DISCUSSION AND CONCLUSIONS

Challenges of Drilling in Overpressured Basins

Expedition 308 was the first expedition in IODP/ODP/DSDP history during which large volumes of weighted mud were used as part of the experimental design to achieve the scientific goals of the expedition (Fig. F19). Prior to the expedition, it was determined that there was a significant probability of encountering shallow-water flow in the subsurface. Shallow-water flow occurs when overpressured and unconsolidated sands flow into the borehole, which is at a lower pressure. Ultimately, these sands can be expelled at the seafloor.

To counter this behavior, weighted muds were used during drilling. The weighted mud is pumped down to the seafloor and ultimately expelled at the seafloor (Fig. F20). The effect of the weighted mud is to create a higher pressure within the annulus, which will offset the overpressure within the formation. During Expedition 308, a weighted mud composed of barite, sepiolite, and seawater was used at Sites U1323 and U1324, two of the sites where shallow-water flow was considered a significant risk. In these locations, 10.5 ppg mud was generally used. Shallow-water flow was, in fact, encountered in Hole U1323A (see "Operations") at ~200 mbsf while drilling with seawater (Fig. F21). An abrupt increase in pressure associated with a thin sand (indicated in the gamma ray log) was identified. After raising the mud weight in the borehole, we drilled forward some distance before ultimately killing the flow with a 13.5 ppg mud, whereupon we cemented the hole with 14 ppg mud. Table T1 illustrates the approximate volumes, durations, and depth ranges that mud was used.

In addition to the need to drill the borehole safely, weighted mud was also used to stabilize the hole for long-term deployments of penetrometers (DVTPP and T2P). In this case, the weighted mud was extremely useful for keeping the hole open. Because the formation is relatively plastic, over time it has a tendency to close on the drill string. Thus, there is a danger when one is not rotating or circulating fluid that the hole will close on the drill string. In Ursa Basin, this began to happen when seawater was in the hole at depths below 50 mbsf. However, when a 10.5 ppg weighted mud was used, this problem was greatly reduced. This allowed us to make prolonged penetrometer measurements. In these cases, the borehole stability was striking. We ran some DVTPP deployments for >1.5 h and experienced no problems with borehole closure. These deployments were far longer than previous deployments of downhole tools in IODP/ODP history. Typically it took 40–70 bbl of weighted mud to be spotted in the hole during a penetration deployment.

A critical goal of Expedition 308 was to measure pressure within mudstones using a penetrometer. Two tools were used for this task: the DVTPP and the T2P (Fig. F22). The DVTPP was deployed previously during ODP Legs 190, 201, and 204 (Moore et al., 2001; D'Hondt, Jørgensen, Miller, et al., 2003; Tréhu, Bohrmann, Rack, Torres, et al., 2003). The T2P is a new tool under development as a cooperative effort between MIT, Pennsylvania State University, and IODP-TAMU. The DVTPP and the T2P are similar devices. The primary difference between the two tools is that the T2P has a 6 mm diameter tip, whereas the DVTPP has a ~23 mm diameter tip that rapidly widens backwards from the tip. The T2P was designed both to dissipate more rapidly and to dissipate with a characteristic pressure profile. Both properties allow the T2P to be deployed for shorter periods than the DVTPP in order to interpret in situ pressures.

These penetration tools induce a pressure pulse as they are inserted into the sediment. The initial pressure response and its decay are defined by the insertion rate of the probe, the modulus of the sediment, and the bulk permeability of the sediment. The pressure dissipation that results after penetration is used to infer in situ pressure and rock properties (Fig. F23A).

Pressure measurements during Expedition 308 were extremely challenging. There were significant successes that allowed us to define the pressure gradient in Ursa Basin. There were a total of 25 T2P deployments and 20 DVTPP deployments (Fig. F24). Of these only 56% of the T2P and 45% of the DVTPP deployments were either fair or good and there were many poor or unsuccessful deployments (Fig. F23). Key problems were threefold. First, in early cases, there was a leak in one of the DVTPP tools and there may have been a leak in one of the T2P tools (e.g., Fig. F23C). The leak resulted in abrupt pressure drops below hydrostatic pressure during the dissipation phase. Second, the T2P was prone to bending due to the very narrow diameter tip. Third, both tools had difficulty remaining coupled to the sediments during deployment in the shallow sections (Figs. F23B, F25). During many of the deployments, after the drill string was raised and subsequent to penetration, there was an abrupt drop in pore pressure. In these cases, there was often a frictional heating pulse associated with the drill string being raised. In addition, the accelerometer in the DVTP suggests movement of the tool when the drill string was raised. These results are interpreted to record the partial dislodgment of the tool due to friction in the colleted delivery system (e.g., Fig. F25). Review of DVTPP records from Legs 201 and 204 suggest that both leakages and tool dislodgement during elevation of the drill string have been a persistent problem with the DVTPP (D'Hondt, Jørgensen, Miller, et al., 2003; Tréhu, Bohrmann, Rack, Torres, et al., 2003).

Summary of Principal Results

Brazos-Trinity Basin #4 contains a 175 m thick succession of sand-rich turbidite fans, mass transport deposits, and hemipelagic deposits laid down within the last ~122 k.y. Prefan deposits dating back to MIS 6 form a conformable succession of laminated and bioturbated clays deposited from distal turbidity currents and/or river plumes. The initial turbidite deposits in the basin are mud rich, with the exception of the very first turbidity currents to enter the basin. This initial pulse, possibly derived from failure of older shelf edge deposits, accumulated an ~8 m thick sand-rich interval. A basin-wide pause in turbidity current influx spans 30–40 k.y. between late MIS 5 and MIS 4/3. During MIS 3–2, a 130 m thick succession of sand-rich fans accumulated in Brazos-Trinity Basin #4, containing 2–25 m thick packets of very fine to lower medium sand beds. A 2–3 m microfossil-rich clay marks the end of turbidity current influx into the basin during the Holocene.

