Current estimates for the amount of methane trapped below gas hydrate provinces remain highly speculative, and explanations for how this methane is injected into the atmosphere are wide-ranging and unverified. Blake Ridge, one of the largest passive margin gas hydrate provinces on Earth, is traditionally characterized as an expansive yet dilute reservoir of methane hydrate with no significant fluid advection. Previous 2-D seismic analysis and Ocean Drilling Program Leg 164 drilling results show evidence for both concentrated zones of hydrate and possible fluid flow; however, the extent of these phenomena remains ambiguous. Here we analyze high-resolution 3-D seismic data collected at Blake Ridge in 2000 and map seismic indicators of concentrated hydrate and fluid flow. We also use the seismic data to map the base of the gas hydrate stability in 3-D. Our analysis demonstrates that the gas hydrate phase boundary varies significantly in areas of high sedimentation and erosion, suggesting a dynamic hydrate system. Furthermore, evidence of localized bottom-simulating reflector shoaling, particularly at a sediment wave bounding surface, indicates ongoing advection. The analysis reveals that the Blake Ridge gas hydrate system is significantly more dynamic than previous studies suggest, and we hypothesize that fluctuating sedimentation and erosion patterns cause hydrate phase-boundary instability that triggers fluid flow.
 Methane hydrate, a solid ice-like substance consisting of methane and water, abounds beneath continental margins and may be the cause of significant past carbon isotope excursions and global warming events [Dickens et al., 1995, 1997a; Dickens, 2001, 2003; Kennett et al., 2000; Kvenvolden, 1993; Milkov, 2004]. The base of the methane hydrate stability zone is usually distinguished in seismic data by a strong bottom-simulating reflector (BSR), which represents the methane hydrate/free gas phase boundary, below which temperatures are too high for hydrate to remain stable and methane gas exists [Sloan, 1998]. The BSR, which can cross-cut strata and usually mimics the shape of the seafloor, results from the high impedance contrast between hydrate-laden sediment above and methane gas below [Holbrook et al., 1996; Shipley et al., 1979; Tucholke et al., 1977].
 Early estimates suggested up to two-thirds of the world's organic carbon is trapped in methane hydrate [Ginsburg et al., 1993], and that rapid dissociation and release of this methane can result in elevated atmospheric carbon concentrations, triggering or enhancing global warming [Dickens et al., 1995; Dickens, 2001; Kennett et al., 2003; Svensen et al., 2004]. Recently, however, several revised estimates based on seismic studies, drilling results, and large-scale statistical analysis indicate that the global inventory of carbon trapped in hydrate may be only a quarter of the original estimate [Dickens et al., 1997b; Holbrook et al., 1996; Milkov et al., 2003]. Accurate estimates of hydrate/free-gas concentrations require more detailed seismic and drilling analysis where hydrate and free-gas concentrations are sampled and directly detected [Tréhu et al., 2004], thereby improving statistical extrapolation [Dickens et al., 1997b; Milkov et al., 2003]. One way of obtaining better constraints on methane quantities at hydrate provinces is by seismically imaging the region in 3-D. By detecting methane hydrate and accurately estimate hydrate concentrations using 3-D seismic images in conjunction with drilling results, we can constrain volumetrically the quantity of methane hydrate in the region.
 To date, there is no evidence for recent large-scale perturbations in temperature or pressure at Blake Ridge. However, sedimentation and erosion can cause significant changes in both pressure and temperature that can affect the gas hydrate phase boundary [Martin et al., 2004]. During sedimentation, buried sediments warm as the thermal gradient equilibrates; conversely, during erosion, sediments cool to equilibrate. If sedimentation and erosion is sufficiently rapid, associated changes in temperature and pressure result in a dynamically shifting hydrate stability zone, thereby shifting the BSR [Townend, 1997; Henrys et al., 2003].
 Here, using new 3-D seismic images of the Blake Ridge BSR, we show that sedimentation and erosion patterns create a dynamic and unstable hydrate stability zone in the region, and may drive advection. Furthermore, with the 3-D images, we make improved volumetric estimates of hydrate below Blake Ridge, and we use these estimates to refine and improve methane hydrate estimates for the region.
