1. Introduction

Following the first discovery of methane hydrate inthe cores of the MITI (Ministry of International Tradingand Industry, Japan) Nankai Trough Wells drilled in theNankai Trough area in 1999, the METI (Ministry ofEconomy, Trade and Industry; formerly MITI) com-menced the national project for methane hydrate devel-opment, named MH21, in 2001. The purpose of the firstsix years of MH21 is to establish a geophysical and geo-logical exploration methodology to evaluate a totalamount of methane hydrate occurred offshore Japan,especially the Nankai Trough area. For reaching thegoal, we first have to delineate the spatial distribution ofmethane hydrate and characterize the physical proper-ties of the methane hydrate-bearing zone, such as poros-ity and saturation. Regarding the delineation of methanehydrate-bearing zones, Hato et al. (2001) showed thevalidity of interval velocities delineation. Hato andInamori (2002) also reported some other seismic meth-ods for characterizing methane hydrate-bearing zones.

Furthermore, the evaluation technology should be inte-grated with geological information in order to under-stand the mechanism of the methane hydrate system thatdescribes the origin of methane, its migration and itsaccumulation in a form of methane hydrate.

In the context of understanding the methane hydratesystem, it is very important to know gas occurrence justbelow methane hydrate-bearing zone, gas distribution,and gas concentration status. There are publishedreports regarding gas occurrence and its amount belowmethane hydrate-bearing zones from a few areas (Paulland Matsumoto, 2000). However, methane hydrate-related gas occurrence in the Nankai Trough is stilluncertain, particularly since there was no distinct evi-dence of gas occurrence from the results of the MITINankai Trough Well drilling.

For the purpose of confirming gas occurrence justbeneath the methane hydrate-bearing zone and at thezones above and below BSR (Bottom SimulatingReflector) which is almost coincident with the base of themethane hydrate-bearing zone, we applied AVO(Amplitude Versus Offset ) analysis to the seismic data

RESOURCE GEOLOGY, vol. 54, no. 1, 105–113, 2004

105

Application of AVO Analysis to Seismic Data for Detection of Gas below Methane Hydrate Stability Zone

in Nankai Trough Area

Masami HATO, Takao INAMORI, Alfian BAHAR*, 1 and Toshifumi MATSUOKA*

Japan Petroleum Exploration, Co., Ltd., 1-2-1 Hamada, Mihama-Ku, Chiba 261-0025, Japan[e-mail: mhato@japex.co.jp]

* Department of Civil and Earth Resources Engineering, Kyoto University, Yoshida Honmachi, Sakyo-Ku, Kyoto 606-8317, Japan

1 Present address: Institute of Technology Bandung, Dago Asri E-6 Bandung, IndonesiaReceived on September 20, 2003; accepted on February 6, 2004

Abstract: For the purpose of development of methane hydrate, occurring in the deep marine subsurface, as a resource, the mostimportant issue is to understand the methane hydrate system (generation, migration and accumulation) as well as to delineatethe methane hydrate reservoir properties. We have applied the Amplitude Versus Offset (AVO) analysis to the seismic dataacquired in the Nankai Trough, offshore Japan, in order to confirm the occurrence of gas just below the methane hydrate-bear-ing zone, assuming that gas will show a so-called Class-3 AVO response. Knowledge of the amount and occurrence of gas inthe sediment below methane hydrate-bearing zone is one of the keys to understand the methane hydrate system.

We have utilized the qualitative analysis of AVO methodology to delineate how gas is located below the BSR, which isthought to be the reflection event from the interface between the methane hydrate-bearing zone and the underlying gas-bearingzone. In the region of MITI Nankai Trough Well PSW-3, we observe two BSRs separated by 25 ms. After AVO modelingusing well data, we applied AVO attribute analysis and attribute crossplot analysis to the seismic data. Finally we applied anoffset-amplitude analysis to CMP gather data at specific locations to confirm the results of AVO attribute analysis. The AVOanalysis shows that there is very little gas located in the underlying sediment below methane hydrate–bearing zone. This resultsupports the fact that we could not obtain any clear evidence of gas occurrence just below the methane hydrate-bearing zone inthe Nankai Trough well drilling.

Keywords: methane hydrate, free gas, BSR, AVO analysis

acquired in the Nankai Trough area, together with the logdata from the MITI Nankai Trough Well.

