( Springer-Verlag 1999

Geo-Marine Letters (1999) 19 : 150d156

J. Posewang ' J. Mienert

High-resolution seismic studies of gas hydrates west of Svalbard

Received: 6 August 1997 / Revision received: 26 January 1998

Abstract A strong bottom-simulating re#ection (BSR) with high-amplitude variations is detectable in highresolution re#ection seismic pro"les west of Svalbard. Above the BSR, anomalously high velocities up to 1840 m/s, calculated from high-frequency ocean-bottom hydrophone (HF-OBH) data, indicate the existence of gas-hydrated sediments. Below the BSR, a low-velocity layer, interpreted as gas-bearing sediments, shows thickness variations from 12 to 25 m. In addition, two other low-velocity layers clearly containing free gas are detected within the classic hydrate stability zone (HSZ) where, a theoretical viewpoint, free gas cannot exist.

Introduction Gas hydrates are ice-like solid crystals that form from gas (mainly methane) and water under conditions of low temperature and high pressure. Due to the low permeability of gas hydrate-cemented sediments, a gas-bearing sedimentary layer can form below the stability zone (e.g., Dillon et al. 1994; Holbrook et al. 1996). The presence of natural submarine gas hydrates is commonly inferred from seismic re#ection data (e.g., Hyndman and Spence 1992). The base of the stability zone for gas hydrates is geophysically identi"ed by the occurrence of a bottom-simulating re#ector (BSR) (Stoll et al. 1971). The BSR is a re#ection at the boundary between the high-velocity gas hydratecemented sediments and the underlying low-velocity gasbearing sediments. Whereas compressional velocity values

JoK rg Posewang ( ) SFB 313, Christian-Albrechts-University, Olshausenstr. 40, D-24118 Kiel, Germany JuK rgen Mienert Department of Geology, University of Troms+, Dramsveien 201, N-9037 Troms+, Norway

of 1700}2400 m/s are known to be typical for gas hydrated sediments (Katzman et al. 1994; Lee et al. 1994; Minshull et al. 1994; Andreassen et al. 1990, 1995), values below 1500 m/s indicate free gas in the pore space. The BSR mimics the shape of the sea #oor, often cuts the dominant stratigraphy, and is characterized by a high, reversed- polarity event (Lodolo et al. 1993; Katzman et al. 1994; Andreassen et al. 1995). In most seismic investigations to date, a signal source with a main frequency lower than 60 Hz was used (e.g., Bangs et al. 1993; Eiken and Hinz 1993; MacKay et al. 1994; Andreassen and Hansen 1995; Boehm et al. 1995). Due to the low resolution, it was impossible to detect the base of the gas re#ection (BGR) below the BSR or the upper limit of the gas hydrated sediments. In this paper we present high-resolution seismic re#ection data from the Fram Strait west of Svalbard to improve our understanding of complex hydrated zones. Seismic sections from multichannel data and detailed velocity information from high-frequency ocean-bottom hydrophones (HF-OBH) data yield new "ndings about the velocity distribution and structure of gas hydrated sediments. The resolution allows us to calculate the thickness of the gas hydrated and gas layers and to detect velocity variations within the gas hydrate stability zone.

Area under investigation The investigation was conducted in summer 1995 on board the R/< A. A. Karpinskiy on the north-west continental slope of Svalbard (Mienert 1994) (Fig. 1). Water depths in the study area vary from 400 m on the upper slope to 2350 m at the northern edge of the Knipovich Ridge. Seven multichannel seismic re#ection pro"les and one HF-OBH pro"le were recorded in order to get information about the seismic structure and velocity of the subsea #oor. Two of these pro"les, K18 (multichannel) and K20 (HF-OBH), are discussed in this paper.