Brazos-Trinity Basin #4 is the terminal basin of a series of bowl-shaped minibasins on the upper–middle continental slope that are linked to late Pleistocene shelf edge deltas by a network of submarine channels offshore Texas (Fig. F4). The fluvial, shelf, and slope portions of the Brazos-Trinity drainage system represent a natural laboratory to investigate the mechanics of sediment transfer across a continental margin from source to sink. Originally studied with sparse seismic records and piston cores (Gardiner, 1986; Satterfield, 1988; Satterfield and Behrens, 1990), the basin has been the subject of numerous detailed studies by the industry and academia (Winker, 1996; Beaubouef and Friedman, 2000; Pirmez et al., 2000; Badalini et al., 2000; Beaubouef et al., 2003; Mallarino et al., in press). Industry studies have focused particularly on the stratigraphic architecture of the intraslope basins because of their similarity with deepwater reservoirs formed on continental margins with a mobile (salt or shale) substrate. The shallow burial depths of such near-seafloor analogs allow for exceptional vertical and spatial resolution through the use of very high frequency (>150 Hz), short-offset seismic profiles. Studies to date have focused on the mapping of sequences using both 2-D grids and 3-D seismic data, resulting in the development of stratigraphic models derived from the interpretation of seismic geometry, acoustic facies, and seismic attributes (Winker, 1996; Badalini et al., 2000; Beaubouef et al., 2003).

Lithologic calibration of these basin-fill models has been limited to short piston cores (e.g., Satterfield and Behrens, 1990). More recently, Mallarino et al. (in press) report on a series of long piston cores in Brazos-Trinity Basin #4, but the recovery was limited to the upper ~5 to ~25 m in the sandy basin fill, although successful cores up to 42 m long were obtained in the basin margins. The basin margin cores enabled Mallarino et al. (in press) to develop a high-resolution chronostratigraphy for the upper part of the basin fill, dating to ~90 ka.

Expedition 308 successfully acquired cores and a full suite of downhole logs that, for the first time, sample the entire infill of Brazos-Trinity Basin #4 and underlying conformable succession along a transect from the basin center to the basin margin (Fig. F6). This unique data set provides a detailed characterization of the sediment gravity flow deposits and hemipelagic successions within the basin fill and allows for the age dating of the various fan sequences.

The Brazos-Trinity Basin #4 transect was drilled on a dip-oriented seismic line from the basin entry point to the east and across the thickest portion of the basin fill (Fig. F26). The short-offset multichannel seismic line has a frequency content in the range of 100–500 Hz with peak frequency at ~300 Hz (Winker, 1996). Reflector R40 separates two distinct units: a conformable succession of subparallel reflections below and an onlapping succession representing the infill of Brazos-Trinity Basin #4 above. The underlying succession appears to thicken gradually toward the northern basin margin. The basin fill succession displays an alternation of acoustically transparent intervals with packages displaying high-amplitude continuous to semicontinuous reflections. Previous workers interpreted these seismic facies as the result of the alternation between muddy mass transport complexes (transparent intervals) and sandy turbidite fans (bedded intervals) (Winker, 1996; Beaubouef and Friedmann, 2000; Badalini et al., 2000).

Well logs from Sites U1319, U1320, and U1321 are posted on the dip seismic line in Figure F26. The integration between logging-core and high-resolution seismic makes it possible to confidently correlate stratigraphic units up to a few meters thick and to estimate the lateral extent of individual mud and sand packets away from the boreholes (Fig. F27). The match between lithostratigraphic boundaries and most key reflectors in the basin using the preliminary shipboard results is quite good (Fig. F26, F27).

Seismic Reflector R40 represents an angular unconformity at the base of the basin infill and can be traced continuously across the entire area. Seismic Reflector R30 marks an angular unconformity within the basin fill. A thin interval of parallel reflections forms a laterally continuous marker locally disrupted by an acoustically transparent/chaotic unit. Seismic Reflectors R10 and R20 are continuous reflections that separate acoustically transparent/chaotic intervals from intervals of high acoustic amplitude intervals with reflections ranging from laterally continuous to discontinuous. Winker (1996) showed that sediments between Reflectors R40 and R10 are sourced from the western feeder channel and sediments above R10 were sourced from the eastern feeder channel.

A preliminary age model was developed on board by integrating microfossil biostratigraphy together with correlation of magnetic susceptibility and NRM intensity data with global curves (Fig. F28). The chronostratigraphy developed by Mallarino et al. (in press) provided additional constraints for the age model developed onboard. Tephra event Y8 (Drexler et al., 1980, Mallarino et al., in press) occurs in the drilled section, providing an absolute correlation and age marker. The entire drilled succession appears to be younger than 150 ka since the last occurrence (LO) datum of Helicosphaera inversa was not observed. The base of the basin infill occurs near the W/X planktonic foraminifer zone boundary, at ~122 ka, and coincides with a ~2 m thick condensed hemipelagic interval. A pause in turbidity current influx also occurs within the basin fill and spans a ~40 k.y. period from ~90 to ~50 ka. The LO of Globorotalia flexuosa at ~68 ka is observed at both Sites U1319 and U1320, but based on seismic correlation and the occurrence of G. flexuosa within the turbidite infill section at Site U1320, it is possibly the result of reworking by turbidity currents.

Integration of the lithostratigraphy, biostratigraphy, logging characteristics, and seismic stratigraphic correlations are summarized in a structural cross section (Fig. F29). Examples of core photographs of the lithofacies encountered in each of the Brazos-Trinity Basin #4 sequences are illustrated in Figure F30.

