By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
 A three-dimensional seismic reflection survey was carried out over a gas hydrate field offshore Vancouver Island in the area of Ocean Drilling Program Leg 146, Site 889/890. The area of investigation is located on the accretionary prism of the Cascadia margin in water depths of 1200–1500 m. Several seismic blank zones were observed in the seismic data over a frequency range from 30 Hz to 3.5 kHz, where the degree of blanking increases with seismic frequency. Time slices and selected horizon slices were used to define the spatial distribution of the blank zones. The zones range from 80 m to several hundred meters in width. One blank zone, almost circular with a diameter of ∼400 m, has a distinct seafloor expression. It shows the characteristics of a mud/carbonate mound and is probably associated with free gas expulsion. Hydrate was found at four sites with a piston corer within this blank zone at depths of 3–8 m below the seafloor. We present a tectonic model that explains the blank zones to be a result of differential uplift along thrust faults. The blank zones represent conduits for fluids and gas migrating upward. Blanking of the seismic energy is believed to be mainly the result of increased hydrate formation in lenses within the faults. The blanking effect is enhanced locally because of scattering at carbonates formed at the seafloor.
 Gas hydrates are solid ice-like structures in which gas molecules (largely methane) are trapped within cages of water molecules. The stability of natural gas hydrates is mainly affected by temperature and pressure, but secondary influences include pore fluid salinity, the nature of gas enclosed, and the grain size of the host material [Clennell et al., 1999; Sloan, 1998]. Gas hydrates represent a large potential energy resource [e.g., Kvenvolden, 1993] but may also play an important role in global climate change and as geological hazard. Estimating the amount of gas venting and hydrate formation at or close to the seafloor is crucial to the question of whether gas hydrates can affect global climate [Kennett et al., 2000].
 Our most recent experiments have focused on high-resolution seismic studies of the shallow sediments (upper ∼300 m). A deep towed multichannel seismic survey carried out in 1997 using Deep Tow Acoustics/Geophysics System (DTAGS) detected several subvertical chimneys of reduced seismic reflectivity. As part of a larger scale three-dimensional (3-D) seismic study in 1999, the area with the most prominent blank zones was chosen to further investigate the nature of those structures. A grid of closely spaced lines with single-channel seismic and high-resolution 3.5-kHz subbottom profiling was acquired over the vent field. On the basis of these surveys, a piston-coring cruise was carried out, and a second high-resolution single-channel 3-D seismic survey over the most prominent blank zone was acquired in July 2000. This zone was also target of a seafloor sampling and video observation survey with the unmanned submersible ROPOS in September 2000.
 The vent field is characterized by a large number of blank zones. We therefore focused our seismic studies over this field to understand the physical and geological processes involved in venting and related seismic blanking. In this paper, we examine the detailed 3-D reflection structure of the uppermost 250 m of sediment and the spatial distribution of the blank zones over the vent field. On the basis of the seismic and geological data available, we demonstrate that the blanking is most likely the effect of (1) intense hydrate formation inside the blank zone and (2) attenuation loss due to surface scattering at carbonate formations. However, the occurrence of free gas as additional source of amplitude loss cannot be completely ruled out based on the data available.
1.1. Tectonic Setting
 The area of investigation is located on the accretionary prism of the Cascadia subduction zone (Figure 1). The Juan de Fuca plate converges orthogonally to the North American plate at a present rate of ∼45 mm yr−1 [e.g., Riddihough, 1984]. Seaward of the deformation front, the Cascadia basin consists of pre-Pleistocene hemipelagic sediments overlain by a rapidly deposited Pleistocene turbidite for a total thickness of ∼2500 m. At ODP Site 889, sediments in the upper 128 m below the seafloor are silty clays and clayey silts interbedded with fine sand turbidites. This sequence was interpreted as little deformed slope basin sediments, deposited in place [Westbrook et al., 1994]. The sediments below this sequence are more deformed, compacted, and cemented and were interpreted as accreted Cascadia Basin sediments. Seismically, the accreted sediments are almost transparent and show no seismic coherency, whereas the younger slope basin sediments show stronger continuous reflectivity.
 Most of the incoming sediment is scraped off the oceanic crust and folded and thrust upward to form elongated anticlinal ridges with elevations as high as 700 m above the adjacent basin. The thrust faults near the deformation front penetrate nearly the entire sediment section [Davis and Hyndman, 1989]. Landward from the deformation front, the seafloor rises rapidly to a water depth of 1400–1500 m, forming a bathymetric bench. The 12 km × 8 km area of detailed investigation is located on this bench around two topographic highs, which rise 200 m over the surrounding seafloor (Figure 2). The area between the topographic highs forms a 350-m-deep trough filled with recent slope basin sediments.