151

varying re#ection strength of the BSR is assigned to constructive or destructive interference between the BSR and the underlying re#ection and to variations in the acoustic impedance contrast along the BSR horizon (Andreassen and Hansen 1995). Velocity analysis of multichannel seismic data shows three distinct layers with di!erent velocities (Andreassen and Hansen 1995). Due to the low resolution, the values give a large-scale velocity trend. A nearly 200-m-thick layer between sea #oor and BSR has an average velocity of 1630 m/s followed by a low-velocity layer 100 m thick (1450 m/s) and a zone of higher velocities. The highvelocity layer above the BSR is interpreted to contain gas- hydrated sediments whereas the low-velocity layer indicates free gas beneath the BSR.

Methods

Fig. 1 Bathymetry and locations of seismic re#ection pro"les on the western continental slope of Svalbard. Pro"les K18 and K20 are discussed in this paper. The inset presents an overview map of the northern North Atlantic with the box showing the study area

In the past, a number of seismic re#ection surveys were carried out in this area. Eiken and Hinz (1993) and Andreassen and Hansen (1995) observed a strong BSR on seismic pro"les that they attributed to gas hydrates. The BSR is traceable from the upper slope down to the deep sea. The subbottom depth ranges from 165 m in a water depth of 860 m to 250 m in a water depth of 2350 m. The

Fig. 2 Correlation between HF-OBH data on the left and the seismic section of K18 on the right. The re#ection horizons are well traceable between both data sets and the BSR as well as the BGR are clearly detectable

A single 2-l airgun source with a shot interval of 10 s was used during the survey. The source wavelet has a main frequency of 80}200 Hz, and the deepest re#ections were recognized in a depth of 450 mbsf. Furthermore, the acquisition system included a towed 6-channel streamer with a 120-m active section and two HF-OBH systems on the sea #oor 700 m apart from each other. HF-OBH pro"le K20 was recorded along the seismic re#ection of pro"le K18 to obtain detailed velocity information about the subsea #oor. In the area around the HF-OBH stations both pro"les are nearly coincident, which is important in order to correlate the velocity information with the seismic section. Figure 2 shows the correlation between HFOBH 5 travel-time curves (left part) and the seismic

152 Fig. 3 Computed and digitized real travel-time curves were brought to match by iteratively varying the layer thicknesses and the interval velocity. Top part shows the successful calculation with correct velocity values, whereas in the bottom part a velocity change of 100 m/s in the layer above the BSR leads to strong di!erences between the curves

section of pro"le K18 (right part). From one section to the other the re#ection horizons are well transferable and the BSR as well as the BGR are clearly detectable. The recorded HF-OBH data represent a series of traveltime curves, i.e., re#ection hyperbolae. The curvature of these hyperbolae is characteristic for the propagation velocity of seismic waves. The calculation of the propagation velocity was carried out by standard trial and error raytracing modeling methods using the algorithm developed by Luetgert (1992). The computed and the digitized real travel-time curves were matched by iteratively varying the model parameters as layer thicknesses and interval vel-

ocities (Fig. 3). In the upper part of the "gure the velocity calculation was successful and the digitized and the computed curves match well. The error in calculating the velocities in the upper 30 mbsf is about $10 m/s, and in depths '300 mbsf about $100 m/s. This is shown in the bottom part of Fig. 3. A velocity change of 100 m/s in the layer above the BSR leads to strong di!erences between digitized and computed curves. The result of the velocity analysis is a detailed velocity} depth model re#ecting the distribution of gas-hydrated and gas-bearing sediments at the HF-OBH stations. In combination with the multichannel pro"les, the BSR is

153

Fig. 4 Section of seismic re#ection pro"le K18 based on a 2-l airgun and a 6-channel streamer. The arrows mark the locations of the HF-OBH stations 5 and 6 at the sea #oor. A strong BSR occurs at a depth of 0.25 s TWT bsf. The BSR parallels the sea #oor, crosses the discordant sedimentary strata, and is characterized by high amplitudes. Below the BSR the base of gas bearing sediments is marked by the BGR

accurately detectable and traceable over large distances along the continental slope. The theoretical vertical resolution of the seismic signals is about 2}6 m, but a vertical resolution of 10 m is more realistic. This resolution enables the detection of the upper and lower boundary of the gas-hydrated sediments, the base of gas-bearing sediments below the BSR, and velocity variations within the gas hydrate stability zone.