The following summarizes the basin stratigraphic architecture and evolution, from oldest to youngest:

All three sites penetrated a succession of laterally extensive subparallel reflectors below seismic Reflector R40 at the base of the basin infill. This succession is composed of terrigenous laminated clay with color banding between grayish green and reddish brown with varying degrees of bioturbation (Fig. F30E). These clays have a low TOC (average = 0.5 wt%) but are rich in CaCO₃ (average = 23 wt%), most of which is associated with fine-grained detrital carbonate and dolomite. The unit is devoid of sand and contains very rare silt lamina.

The 125 m thick prefan deposits cored at Sites U1319 and U1320 are interpreted to be younger than 150 ka and were deposited at an average rate of >4 m/k.y. This succession is interpreted to represent deposition from distal turbidity currents overspilling from basins adjacent to Brazos-Trinity Basin #4 (laterally and/or updip), possibly with a significant contribution from sedimentation from surface plumes of coastal rivers. It is unclear at this stage whether these slope deposits contain a record of spillover from updip basins or whether they represent distal turbidity currents from adjacent basins at a time when shelf edge deltas of the Brazos-Trinity drainage system were located to the east and/or west of Basin #1.

Capping the prefan deposits there is a ~2 m thick intensely bioturbated microfossil-bearing clay recovered at Site U1319 (Fig. F30D). At Site U1320, the same interval occurs within a zone of poor core recovery. This condensed sedimentation interval marks a dramatic reduction of terrigenous sediment flux to the basin and occurs at the base of the onlap fill of Brazos-Trinity Basin #4. We interpret this condensed interval to represent the sea level highstand during MIS 5e. Seismic correlation of the regional seismic Reflector R40 between the two sites indicates that this condensed interval represents the base of Brazos-Trinity Basin #4 infill.

The lower fan unit in Brazos-Trinity Basin #4 thins very gradually onto the basin margin, from 30 m at Site U1320 to ~10 m at Sites U1321 and U1319. The sequence is dominated by laminated and bioturbated muds with thin beds of silt and sand (Fig. F30C). Sand content decreases from ~23% at Site U1320 to only a few percent at Site U1319. Most of the sand occurs in a ~8 m thick interval of poor recovery at the base of the lower fan at Site U1320, as interpreted from the well logs. This basal sand displays a significant lateral variation in thickness along the 20 m separating Holes U1320A and U1320B and probably represents the infill of an erosional scour or channel. Excluding this basal sand, the lower fan has a sand content of only 8%.

The mud-rich lower fan was deposited between ~120 and ~90 ka. The first pulse of sandy turbidity currents appears to have occurred within MIS 5e or at the rapid sea level fall event marking the MIS 5e–5d transition. At that time, sea level was still higher than today and the coastline was far landward of the modern shelf edge. The source of this initial pulse of turbidites is inferred to result from remobilization of shelf edge sediments deposited from previous sea level lowstands or from submarine failures in the updip basins.

The lower fan is capped by an ~8 m thick interval of foraminifer- and nannofossil- bearing clay displaying intense bioturbation, similar to the hemipelagic drape at the base of the lower fan (Fig. F30B). Near the base of this hemipelagic drape, a 2 cm thick layer of volcanic glass shards is observed at both Sites U1319 and U1320. This ash layer provides an independent correlation and age marker confirmed by the Emiliana huxleyi acme in close proximity to the ash. This ash layer is interpreted as tephra event Y8, representing the outfall from the Los Chocoyos (Guatemala) eruption at 84 ka (Drexler et al., 1980; Mallarino et al., in press). Seismic Reflector R30 occurs at the top of this interval and can be laterally traced over most of the basin.

The duration of this basin-wide pause in sediment flux to Brazos-Trinity Basin #4 cannot be determined precisely without additional shore-based analyses of the material recovered, particularly magnetic and isotope stratigraphies. Mallarino et al. (in press) indicate that the pause in turbidite sedimentation may have lasted into early MIS 3, but earlier MIS 4 turbidites may not have reached the basin margin core site that Mallarino et al. (in press) studied. The paucity of terrigenous input into the basin during this interval, despite significant oscillations in sea level, is interpreted to indicate that either fluvio-deltaic input was directed away from the head of the system near Basin #1 or that turbidity currents were completely trapped in updip Basins #1 and #2.

The middle fan comprises a succession of sediment gravity flow deposits including massive and normally graded fine to lower medium sand beds up to several meters thick interbedded with thinly laminated muds and silts.

The middle fan thins from ~110 m at Site U1320 to ~50 m at Site U1321 and to ~12 m at Site U1319, illustrating the strong effect of basin margin topography on sand distribution. The lower part of the middle fan is represented by a very thin interval at Sites U1321 and U1319, whereas the upper part of the fan succession reaches higher onto the basin margin. At the base of the middle fan, an acoustically transparent interval corresponds to a slump/debris flow deposit containing contorted beds and mud clasts. This slump unit originates from the basin margin to the east (Winker, 1996).

Average sand content of the middle fan unit is ~40% at Site U1320, ~80% at Site U1321, and only minor amounts of sand were recorded at the basin margin Site U1319. Sand distribution within the middle fan, however, is quite variable both vertically and spatially. At Site U1320, sand beds are organized in 2–10 m thick packets with an overall increase in sand content upward, particularly above seismic Reflector R20. The sand packets are capped by muds and intervals of laminated mud with thin bedded silts and sands. Some of these muddy intervals can be correlated across the transect, whereas others are either eroded away or pinch out between boreholes (Fig. F29).

The upper portion of the middle fan, above seismic Reflector R20, is the sandiest interval of the entire basin fill with amalgamated sand beds forming a 25 m thick unit at Site U1321 (Fig. F29). This interval is also remarkably transparent on seismic data reflection profiles (Fig. F26). At Site U1320, a sharp contact between a thick-bedded sand packet above a laminated, partly contorted interval of mud with thin beds of silt and sand correlates with a subtle seismic reflection. At Site U1321, this acoustically transparent interval is composed of ~100% sand. Subtle reflections on the seismic lines suggest that the basal mud unit at Site U1320 is most likely eroded by channels and scours and is completely absent at Site U1321 (Figs. F26, F27).