 The area to the east of the two ridges forms a broad slope basin filled with up to 0.5 s two-way time (TWT) of sediment. The southwestern margin of this basin lies at an eastward facing scarp in the hanging wall of a thrust fault (Figure 3). This thrust seems to be westward dipping based on the asymmetry of the deformation of the basin sediments and the uplift of the ridge [Westbrook et al., 1994]. However, the fault plane is not imaged in this high-resolution seismic survey.
 The sediment-filled trough between the topographic highs shows complicated deformation structures (Figure 4). The inner portion of the trough, in which the blank zones were observed, is uplifted relative to the surrounding sediments by up to 50 m.
1.2. Fluid Venting and Carbonate Formation
 The overall accretion process results in shortening and tectonic thickening of the incoming sediment column. The resulting underconsolidated section reestablishes an equilibrium porosity-depth relation through consolidation and fluid expulsion [Hyndman and Davis, 1992]. Fluid venting and associated mud volcanism have been reported from many accretionary prisms [Suess et al., 1999, 1998; Elderfield et al., 1990; Henry et al., 1990; Le Pichon et al., 1990]. The fluids are often inferred to be discharged along major faults (including the decollement) and only a small amount expelled pervasively. This is in contrast to the accretionary prism offshore Vancouver Island where the lack of strong local thermal anomalies at the deformation front and the lack of pronounced BSR depth disturbances suggest that only minor concentrations of fluid flux move up faults and other hydrologic conduits [Davis et al., 1990; Ganguly et al., 2000].
 Fluid expulsion and venting are often associated with near-surface diagenetic processes such as carbonate formation. Widespread carbonate sedimentation in the general area was interpreted from high-amplitude backscattering on GLORIA side-scan images [Carson et al., 1994]. Carbonate slabs, nodules, and chimneys were collected on the Cascadia margin off Oregon by submersibles or grab samplers [Kulm and Suess, 1990; Bohrmann et al., 1998]. During seafloor video observations with ROPOS in September 2000, widespread carbonates were found around the most prominent vent site [Chapman and Riedel, 2000]. Carbonates formed either as several centimeter thick plates covering larger areas or can be found as chunks embedded in the seafloor.
2. Data and Observations
 Several subvertical zones of reduced seismic reflection amplitude have been observed on several seismic data sets using different recording systems and source frequencies. There are four prominent blank zones in this area labeled 1–4 in all following data examples. The first observation was made during a deep towed multichannel DTAGS cruise (Deep-Tow Acoustics/Geophysics System) in 1997 [Chapman et al., 2002]. This area was further investigated as part of a 3-D seismic reflection survey on CCGS J. P. Tully in 1999. The main 3-D survey consisted of 40 lines spaced at 100 m plus eight crossing lines spaced at 1000 m. The seismic source was a single 40 cubic inch (0.65 L) air gun with a wave shape kit to reduce air gun bubble energy. Over each line, single channel (Teledyne Ltd. hydrophone streamer) and multichannel (Canadian Ocean Acoustics Measurement System (COAMS)) seismic data and 3.5-kHz subbottom profiler data were recorded. The vent field was further investigated with a grid of 31 seismic lines spaced at 25 m with single-channel seismic data and 3.5-kHz subbottom profiler data.
 In July 2000 a second high-resolution single channel 3-D survey was carried out over the most prominent blank zone. This survey was acquired with the same air gun and streamer configuration as used in 1999. However, shot spacing was set to 12 m to increase horizontal resolution. A set of 21 lines spaced at 25 m was acquired in northeast–southwest direction and covered the entire blank zone.
2.1. The 3.5-kHz Subbottom Profiler
 The hull-mounted 3.5-kHz subbottom profiler (ORE Model, 140 transceiver) on the CCGS J. P. Tully gave a high-resolution image of the uppermost 30–40 m of the sediment column (Figure 5). No data processing was applied, and the amplitude envelope was used for plotting purposes. Four blank zones are clearly visible, sharply interrupting the sediment layers. Blank zone 1 (∼400 m in diameter) has a distinct surface expression, but the center part could not be resolved completely with this system. The maximum elevation of this surface expression is estimated to be ∼6 m, but the actual height of the central portion is uncertain. Blank zones 2–4 do not extend completely to the seafloor and do not show a distinct surface expression. They interrupt deeper sediment layers, but their upward development is apparently stopped at a high-amplitude layer at about 10 ms TWT below seafloor (bsf) (6–8 m).