Results A well-de"ned, high-amplitude re#ection, the BSR, crosses the discordant sedimentary strata and can be traced continuously along the entire slope west of Svalbard. Figure 4 shows the western part of pro"le K18 (Fig. 1). The arrows mark the locations of HF-OBH stations 5 and 6 at the sea #oor. Due to the small distance to the Knipovich Ridge, which acts as a heat source, the subsea #oor temperature increases downslope. Therefore, the subbottom depth of the BSR decreases with increasing water depth from 0.26 s TWT bsf upslope to 0.22 s TWT

bsf downslope. The BSR exhibits strong amplitude variations. Downslope, it is typically a continuous and strong re#ector. Upslope, it is very weak and parts of it show high-amplitude re#ections below the BSR. Furthermore, the seismic section (Fig. 4) shows a seismic signal below the BSR that changes to lower frequencies. Detailed frequency studies revealed that sedimentary layers below the BSR act as a low-pass "lter on seismic signals. Above the BSR, frequencies up to 170 Hz predominate; below the BSR, frequencies of less than 80 Hz prevail (Posewang 1997).

High-resolution velocity pro"les Figure 5 presents the velocity-versus-depth models of HFOBH station 5/6 in correlation with the empirical model for marine sediments by Hamilton (1980). The water depth at the HF-OBH locations is 1531/1509 m, and with a signal penetration of 300 mbsf, the models resolve 11 layers. The velocity-versus-depth pro"les inferred from the HF-OBH data show several zones of alternating high and low velocities. At HF-OBH station 5 the highest velocities appear at a depth between 50 mbsf and 230 mbsf with values up to 1840 m/s. In comparison to the velocity model of Hamilton (1980), these velocity values are anomalously high, indicating the existence of gas-hydrates. Below these gas hydrated sediments, a pronounced velocity decrease follows. The velocity drops to 1380 m/s, a value lower than

154

Fig. 5 Velocity depth models of HF-OBH station 5/6 correlated with empirical model for marine sediments by Hamilton (1980). The velocity depth models show several zones of alternating high and low velocities. The anomalously high velocity values up to 1840 m/s are an indicator for gas-hydrated sediments, whereas values below 1500 m/s are characteristic of gas-bearing sediments

140 m in the east to 180 m in the west. This zone is interrupted by a 40-m-thick low-velocity layer containing a distinct gas lens. Second, a low-velocity layer follows below the gas-hydrated zone, indicating the existence of free gas. The transition between the high- and low-velocity zone corresponds with the appearance of the BSR. Therefore the BSR clearly represents the boundary between gas- hydrated to gas bearing sediments. Third, strong re#ection amplitude variations occure along the BSR. These variations correlate with thickness variations of the free gas layer below. At HF-OBH station 5, a thin free gas layer appears and the BSR shows small amplitudes, whereas at HF-OBH 6, a thick free gas layer and a strong BSR with high amplitudes can be observed. Fourth, strong lateral velocity variations were detected directly beneath the sea #oor within a 20-m-thick layer. Velocity values change from 1520 m/s in the west to 1220 m/s in the east, indicating the transition from marine to gas- bearing sediments.