Previous basin models derived from seismic facies analyses interpreted this transparent unit as muddy mass transport deposits (e.g., Winker, 1996; Badalini et al., 2000; Beaubouef and Friedmann, 2000; Beaubouef et al., 2003). Detailed logging-seismic-physical property analyses on shore are needed to unravel the low impedance contrasts within these shallowly buried muds and sands.

Capping the middle fan is a ~10 m thick organic-rich homogeneous dark green to black clay with a sharp base and top. This unit thins onto the basin margin but appears to extend across the entire basin fill (Figs. F26, F27). It is interpreted to represent a debris flow deposit.

The upper fan comprises a sand-rich unit forming a tapered wedge across the basin. At Site U1320 it is ~25 m thick and contains thick and medium beds of fine and very fine sand. The upper fan thins to ~18 m at Site U1321 and has similar logging characteristics but a slightly lower sand content than at Site U1320. The correlative unit at Site U1319 is only ~3 m thick and is mostly mud with some thin beds of silt and sand. Sand beds in the upper fan are organized in bed packets ranging in thickness between 2 and 8 m capped by intervals of mud with thin beds. Correlation to seismic profiles (Fig. F27) shows that the sand packets represent mounded seismic bodies with internally discontinuous reflections representing fan lobes extending laterally for several kilometers. Beaubouef et al. (2003) show spectacular high-resolution 3-D images from these channelized fan lobes. The upper fan is capped by a microfossil-rich clay, indicating that turbidity current deposition ceased in the basin during the last sea level rise.

The upper and middle fans were deposited between ~47–60 and ~10 ka and represent the main pulse of turbidity current influx into Brazos-Trinity Basin #4. This implies an average accumulation rate between 2.5 and 3.5 m/k.y. for the 130 m thick succession at Site U1320, compared with an average accumulation rate of ~0.2 m/k.y. at the basin margin Site U1319. Terrigenous influx was reduced dramatically throughout Brazos-Trinity Basin #4 during the Holocene as indicated by the presence of a microfossil-rich clay in the upper 2–3 m of the sediment column at both Brazos-Trinity Basin #4 sites.

Sedimentation in Brazos-Trinity Basin #4 is the result of a complex interaction between fluvio-deltaic dynamics, sea level changes, and interactions between turbidity currents and submarine topography. During the low sea level period corresponding to MIS 6, the basin received significant input of terrigenous sediments, but a complete absence of sand and silt indicates that either turbidity currents were filling basins updip or that deltaic systems were positioned in areas adjacent to the Brazos-Trinity slope at the time.

During the stepwise sea level fall between MIS 5e and 2, the basin received as much as 175 m of sediment gravity flow deposits comprising turbidites, slumps, and debris flows. A pause in turbidite deposition occurred from MIS 5a to ~MIS 4. This period comprised both a relative rise and a relative fall in sea level. Therefore the lack of turbidite influx into Brazos-Trinity Basin #4 must be the result of factors other than sea level changes, including lateral shifts in the sediment source on the shelf or trapping of sediments in Basins #1 or #2 updip preventing the spillover of turbidity currents into Brazos-Trinity Basin #4.

Seismic facies-based interpretations of the basin fill architecture by previous authors are often contradictory, and our results provide the needed calibration to validate these interpretations. Seismic intervals that show abrupt onlap onto the basin margins generally correspond to sand-rich turbidites (e.g., middle fan above Reflector R30), whereas those seismic units that have a gradual thinning pattern onto the basin margin tend to have lower sand content (e.g., lower fan between Reflector R40 and R30). The acoustically transparent units observed in the basin fill are composed of very high sand content, ranging from 50% to ~100%, contradicting previous interpretations (e.g., Badalini et al., 2000; Beaubouef and Friedman, 2000; Beaubouef et al., 2003).

Mud-rich turbidite intervals appear to extend laterally for significant distances across the basin and rise as much as ~50 m above the basin floor. This could result from turbidity current run-up onto the basin margins or represent a measure of the thickness of turbidity currents entering the basin. However, some of the relief, perhaps as much as 20 m, appears to result from subsidence at the basin center since the onset of basin filling.

Overall, the sand content in the basin fill increases upward with the lowest sand content observed in the lowermost fan (~23%), but there is significant spatial variability in the profiles. The highest sand content is encountered in the southernmost edge of the basin at Site U1321, in the upper part of the middle fan. The sand content of the upper fan is high (~60%–70%), but not as high as the upper part of the middle fan (77% to ~100%). Although the general increase in sand content could be interpreted as the result of progressive sea level fall and advance of the Brazos-Trinity fluvial systems toward the shelf edge (e.g., Mallarino et al., in press), it is clear that lateral shifts in deltaic depocenters and trapping of sands in the updip basins also have an important influence in sand influx into Brazos-Trinity Basin #4.

Drilling at Sites U1322, U1323, and U1324 investigated a sedimentary wedge flanking a buried submarine channel of the modern Mississippi Fan, the Southwest Pass Canyon channel-levee system (Fig. F31). Of the two cored sites, Site U1324 penetrated the thickest part of the east levee. Site U1322, 12 km east, penetrated the thinner portion of the levee. The core descriptions and age results of these two sites, along with the lithostratigraphic column interpreted from the LWD data at Site U1323, provide the basis to evaluate the diverse sedimentary history of this system.

Nannofossils and planktonic foraminifers indicate that the sediment sequence recovered at Sites U1322 and U1324 was mainly deposited over the last 60 k.y. A good correlation can be made between the two sites using faunal assemblage zones and species datums (Fig. F32).