 The DTAGS system consists of a Helmholtz transducer source emitting a chirp-like sweep signal of frequencies from 250 to 650 Hz and a 600-m-long hydrophone streamer with two subarrays containing 24 hydrophone groups each [Gettrust and Ross, 1990; Gettrust et al., 1999; Chapman et al., 2002]. The original sweep was not recorded and a synthetic sweep was used for cross correlation. Source waveform and amplitude after correlation proved to be quite unstable. The tilt of the DTAGS streamer varied considerably so careful source, and receiver geometry corrections corresponding to the short wavelengths of 2–6 m were required [Walia and Hannay, 1999]. In this study only near offset traces were used to create a stack without applying normal move out corrections.
 The largest blank zone (zone 1) penetrates to the seafloor and has a dimension of ∼250 m at the seafloor and widens to ∼300 m at ∼170 ms TWT (∼140 m) bsf (Figure 6). The seismic line shown in Figure 6 does not cross the blank zone on a diameter. A prominent high-amplitude reflector at ∼10 ms TWT (∼6–8 m) bsf is an apparent upward barrier for blank zones 2–4. Several small-scale (20–40 m wide) blank zones are also visible but stop at deeper layers.
2.3. COAMS Multichannel Seismic Data
 The multichannel 3-D seismic survey was carried out using the COAMS system consisting of 102 hydrophones over an active range of 1100 m. A localization technique for the multichannel COAMS streamer was developed based on observed travel times of the direct wave and seafloor reflection and a priori information from five depth sensors along the array [Dosso and Riedel, 2001]. The COAMS array also included one compass at the end of the active section of the streamer at an offset of 1100 m. Statistical analyses of the compass data indicate no significant feathering of the streamer by comparing the compass reading at lines acquired in different directions. After a turn onto a new line the streamer generally equilibrated to the new direction within the first kilometer of the line and afterward followed the direction of seismic acquisition within a limit of 1–2°. After hydrophone localization (offset and depth) the data were processed with the INSIGHT/ITA system. Processing included spherical divergence correction, band-pass filter (20–200 Hz), deconvolution for bubble and ghost reduction, slant stack for noise reduction, common midpoint (CMP) sorting, semblance velocity analysis, normal move out (NMO) correction, and stacking of the near-offset traces (channels 1–64). Far-offset traces did not contribute constructively to the stack and were used only for velocity definition and for amplitude versus offset (AVO) effects.
 Six parallel multichannel seismic inlines from the main 3-D grid cover the area where blank zones are observed. As an example, inline MC-27 and line MC-X7, additionally acquired in the perpendicular direction, intersect near blank zone 1 (Figures 7 and 8). This zone shows a strong diffraction hyperbola on inline MC-27, whereas no similar feature is observed in the perpendicular direction. At the edges of blank zones 2–4, diffractions occur that locally enhance the reflectivity of sediment layers (Figure 7). On both lines, a BSR can be identified, and the BSR amplitudes are enhanced around blank zone 1. The increase in BSR amplitude is asymmetric and localized over a section ∼400 m wide to the NW side of the blank zone. The crossing line MC-X7 also indicates enhanced reflection amplitudes of sediment layers ∼80 ms TWT above and below the BSR. All sediment layers in the uppermost 250 ms TWT throughout the blank zone are pulled up by ∼5–10 ms.
2.4. Single-Channel Seismic Data 1999
 The single-channel streamer is composed of 50 hydrophones spaced at 0.5 m over an active section of 25 m length, and all hydrophones are summed to one trace. The midpoint of the active section was ∼30 m behind the seismic source giving almost vertical incidence data in water depth of 1200–1400 m. Data processing included spherical divergence correction, band-pass filtering (20–180 Hz), deconvolution, and migration. These lower-frequency data image the entire slope basin sediment section deposited on top of accreted sediment (Figure 9a). The contact between both sediment types is not clearly visible and is probably gradational. Sedimentation in the southern half of the area is almost parallel to the seafloor without any indication of near-surface faulting. The sedimentation is interrupted only by blank zone 1 which penetrates the entire sediment column. The northern part of the area shows evidence of near-surface folding and faulting associated with the occurrence of blank zones 2–4. Wherever the tilt of the sediment layers changes in this area, blank zones occur. All of these blank zones appear as sharply bounded chimneys penetrating through the whole sediment column. Within the data resolution the chimneys extend up to the seafloor. In this data set the high-amplitude layer at 10 ms TWT bsf is not resolved due to a receiver ghost, which follows the primary reflections at around 6–8 ms and merges with the reflection of interest.