2. Geologic Setting
 The Blake Ridge contourite drift, located off the southeastern US coast, represents one of the largest methane hydrate provinces on Earth [Holbrook et al., 1996] (Figure 1). The surface morphology of Blake Ridge is controlled by the western boundary undercurrent, which erodes sediment from the eastern flank of the ridge and redeposits it on the western half [Jenkins and Rhines, 1980] (Figure 1). The northern ridge crest is overlain with kilometer-scale eroded sediment waves [Holbrook et al., 2002]. Typical sediment waves appear roughly symmetrical in 2-D cross section and migrate upslope toward a current source [Lee et al., 2002; Normark et al., 1980]; however, the sediment waves on the crest of Blake Ridge appear asymmetric and migrate to the southwest, nearly perpendicular to the direction of the western boundary undercurrent (Figure 1) [Hackwith, 2002; Holbrook et al., 2002]. The unusual shape of the sediment waves along the southern edge of Blake Ridge (as compared to the more continuous and evenly spaced sediment waves further north) likely results from complex ocean currents created by the overriding Gulf Stream mixing with the Western Boundary Undercurrent at intermediate bottom waters [Bryan, 1970] (Figure 1). Three ODP Leg 164 drill sites (Figure 1) drilled in 1995 are located approximately 10 km southeast of the sediment wavefield. Drilling results reveal near-uniform fine-grained silt and mudstones to depths of 700 mbsf and coring and seismic results indicate these sedimentary conditions exist across the study area [Paull et al., 1996]. ODP Leg 164 results suggest that no first-order changes in sediment grain-size, porosity, or seismic velocity exist across the region, except where hydrate or methane gas was present [e.g., Collett and Ladd, 2000; Holbrook et al., 1996]. Comparison of seismic data collected at the drill sites with Leg 164 drilling results suggests first-order seismic anomalies can be directly linked to changes in concentration of methane hydrate and free gas [Holbrook et al., 1996; Collett and Ladd, 2000; Paull et al., 1996, Lee and Dillon, 2001].
 Previous seismic studies document a clear BSR below most of the crest of Blake Ridge, particularly where sediment waves are absent (Figure 2) [Dillon et al., 1998; Holbrook, 2001]. In contrast, the BSR is patchy or absent below the sediment wavefield and below the eastern erosional flank where kilometer-wide eroded sediment waves extend below the hydrate stability zone [Holbrook et al., 2002]. Several hypotheses have been proposed to explain why less gas exists below the eroded sediment waves. On the basis of single-channel seismic data, Dillon et al.  interpreted the sediment waves as normal faults, and suggested that these faults represent a zone of seafloor collapse caused by catastrophic gas escape. More recently, Holbrook et al.  used high-resolution 2-D multichannel seismic data to show that the previously interpreted “faults” are eroded sediment waves, and also suggested that the area represents a region of significant gas escape. Holbrook et al.  suggest that fluid flow and gas escape occur along high-permeability pathways controlled by bedding orientation and exposed by erosion. Though the two theories differ in geological interpretation, both suggest significant quantities of methane escaped from beneath Blake Ridge [Dillon et al., 1998; Holbrook et al., 2002]. Beside the lack of clear evidence that gas once existed at this site, an unresolved problem with each hypothesis is that no direct evidence exists for ongoing or recent fluid flow below sediment waves, with both theories assuming gas escaped simply because less gas is observed below the region.
3. Data Collection and Processing of the Blake Ridge 3-D Volume
 We collected high-resolution 3-D multichannel seismic data over a ∼370 km2 area of Blake Ridge on the R/V Maurice Ewing during the fall of 2000. The location of the 3-D volume, which images the southeast corner of the Blake Ridge sediment wavefield, the eroded eastern flank of the ridge, and the southeast crest of Blake Ridge, was chosen because it offers unique characterization of several geological features (Figure 1). Specifically, the volume enables 3-D comparison of the geologically more simple and well studied southeast crest of Blake Ridge characterized by continuous reflections and a strong BSR, with the more geologically complex eroded eastern flank and the sediment wavefield, characterized by significantly less gas and an intermittent BSR [Dillon et al., 1998; Holbrook, 2001; Holbrook et al., 2002]. Notably, the 3-D volume images regions previously interpreted to have both high and low methane gas concentrations [Dickens et al., 1997b; Gorman et al., 2002; Holbrook et al., 1996; Hornbach et al., 2003], as well as the transition between these zones. The volume was collected less than 10 km north of three ODP Leg 164 drill sites, and 2-D MCS lines tie ODP borehole data to the 3-D volume (Figure 1).