2. Seismic Data and Well Data

In the study, the 2D seismic data of line N96-FH,acquired in the Nankai Trough in 1996, were used forAVO analysis. In 1999, well log data wereacquired in the MITI Nankai Trough Well(PSW-3: Post Survey Well 3), located onthe seismic line N96-FH (Fig. 1).

2.1. Seismic data

To maintain relative amplitude and highresolution, no filtering and no scaling opera-tions were applied. Figure 2 shows a stackedsection of N96-FH, with the amplitudecolor-coded red and blue. In the section, wecan clearly see a double (upper and lower)BSR almost parallel to the seafloor. At thelocation of the MITI Nankai Trough Well,the upper-BSR is about 300 ms below theseafloor and the two-way time between thetwo BSRs is about 25 ms. Because positiveamplitude is red and negative is blue, theseafloor reflection is red, and the two BSRs,whose polarities are reversals of the seafloorreflection, are blue.

2.2. Well data and identification of methane hydrateand related gas

The distribution of methane hydrate and physicalproperties such as methane hydrate saturation wereinvestigated mainly by using the well log data. From theMITI Nankai Trough Well drilling report (JNOC,

M. HATO, T. INAMORI, A. BAHAR and T. MATSUOKA106 RESOURCE GEOLOGY :

Hamamatsu-city

10 km

Tenryu River

100

300

500500

500

1700 2000

1500

100300

500

700

900

1000

1200

12001000

700

900

1000

1500

Omaezaki

1200

MITI NankaiTrough Well

MITI SeismicSurvey ‘N96-FH’

700

Window

1400m

Upper BSR

Lower BSR

PSW-3

500

1000

1500

Tw

o w

ay t

ime

in m

s

Fig. 2 Stacked section of line N96-FH in the Nankai Trough: Red color corresponds to peak (positive) and blue one to trough(negative). Dotted square zone is a window for a detailed AVO analysis. Well PSW-3 occurs at CMP 3878.

Fig. 1 Index map of MITI Nankai Trough Well and seismic survey lineN96-FH. The well PSW-3 located just on the N96-FH was used for thestudy. The contour and number means the water depth of seafloor.

2000), the zone of methane hydrate-bearing sand layers(HBSZ) and high concentration of methane hydrate-bearing sand layers (HCZ) were identified (Fig. 3).HBSZ stands for the interval of methane hydrate-con-taining sand layer, and HCZ stands for the interval ofhighly- concentrated methane hydrate-containing sandlayer. These zones are summarized in Table 1.

A high concentration of methane hydrate in the sedi-ment is characterized by high P- and S-wave velocitiesand high resistivity. In the interval of HCZ, an especial-ly high concentration of methane hydrate is located nearthe bottom of the zone (250–265 mbsf; 1195–1210mbsl). On the original sonic log (P-wave), there weretwo intervals at around 265 mbsf (1210 mbsl) and 287mbsf (1232 mbsl) in depth (indicated by ‘LVZ’ in Fig.3) where the data were not acquired due to unknownreasons. Although a recovered velocity of data-droppedinterval becomes lower than the water velocity afterbeing compensated by VSP data, the velocity in theLVZ is set to be a constant (1500 m/s) through thisstudy. These LVZs are thought to result from theDomenico effect (Domenico, 1976) due to small

amounts of gas occurrence in the pore space. On theother hand, we have not obtained any direct and clearevidence of gas existence from the well drilling. So theexistence of LVZs, although the exact velocities ofthese intervals are still unknown, is the only piece ofcircumstantial evidence for gas occurrence in the MITINankai Trough Well. By taking into consideration therelatively low resistivity log in these zones, it is thoughtto exist a very small amount of free gas in this zone.

Based on the time-depth information of the VSP data,the top of these low velocity zones are approximatelycorrelated to near the upper- and lower-BSR depth onthe seismic section. Furthermore, since there is no den-sity log available, we set the density to be a constantvalue of 1.0 g/cc through the all processing steps.