 Blank zone 1 shows different characteristics than the other zones. In the unmigrated sections, blank zone 1 shows a strong diffraction hyperbola (Figure 9b), probably a result of the topography of the seafloor, visible in the 3.5-kHz subbottom profiler data. Deeper layers (e.g., at 1.8–1.9 ms TWT on line SC-18) are pulled up by ∼5–10 ms TWT, which might also be related to the topography. The BSR, which can be identified in this area where it crosscuts the sedimentary reflectors on some of the lines, marks the top of a high-reflectivity zone. On most of the other lines, no single BSR reflection can be identified, but below the expected depth of the BSR an increase of mostly chaotic reflectivity is observed.
2.5. The 3-D Analysis
 The 31 lines of closely spaced single-channel seismic data from the 1999 survey have been used to generate a 3-D data cube. Each line was processed separately as explained above and then migrated using velocities defined from the multichannel COAMS data. The 3-D or 2.5-D migration proved to give poor results due to the limited cross-line dimension of only 750 m. However, the geology of this area does not show any strong variation, and no significant change in the velocities was observed on adjacent multichannel lines over the area of interest. Therefore the same velocity-depth function was used to migrate all single-channel seismic lines.
 The single-channel data were acquired during varying weather conditions yielding strong static shifts (striations) among the individual lines. To remove the cross-line statics, the data were flattened to the seafloor, and a lateral coherency filter was used to remove residual swell. After swell removal the data were shifted back onto the smoothed bathymetry as defined from the seismic data and regional bathymetry available in this area. From this data cube, amplitude time and horizon slices were created (1) to define the spatial extent of the blank zones and (2) to define the detailed sediment structure and nature of deformation within the area.
 In the southeast part of the 3-D area the sediments generally follow the seafloor topography (Figure 10a) and have a slight SSW dip. The slices of seismic amplitude were generated at 1.79 s and 1.802 s (Figures 10b and 10d). A regular pattern of color change can be identified indicating the dip of the reflectors. In the southeast corner of the slices, sedimentation is interrupted by blank zone 1, and only chaotic reflectivity can be observed. The dip of the layers changes abruptly at around cross-line position 65–70. The sediments appear to be curved toward the west. Between cross lines 60 and 80 the sediments form a plateau before they steeply dip toward the northeast at cross-line position 100. The spatial extent of the blank zones can best be visualized using slices of instantaneous amplitude, i.e., the envelope of the seismic trace (Figures 10c and 10e). Blank zone 1 in the southeast corner is almost circular and has a diameter of ∼400 m. The inner low-reflectivity zone (dark shaded color) is surrounded by a high-amplitude ring, which is up to 40 m wide. Blank zones 2–4 are more linear features and appear to follow an east–west trend. They are also surrounded by high-amplitude rims, but these rims are not always well developed.
 To test whether the high-amplitude rims are artificially generated by the time slice technique, one selected horizon slice was generated (Figures 10f and 10g). This horizon slice reconfirms the high-amplitude rims as well as the east–west trend of the blank zones 2–4. The topography of this horizon is almost identical to the seafloor with the same steep increase in dip at cross-line position 65.
2.6. Hydrate Occurrence
 During a cruise on CCGS J. P. Tully in July 2000, massive hydrate was recovered in piston cores around blank zone 1 at depths of 3–8 m below the seafloor. Hydrate was usually found at the bottom of the cores or the core catcher and may have stopped deeper core penetration. Samples varied between 2 and 15 cm in size. At one position the recovered sample was entirely composed of hydrate with no sediments imbedded. Several samples have been preserved in liquid nitrogen for future analyses.
 The top of the hydrate was mapped during the 2000 3-D single channel seismic survey as a shallow reflector around the center of the blank zone. The depth of the reflector was confirmed by the depth of the hydrate recovery at the four piston core sites. The hydrate forms a cap, which dips away from the center of the vent (Figure 11). Below the hydrate cap the amplitude of the seismic data is reduced, probably an effect of increased transmission loss at the high reflectivity cap. However, blanking extents over even a larger region than just where the hydrate cap was mapped. In a horizon slice at a depth of ∼30 m bsf, a high-amplitude rim is observed around the center part of the vent only (Figure 12a). Time slices of instantaneous amplitude instead show a partially closed ring of higher amplitudes around the entire blank zone (Figures 12b and 12c).