 To create the 3-D volume, we shot 88 2-D seismic lines spaced 75 meters apart over the survey area. The seismic receiver consisted of a 4 km streamer with 324 channels grouped in 12.5 m intervals. We towed the streamer ∼5 m below sea level behind the ship, which steamed at ∼5 knots. The seismic source consisted of two high-frequency generator-injector (GI) air guns, configured at 105/105 in3, which were also towed at ∼5 m depth and fired every 37.5 m.
 We ensured good seismic coverage of the 3-D survey by implementing real-time graphic displays of streamer location during shooting, which we obtained by attaching differential GPS units to the tail-buoy of the streamer and the stern of the ship. Using the GPS units in conjunction with multiple compasses located on streamer birds, we were able to reconstruct in real time the location of the streamer during 3-D shooting and obtain accurate navigation, fold, and geometry information.
 We created the 3-D seismic volume with bin sizes of 37.5 m in the inline direction and 75 m in the crossline direction (perpendicular to ship track). We normalized shot gather traces and sorted them into bins. The fold for 37.5 × 75 m binned volume exceeds 40 for >95% of the survey area, and this high fold was vital for proper velocity analysis. We determined stacking velocities by semblance analysis, picking approximately every 25th common midpoint in the inline direction every other inline. We used these velocity picks to create the initial smoothed 3-D velocity model. We then created an initial prestack depth migration of the volume using Paradigm Geophysical software. Migration aperture in the inline and crossline direction was 2000 m and 150 m, respectively. Larger apertures resulted in reduced signal-to-noise caused by spatial aliasing. The 75 m crossline trace interval produced an acceptable compromise between high fold data and limited spatial aliasing—the fold often exceeds 40 traces per bin, yet features with dips of 10 degrees or less are properly imaged [Yilmaz, 1999]. After migration, we band-pass filtered and normalized all traces to help correct trace imbalances.
4.1. Three-Dimensional Structure of Buried Sediment Waves at the Eastern Flank of Blake Ridge
 Though multibeam data (Figure 1) offer valuable insight into the surface structure of sediment waves at Blake Ridge, the new 3-D seismic images allow us to track and map sediment waves extending hundreds of meters below the seafloor. Using high-resolution multichannel 2-D seismic Data, Holbrook et al.  show that many sediment waves at Blake Ridge extend below the gas hydrate phase boundary (>400 m below the seafloor). We make similar observations, particularly east of the sediment wavefield along the eastern flank of Blake Ridge, where the 3-D seismic data clearly show several buried sediment waves (Figure 2). Like the sediment waves described by Holbrook et al. , the buried waves on the eastern flank of the ridge display sigmoidal shapes that are erosionally truncated, with onlapping overlying strata. Although these features were originally interpreted as faults [e.g., Dillon et al., 1998], the true dip of these features, which we can accurately measure with 3-D data, is <7o, and the boundaries merge into continuous strata at their base, suggesting a sedimentary origin (Figure 2). The buried sediment waves along the eastern flank of the ridge generally trend northwest-southeast similar to those observed on the crest of the ridge. Age-depth constraints on the 3-D stratigraphy obtained from ODP Leg 164 indicate that the sediment waves on the eastern flank of the ridge are more than 6 Ma old, and that these long-lived features have likely been migrating east-to-west across the region [Hackwith, 2002]. The seismic images therefore demonstrate that buried sediment waves extend well east of the crest of the ridge and that the sediment wavefield is much larger than seafloor bathymetric data suggest (Figures 1 and 2). Seafloor erosion, by smoothing the surface of the eastern flank, has simply masked the surface expression of the eastern sediment waves.