Discussion and conclusions the sound velocity of seawater (SVS) caused by gas-bearing sediments. The base of this 12-m-thick layer shows a velocity increase from 1380 m/s to 1600 m/s related to a well-pronounced event in re#ection seismic pro"les, the so-called base of the gas re#ection (BGR) (Camerlenghi and Lodolo 1994) (see also Fig. 2). A special feature in this velocity-versus-depth model is a low-velocity layer, which is sandwiched by gas-hydrated sediments. At a depth of 80 mbsf, the velocity decreases from 1720 m/s to 1660 m/s, a velocity that is not anomalously high according to Hamilton (1980). Consequently, a layer without gas hydrates exists within the classic hydrate stability zone (HSZ). At HF-OBH station 6 anomalously high velocities, with values between 1720 m/s and 1820 m/s, also indicate the existence of gas hydrates down to a depth of 210 mbsf. The base of the low-velocity layer below the gas-hydrated zone, with a thickness of 25 m, is marked by the slightly gradual increase of the velocities to values higher than 1600 m/s. The noteworthy observation in this model is the interruption of the gas-hydrated zone by a low-velocity layer. Related to gas-bearing sediments, the velocity decreases from 1720 m/s to 1400 m/s at a depth of 80 mbsf. Hence, a free gas zone is sandwiched between the gas hydrate-cemented sediments. A further remarkable observation is a distinct low-velocity layer adjacent to the sea #oor. The value of 1220 m/s is the lowest calculated velocity in the working area and is also explained by the existence of free gas in the pore space. Summarizing the results, four prominent features are conspicuous. First, gas hydrate cemented sediments with high velocities exist in the zone between 30 and 210 mbsf. The thickness of the high-velocity unit increases from

The high-resolution velocity analysis reveals an average velocity of 1645 m/s for a 210-m-thick HSZ. These results correspond to the low-resolution records of Andreassen and Hansen (1995), who estimated an average velocity of 1630 m/s between the BSR at 200 mbsf and the sea #oor. The improved knowledge is clearly documented by the high-resolution data and a combination of re#ection seismic and HF-OBH data. According to the experimental gas hydrate stability curves (Sloan 1990), at a water depth of 1530 m and a bottom water temperature of !0.73C (Dietrich 1969), the classic hydrate stability "eld must extend from the sea #oor to the BSR. On the other hand, two low-velocity layers were detected within the classic HSZ. The upper one is directly located below the sea #oor, showing a limited lateral extension. In the transition area of marine to gas-bearing sediments, phase analyses reveal no indications of a phase reversal at the sea #oor. Hence the low-velocity layer is located within the upper 20 m of the sediments. This result is illustrated in the detailed seismic section of Fig. 6. The sea #oor amplitude is in phase over the complete distance of the section, but at a depth of 20 ms TWT below the sea #oor, marked by a box, a clear phase reversal is detected. The observation of gas-bearing sediments close to the sea #oor corresponds to a pockmarked "eld at the sea #oor observed by Crane et al. (1995). SEAMARC II data show that the pockmarked "eld extends from the area around HF-OBH station 6 to the northeast. Our interpretation is that free gas migrates upward from lateral limited gas-bearing layers to the sea #oor and escapes into the oceanosphere. A second low-velocity layer is located at a depth of 80 mbsf with a thickness of 40 m. It is sandwiched between layers of gas-hydrate cemented sediments and contains

155

Fig. 6 Detailed seismic section of pro"le K18 showing the upper 200 ms TWT of the sea #oor between HF-OBH station 5 and 6. At a depth of 20 ms TWT below the sea #oor a clear phase reversal was detected (black box) indicating the transition between marine and gas-bearing sediments

a gas lens. The extent of the gas lens is too low to locate it by phase analysis. The detected velocity variations indicate the existence of gas-hydrated sediments beneath gas-bearing sedimentary layers and might point towards small-scale vertical changes in the gas composition within the gas hydrate reservoir. The origin of free gas within the classic HSZ is not yet well understood. In this context, Zatsepena and Bu!et (1997), who estimated hydrate formation times, suggested the possibility of free gas migrating into the HSZ (Mienert and Posewang, 1998) (on the bases of phase stability models) as one possible explanation.