Sediment sections corresponding to lithostratigraphic Subunits Ib (lower part) to Id at both sites (30–130 mbsf at Site U1322 and 30–170 mbsf at Site U1324), between seismic Reflectors S10 and S30, are dated by planktonic foraminifers belonging to MIS 2 (10–25 ka). The age control points are rare from the lower part of Holes U1322B and U1324B. However, fairly reliable planktonic foraminifer data indicate an MIS 3 age for sediments recovered from 130 to 190 mbsf at Site U1322 and from 170 to 329 mbsf at Site U1324. Farther downhole, biostratigraphic constraints are inadequate to correlate between the two sites. The nonoccurrence of some datums older than 65 ka provides negative evidence that the sediment from the lowermost part of Sites U1322 and U1324 is younger than 65 ka. The rarity of microfossils in the lower part of the drilled section provides little specific information regarding species datums but does suggest that the sediment interval was deposited during sea level lowstands belonging to MIS 4 (~60–65 ka). Results from previous studies, including Winker and Booth (2000), indicate that the last occurrence datum of G. flexuosa (68 ka) lies at the base of the Blue Unit, which is ~200 m deeper than the bottom of either Hole U1322B or Hole U1324B. Therefore, we believe that the sediment sequence recovered at both sites in Ursa Basin was deposited in the last 60 k.y., during MIS 1–4.

The upper 160 m of Site U1324 and the upper 135 m of Site U1322 are remarkably similar. They consist of mud, clay, and two mass transport deposits (Fig. F31). The base of these intervals ties closely with the regional seismic Reflector S30 and is dated at 24 ka. Sedimentation rates during the accumulation of these correlative intervals at both Sites U1324 and U1322 were similar, ranging from 10 to 12 m/k.y., and declining to lower rates in the uppermost Holocene (Fig. F32).

The base of the cored sections of Sites U1324 and U1322 tie closely with seismic Reflector S60 (Fig. F33), which we believe to be ~60 ka based on shipboard biostratigraphic and magnetostratigraphic analyses (Fig. F32). S60 is a reflector within the Ursa Canyon east and west levee and therefore should have been deposited approximately synchronously.

Although the upper portions of Sites U1322 and U1324 are similar, sections below show lithologic contrasts. Between seismic Reflectors S60 and S40-1324 at Site U1324, lithostratigraphic Unit II contains interbedded sand, silt, and mud, representing relatively unconfined flow turbidites deposited by a developing channel system. Above seismic Reflector S60 at Site U1322, lithostratigraphic Unit II contains a series of stacked clay- and mud-rich mass transport deposits with a marked paucity of sand and silt. Apparently, the Ursa Canyon channel-levee system acted as a barrier that confined sand and silt at Site U1324 and effectively shielded Site U1322 from deposition. The Ursa Canyon system would have had considerable topographic relief on the seafloor which would have confined sands, silt, and mud to the west during the accumulation of lithostratigraphic Unit II at Site U1324. Many seismic reflectors within the upper part of lithostratigraphic Unit II climb to the east up and over the top of the Ursa Canyon channel-levee system; therefore, we interpret there was very little accumulation at Site U1322 before sediment breached the height of the Ursa Canyon system.

During Expedition 308, we examined numerous cores containing mass transport deposits. Inclined bedding planes, folds, and faults commonly characterize mass transport deposits (Fig. F34). Mass transport deposits are recognized seismically by their discontinuous to chaotic reflections of transparent to variable amplitude (Fig. F34). Logging curves typically show positive anomalies in density and resistivity at the top of, and within, mass transport deposits (Fig. F34). The resistivity-at-the-bit logging images reveal striking folds and faults associated with mass transport deposits and suggest north to south sediment transport.

The seismic cross section and the contrasting lithostratigraphic columns at Sites U1322 and U1324 suggest mass transport deposition was more active earlier to the east and progressively migrated upsection and to the west. The more rapid accumulation of sediments in the basal portion of Site U1324 relative to Site U1322 may have created a thickness imbalance that could have expelled pore fluid to the east, initiating failure. The overpressured sands that created operational problems while drilling Site U1323 may represent a fingerprint of this process of lateral expulsion of fluids.

Analysis of bulk density and porosity profiles suggests overpressure in the lower lithostratigraphic units near the depocenter of Brazos-Trinity Basin #4 (Site U1320) and throughout Ursa Basin (Sites U1322, U1323 and U1324). The magnitude of the overpressure is here defined as

^* = (P – P_h)/(_v – P_h), (1)

where

P = pore pressure,
P_h = hydrostatic pressure, and
_v = total vertical stress.

Brazos-Trinity Basin #4

Lithostratigraphic Unit V at Site U1320 and lithostratigraphic Unit VI at Site U1319 (both below Reflector R40) are equivalent stratigraphic sections that have been subject to different burial histories. Above Reflector R40 at Site U1320, there is 178 m of overburden represented by the ponded infill of the Brazos-Trinity Basin #4. In contrast, at Site U1319, located on the basin's flank, there is only a few meters of overburden. Immediately below Reflector R40, porosities are ~65% at Site U1319 and ~50% at Site U1320. At the bottom of the holes at both sites porosities are ~40%, although the overburden is higher at Site U1320 than at Site U1319 (Fig. F35).

Porosity () relates to void ratio (e) by the formula

e = /(100 – ). (2)

The pore pressure at Site U1320 is estimated from the differences in void ratio with respect to the hydrostatic effective stress (Fig. F36) using the formula

P = _v – 10^{(e0 – e)/Cc}, (3)

where

P = pore pressure predicted from the void ratio,
_v = total vertical stress,
e = measured void ratio,
e₀ = reference void ratio (3.49), and
C_c = compression index (0.89).

Reference void ratio and compression index constants are derived from the curve fit of the data from Site U1319 in Figure F36. This approach is derived from standard geotechnical practice (e.g., Lamb and Whitman, 1969). We predict the difference in pore pressures between Sites U1319 and U1320 based on the assumption that both locations have similar stress-strain properties, that sediments at Site U1319 are normally pressured, and that sediments at both locations are normally consolidated (i.e., their in situ effective stress is their maximum past effective stress).