3. AVO Analysis

AVO analysis has become widely used in oil explo-ration as a powerful delineation tool for lithology andpore fluid type within the reservoir by analyzing the P-wave reflections from the interfaces of the reservoir andsurrounding sediment. The AVO technique has beendeveloped by many researchers such as Ostrander(1984) and Rutherford and Williams (1989) and all ofthem were started from the Aki-Richards equation (Akiand Richards, 1980), which is a practical approximationto the Zoeppritz equation (Zoeppritz, 1919) for thereflection coefficient at the reflection interface. Shuey(1985) developed a simple and easy-to-handle equationfor the P-wave reflection as a function of the incidentangle, shown below.

R (θ ) = P + Gsin2 θ (1)

1 ∆vp ∆ρP = +

2 vp ρ

1 ∆vs vs 2 ∆vs ∆ρG = –2 2 +

2 vp vp vs ρ

where Vp, Vs and ρ are the average P-wave and S-wavevelocity and density of the upper and lower layerbounded by the reflector.

A schematic figure is shown in Figure 4, whichdescribes the basic concept of AVO analysis; the rela-tions of the incident transmitted and reflected waves. Ascan be seen in equation (1), the amplitude of the reflectedwave R(θ) can be decomposed into the so-called ‘AVOattributes’, P (intercept) and G (gradient). P is the zero-

vol. 54, no. 1, 2004 Application of AVO Analysis 107

LVZ

HCZ

Vp Vs GR LLD Vp/Vs

(a) (b) (c) (d) (e)

1180

1200

1220

1240

1260

Dep

th in

met

er

HBSZ

UpperBSR

LowerBSR

Fig. 3 Major well log data of the well PSW-3 (the depthat the well location is 945 m). (a) P-wave velocity, (b)S-wave velocity, (c) Gamma Ray, (d) Resistivity, (e)Ratio of Vp and Vs. Bold dashed interval is thehydrate concentrated zone (HCZ) whose interval isfrom 1135 to 1213 mbsl. HBSZ is the interval of distri-bution of hydrate-bearing sand layers. Two regulardotted intervals are low velocity zones (LVZ). Twoarrows are the depth of upper and lower BSR.

Table 1 Two intervals of hydrate identified by log interpretation.

in Fig. 3 interva l unit Main logging tools for identification

Distribution of methane hydrate-bearing sand HBSZ 1095–1229 mbsl GR, FMIHigh concentration of methane hydrate-bearing sand HCZ 1135–1213 mbsl GR, Resisitivity, Sonic

offset change in acoustic impedance (named as ‘truenormal incident P-wave reflectivity’) and G is relatingto the Poisson’s ratio or Vp/Vs ratio. P and G can becomputed by the Least Square method applied to thepre-stack seismic data (Smith and Gidlow, 1987). Bysimple mathematical operation such as a product and/ora linear combination using P and G, we can reconstructphysically meaningful and gas-identifiable sections suchas ‘pseudo Poisson’s ratio’ section. These reconstructedsections are called ‘AVO attribute sections’. How weapply AVO analysis to the seismic data to identify thegas will be described in the next section in detail.

Based on the AVO theory, for a water-saturated shalelayer above a sand layer containing gas, R(θ) has nega-tive polarity and its magnitude increases as the incidentangle increases. This response is caused by a lowerPoisson’s ratio and a lower impedance of the gas sandlayer relative to the upper layer. The typical AVOresponse from the top of a gas-saturated sand layer iscalled ‘Class-3’ AVO response (Fig. 4(b)). Rutherfordand Williams (1989) identified three basic types ofAVO responses from Class-1 to Class-3 (Fig. 4(b)).Later, a Class-4 response was identified (Castagna et al.,1998) as resulting from a similar Poisson’s ratio of theupper and the lower layer.

4. Application of AVO Analysis and Identification ofGas

According to the AVO theory, changes in the AVOresponses can be directly correlated to changes inPoisson’s ratio, which indicates pore fluid variation,especially gas existence. For the case of gas-saturatedsediment below the methane hydrate layer, Poisson’sratio and the P-wave velocity of the gas-saturated sedi-ment should be lower than the methane hydrate layer.

Based on the reasoning above, weapplied AVO analysis to identify thefluid type by assuming Class-3 gas if it islocated beneath the methane hydrate-bearing zone. In this study, followingAVO analysis flow was adopted. 1)Generation of synthetic seismogram tocompare it with observed seismic dataand AVO modeling, 2) AVO attributeanalysis using P and G attributes, 3)AVO cross-plot analysis, and 4) detailedinvestigation of offset to amplitudeanalysis.