Fig. 7 Detailed seismic section of pro"le K18 showing the rising BSR. The box marks the area of constructive and destructive interference of the BSR amplitude with amplitudes of crossed re#ection horizons

The detailed analysis of the data clearly indicates that the BSR is a re#ection from the base of the HSZ. Above the BSR, a 90- to 120-m-thick layer with stable velocity values of 1800}1840 m/s has formed. A homogenous gas hydrate concentration and a possibly disseminated distribution within the layer may justify these nearly constant velocities. Free gas, which is trapped below the BSR, explains the abrupt decrease in velocities, the high re#ection amplitudes in this subbottom depth, and the existence of a lowpass "lter to seismic signals. The sealed free gas migrates into the overlaying strata with the rising BSR. Therefore the thickness of the free gas layer below the BSR varies. The re#ection amplitude of the BSR also varies according to the thickness. Due to the high-resolution character of the data, the base of the gas re#ection (BGR) could be detected and the thickness variation could be calculated. In the area around HF-OBH station 5, the free gas layer has a thickness of 12 m and the amplitude of the BSR is weak. In the subbottom around HF-OBH station 6, the thickness of the unit runs up to 25 m and the BSR appears strong. The reason for the correlation between high BSR amplitudes and thickness of the free gas layer is not the thickness itself but the interference of the BSR amplitude with the amplitudes of crossed horizons. Due to constructive interference, an anomalously strong re#ection appears, whereas destructive interference leads to extinction. This result is illustrated in the detailed seismic section of Fig. 7. The black box marks an area of 11 shot points, where the rising BSR crosses some re#ection horizons. In the left part of the box, the BSR amplitude is in constructive interference with amplitudes of the re#ection horizon and a strong amplitude appears. In the central part, destructive interference takes place and the amplitudes are extinguished. The combination of the constructive

156

interference and high-re#ection amplitudes from the pronounced impedance contrast between gas-hydrated and gas-bearing sediments leads to anomalously high amplitudes, and a strong BSR can be observed. On the other hand, if the BSR is running parallel to the strata, as in the area around HF-OBH station 5 (Fig. 4), the amplitude is weak. The main conclusions are: f

f

f

f

f

f

The observed velocity variations indicate the existence of free gas within the classic HSZ. A low-velocity zone, indicating the existence of free gas in the pore space, is located close to the sea #oor and corresponds to a pockmarked "eld observed at the sea #oor. A 100-m-thick layer with stable velocity values of 1800}1840 m/s has formed. These nearly constant velocities possibly indicate a homogenous gas hydrate concentration and possible disseminated distribution throughout the layer. The combination of high-resolution velocity analysis and seismic sections clearly indicate that the BSR is a re#ection from the base of the HSZ. The amplitude of the BSR varies. If the BSR is running parallel to the strata, the amplitude is weak. If the BSR crosses the strata constructive interference leads to high amplitudes. Due to the high-resolution seismic data, the base of the gas-bearing layer (BGR) below the BSR is detectable. This layer shows variation in thickness from 12 m to 25 m.

Acknowledgments We are grateful to the o$cers and crew of the R/< A. A. Karpinskiy for their cooperation during the "eld work. Special thanks are extended to the scienti"c crew of this cruise. We thank the Deutsche Forschungsgemeinschaft and the European Commission MAST II for funding the project under the SFB 313 and ENAM (MAS3-CT96}0003).

References Andreassen K and Hansen T (1995) Inferred gas hydrates o!shore Norway and Svalbard. Norsk Geologisk Tidsskrift 45 : 10}34 Andreassen K, Hogstad K, and Berteussen KA (1990) Gas hydrate in the southern Barents Sea, indicated by a shallow seismic anomaly. First Break 8 : 235}245 Andreassen K, Hart PE, and Grantz A (1995) Seismic studies of a bottom simulating re#ection related to gas hydrate beneath the continental margin of the Beaufort Sea. Journal of Geophysical Research 100 : 12659}12673 Bangs NL, Sawyer DS, and Golovchenko X (1993) Free gas at the base of the gas hydrate zone in the vicinity of the Chile Triple Junction. Geology 21 : 905}908 Boehm G, Camerlenghi A, Lodolo E, and Vesnaver A (1995) Tomographic analysis and geological context of a bottom simulating re#ector on the South Shetland Margin (Antarctic Peninsula). Bollettino di Geo"sica Teoricaed Applicata 37(145) : 3}23