4.2. Extent of BSRs
 The BSR is recognized in the 3-D volume as a clear negative impedance reflection that generally mimics the shape of the seafloor (Figures 2 and 3) . The data show a strong BSR with a thick free gas zone below the southwestern corner of the volume as indicated by the high-amplitude anomalies below the BSR, particularly across the crest of Blake Ridge in areas where no sediment waves exist (Figure 3). In contrast, the BSR is patchier across the eastern flank of the ridge and below the sediment wavefield (e.g., Figure 2). With strong BSR reflections across the crest, and weaker or no BSR reflections along the eastern flank and in the sediment wavefield, the 3-D volume embodies the general character of the BSR across the ridge [e.g., Dillon et al., 1993; Holbrook et al., 2002; Hackwith, 2002].
 Previous studies indicate that Blake Ridge, with its generally homogenous stratigraphy [Paull et al., 1996], simple geologic structure [e.g., Dillon et al., 1993; Hackwith, 2002], and limited geophysical variability [e.g., Lee et al., 1994; Ruppel, 1997], represents an ideal natural laboratory to isolate and study gas hydrate [Holbrook et al., 1996; Holbrook, 2001; Paull et al., 1996]. Covering the central crest, western flank, and southern boundary of the sediment wavefield, the 3-D volume captures the geology of Blake Ridge and is representative of the region. Previous estimates of BSR extent, based on statistical interpolations of widely spaced single-channel 2-D seismic lines, suggest a BSR extends over approximately 50% of the hydrate province [e.g., Dillon et al., 1993; Holbrook et al., 1996]. We determined the areal extent of the BSR within the 3-D volume by picking the BSR horizon along every 3-D inline wherever it was visually detectable (Figure 4). The total area of the 3-D survey is 376 km2, and the BSR extends over at least 286 km2, or ∼76% of the volume (Figure 4b). Though the volume admittedly covers only a small area of the Blake Ridge hydrate province, the analysis, if representative of the region, suggests that the extent of the BSR at Blake Ridge may be 50% higher than previous estimates.
4.3. Three-Dimensional Images of Concentrated Gas Hydrate Zones: Lens and Stringer
 Using the 3-D data set, we map the Blake Ridge “lens” [Hornbach et al., 2003] and “stringer” [Anton et al., 1999], two features interpreted as zones of concentrated hydrate (Figures 3 and 5) . Like the BSR, the lens is a bright reflector that crosscuts strata, however, it extends upward from the BSR into the hydrate stability zone, and detailed velocity analysis suggests the lens may consist largely of hydrate [Hornbach et al., 2003]. One explanation for lens formation is that it formed in place by seafloor erosion which converted free gas below the BSR to gas hydrate as the temperature equilibrated [Hornbach et al., 2003].
 The lens clearly extends from north to south across the eastern flank of the 3-D volume, covering an area of ∼30 km2 (Figure 5). Assuming an average vertical thickness of 75 m, and an average bulk hydrate concentration of 17% by volume (previous estimates suggest 13–22% bulk hydrate; Hornbach et al. ), the lens contains 4.5 × 1010 kg of methane.
 The stringer, a feature that covers a significantly greater area than the lens (Figures 3 and 5) is also characterized by anomalously high amplitudes within the hydrate stability zone. The cause the stringer reflection remains unclear. Although evidence exists that the stringer represents a zone of concentrated hydrate [Anton et al., 1999; Nealon, 2005], we cannot rule out the possibility it represents an unconformity or a zone of enhanced carbonate concentrations. The stringer dips ∼1° to the northwest (toward the sediment wavefield), and intersects the BSR along its western edge (Figures 3 and 5). As noted in previous work [Anton et al., 1999], where the stringer intersects the BSR, a continuous high-amplitude gas-charged reflector extends below, and both the stringer and gas-charged zone follow the same stratal boundary, indicating that gas may be migrating laterally along high-permeability strata into the hydrate stability zone (Figure 3).
 Though hydrate concentrations in the stringer are poorly constrained, waveform inversion analysis of similar amplitude anomalies in the region suggest hydrate concentrations as high as 30–42% [Gorman et al., 2002; Korenaga et al., 1997], and preliminary waveform inversion analysis on the stringer suggests 10% bulk hydrate [Nealon, 2005]. The stringer covers 70 km2 in the 3-D volume (Figure 5), extends north-south across the volume and has an average thickness of 20 (±10) m [Nealon, 2005, Gorman et al., 2002]. Assuming an average of 10% bulk hydrate exists within the stringer and that the stringer reflection indeed represents an increase in hydrate concentration, we estimate the total methane in the stringer at ∼2 × 1010 kg.
