The experimental results are given in Table 2. Figure 4 shows the results for Section 149-897C-34R-1, which is a typical example. During the undrained test, the traveltime initially decreases sharply with increasing effective pressure (Fig. 4A). This is typical behavior for a poorly consolidated sediment undergoing compaction at pressures below the maximum pressure to which it has been previously subjected. The rapid decrease in traveltime is attributed to compaction of the sample arising from collapse of pore space, which developed during decompaction of the sample following recovery (see discussions by Tschebotarioff, 1952; Lambe and Whitman, 1969; Das, 1983). The traveltime continues to decrease above 4.5 MPa, but at a decreasing gradient. This is interpreted as compaction at pressures greater than the highest pressure the sample has previously experienced. Post-recovery pore space has largely collapsed, and continued compaction requires grain re-arrangement and pore-fluid expulsion (Lambe and Whitman, 1969). The point of maximum curvature in the traveltime vs. pressure curve indicates the maximum pressure that the sample has previously experienced. This must occur somewhere between the second and third data points (at 5 MPa and 10 MPa, respectively) in this experiment, so the effective pressure is taken to be 7.5 MPa ± 2.5 MPa. Assuming no great thickness of sediment has been eroded from the overlying strata, this is the in situ effective pressure Pe'. Above 26 MPa, the traveltime curve becomes nearly linear. This is interpreted to indicate nearly complete closure of post-recovery pore space. Further compaction requires fracture and reorientation of the sample grains (e.g., Lambe and Whitman, 1969; Das, 1983). Following the measurement at 50 MPa, the confining pressure and pore pressure were reduced incrementally until the effective pressure reached 6 MPa. The traveltime increases during this sequence of measurements, but only slightly (see the lower traveltime leg in Fig. 4A). This "hysteresis curve" is typical of sediments, and it indicates that the sample has been compressed beyond its most compact previous state during the experiment, resulting in irreversible grain rearrangement (Das, 1983). The measurement taken at an effective pressure of 6 MPa had a confining pressure of 10 MPa and a pore pressure of 4 MPa. After this measurement was taken the pore pressure was released, allowing the effective pressure to rise to 11 MPa (equal to the confining pressure). The drained test was then conducted at confining pressures up to 75 MPa. The traveltimes during this sequence of measurements were generally close to those obtained during the undrained test, indicating that velocity changes are mostly due to changes in the effective pressure, as expected.
The variation of velocity with pressure for Section 149-897C-34R-1 is shown in Figure 4B. The "uncorrected" curve is the velocity calculated by assuming that no shortening of the sample occurred during the experiment. The "corrected" curve is determined by interpolating for the length of the sample at the pressure at each measurement point as discussed previously. The corrected curve shows no increase in velocity above 50 MPa, suggesting that this sample underwent little rebound as the pressure was lowered. If it had undergone rebound, the length of the sample at the highest pressure would be less than the length measured after the experiment, and the true velocity at high pressure would be less than that shown in Figure 4B. This would require the velocity curve to decrease at high pressure. This is unlikely, so it is inferred that the length of the sample measured after the experiment represents the sample length at high pressure.
The velocity curve mimics the traveltime curve, showing a dramatic increase in velocity with increasing pressure below Pe', maximum curvature at Pe', and a decrease in slope above Pe'. The velocity at Pe' is the expected in situ velocity, and it is picked as the intersection of straight lines fitting the high-pressure and low-pressure portions of the velocity curve. The nearly flat velocity curve above ~25 MPa suggests that the lithology represented by Section 149-897C-34R-1 will have a maximum velocity of not much greater than 3375 m s-1 at depths where the effective pressure is greater than 25 MPa. This velocity is similar to that determined for deeper portions of the sedimentary sequence on the Iberia Abyssal Plain determined from refraction surveys (Whitmarsh et al., 1990).
The corrected velocity curves for the other samples are shown in Figure 5. Not all of the samples have velocity curves with Pe' and Vmax as clearly defined as in Section 149-897C-34R-1. For example, Section 149-898A-20X-3 continues to show a moderate increase in velocity at high pressure and the point of maximum curvature is less obvious (Fig. 5C). In this case, Pe' has relatively large error bars associated with it. The continued increase in velocity at high pressure may indicate either that:
1. the sediment is more compressible at high pressure;
2. the sample had abnormally high in situ pore pressure, so significant porosity and permeability existed. As a result, consolidation and the associated steep velocity increase continues at pressures above the highest pressure previously experienced; or
3. the sample underwent some rebound before it was removed from the pressure vessel, and so the velocity at high pressure is overestimated. It is not possible to differentiate between these three possibilities with the available data, so Vmax is poorly constrained.
Several experimental problems that are worth further comment are evident in the curves in Figure 5. For Section 149-898A-13H-5 (Fig. 5B) the low-velocity leg of the curve indicates measurements taken during the undrained test. The high-velocity leg of the curve indicates measurements taken during the drained test. It was discovered after the undrained experiment that the pore-pressure outlet had become clogged by clay, preventing expulsion of pore fluid from the sample during the latter stages of the undrained experiment. As a result, pore pressures were high, and the effective pressure on the low-velocity leg was much lower than indicated on the figure. The low-velocity leg of Figure 5B is erroneous and should not be considered. The pore-fluid outlet was cleared before running the drained test (high-velocity leg), so the high-velocity leg of the curve is considered accurate. On Section 149-899B-1R-2 (Fig. 5D), a leak in the hydraulic-fluid reservoir developed before measurement of the last data point (P = 12 MPa, V = 2721 m s-1). The last data point is erroneous, and the experiment was terminated after this measurement (as a result, there is no undrained series of measurements on this sample).