The overpressure at Site U1320 that is estimated from the differences in void ratio with Site U1319 is approximately ^* = 0.7 (Fig. F36). This contrasts with the last value measured by the T2P, which recorded an overpressure ^* = 0.2. Further consolidation tests in the laboratory and refined analysis of the T2P and DVTPP pore pressures will refute, refine, or confirm these estimates.

In Ursa Basin, lithostratigraphic Subunits Ia–Id are the same lithostratigraphic units at Sites U1322, U1323, and U1324. Lithostratigraphic Subunits Ib and Id are slump deposits, whereas lithostratigraphic Subunits Ia and Ic are not disturbed. The lithostratigraphic units below Reflector S30 are not equivalent at each site, and thus comparison is difficult. The thickness of the strata between the seafloor and Reflector S30 is 124 m at Site U1322, 197 m at Site U1323, and 165 m at Site U1324. Lithostratigraphic Subunits Ia–Id have a similar overburden at all three sites and have comparable lithology and depositional history, although the accumulation rate is higher at Site U1322 than at Site U1324.

Sites U1322, U1323, and U1324 have similar porosity and bulk density in lithostratigraphic Unit I and therefore similar consolidation states (Fig. F37). The porosity profiles from the three sites show similar trends with a relatively rapid decrease in porosity from 80% at seafloor to 55% in lithostratigraphic Subunit Ia (above seismic Reflector S10). Then, a gentler gradient is observed down to the bottom of the hole, with the lowest porosities at ~45%.

It is assumed that lithostratigraphic Subunit Ic at Site U1324 is hydrostatically pressured. The amount of overpressure at Sites U1322 and U1324 can be estimated through equation 3. The reference void ratio (e_{0 =} 3.32) and compression index (C_c = 0.81) are obtained from the void ratio versus vertical effective stress plot of lithostratigraphic Subunit Ic at Site U1324 (Fig. F38). The estimated overpressure is ^* = 0.4–0.5 of the vertical effective stress below Reflector S30 at Site U1322 and is ^* = 0.6–0.7 at Site U1324 (Fig. F38). The predicted pore pressures fit along the last recorded in situ pore pressure measurements from the T2P and DVTPP measurements. Further consolidation tests in the laboratory and refined analysis of the T2P and DVTPP pore pressures will refute, refine, or confirm those estimates.

At Site U1322, bulk densities are greater in slumps than in material that is not a part of a slump deposit (Fig. F37). The bases of slump deposits typically show lower porosities than the nonslumped intervals immediately below. These variations in porosities are accompanied by differences in undrained shear strength (e.g., Fig. F16), with slumps typically showing higher values than nonslumped deposits. The higher consolidation is inferred to result from the reformation of the originally loose sediments into a more dense structure by dewatering during the landslide process. It is also observed that porosity and undrained shear strength at the top of the slumps does not vary significantly with respect to the in situ sediments above. This might imply that dewatering and consolidation preferentially takes place at the base of the slump where shearing is most likely.

Subunits Id and Ib correspond to the same slump intervals at Sites U1322 and U1324. The general trend in porosity and undrained shear strength was similar at both sites. However, the porosity and undrained shear strength profiles at Site U1324 (Fig. F18) show more subdued variations (or no variation at all) between slumped and nonslumped units than those at Site U1322 (Fig. F16). We speculate that at Site U1324, upslope on the Mississippi Canyon levee, the total shear deformation may have been less than at Site U1322, in the center of the basin. The higher shearing at Site U1322 might then translate into a higher degree of consolidation.

Within the nonslumped units, some layers appear to have high porosity and water content. An example is the interval that correlates with Reflector S30 (~160 mbsf at Site U1324) at the base of lithostratigraphic Subunit Id, a 35 m thick slump deposit (Fig. F37). These layers might have significant overpressure and then have acted as weak layers along which landslides initiated.

The porosities in Brazos-Trinity Basin #4 and Ursa Basin show similar trends with depth (Fig. F39). The porosities in Ursa Basin are slightly lower in the uppermost 100 m of the sediment column and then decrease more gradually with depth than those in Brazos-Trinity Basin #4. It is not clear if the differences in the porosity trends are due to the differences in lithology at the two different locations, the difference in vertical effective stress, or a combination of these factors.

The results that can be obtained now are quantitative, but relative estimates are based on several assumptions. The processing of the pore pressure measurements (see "Downhole Logging") from the T2P and DVTPP, as well as the consolidation tests in the laboratory, will provide a better perspective of the overpressures at both basins and their hydrogeological regimes.

The geochemistry of sediment pore water is indicative of the composition of the seawater trapped during sedimentation and is influenced by post depositional diagenetic transformation and mixing with new water masses. As such, it is a potentially useful means of tracing fluid flow in Brazos-Trinity Basin #4 and Ursa Basin. The 93 interstitial water samples collected from Brazos-Trinity Basin #4 (51 at Site U1319 and 42 at site U1320) and the 124 samples from Ursa Basin (49 from Site U1322 and 75 from Site U1324) were analyzed shipboard for alkalinity, pH, salinity, chlorinity, sulfate, phosphate, ammonium, silica, Na, K, Mg, K, B, Li, Sr, Ba, Fe, and Mn (Figs. F40, F41).

Based on the fact that the maximal variation of the different tracers (Figs. F40, F41) is limited to the sediments of the first 30 mbsf in Brazos-Trinity Basin #4 and the first 100 mbsf in Ursa Basin, we infer that organic degradation and microbially mediated reactions were restricted to these intervals. The pattern of anaerobic degradation of organic matter and the pore water redox conditions in both basins are characteristic of deep marine sediment diagenesis (e.g., Schulz, 2000). The sulfate-methane interface (SMI) occurs at shallower depths in Brazos-Trinity Basin #4 Sites U1319 (SMI = 15 mbsf) and U1320 (SMI = 22 mbsf) than in Ursa Basin Sites U1322 (SMI = 74 mbsf) and U1324 (SMI = 94 mbsf). We infer that this difference is mainly driven by the higher sedimentation rates at Ursa Basin compared to Brazos-Trinity Basin #4. Pore water SO₄ and Mn concentrations within the sulfate reduction zone show a strong antithetical relationship at all sites. Ammonium concentration at the Ursa Basin sites increases with depth, suggesting that more reducing conditions may have existed at these locations relative to sites in Brazos-Trinity Basin #4.