 Previous methane estimates for the Blake Ridge extrapolated drilling results, single channel 2-D seismic data, and vertical seismic profile data across 10,000 km2. These area estimates suggest an average hydrate concentration of 300 kg/m2 across Blake Ridge, with the majority located in 20- to 50-m-thick zones found directly above the BSR [Holbrook et al., 1996; Holbrook, 2001; Paull et al., 1996]. In contrast, we estimate an average hydrate concentration within the 3-D volume from the stringer and lens alone of 450 kg/m2. If the stringer and lens do represent concentrated zones of hydrate, they constitute the largest sources of methane within the volume. It may be plausible to add the previous estimate of 300 kg/m2 to our estimated concentration, because the previous estimate represents an average background value in areas where hydrate is not directly detected in seismic data. Thus, we suggest that the average hydrate concentration within the volume may be as high as ∼750 kg/m2.
 There remains some debate regarding whether the stringer and lens represent true concentrated zones of hydrate or another form of sediment diagenesis that cements sediment and increases p-wave velocity. Though it is difficult to determine remotely (i.e., without drilling) whether the lens and stringer consist of hydrate, previous drilling results and logging at the Blake Ridge ODP sites (994, 995, 997) indicate that anomalously high seismic velocities at Blake Ridge correlate directly with high hydrate concentrations, and do not correspond to increases in carbonate content [Collett and Ladd, 2000; Lee, 2000]. Nonetheless, ODP site 104, drilled ∼100 km south of site 994, and the only site drilled on the eastern flank of the ridge, encountered very hard ankerite just above the BSR [Hollister et al., 1970]. Future drilling will ultimately determine whether high-amplitude high-velocity seismic features like the lens and stringer on Blake Ridge are indeed concentrated hydrate deposits.
5. Blake Ridge Geomorphology: Constraining Erosion and Sedimentation
 By analyzing 2-D seismic profiles that tie stratigraphic horizons of known age from ODP drill sites to the 3-D volume, we place constraints on sedimentation rates at the Blake Ridge 3-D volume and their spatial distribution (Figure 6). ODP Leg 164 drilling shows that sedimentation rates for the past 3 Ma south of the sediment wavefield average from 40 to 70 m/Ma [Paull et al., 1996]. Here, we track the location of the 1 Ma and 3 Ma sedimentary horizons, which are continuous and distinguishable across the western half of the 3-D volume. These horizons can only be mapped through the southwestern portion of the volume because erosion has removed them from the eastern flank and parts of the sediment wavefield [Hackwith, 2002]. Nonetheless, where they are mapped, the dated horizons place important constraints on sedimentation rates.
 Isopachs between the seafloor and dated subsurface horizons provide a way to visualize sedimentation rates. The isopach map between the seafloor and the 3 Ma horizon shows that the thickest sediment (and therefore areas of highest average sedimentation) occurs along the northwest region of the 3-D volume, at the extreme southern edge of the sediment wavefield (Figure 6). Sediment thickness between the seafloor and the 3 Ma horizon exceeds 400 m in this region, indicating average sedimentation rates of 140 m/Ma. This rate is 2–3 times that observed 10–15 km south at ODP sites. The variable sedimentation rates within the sediment wavefield likely result from complex ocean currents in the region [Bryan, 1970].
 Erosional scours are identified in the isopach maps, particularly along the northwest corner of the volume (Figure 6). These scours correspond to truncated eroded sediments that outcrop at the seafloor (Figures 2 and 6). Though we cannot constrain erosion rates on Blake Ridge using the seismic data, we can determine where erosion occurs (Figures 2 and 6). The 1 Ma horizon is truncated and terminates approximately midway between the eastern and western boundaries of the 3-D survey, indicating that seafloor morphology along the eastern half of the 3-D volume is generally controlled by erosion (Figure 6). The termination pattern of the age-date horizons in the sediment wavefield, especially at the northern edge of the 3-D survey, also indicates that erosion shapes much of the sediment wavefield (Figures 6b and 6c). The sediment wavefield therefore represents a dynamic area currently undergoing both rapid sedimentation and significant erosion.
