Site 1039 was drilled through the trench floor as a reference site, and a continuous set of cores was acquired through the sedimentary section and bottomed in 22 m of intrusive gabbro. The reference section was not deformed in any way, providing a true reference for this program. Drilling recovered about 5 m of terrigenous turbidites, overlying a thick sequence of diatom ooze with ash. Silty clay with ash underlies the diatom ooze, and this sequence rests on calcareous clay and siliceous nannofossil ooze. A few tens of meters of calcareous ooze and breccia overlies the igneous gabbro intrusives. The sedimentary section is just under 400 m thick. The sequence is sufficiently distinctive that most of it should be recognizable if present within the continental margin. Geochemistry of pore waters show gradients of Ca, K, and Si approaching those of seawater in the lowermost few tens of meters in this section, implying a source of seawater at the base of, or beneath, the sedimentary section. Relatively rapid communication between the seawater and the oceanic crust could provide a clear explanation for the very low values of heat flow determined by surface data and by downhole measurements on Leg 170.
Site 1040 was cored through the deformed sedimentary wedge, the décollement, and the underthrust sedimentary section beneath the décollement. Major discoveries were made in this site. First, the material making up the deformed sedimentary wedge does not resemble that of the incoming strata in Site 1039. If anything, these sediments more closely resemble those of the slope apron, cored at Site 1041 (discussed below). Secondly, the underthrust section beneath the décollement appears to be nearly intact. Only the top few meters of the turbidites seen at Site 1039 are missing, but this difference could be explained by winnowing of bottom currents, because a 3.5 kHz seismic record taken between the two sites shows the upper 5-8 m of sediment pinching out toward the trench axis. Third, the underthrust section at Site 1040 is reduced in thickness relative to that at Site 1039. The upper 180 m at Site 1039 are reduced to approximately 67% of original thickness at Site 1040 and the lower 200 m are reduced to about 80% of original. These are approximate numbers and the variations occur within these intervals. In addition, relatively steep dips noted in the cores from the highest 100 m of the underthrust section at Site 1040 could result in some thickening of that section over what it would be without the dips. Quantitative determinations of these differences must be carried out post-cruise. Finally, the geochemistry has documented two major fluid conduits. One is at a depth of about 170 mbsf within the deformed wedge and the second is associated with the décollement.
Two problems occurred during the drilling of Site 1040. First, many of the cores in the lower part of the deformed wedge underwent extreme torsion due to RCB coring of a ductile clay. The induced deformation made it extremely difficult to discern natural deformation fabrics in these cores. A second problem was that the LWD logging was not able to pass through the décollement, in spite of two efforts with different drill bits to carry it out. For these reasons, we developed a shipboard proposal to drill an additional site (Site 1043) through the wedge, décollement, and underthrust section at a location just 400 m landward of the toe of the margin. At Site 1043, the depth to the décollement is about 150 mbsf, making it significantly more accessible to both XCB coring and LWD.
Site 1043 penetrated the deformed wedge, décollement, and hemipelagic section. Coring stopped just above the contact with the nannofossil ooze. A complete set of LWD logs was acquired at this site. We obtained a less disturbed cored section above and through the décollement than at Site 1040, and the geochemistry data is of high quality at both sites. The LWD logs showed fine detail in the physical properties downhole, including several thin (1-2 m), low-density layers within the décollement zone that are very likely faults. Another fault zone ~80 m above the décollement is clearly marked as a low density zone in the logs, and both of these faulted regions are associated with chemical anomalies requiring fluid conduits. First examination of the combined dataset suggests that as much as 10 m may be missing from the top of the incoming section relative to that at Site 1039. A careful post-cruise comparison of Sites 1039, 1043, and 1040 will be necessary to determine how much material is removed from the lower plate during subduction. Very preliminary analyses of the LWD logs suggest that roughly 9 m of the section cored at Site 1039 may be missing from Site 1043 beneath the décollement. Whether all or only part of this 9 m has been accreted to the upper plate remains to be determined.
Site 1041 was designed to answer questions about the nature of the sediment apron and the composition of the upper part of the prism. We made two attempts to drill through the apron but did not succeed in getting much below 400 mbsf (high-amplitude top-of-prism reflection was at about 550 m there). Nonetheless, we succeeded in acquiring an excellent section through the apron, including a number of samples of gas hydrate. The apron is largely composed of silty clay with rare nannofossils and microfossils, very different from the pelagic oozes resting on the Cocos Plate at Site 1039. The inability to penetrate through to the prism reflector prompted adding yet another site originating on the ship. Site 1042 was located 7-km landward of the toe-of-slope at the seaward edge of what is still clearly the high-amplitude, top-of-prism reflection. At Site 1042, the reflecting surface lies at a depth of 310-320 mbsf. We spot-cored to 200 m but were unable to reach the reflecting surface on the first try. We then drilled to the surface with a larger diameter bit, then reentered the hole with the RCB bit. We cored successfully to 390 mbsf, initially penetrating the high-amplitude reflecting surface at 316 mbsf. The material was a carbonate-cemented breccia containing sandstone clasts, basalts, chert fragments, and cummulate gabbro or ultramafic rock. Many of the fragments were angular, suggesting little abrasion during deposition. The breccia matrix and some of the sandstone clasts were dated as middle Miocene. These breccias and their clasts resemble rocks known to occur onshore in Costa Rica, and the measured seismic velocity and density of the breccia can explain the seismic velocities and reflectivity measured for that surface. We feel reasonably confident that we penetrated the prism surface.
Finding material with affinities to onshore Costa Rica but not with the incoming Cocos Plate makes it very likely that much of the incoming sediment on the Cocos Plate bypasses the toe region of the continental margin. Seismic data show evidence of some underplating beneath the prism, but the rate of underplating is unknown at present. Paleodepth determinations on Deep Sea Drilling Project (DSDP) Leg 84 (Site 565) show evidence of uplift of the middle to lower slope region. Onland exposures attest to coastal uplift during the Pleistocene. Post-cruise science will address the question of how much underplating is possible. Nonetheless, much (from 90% to 99%) of the incoming sediment is bypassing the toe region through underthrusting.