 In contrast to the single-channel seismic data acquired in the perpendicular direction in 1999, this data sets shows no indication of pull up. Most of the sediments follow the general seafloor topography or were completely disrupted and no coherent reflectivity was observed.
2.7. Frequency-Dependent Blanking
 Blanking was observed over a wide frequency range in different seismic acquisition systems: (1) 3.5 kHz subbottom profiler, (2) vertical incidence DTAGS (250–650 Hz), (3) vertical incidence air gun (Teledyne, 20–180 Hz), and (4) multichannel air gun (COAMS, 20–180 Hz). The amount of blanking increased with seismic frequency (Figure 13). For a time window of 14 ms the RMS amplitude in representative data examples was calculated at a subseafloor depth of ∼20 ms TWT. The amount of amplitude reduction shown in Figure 13 was calculated by dividing the average amplitude inside a blank zone by the average amplitude outside the blank zone. For this calculation, only blank zones 2–4 have been used.
 Amplitudes in the 3.5-kHz subbottom profiler were variable resulting in an amplitude reduction varying from ∼75% to almost 100%. For vertical incidence DTAGS data, amplitude reduction ranged from 50% to 100%. Amplitudes of the DTAGS data were unstable because the original source sweep was not recorded and a synthetic sweep was used in the cross correlation. Amplitude reduction in the lower-frequency air gun data was observed to be dependent on the angle of incidence. The vertical incidence Teledyne data show higher-amplitude reduction (45–65%) than the multichannel COAMS data (25–50%).
3. Amplitude Versus Offset (AVO) Analysis
 To further investigate the amplitude behavior across a blank zone an AVO analysis (Figure 14) was carried out for a selected horizon on inline MC-27 at a depth of ∼100 ms TWT bsf (see Figure 7). RMS amplitudes were calculated for a window length of 10 ms, representing the length of the wavelet. Amplitudes were then corrected for streamer directivity. Amplitudes for the air gun source were assumed not to vary with angle, and no source directivity correction was performed. The near-offset traces were in general noisier than the far-offset traces due to noise generated by the tow cable of the streamer. AVO trends were calculated at three different positions around blank zone 2. The AVO behavior is nearly symmetrical about the center of the blank zone, and we show the results from one side only.
 Seismic amplitudes were calibrated to represent a seafloor reflection coefficient of 0.2 for vertical incidence. This reflection coefficient was confirmed by analyzing the ratio of primary-to-multiple reflection amplitudes [Warner, 1990] and by physical properties obtained during the piston-coring cruise in July 2000.
 Because of the limited range of offsets available with the COAMS array, the maximum angle of incidence at the reflector is ∼30°. Simple straight path ray tracing was used to determine the angle of incidence. AVO modeling was done for a simple two-layer model only, and transmission loss at additional layers between the seafloor and the target horizon was assumed to be small and therefore neglected.
 Velocities determined from the multichannel NMO analysis provided the basic constraints on the AVO modeling. Density and S wave velocity were allowed to vary within reasonable bounds determined from Hamilton's  equations and the ODP downhole logs [Westbrook et al., 1994] (Table 1).
Table 1. Range Limits for Elastic Inversion
P Wave Velocity, m s−1
S Wave Velocity, m s−1
Density, g cm−3
 Three locations around blank zone 2 were modeled representing (1) the area outside of the blank zone (Figure 14a), (2) inside the blank zone (Figure 14b), and (3) outside but close to the left boundary (Figure 14c). Locations 1 and 2 show a similar AVO behavior with amplitudes increasing with offset. However, the overall magnitude of the reflection coefficient is decreased inside the zone by ∼50%. Location 3 shows a different AVO trend, i.e., a decrease of reflection coefficient with offset. At location 3 the vertical incidence reflection coefficient (0.1) is the same magnitude as that outside of the zone; that is, the vertical incidence rays are not affected by the blank zone. Rays that have to travel through the blank zone are reduced in their amplitudes, and the amount of amplitude reduction increases with ray path length.
3.1. Elastic 1-D AVO Inversion
 The inversion scheme used is a sequential-quadratic programming technique [Riedel and Theilen, 2001], i.e., an iterative linearized inversion procedure to find the minimum misfit between observed and modeled AVO data. The forward model is based on the full Zoeppritz  equation for two solid, elastic layers. The inversion was constrained to accept only those models that fall inside the predefined range of meaningful values (Table 1).