 The isopach map between the seafloor and 1 Ma horizon reveals a similar pattern as the 3 Ma isopach, with highest sedimentation rates along the southern edge of the sediment wavefield, and discontinuous but significant sedimentation in the northwest corner of the volume. The 1 Ma isopach at its thickest point approaches 200 m, indicating sedimentation rates of ∼200 m/Ma at some locations along the southern edge of the sediment wavefield (Figure 6). Given that the 3 Ma isopach exhibits significantly less variability than the 1 Ma isopach, sedimentation and erosion appear to vary over short (<1 Ma) timescales.
6. Evidence for Dynamic BSRs and Focused Advection at Blake Ridge
 As noted in previous studies, where erosion occurs on Blake Ridge, less gas exists below [Holbrook, 2001; Holbrook et al., 2002]. Holbrook et al.  noted that the BSR is discontinuous or weaker where eroded sediment waves intersect the gas hydrate phase boundary. Here, we observe the same phenomena both in the sediment wavefield and along the eastern flank of the ridge. In general, where eroded sediment waves extend below the hydrate stability zone the BSR is weak or absent (Figures 2, 3, and 6).
 Understanding why the BSR is generally discontinuous or nonexistent where sediment waves intersect the gas hydrate phase boundary remains a challenge. Previously, both Holbrook et al.  and Dillon et al.  suggested that the lack of gas below these regions indicates recent fluid flow and gas escape. To date, however, there is no direct evidence for fluid flow along sediment wave bounding surfaces, and this hypothesis has been difficult to test. By analyzing the shape, depth, continuity, and stability of the gas hydrate phase boundary at Blake Ridge in 3-D, we evaluate whether steady state conditions prevail at the BSR, or if the phase boundary is dynamic and potentially conducive to fluid flow. Specifically, we explore the effects of sedimentation and erosion on the BSR to investigate whether it is at steady state.
6.1. Effects of Erosion and Sedimentation on BSR Depth
 With sedimentation rates in the 3-D volume constrained for the past 3 Ma and the location of seafloor erosion known (Figure 6), we can calculate the effect sedimentation and erosion have on subsurface temperatures and BSR depth. Qualitatively, we expect that in regions where ongoing sedimentation outpaces conductive heat transfer, sediments will be anomalously cool, and the BSR may be deeper than predicted by a steady state geotherm (see left and center panels of Figure 7a) [Townend, 1997; Henrys et al., 2003; Martin et al., 2004]. Conversely, in regions undergoing erosion, the subsurface temperature will be higher than expected, resulting in an anomalously shallow BSR (see left and center panels of Figure 7b).
 We can make these assumptions because the methane hydrate phase boundary (i.e., BSR) is dependent primarily on temperature and to a lesser extent on pressure and salinity [Dickens and Quinby-Hunt, 1994, 1997; Sloan, 1998]. For example, at Blake Ridge, a 1% change in salinity and a 100 m change in seafloor depth results in only a ∼1.5 and ∼5 m shift in the BSR, respectively [Dickens and Quinby-Hunt, 1994, 1997; Sloan, 1998]. However, a 1°C change in the thermal gradient at Blake Ridge shifts the BSR a significant ∼15 m [Sloan, 1998].
 Here, we use a transient 1-D thermal model adapting the methods of Martin et al.  and Benfield  to calculate temperature as a function of depth in the presence of sedimentation or erosion [Benfield, 1949; Martin et al., 2004]. We assume an initial starting depth of 2800 meters below sea level (mbsl), with a constant seafloor temperature of 3.5°C and an initial thermal gradient of 34.3°C/km, defined as the average of measured values during ODP Leg 164 [Ruppel, 1997]. We define sediment thermal conductivity, porosity, and density from ODP Leg 164 results. We consider a range of sedimentation (and erosion) rates from 0 to 500 m/Ma, for a time period of 0–3 Ma (e.g., Figure 8).
 The greater the sedimentation rate, the lower the temperature at a given depth below the seafloor. For example, if no sedimentation occurs, the temperature 500 m below the seafloor is ∼21°C, whereas the temperature at this same depth in a region that has experienced 500 m of sedimentation over the past 1 Ma is ∼17°C (Figure 8). The analysis shows that sedimentation and erosion at Blake Ridge can have real and observable effects on BSR depth (Figures 7 and 8). For example, low sedimentation rates (∼75 m/Ma) over short time periods (∼1 Ma) can cause significant subsurface temperature changes that result in observable (∼20 m) BSR shifts (Figure 7).