4.1. Synthetic seismogram and AVOmodeling

Both pre-stack and post-stack syn-thetic seismograms were generated by

applying the Zoeppritz equation to the well log data.The synthetic data and the pre- and post-stack seismicdata at the well location (CMP 3878; Common MidPoint) are shown together with some elastic parametersin Figure 5. Two BSRs are indicated by the dotted linein Figure 5. Depth to time domain conversion was doneusing the time-to-depth information of VSP data. Fromcomparing synthetic with observed seismic trace of bothpre- and post-stack processing, we could say there is agood match between the synthetic and observed trace,except some unmatched reflections shown in the figureby E1 and E2. As for the synthetic reflection at E1, it isunknown why the real data show no equivalent ampli-tude high. One possibility is that the high velocity fea-ture on sonic log data is very small in lateral extentcompared to the Fresnel zone.

We performed AVO modeling by using the sonic logto produce a theoretical AVO response and then com-pared this with the observed BSR AVO response at thewell location. In this modeling, we should emphasizethat the velocity and density in the LVZs was set to theconstant values of 1,500 m/s and 1.0 g/cm3, respective-ly. The results of comparison of the AVO response fromAVO modeling and the observed data of both BSR's areshown on Figure 6. From this figure, we can see theAVO response of the observed data tends to remainconstant with increasing offset, while the syntheticAVO curve at the BSR shows increasing negativeamplitude with increasing offset. The reason of thisinconsistency of the response is still under considera-tion, but there are some possibilities such as locality ofgas existence like methane hydrate occurrence describedabove by involving with Fresnel zone.

4.2. AVO attribute analysis

In order to analyze the gas-related AVO response

M. HATO, T. INAMORI, A. BAHAR and T. MATSUOKA108 RESOURCE GEOLOGY :

θ1

VP1

VP1

ρ,VS1

VS1VS VP

, 1

VP2

VP2

ρ,VS2

VS2

,

θ

2

2

θR( )

= +( ) /1 2 2

(a)(b)θR( )

Incident angle

0 0 10 20 30

Ref

lect

ion

ampl

itud

e

Class-2

Class-3

Class-4

Class-1+

+ +2 2

= =

θ θθ

—

Fig. 4 (a) Concept of AVO analysis; (b) The nature of the variation ofreflected amplitude (R(θ)) with offset depends on the relative values ofthe elastic properties in the media. The types of response have been divid-ed into four classes.

along the BSR precisely, we first calculated basic AVOattributes, P and G, and then constructed two AVOattribute sections: the product of P and G, and the pseu-do Poisson’s ratio. The product of P and G (hereaftercalled ‘P*G’) is always positive valued when the top ofgas-saturated layer will show Class-3 response. Fromthe AVO theory, a linear combination of P and G (here-after called ‘aP+bG’) is known as an equivalent valueto change in Poisson’s ratio. In our study, the coeffi-cients a and b are set to 0.5. Thus, when a gas layer,

whose Poisson’s ratio is low,is located below a methanehydrate-bearing zone, thevalue of P+G of the BSRamplitude becomes negative.Because the acoustic imped-ance of a gas layer is alwayslower than that of non-gaslayer, the P attribute (changein acoustic impedance) wasused in addition to the otherAVO attributes to identifygas occurrence. This hybriduse of AVO attributes canimprove the accuracy of thefinal result for gas existencecompared to a regular AVOanalysis.

The AVO attribute calcula-tion was performed using lineN96-FH data over the selectedwindow shown in Figure 2.The P-reflectivity section isshown in Figure 7(a) by beingcolor-coded by red (positivereflectivity) and blue (negativereflectivity). The P-reflectivityis also underlaid by the P*Gattribute and aP+bG attributecolor sections (Fig. 7(b)).Since a model with methanehydrate overlaying gas-saturat-ed sediment would show nega-tive P and negative G, weexpected negative P (bluecolor in Fig. 7(a)) and positiveP*G (red color in Fig. 7(b))along the upper-BSR as gasindicator. On the other hand,in case of blue color, whichmeans negative value, thereseems to be some different sit-uation showing Class-4response.