 The location outside the blank zone was used first to determine a reasonable set of physical properties for the two layers of interest since this location is not affected by attenuation loss due to the blank zone (Table 2). The inverted values for P wave velocity and density in both layers are reasonable and fall well within the expected range for the type of sediments involved. Values obtained for the S wave velocities are, in contrast, not very reliable. However, the best misfit to the observed AVO data could only be found if the S wave velocity was allowed to decrease significantly. Constraining the inversion to accept only an increase in S wave velocity resulted in a misfit that was up to 5 times higher.
Table 2. Elastic 1-D Inversion Results Inside and Outside of Blank Zone
P Wave Velocity, m s−1
S Wave Velocity, m s−1
Density, g cm−3
Outside blank zone
Inside blank zone
 An elastic 1-D AVO inversion was also performed at location 2 inside the blank zone. The same constraints as used at location 1 outside of the zone were applied to this inversion. The resulting P wave velocities were similar to the values obtained outside of the zone; however, to fit the observed lower level of reflection coefficients, the density contrast was reduced to almost zero (Table 2).
3.2. Viscoelastic 2-D AVO Modeling
 Viscoelastic 2-D AVO modeling (VAVO) was used to determine the amount of attenuation needed to explain the observed amplitude decrease for rays traveling through the blank zone. The modeling was based on the ray-tracing code HARORAY [Pignot and Chapman, 2001], which includes attenuation loss along ray paths. The subsurface was parameterized into cells with varying physical properties. The properties obtained from elastic 1-D inversion at location 1 outside of the blank zone were used as background elastic properties. This assumption is based on the observation during NMO analyses that the P wave velocity does not vary across the blank zone. Since density and S wave velocity are both closely related to the P wave velocity [Hamilton, 1980], these parameters cannot vary independently. For simplicity, only 11 representative angles were chosen in all modeling cases.
 To first achieve a reasonable fit to the vertical incidence reflection coefficient inside the blank zone, an attenuation coefficient αp of 0.085 dB m−1 kHz−1 (Q = 200 for a P wave velocity of 1650 m s−1 and frequency of 150 Hz) for the P wave was found to be sufficient (Figure 14b). This parameter was subsequently used to calculate the corresponding reflection coefficients for larger angle of incidence. The increase of reflection coefficient with angle of incidence in the VAVO modeling is not as strong as observed for the elastic background AVO function defined from the location outside the blank zone. This is the result of a competing effect between larger attenuation loss at larger ray path length and increase of the reflection coefficient of the background AVO function. The elastic 1-D inversion result at this location and the 2-D VAVO modeling both fit the data well, but the 2-D VAVO modeling is based on a more reasonable physical model and is therefore preferred.
 Next, the location close to the boundary of the blank zone was modeled (location 3). At this location, only nonvertical rays (larger than ∼8°) travel through the blank zone. The vertical incidence reflection coefficient is the same magnitude as determined at location 1. The outside AVO function is plotted for reference (Figure 14c). For rays that pass through the blank zone the resulting reflection coefficient is reduced. The attenuation coefficient of 0.085 dB m−1 kHz−1 (Q = 200) was not sufficient to explain the observed decrease. A reasonable fit to the data was achieved only by increasing the attenuation coefficient to 0.3 dB m−1 kHz−1 (Q = 55). This apparent mismatch between the attenuation can be explained with an additional attenuation loss for rays passing through the boundary of the blank zone. The boundary of the blank zone might be a relatively rough surface and the additional attenuation is due to scattering.
4. Interpretation and Discussion
 A large number of subvertical blank zones, interpreted as vents, were observed in the area around ODP Site 889/890. These zones were investigated using different seismic systems with varying frequencies and by piston coring and remote operated vehicle (ROV) video observations. The vent field is characterized by at least four prominent blank zones over an area of roughly 1 km × 3 km. About 3 km north of this vent field, another blank zone was observed. However, we focused our studies on the vent field with the higher density of blank zones to understand the physical and geological processes involved with venting and related blanking.