 A fundamental problem with determining if erosion and sedimentation cause anomalous BSR shoaling or deepening at Blake Ridge is that the true steady state equilibrium BSR depth is poorly constrained [Henry et al., 1999; Ruppel, 1997]. To address this problem, we compare observed BSR depths to a “baseline” BSR, which we calculate by taking the average measured thermal gradient from nearby ODP Sites 994, 995, and 997 [Ruppel, 1997]. Measured thermal gradients show generally uniform background heatflow throughout the region with an average value of 34.3 ± 2.3°C/km, equivalent to ∼30 m of error in BSR [Ruppel, 1997]. Therefore, if sedimentation or erosion do not significantly affect subsurface temperature, the observed BSR should be in equilibrium and run approximately parallel to, and within 30 m of, the baseline BSR (Figure 9). Deviations from this behavior indicate variations in gas hydrate stability. If sedimentation and erosion are occurring at high enough rates to impact subsurface temperature, the BSR below these areas will not be in equilibrium and the observed BSR will deepen or shoal relative to the “baseline” BSR (Figure 9).
6.2. Evidence for a Dynamic BSR Within the Blake Ridge 3-D Volume
 Assuming a thermal gradient of 34.3°C/km, a bottom water temperature of 3.5°C, and a constant salinity of 33.5‰, we use Sloan's  technique to predict a steady state BSR depth for pure methane hydrate. In areas where erosion and sedimentation dominate, the BSR shallows and deepens, respectively, to the baseline steady state BSR (Figure 9). For example, the BSR shallows strikingly along the eastern half of the volume with respect to the baseline BSR and the seafloor, coinciding exactly with the observed trend of decreasing sedimentation and the onset of erosion from west-to-east (Figure 9). In areas where sedimentation is highest, BSR depth is deepest, and where erosion dominates (along the far eastern flank of the ridge), the BSR shallows most. Here, the BSR shallows with respect to both the baseline BSR and the seafloor (Figures 9b and 9d). Though changes in thermal conductivity or topographically driven fluid advection [e.g., Dugan and Flemings, 2000] cannot be ruled out as possible explanations for the observed BSR shoaling along the eastern flank of the ridge, the fact that erosion dominates where BSR shoaling occurs and sedimentation prevails where the BSR deepens, is consistent with erosion/sedimentation controlled heat flow fluctuations (Figures 7–9). We calculate that lateral heat flow (thermal refraction) caused by the slightly (∼2 degree) dipping eastern flank of the ridge is less than 1 degree/km, well below the estimated error in the vertical thermal gradient of ± 2.3 degrees. Thus, thermal refraction affects cannot easily explain this discrepancy. The high degree of variability in these areas implies that the BSR, and therefore, the gas hydrate system is not in steady state, but instead, fluctuating as erosion and sedimentation shift the thermal gradient. Our analysis therefore suggests temperature instabilities evolve directly from variations from sedimentation and erosion, and that these variations play a primary role in creating a dynamic gas hydrate system at Blake Ridge.
6.3. Evidence for Focused Advection on the Eastern Flank of Blake Ridge
 We focus on the eastern flank of the 3-D volume, where the seafloor dips roughly uniformly <2 degrees to the east-northeast and lacks significant variations in seafloor bathymetry that may cause thermal refraction. Thus, as with past studies in the region [Wood et al., 2002], any anomalous small-scale (tens of meters of vertical offset over <1 km2) BSR variation may be attributed to advection.
 Though the BSR is intermittent and weak across parts of the eastern flank of the ridge, we were able to map the BSR in detail in the vicinity of the hydrate lens, where the BSR is a strong continuous reflector (Figure 10). We picked the BSR along every inline and merged these picks to produce a high-resolution gridded 3-D image of the BSR across the region (Figure 10a). The 3-D image shows the significant topographic variability of the BSR below the eastern flank of Blake Ridge in high resolution (Figure 10).