The estimated in situ velocities and effective pressures determined from Figure 5 are given in Table 3. The in situ velocity generally increases monotonically with increasing effective pressure, consistent with a decrease in porosity and increasing sediment compaction during burial. Plotted as a function of depth, the estimated Pe' and velocity data show a reasonably uniform trend between all four sites (Fig. 6). Note that the estimated velocity for Section 149-899B-1R-2 at 232 mbsf falls off the trend of the other data and has large error bars associated with it. This data point is considered suspect because poor transducer/sample coupling prevented measurement of the velocity at low pressures. The first data point obtained for this sample was at an effective pressure of 6 MPa, much higher than the lithostatic pressure at this depth. The uniform trend in velocity as a function of depth for the other samples is not surprising for Sites 897, 898, and 899. The shipboard velocity measurements showed a consistent linear increase in velocity with depth of burial in the sediments, with a similar slope observed in data from all three sites (Fig. 1; Sawyer, Whitmarsh, Klaus, et al., 1994). Two aspects of the data shown in Figure 6 were unanticipated, however. First, the data point for Site 900 falls within the trend of the data from the other three sites. The shipboard velocity measurements for Site 900 show a markedly different slope when plotted as a function of depth of burial than observed at Sites 897, 898, and 899. This difference is not apparent in the velocity measurements taken under pressure. One explanation of the data observed shipboard is that the sediments at Sites 897, 898, and 899 have a smaller compressibility coefficient than the more proximal sediments at Site 900. If so, sediments recovered from Site 900 would have undergone less rebound following recovery, resulting in the greater velocity vs. depth trend, which was observed at Site 900. Alternatively, pore pressure may be lower at Site 900 than at comparable depths at the other three sites, resulting in a higher effective pressure, more compaction during burial, and a higher in situ velocity gradient. The data from the shore-based experiments suggest that the former explanation is more likely. Apparently, the in situ velocities of sediments at all sites are comparable at similar depths. The higher velocity gradient measured aboard ship may thus be an artifact of differing amounts of decompaction following recovery, and it does not necessarily indicate different in situ velocity structure.
The estimated in situ velocity trend is in reasonable agreement with the velocity structure estimated from sonobuoy data (Whitmarsh et al., 1990), although the laboratory data appear to consistently exceed the seismically determined velocities (Fig. 6A). This may partially be a result of the constant velocity layer model determined from traveltime modeling of the seismic data, but it probably also reflects the large uncertainties involved in estimating the in situ effective pressure from the laboratory data. If the in situ effective pressure was systematically overestimated, the in situ velocities determined from the laboratory data would similarly be overestimated. The laboratory data are more consistent with the sonobuoy data if a linear increase in velocity is adopted in the upper acoustic layer (dotted line, Fig. 6A). A similar velocity gradient was observed in the shipboard velocity measurements in the upper 400 mbsf (Fig. 1). However, the sonobuoy data provide remarkably accurate estimates of the depth to major lithostratigraphic boundaries encountered during Leg 149. The in situ velocity structure must therefore not be greatly different than that interpreted by Whitmarsh et al. (1990) on the basis of the sonobuoy data, although traveltime modeling of the sonobuoy data does permit modest gradients in the velocity structure. Furthermore, the linear velocity gradient shown by the dotted line in Figure 6A does not result in a velocity discontinuity between acoustic Layers 1 and 2. The presence of a reflection in the sonobuoy data at the base of acoustic Layer 1 requires such a discontinuity (Whitmarsh et al., 1990). There is some evidence in the laboratory data for a rapid increase in velocity over a small depth interval, which may produce the reflection, although the laboratory data would place the change in velocity at a depth of about 335 to 370 mbsf, rather than 420 mbsf as indicated on the sonobuoy profile (Fig. 6A). This is not surprising, as the sonobuoy line and the various drill sites used in this study were distributed over a ~100 km distance. In this latter interpretation of the laboratory results (dashed line, Fig. 6A), acoustic Layer 1 is interpreted to show a modest increase in velocity to a depth of about 320 mbsf. A strong velocity gradient between the data points collected at 336 mbsf and 369 mbsf is interpreted to bound the layer that produces the reflection. This velocity gradient is somewhat apparent in the shipboard data (Fig. 1). The dashed line in Figure 6A is the preferred interpretation of the laboratory data. Nonvertically incident seismic rays will probably travel at velocities intermediate between the velocities given in the constant velocity model derived from the sonobuoy data and the linear velocity profile (derived from vertically incident data) shown by the dashed line in Figure 6A.
The second unanticipated aspect of the data shown in Figure 6 is the apparent increase in the in situ effective pressure from 2-3 MPa above about 200 mbsf to about 6 MPa below 200 mbsf. This is interpreted to indicate a transition from poorly consolidated sediments above 200 mbsf to consolidated sediments below 200 mbsf. Above 200 mbsf, grains are inferred to be in loose contact, allowing free pore-fluid migration. As a result, pore fluid is expelled with increasing depth of burial and effective pressure increases. Below 200 mbsf, grains are inferred to be in close contact and pore-fluid migration is restricted. Below this depth, the increase in confining pressure that results from the increasing depth of burial is largely balanced by an increase in pore-fluid pressure. As a result, effective pressure (and velocity) show little increase below this depth.