Alkalinity has a maxima at ~35 and 25 mbsf at Sites U1319 and U1320, respectively, in Brazos-Trinity Basin #4 and at ~26 and 47 mbsf at Site U1322 and U1324, respectively, in Ursa Basin. This maxima in alkalinity correlates with pore water increases in Ca, Sr, B, and Li concentrations (Figs. F40, F41). We interpret this relationship as indicating diagenetic carbonate dissolution at shallow depths.

The initial pore water chemistry results obtained during Expedition 308 seem to indicate the presence of hydrogeologic flows in the overpressured sediments of Ursa Basin and Brazos-Trinity Basin #4. The strongest evidence supporting the presence of fluid flow is that seismic reflectors are often associated with pronounced changes in pore water chemistry, both in Brazos-Trinity Basin #4 and Ursa Basin (Figs. F40, F41). This suggests that seismic surfaces occur along permeable stratigraphic horizons that acted as conduits focusing lateral fluid migration. At Site U1320 in Brazos-Trinity Basin #4, the changes in pore water chemistry are mainly related to seismic Reflectors R10 and R30, whereas at Site U1319, which is a more condensed sedimentary section located at the steep margin of the basin, these changes are restricted mostly around Reflector R30 (Fig. F40). Below Reflector R30, some of the element concentrations also show similar spikes associated with seismic reflectors such as the Cl minimum between Reflectors R50 and R60. This minimum may indicate lateral migration of slightly fresher fluid at this particular strata (Fig. F40). At Site U1324 in Ursa Basin, changes in pore water chemistry equivalent to those described above in Brazos-Trinity Basin #4 are centered around Reflector S10 and below Reflector S30 (Fig. F41). At Site U1322, the chemical changes around Reflector S10 are similar to those observed at Site U1324, but the changes below Reflector S30 are less pronounced (Fig. F41).

Further evidence of fluid flow is provided by downhole variations in chloride concentration within Ursa Basin. Site U1322 exhibits a near linear decrease in chlorinity from the seafloor to terminal depth (y = 3.7e³ – 6.5x; R² = 0.68) (Fig. F41). Initial concentrations are identical to seawater concentration of 559 mM, increase rapidly to 578 mM at 15 mbsf and then decrease to 542 mM at the base of the hole. This gradual pore water freshening may indicate mixing between overlying seawater and fluids from the Blue Unit. Alternatively, dewatering of clay-rich lithologies during diagenesis may also produce freshening of pore waters. However, we suggest this latter possibility is less likely as we did not observe a similar decrease in chlorinity with depths at Site U1324, which must have experienced more intense burial diagenesis as it was buried deeper (maximum drilling depth was 600 mbsf versus 238 mbsf at Site U1322).

In summary, chemical pore water profiles in Brazos-Trinity Basin #4 and Ursa Basin record a combination of processes, which includes anaerobic degradation of organic matter, carbonate dissolution and precipitation, and lateral fluid migration. We propose that most of the chemical changes in pore waters discussed above are driven by lateral fluid flow. Though postcruise work will provide higher resolution data and better constraints, we interpret the data as showing two fluid flow pathways: 1) above Reflector S30, a shallow fluid flow system possibly related to recharge of seawater or other shallow circulated fluids in the basin, and 2) a deep-seated fluid flow that took place below Reflector S30 and that could be linked to an overpressured unit such as the Blue Unit in Ursa Basin. We stress, however, that this interpretation is preliminary, and postcruise work will serve as a test for this working hypothesis.

Overview

In the mudrocks above the Blue Unit in Ursa basin, porosity declines rapidly in the uppermost 100 m and subsequently declines only slightly. Pressure predictions based on porosity suggest overpressures with a normalized overpressure ratio (*) >0.6. Direct pressure measurements also record overpressures with * > 0.5. At equivalent depths, pressures (as recorded by both consolidation and direct measurements) are slightly greater to the east at Site U1322 where the overburden is thin than to the west at Site U1324 where the overburden is thick. The temperature gradient is significantly greater at Site U1322 (26°C/km) than to the west at Site U1324 (18.4°C/km). Sedimentation rates at Site U1324 were almost 3 times greater than at Site U1322. A conceptual model that links these observations is that there is upward flow everywhere within the overburden above the Blue Unit; the flow rate is constant, reflecting a constant overpressure gradient, at each site. However, significant lateral flow within the Blue Unit must also be present to account for the similar overpressure gradient at the two locations (despite the threefold difference in sedimentation rate) and the elevated heat flow emanating from the Site U1322 location.

Sediments deposited in the last ~70 k.y. in Ursa Basin can be divided into three successive depositional units: the Blue Unit, the Ursa and Southwest Pass Canyons channel-levee systems, and distal fan and hemipelagic deposition (Fig. F10).

The Blue Unit forms a regional permeable unit that is composed of interbedded thick sands (some >50 m thick) and mudstone. Its base is the base of the deepest sand, which ties to a weak negative amplitude within the shallow sedimentary section (Fig. F9, F10). Its top ties to a strong positive seismic amplitude that marks an increase in impedance with depth at the top of sands within the Blue Unit. In places, the thick Blue Unit sands are truncated by the Ursa channel (Fig. F10).

The Ursa Canyon channel-levee system has high-amplitude chaotic seismic reflections that most likely are permeable sands. However, the bounding deformation zone is dominated by mudstone and contains rotated slump blocks within the Blue Unit where the channel has incised into the Blue unit.