 The above analyses showed that blank zones 2–4 are associated with near-surface faults following an east–west trend. They are 80–300 m wide linear features, which do not penetrate completely to the seafloor. They can be traced through the sediments down to the BSR for a total depth of ∼250 m. Blank zones 2 and 4 are surrounded by high-amplitude rims, typically 20–50 m wide. Blank zone 1 appears to be of a different nature than the others. It is not related to shallow tectonic structures (faults) but shows a distinct surface expression. This zone also is marked by a ring of increased seismic amplitude with a diameter of ∼400 m, which can be traced to a depth of ∼100 ms TWT bsf. The center of blank zone 1 coincides with the sites where massive hydrate was found.
 The occurrence of blank zones in this area can be explained with a simplified tectonic model. The area of investigation is located around two topographic highs. The eastern ridge was interpreted as being formed by a westward dipping thrust fault based on the observed asymmetry in sediment deformation [Westbrook et al., 1994]. The western ridge appears to be formed at a similar but eastward dipping thrust fault. The western margin of the sediment-filled trough shows similar deformation features as seen in the eastern slope basin. Both ridges follow a southeast–northwest trend, but the uplift is asymmetric. The eastern ridge appears to be uplifted by a larger amount than the western ridge. This results in a more southerly dip of the sediment layers in the trough. Differential uplift of the ridges and the overall compression of the site due to the accretion process result in complicated deformation structures and fault generation in the sediment-filled trough. The blank zones are associated with these faults, which penetrate the entire sediment column down to the BSR. The deeper roots of these faults cannot be resolved because deeper sediments generally show no coherent reflectivity due to pervasive deformation. No direct relation between the major thrust faults and the faults associated with the blank zones can therefore be detected. Since the sediment layers generally dip toward the south in the area of the vent field, blank zones have an apparent E-W trend.
 In general, uplift of the entire area of investigation destabilizes the lowermost gas hydrates because of the strong pressure effect on stability. The highest concentration of hydrate is found in the ∼50 m just above the BSR [Hyndman and Spence, 1992]. If a sediment column is uplifted, the base of hydrate stability must move up through the sediment. Hydrate dissociation occurs, leaving free gas behind. Beneath the center portion of the sediment valley an increase in reflection amplitude is observed below the BSR suggesting the presence of free gas (Figures 3 and 4). This gas cannot easily migrate upward through the very low permeability sediments containing gas hydrate. Compression and internal deformation of the sediments form faults, which can act as conduits for the free gas and fluid upward migration.
 Preliminary analyses of the sulfate gradients acquired during the piston-coring cruise in July 2000 suggest a much higher supply of methane inside blank zone 1 compared to the area outside of the vent. The formation of massive gas hydrate at blank zone 1 is probably the effect of this intensified methane supply through the vent. The top of the hydrate was mapped in the single-channel 3-D seismic data and showed a deepening of the cap away from the center. However, blanking occurs also outside of the area of main hydrate recovery, where no anomalous coherent reflector was found. Venting might have been active at an earlier stage but along a different pathway. Massive hydrate might have been present but subsequently dissociated due to the concentration gradients toward the seawater after the main supply jumped to the new present location. Hydrate still might be present but at lower concentration insufficient to be detected in the seismic data.
 Several possible mechanisms exist to explain seismic blanking: (1) occurrence of free gas, (2) increased porosities, (3) destruction of sediment layers due to fluid flow, (4) formation of hydrate and/or (5) surface-carbonate related transmission loss.
The occurrence of free gas is relatively unlikely although active gas bubbling at vents was observed at several other locations [e.g., Suess et al., 1999]. The multichannel seismic data showed that several layers could be traced through the blank zones and considerable coherent reflectivity is also observed below the blank zones. This would be difficult to explain if free gas were present inside the blank zone. Since the entire sediment column lies well within the hydrate stability field, free gas would be transformed immediately into hydrate. Free gas can only be present if hydrate formation depleted all the available pore water in the sediments. If this were the case, the increased hydrate formation would be detectable with a large increase in P wave velocity. A strong pull up of sedimentary layers would be the effect, if they can be traced through the zone. No such strong increase in P wave velocity was observed, and the apparent pull up seen in some seismic data is more likely related to topography around blank zone 1 and related diffractions. This pull up was observed only for seismic lines crossing the vent in a NW-SE direction but not in the perpendicular direction.
An alternative general explanation of the blanking effect of seismic energy and the increase of amplitude reduction with increasing frequency is an increase in porosity [Zühlsdorff et al., 1999]. However, an increase in porosity of ∼25% as suggested by Zühlsdorff et al. to explain the amount of blanking would significantly decrease the P wave velocity. No such decrease was detected from the multichannel seismic velocity analysis.