In the P*G attribute section of Figure7 (b), we canidentify two specific zones in which red color (gas indi-cation) can be seen. Along the upper-BSR in Region 1(circled by white line and indexed by number 1), asmall red color event appears. Below the upper-BSR inRegion 2 (circled by white line and indexed by number2), some red color events are seen. Almost all theremaining zone along and below the BSR is dominatedby negative values (blue color) of the P*G attribute,which means no gas indication there.

vol. 54, no. 1, 2004 Application of AVO Analysis 109

@ Upper BSR

Curve: well

Dot: seismic Dot: seismicCurve: well

@ Lower BSR

0.0

-0.05

-0.1

Ref

lect

ivit

y

0.0

-0.05

-0.1

500 1000 1500Offset (m)

500 1000Offset (m)

1500

Ref

lect

ivit

y

Fig. 6 Amplitude variation with offset along BSRs of both synthetic (solid curve) and seis-mic data (dot) at PSW-3. Synthetic data for both BSRs show possibility of Class-3 gasoccurrence below BSRs. Seismic data do not show evidence of Class-3 behavior.

Upper-BSR

Lower-BSR

Seismic

Synthetic

ρρρρ VSVP

E1>>

E2>>

PoissonísRatio

1500

1450

1600

1650

Tim

e in

mill

isec

ond

1550

Corresponding depth in m

eter

Vp Density Vs Synthetic Seismic

1100

1200

1225

1250

1150

5

(a) (b) (c) (d) (e) (f) (g) (h)

PoissonísRatio

Synthetic stackedSeismic stacked

Fig. 5 Well log data; (a) P-wave velocity, (b) defined density, (c) S-wave velocity, (d) cal-culated Poisson’s ratio, (e) synthetic pre-stack seismogram, (f) synthetic post-stack seis-mogram, (g) pre-stack seismic data, (h) stacked data. Double BSRs whose polarities arenegative can be seen (arrow).

M. HATO, T. INAMORI, A. BAHAR and T. MATSUOKA110 RESOURCE GEOLOGY :

3683

1

2

↑38

78(P

SW

#3)

3719

u-B

SR l-

BS

R

Fig

. 7 (

b)

↑ W3878

1

3

5

3683

↑ 3

878

(PS

W#3

)37

19

2

1u-

BS

R l-BS

R

Fig

. 7 (

a)

3683

↑ 3

878

(PS

W#3

)37

19

2

1u

-BS

R l-B

SR

Fig

. 7 (

c)

3878

(P

SW

-3)

3683

3719

(b)u-B

SR

l-B

SR

1

2

100 50 0 -50

-100

2010

0-1

0-2

0

Bas

e of

gas

laye

r

Top

of g

as la

yer

(a)

↑

psw

-3 M

odel

2

Zone 2 Zone 1

Fig

. 8

In the P-attribute section, Figure 7(a), the blue coloris seen in Region 1, which means that acoustic imped-ance decreases with depth. This result is in agreementwith the P*G attribute result. In the aP+bG (pseudoPoisson’s ratio) attribute section (Fig. 7(c)), a very smallorange color event is seen in Region 1 along the BSR,which represents a reduction in Poisson’s ratio. Becausesediment with gas has a Poisson’s ratio lower than thatof surrounding sediment, this result also indicates gasoccurrence below the BSR.

On the other hand, inside Region 2 below the BSR,we can see some red color (positive value of P*G) inFigure 7(b) but red color (positive value of P) in Figure7(a). In addition, yellow color (positive change inPoisson’s ratio) can be seen in Region 2 on Figure 7(c).Integration of these attribute results leads to the conclu-sion that these specific events inside Region 2 are notthe reflection from the gas top but rather from the bot-tom of a gas layer.

4.3. AVO attribute cross-plot analysis

To get additional information for determining theAVO attribute analysis results, the AVO attribute cross-plot analysis was also applied. This is a mapping andclustering of P-values against G-values to identify thereflection events from the top and the bottom of sandlayers containing gas. All calculated P and G values (P-G pairs) were mapped in P-G space over the data rangefrom CMP 1000 - 9000 and ±200 ms from the upperBSR (Fig. 8(a)). Two specific zones shown in Figure8(a) were identified. A color code was specified foreach cluster, and the events corresponding to each colorwere projected back to the original seismic section.Figure 8(b) shows the cross-plot section with only theclustered events displayed. The two zones describing byRegions 1 and 2 of Figure 7, for investigating gas indi-cation, are also shown in Figure 8(b). The red colorevent in Region 1 indicating the gas top along the BSRand the blue color events in Region 2 are fairly consis-tent with results of all other AVO attributes.