Camerlenghi A and Lodolo E (1994) Bottom simulating re#ector on the South Shetland Margin (Antarctic Peninsula) and implications for the presence of gas hydrates. Terra Antarctica 1 : 154}157 Crane K, Vogt PR, Sundvor E, Shor A, and Reed T (1995) SeaMARC II investigations of the northern Norwegian-Greenland Sea. In: Crane K and Solheim (Eds.), Sea#oor Atlas of the Northern Norwegian-Greenland Sea. Norsk Polarinstitutt Meddelelser 137 : 32}34 Dietrich G (1969) Atlas of the oceanography of the northern North Atlantic Ocean: Based on the polar front survey of the international geophysical year, winter and summer 1958. In: Conseil International pour L'Exploration de la mer, Service Hydrographic, Charlottenlund Slot, Danemark, 173 pp Dillon WP, Lee MW, Fehlhaber K, and Coleman DF (1994) Gas hydrates on the Atlantic continental margin of the United States controls on concentration. In: Howell D (Ed.), US Geological Survey Professional Paper 1570 : 313}330 Eiken O and Hinz K (1993) Contourites in the Fram Strait. Sedimentary Geology 82 : 15}32 Hamilton EL (1980) Geoacoustic modelling of the sea #oor. Journal of the Acoustic Society of America 68 : 1313}1340 Holbrook WS, Hoskins H, Wood WT, Stephen RA, Lizzaralde D, and Leg 164 Science Party (1996) Methane hydrate and free gas on the Blake Ridge from vetical seismic pro"ling. Science 273 : 1840}1843 Hyndman RD and Spence GD (1992) A seismic study of methane hydrate marine bottom simulating re#ectors. Journal of Geophysical Research 97 : 6683}6698 Katzman R, Holbrook WS and Paull CK (1994) A combined vertical}incidence and wide-angle seismic study of a gas hydrate zone, Blake Ridge. Journal of Geophysical Research 99: 17975}17995 Lee MW, Hutchinson WF, Agena WF, Dillon WP, Miller JJ, and Swift BA (1994) Seismic character of gas hydrates on the southeastern US continental margin. Marine Geophysical Research 16 : 163}184 Lodolo E, Camerlenghi A, and Brancolini G (1993) A bottom simulating re#ector on the South Shetland Margin, Antarctic Peninsula. Antarctic Science 5 : 207}210 Luetgert JH (1992) MacRay } Interactive two-dimensional seismic raytracing for the Macintosh. US Geological Survey Open File Report 92}356 MacKay ME, Jarrard RD, Westbrook GD, and Hyndman RD (1994) Origin of bottom-simulating re#ectors: Geophysical evidence from Cascadia accretionary prism. Geology 22 : 459}462 Mienert J (1994) Gas hydrates on the northern European margin. Cruise report, GEOMAR Report 30 : 137}186 Mienert J and Posewang J (1998) Evidence of shallow- and deepwater gas hydrate destabilizations in North Atlantic polar continental margin sediments. Geo-Marine Letters (in press) Minshull TA, Singh SC, and Westerbrook GK (1994) Seismic velocity structure at a gas hydrate re#ector, o!shore western Colombia, from full waveform inversion. Journal of Geophysical Research 99 : 4715}4734 Posewang J (1997) Nachweis von Gashydraten und freiem Gas in den Sedimenten des nordwesteuropaK ischen Kontinentalabhangs mit hochau#oK senden re#exionsseismischen Methoden und HFOBS-Daten. Berichte aus dem Sonderforschungsbereich 313, 68 PhD thesis. University of Kiel, 137 pp Sloan ED (1990) Clathrate Hydrates of Natural Gas. New York: Marcel Dekker, 641 pp Stoll RD, Ewing J, and Bryan GM (1971) Anomalous velocities in sediments containing gas hydrates. Journal of Geophysical Research 76 : 2090}2094 Zatsepena OY and Bu!et BA (1997) Phase equilibrium: Implications for the formation of hydrate in the deep sea #oor. Geophysical Research Letters 24 : 1567}1570