs 10,500-19,000) and the large step in Moho topography (CDP 15,000; in line 1293, CDP 19,000). In line 1302, the transition (Fig. 6, CDP 5200) is quite gradual, the flows having apparently run around the end of line 1300's rotating block. In both cases, thin, upper flow units appear to have traveled in and out of the profile, leaving thin edges facing both north (CDP 10,900) and south (CDP 6900). The rough-smooth transition as seen in line 1302 is gradual, as the expression of the normal faults is only slight there. As a result, the rough B" surface can be seen beneath the distal edge of the smooth B" flows. A pair of the previously described throughgoing listric crustal 'faults' characteristic of rough B" crust can also be seen (CDPs 6500 and 7500). Once it is covered by smooth B" volcanics, the reflection from the top of the thin crust is quite weak, which is probably the result of three things, i.e. a small velocity contrast, a rough interface, and the absence of an intervening sedimentary layer. Some sonobuoys from this and previous studies have produced complete crustal sections, particularly in an area where sonobuoy records (e.g. Fig. 7) show that strong postcritically refracted energy is frequently returned from the Moho. This area is crossed by

~o

0 0 cO

0 0

0 0

0 0 ¢0

0 0

0 0

0 0

0 0 0

0 0

13.. T O

o

CO

oes 'ewp,/~eN~-OM± o~

0

iI ,

.

•

. . . . . .

,.

0 0

0 0 co

0

.

..

2 - " - _- -.

. ~ . N

...

....

-..~.

.-..r

.-

.

•

.. .

~.

7._-. ~. >. . . . _-. _

.

.... ...,- ::. ....

.~ .: . . . . -_

-