4.4. AVO (offset-amplitude) analysis

For the purpose of confirming the results of well loganalysis and AVO attribute interpretation, three loca-tions (CMP 3683, 3719 and 3878) were selected andoffset-amplitude analysis was applied to these CMP

gather data by targeting the upper-BSR. One location(CMP 3719) is over the region where gas occurrencealong the upper-BSR was inferred by attribute analysis.One of two other locations (CMP 3683) is located out-side the gas-indicated zones.

The CMP 3878 is located just at the well (PSW-3)and shows no gas indication when single gather datawas used for AVO analysis, and the theoretical AVOresponse using the well data is showing Class-3response. The offset-amplitude analysis was carried outusing the following procedures; 1) generate super-gatherCMP gather by stacking 11 consecutive CMP gathers atthe location of both CMP 3683 and 3719, 2) identify theupper-BSR on the CMP super-gather, 3) pick ampli-tudes along both the constant BSR-time and troughreflection event picking, and 4) plot these picked valuesagainst offset, which is limited to 1500 m. The reasonwhy we adopted two types of picking is to prevent pick-ing errors due to imperfect alignment of the BSR eventafter the NMO (Normal Moveout) correction.Consistency in the AVO trends for the both pickingmethods gives confidence in the reliability of theresults.

The results are shown in Figure 9(a) to (c) togetherwith NMO-corrected super-gather data. At CMP 3719,both AVO responses clearly show a typical Class-3response, which is an increasing magnitude of BSRamplitude with increasing offset. In contrast, at CMP3683 and 3878 locations, the AVO responses exhibit aClass-4 response that is rarely seen in AVO analysis onoil exploration.

5. Discussion

In our study, a Class-4 AVO response appeared atlocation CMP 3683 where there is no gas indication onany AVO attribute sections. Recent rock physics studyindicates that a Class-4 response appears according tospecial physical properties of the upper sediment(Carcione and Tinivella, 2000) and a Class-4 responsemay occur when the Poisson’s ratio of the overlyingsediment is closer to that of the underlying sediment (J.Dvorkin, pers. comm., 2003). Due to the low Poisson’sratio of methane hydrate-bearing zone (Lu andMcMechan, 2004), the Class-4 AVO response in ourstudy may be caused by following situation;

vol. 54, no. 1, 2004 Application of AVO Analysis 111

Fig. 7 ( a) P-wave reflectivity. Red color is positive amplitude and blue color is negative amplitude. Yellow dotted lines are upperand lower BSRs. (b) Product of P and G (P*G attribute). Red color event shows gas top. Yellow dotted lines are upper and lowerBSRs. (c) Pseudo Poisson’s ratio (aP+bG attribute). Orange color is negative and yellow color is positive pseudo Poisson’s ratio.Blue dotted line are upper and lower BSRs. 3878 is the CMP location of the MITI Nankai Trough Well (PSW-3).

Fig. 8 AVO cross-plot analysis. (a) Clustering of P-G pair on P-G space. Bold dotted line is the trend line. (b) The back-projectedseismic-equivalent section based on the P-G clustering result. 3878 is the CMP location of the MITI Nankai Trough Well (PSW-3). Blue dotted lines are upper and lower BSRs.

M. HATO, T. INAMORI, A. BAHAR and T. MATSUOKA112 RESOURCE GEOLOGY :

Super Gatherat CMP 3719

Picked amplitude of peak &trough (BSR) along the events

Trough (BSR) @ 1569 ms

Peak @ 1575 ms

Picked amplitude of peak &trough (BSR) at constant time

Trough(BSR)

Peak

Trough (BSR) @ 1576 ms

Super Gatherat CMP 3683

Picked amplitude of peak &trough(BSR) at constant time

Picked amplitude of peak &trough(BSR) along the events

Peak @ 1580 ms

Trough(BSR)

Peak

Super Gatherat CMP 3878

Picked amplitude of peak &trough(BSR) at constant time

Picked amplitude of peak &trough(BSR) along the events

Trough (BSR) @ 1572 ms

Trough(BSR)

Peak @ 1578 ms

Peak

Fig. 9 Offset-amplitude analysis at locations (a) CMP 3719, (b) CMP 3683 and (c) CMP 3878 (location of Nankai TroughWell, PSW-3). (Left) result of constant time picking, (right) result of event picking, (center) super-gather data. The arrow onthe super-gather is the depth of picked event (BSR).