Enhanced fluid migration can result in destruction of sediment layers. An often observed feature related to fluid/gas migration is the formation of a diatrema [Brown, 1990] and/or collapse structure (pockmarks). Several layers could be traced through the blank zones and with an exception of blank zone 1 no surface expression (mound or pockmark) was observed excluding this mechanism as a possible explanation for blanking.
Significant formation of massive hydrate throughout the entire blank zone or fault is unlikely; otherwise, it would result in a strong velocity effect as mentioned above. However, not the entire blank zone has to be filled with hydrate to create blanking. A clear association of hydrate formation with sandier layers was observed at the Mallik gas hydrate well [Dallimore et al., 1999] probably because the sandier layers usually have higher permeability than the surrounding silts and clays. Hydrate formation generally increases the impedance of the sediment because of an increase of P wave velocity by hydrate replacing the pore fluid and only little change in density. Several sandier layers of mostly 5–10 cm thickness have been observed in piston cores and during ODP drilling at Leg 146 [Westbrook et al., 1994]. It can be speculated that a series of those thin layers (lenses) filled with relatively more hydrate than the surrounding material results in an overall transmission loss and blanking. An example of such a hydrate layer/lens was observed as a cap reflector at the mud mound.
During the video observation survey extensive carbonate pavements were observed around blank zone 1. This results in an increased transmission loss at the surface. However, carbonates have not been observed at the other blank zones. Surface related transmission loss could therefore be only a local enhancement of blanking.
 A prominent high-amplitude layer was observed in the 3.5-kHz and DTAGS data at depths of 6–8 m bsf. The high-amplitude layer might indicate a single turbidite event with different physical properties. Initial analyses on piston cores recovered in July 2000 [Spence et al., 2000; Novosel et al., 2000] suggest a possible relation between the high-amplitude layer and (1) the occurrence of an organic carbon-rich layer, (2) the depth at which extensive gas expansion cracks occurred in the cores after recovery, and (3) the base of the sulfate reduction zone as determined from pore fluid geochemistry analyses.
 The 2-D viscoelastic AVO (VAVO) modeling explained the observed AVO trend as effect of attenuation inside the blank zone and at its vertical boundaries. The attenuation (Q = 200) used to model seismic wave propagation completely inside the blank zone was not sufficient to explain the observed decrease in reflection coefficient for the location at the boundary of the blank zone. At this site the observed AVO trend required attenuation that was ∼4 times higher (Q = 55). This increase in attenuation could be the result of additional loss for those rays crossing the boundary. The boundary of the blank zone might be a rough surface resulting in additional scattering loss.
 In the vertical incidence seismic data, blank zones were often surrounded by high-amplitude rims. Wood et al.  suggested that the increase of seismic amplitude could be the effect of edge diffractions at the boundary of the blank zone, where the reflectors disappear. However, particularly for the most recent 2000 3-D data, not all blank zones are surrounded by high-amplitude rims, which may indicate that there is not a sharp transition. These structures could be artificial effects of the seismic data acquisition.
 Several blank zones of reduced seismic reflection amplitude have been observed on seismic data sets using different recording systems and source frequencies. The amount of amplitude reduction increases with seismic frequency but also depends on the angle of incidence of the seismic rays passing through the blank zone.
 One blank zone showed a distinct surface expression, a circular mound with a height of ∼6 m. On the basis of piston coring and seafloor video observations this feature is interpreted as mud/carbonate mound formed as a result of fluid and methane venting. Below the seafloor this blank zone is associated with a circular high-amplitude rim with a diameter of ∼400 m. Hydrate was recovered in piston cores at the center of the mud mound, and the top of the hydrate occurrence was mapped in the 2000 3-D seismic survey. The hydrate forms a cap with the shallowest parts of ∼3–4 m in the center of the zone.
 The other blank zones are linear features, following an east–west trend across the area of investigation. Two of these zones are ∼80–120 m wide, whereas one extends nearly 300 m in width. These zones are associated with near-surface faults resulting from differential uplift of the sediments along westward dipping thrust faults.
 The blanking of seismic energy is believed to be mainly an effect of (1) formation of hydrate in lenses or layers and (2) surface related scattering and transmission loss at seafloor carbonate formations (for blank zone 1).
 The authors want to express many thanks to Seismic Micro Technology for donating the 3-D interpretation software package “The Kingdom Suite” to the University of Victoria. We also would like to thank Bob MacDonald and Ivan Frydecky for their support during data acquisition. Contribution 2000 080 of the Geological Survey of Canada.