Low permeability mudstones cap the channel-levee systems and are thicker to the west (Site U1324) than to the east (Site U1322). Multiple slope failures are present in this section with more to the east in the vicinity of Site U1322 (Fig. F33).

Sedimentation drives overpressure. Sedimentation rates in Ursa Basin are extremely rapid and have significant lateral variation (Fig. F32). On average, the sedimentation rate across the entire interval is ~2.6 times greater at Site U1324 than at Site U1322 (10 versus 3.8 mm/y). Late Pleistocene sedimentation rates varied ~2.8-fold (9.6 versus 3.5 mm/y). The detailed age model at both sites will be strengthened with further shore-based work. However, this work suggests that at the base of Site U1324 sedimentation rate exceeded 25 m/k.y. and at Site U1322 it exceeded 16 m/k.y.

A striking feature of the three sites (U1322, U1323, and U1324) is that the porosity at equivalent depths is similar (Fig. F42). For ~50 m there is a rapid decline in porosity from ~80% to ~60%. Thereafter porosity declines at a lower rate with depth to the bottom of the hole. We predict pressure from porosity (see "Physical properties in Brazos-Trinity Basin #4 and Ursa Basin") and find elevated pressures beneath 125 mbsf at Site U1322 and beneath 150 mbsf at Site U1324 (Fig. F43). The overall pressure ratio (*, the fraction of the hydrostatic stress [equation 1]) is ~0.5–0.6 at each location.

Pressure

We made multiple penetrometer measurements with the T2P and the DVTPP (Fig. F44). We have not processed these data and we have only posted the final pressures recorded at the end of the deployment. There is considerable scatter in these data (see "Challenges of Measuring Pressure"). None the less, several clear trends are present. At both locations, multiple measurements of in situ pressure record significant overpressures. Second, pressures between 100 and 250 mbsf are somewhat higher at Site U1322 than they are at Site U1324 (Fig. F44). This is intriguing because the porosity at these two sites is very similar at equivalent depths; thus, this difference is not predicted from the porosity-based pressure prediction. Deep at Site U1324, pore pressures rise to a * value between 0.5 and 0.6.

A striking observation is that the temperature gradient at Site U1324 is 18.4°C/km, whereas it is 26.2°C/km at Site U1322 (Fig. F45). These data were acquired by APCT, T2P, and DVTPP measurements. The APCT measurements were corrected to estimate the actual temperature from the measured temperature; the DVTPP data and the T2P data have not been corrected. Thermal conductivities at the base of Site U1324 are ~1.2 W/m·K, whereas they are ~1.15 W/m·K at Site U1322. If the heat flow is vertical and conductive, this implies a heat flow of ~22 mW/m² at the base of Site U1324 versus ~30 mW/m² at the base of Site U1322.

There are several possible explanations for the ~1.4-fold increase in heat flux at the base of Site U1322 relative to that at Site U1324. First, sedimentation is inversely correlated to heat flow (Wang and Davis, 1992). Thus, the increased thermal gradient at Site U1322 relative to Site U1324 may result from the fact that at Site U1322 the system is closer to thermal equilibrium, whereas at Site U1324 sedimentation was so rapid that it has outpaced the heat flow. A second possibility is that local variations in the proximity to salt bodies have significantly affected the local geothermal gradient. A final possible interpretation is that the heat flow entering the mudstone above the Blue Unit is greater at Site U1322 than at Site U1324 because there is lateral transfer of heat as described below.

Figure F46 illustrates our initial conceptual model for flow within Ursa Basin. After deposition of the Blue Unit, it was incised by the Ursa channel-levee system and then rapidly buried by mud. Sedimentation rates were extraordinarily rapid and the sedimentation rate at Site U1324 was almost three times that at Site U1322. One-dimensional flow modeling would suggest that the overpressure and upward flow rate should be greater at Site U1324 than at U1322 because the sedimentation rate is so much higher at Site U1324. However, the presence of similar overpressure gradients within rocks of similar lithology (and presumably similar hydraulic conductivity) implies that the upward flux of water is approximately equal at each location. To equalize these pressure gradients, given the very different sedimentation rates, we infer some flow must occur laterally within the Blue Unit (Fig. F46). The temperature gradient at Site U1322 is significantly greater than at Site U1324, which implies a conductive heat flux that is 1.4 times greater at Site U1322 than at Site U1324 (Figs. F45, F46A, F46B). The elevated temperature gradient at Site U1322 may reflect lateral transfer of heat by advection within the Blue Unit (Fig. F46).

The overpressure encountered at the base of Site U1324 is significantly larger than at Site U1322 (3.1 versus 1.3 MPa) (Fig. F46). Early models suggested that the entire Blue Unit might be in hydrologic communication and that there would be a single overpressure encountered in both locations. More detailed analysis illustrates that the Blue Unit is composed of multiple sand bodies that have significant lateral extent (Fig. F46). These sands are truncated by the Ursa channel deformation zone, which has deformed the underlying Blue Unit. In the western location (Site U1324), the top of the Blue Unit has been eroded and only the deeper beds are present.

Based on pressure data analyzed here and more detailed regional seismic mapping, we interpret that only the basal sand within the Blue Unit has communicated laterally from Site U1324 to Site U1322. In this basal sand, flow is driven laterally toward Site U1322, underneath the Ursa channel-levee system. Within the Blue Unit at Site U1322, vertical flow will occur between the sand beds; however, there will be a large vertical pressure gradient within the shales between the sand beds. The impedance of these shale layers reduces the pressure from ~3.1 MPa at the base of the Blue Unit to ~1.3 MPa at the top of the Blue Unit (Fig. F46). These observations illustrate the critical importance of being able to sample the pressure field within the Blue Unit in order to fully understand this hydrodynamic system. This is the fundamental goal of the second component of the Gulf of Mexico Hydrogeology proposal.