 With no advection and relatively uniform erosion, the BSR should exhibit little (<10 m) relief over small (<1 km) distances. Though much of the BSR appears continuous and relatively uniform across the eastern flank of the ridge, our map reveals that the BSR is both discontinuous and varies significantly in depth over short offsets (Figures 10b, 10e, and 10f). In particular, a sharp (>20 m) continuous northwest-southeast running step in the BSR occurs across the eastern third of the mapped region. Seismic lines indicate that slightly dipping (∼4 degrees) high-amplitude gas-charged sediments converge directly at this step in the BSR. Thus, one possible explanation for the BSR step is that these gas-charged sediments have higher porosity/permeability than the surrounding strata, such that fluids are preferentially channeled and flow along these sequences, resulting in an anomalously shallow gas hydrate phase boundary within these sediments (Figure 10).
 In addition, we observe anomalous BSR shoaling where a sediment wave boundary intersects the hydrate stability zone just east of the BSR step (Figures 10b–10h). At this intersection, the BSR generally shallows or disappears (Figures 10a, 10b, 103, and 10f). The BSR is, however, cleanly mapped across the sediment wave boundary at the north and south ends of the volume, and at both locales, the BSR shallows significantly (>20 m), implying higher temperatures, and therefore, possible upward fluid flow along the boundary (Figures 10b, 10g, and 10h).
 Based in part on the apparent evacuation of gas below them and on the premise that higher permeability may exist parallel to bedding [e.g., Clennell et al., 1999], Holbrook et al.  suggested that sediment wave boundaries buried below the Blake Ridge sediment wavefield may channel fluids due to hydraulic conductivity contrasts. Here, observations of BSR shoaling imply upward advection and ongoing fluid flow along sediment wave boundaries.
 The 3-D data, which to date offer the most thorough assessment of gas hydrate and free gas on Blake Ridge, reveal widespread and relatively continuous high-amplitude seismic reflections associated with hydrate and free gas. Our analysis implies that past estimates for the density of hydrate in the region may be low by as much as a factor of 2, assuming the observed seismic indicators are in fact hydrate, however, we caution that significant uncertainty in hydrate concentrations at this site remains. Future drilling of these proposed high-concentration gas-hydrate sites will ultimately verify our estimate of methane hydrate in the region.
 Though Blake Ridge is perhaps considered the best example of simple steady state hydrate province, our analysis reveals that moderate sedimentation and erosion creates a dynamic and heterogeneous hydrate system. Furthermore, our analysis suggests ongoing advection at the gas hydrate phase boundary, particularly along the eroding eastern flank Blake Ridge. BSR shoaling occurs along buried sediment wave bounding surfaces and gas charged sediments, implying that these features act as fluid conduits.
 We observe significantly less gas, scattered BSRs, dynamic gas hydrate phase boundary conditions, and evidence for advection in areas undergoing significant sedimentation or erosion. Given that (1) varying sedimentation patterns can cause changes in subsurface pressure and temperature that lead to mechanical failure [Flemings et al., 2003; Hornbach et al., 2004], and (2) hydrate dissociation rates caused by erosion impact the amount of excess pore pressure generated in the subsurface [Xu and Germanovich, 2006], it seems reasonable to infer that erosion or sedimentation can instigate fluid migration and that changes in sedimentation and erosion patterns may ultimately trigger gas migration at Blake Ridge. The clear correlation between seafloor morphology changes and gas hydrate phase-boundary variability suggests that regions experiencing the greatest changes in seafloor geomorphology have dynamic and unstable BSRs, and as a result, are likely the areas most conducive to advection.
 We thanks the officers and crew of the R/V Maurice Ewing for a successful cruise. Helpful discussions with Donna Shillington, John Goff, and Jeff Nealon improved this manuscript. We also thank A. Tréhu, M. Huuse, and I. Pecher for constructive reviews. The initial 3-D stack was produced at the University of Wyoming with computer support supplied by Jeff Lang. The final prestack depth-migrated image was produced at the Institute for Geophysics, University of Texas, with the help of Nathan Bangs and Mark Wiederspahn. Funding for this work was provided by the National Science Foundation and the U.S. Department of Energy. Additional funding for completion of this work was provided by the Geology Foundation at the University of Texas at Austin.