(a)

(b)

(c)

1) a high methane hydrate concentration and a smallamount of gas beneath,

2) a regular methane hydrate concentration and brine-saturated (no gas) sediment beneath.

As for the inconsistency between the AVO response ofthe well data and the observed seismic at the well location(CMP 3878), we have not known exact reason. However,we currently guess that Class-3 response of the wellcomes from high locality of gas occurrence and that of theseismic is showing the averaged AVO response from thewidth of Fresnel zone.

6. Conclusion

The well log interpretation, AVO attribute analysisand offset-amplitude analysis on the well and seismicdata in the Nankai Trough lead to the following conclu-sions.

1) There is very low possibility of gas occurrencebelow methane hydrate-bearing zone, and thus almost allthe region below the methane hydrate- bearing zone isinferred to be brine-saturated sediment.

2) The inconsistency of theoretical AVO response ofwell log and observed seismic data at the well locationshows us possibility of high locality of gas occurrencebelow methane hydrate-bearing zone even if the gasoccurs there.

3) The BSR emerged in the Nankai Trough area doesnot appear to be the result of an upper layer of gas-satu-rated sediment but rather due to the high acousticimpedance of a methane hydrate-bearing zone.

4) The result of seismic AVO analysis supports theresult of drilling and well log analysis that didn’t showany clear evidence of gas occurrence.Acknowledgments: We express great thanks to JapanNational Oil Corporation to give a permission to use thewell and seismic data acquired in the MITI NankaiTrough area for our research and disclose the data in

this article.

References

Aki, K. and Richards, P. G. (1980) Quantitative Seismology.W. H. Freedman and Co., 932p.

Carcione, J. M. and Tinivella, U. (2000) Bottom-simulatingreflectors: Seismic velocities and AVO effects. Geophysics,65, 54–67.

Castagna, J. P., Swan, H. W. and Foster, D. J. (1998)Framework for AVO gradient and intercept interpretation.Geophysics, 63, 948–956.

Domenico, S. N. (1976) Effect of brine-gas mixture on veloci-ty in an unconsolidated sand reservoir. Geophysics, 41,882–894.

Hato, M. and Inamori, T. (2002) Delineation of methane hydratezone using seismic methods on Nankai Trough. Butsuri-Tansa, 55, 435–445 (in Japanese with English abstr.).

Hato, M., Inamori, T. and Takayama, K. (2001) Investigationof methane hydrate in Nankai Trough. Butsuri-Tansa, Proc.104th SEGJ Conf., 2001 (in Japanese with English abstr.).

Japan National Oil Corporation (2000) MITI ‘Nankai TroughWell’ Drilling Report, July, 2000 (in Japanese).

Lu, S. and McMechan, G. A. (2004) Elastic impedance inver-sion of multichannel seismic data from unconsolidatedsediments containing gas hydrate and free gas. Geo-physics, 69, 164–179.

Ostrander, W. J. (1984) Plane-wave reflection coefficients forgas sands at non-normal angles of incidence. Geophysics,49, 1637–1648.

Paull, C. K. and Matsumoto, R. (2000) LEG164 Overview.Proc. Ocean Drilling Program, Science Results, 164, 3–10.

Rutherford, R. H. and Williams, R. H. (1989) Amplitude-versus-offset variations in gas sands. Geophysics, 54, 680– 688.

Shuey, R. T. (1985) A simplification of the Zoeppritz equa-tions. Geophysics, 50, 609–614.

Smith, G. C. and Gidlow, P. M. (1987) Weighted stacking forrock property estimation and detection of gas. Geophys.Prosp., 35, 993–1014.

Zoeppritz, K. (1919) Erdbebenwellen VIII B, Uber Reflexionund Durchgang seismischer Wellen durch Unstetigkeisfla-chen. Gottinger Nachr., 1, 66–84.

vol. 54, no. 1, 2004 Application of AVO Analysis 113