INTRODUCTION

The reservoir characterization of methane hydrate structures plays an important role in evaluating the potential of hydrate fields as future energy reserves. Key parameters that are of interest are the amount of hydrate present, its mobility and recoverability. Increasingly, these are attempted to be inferred from drilling MacKay et al. (1994); Matsumoto et al. (1996). However, the risk of heating and destabilizing the initial hydrate conditions is considerably high. Thus, core-samples and well-log data do not necessarily reflect the correct in-situ hydrate conditions. Consequently, it is important to infer elastic and mechanical properties from surface seismic data Brown et al. (1996); Ecker et al. (1996); Ecker and Lumley (1994); Laberg and Andreassen (1996); Yuan et al. (1996) or VSP Holbrook et al. (1996). Most of these investigations were based mainly on AVO responses and synthetic modeling using P-wave velocity information, accessible directly from the seismic data, and neglected possible important S-wave velocity effects entirely. However, as Ecker et al. 1996 have shown, different S-wave velocity behavior might have significant impacts on the hydrate properties, thus influencing assessments of the amount of hydrate, its position in the pore space and its effect on the surrounding sediment strength.

The recent ODP (Ocean Drilling Program) drilling cruise at the Blake Bahama Ridge and Carolina Rise implemented for the first time a dipole shear wave tool in in-situ hydrates structures Matsumoto et al. (1996). This might give some first order insight in the S-wave velocity behavior of methane hydrates. However, since hydrate is considerably unstable for already small changes in pressure and temperature conditions, the process of drilling might have perturbed the actual in-situ properties. Therefore, it will be important to tie the well information to information obtained from surface seismic which does not influence the hydrate properties. But how good can S-wave velocity behavior be estimated from marine P-wave data?

AVO analysis and inversion is a common tool for attempting to extract S-wave velocity information from seismic data Ostrander (1984); Spratt et al. (1993). The precision of the result and its resolution, however, strongly depends on the angle coverage of the data, data quality, noise contamination and data processing Swan (1993); Tura and Hanitzsch (1995); van Wijngaarden and Berkhout (1996). Considering these effects, what kind of S-wave velocity contrasts can be resolved from surface seismic with regard to the transition from hydrated to non-hydrated sediment? How small can they be and still be distinguishable?

In this paper, I evaluate the effect of different S-wave velocity contrasts at the bottom simulating reflector (BSR) on seismic amplitudes. The BSR is the reflection off the base of the hydrate stability field, thus marking the transition from hydrated to non-hydrated sediment. I use P-wave sonic log and density data from one of the wells drilled during the ODP drilling cruise at the Blake Outer Ridge to obtain a blocked velocity and density model. An S-wave velocity model was obtained by fitting an approximate linear trend to preliminary published shear dipole data Matsumoto et al. (1996). Changing the S-wave velocity behavior at the BSR reflection, I forward model different seismic scenarios that resemble different S-wave velocity contrasts across the BSR. AVO analysis of single modeled CMP gathers data shows that even very small S-wave velocity contrast of about $\pm$ 17 m/s can still be uniquely distinguished in their amplitude behavior. Subsequently, I simulate a more realistic data environment by adding random Gaussian noise with a S/N ratio of 2:1. This introduces significant variations in the seismic amplitude responses, and thus causes the AVO trend of the different models to overlap. The unique resolution of the models depends strongly on the standard deviation of the data. Even a qualitative analysis to enable the determination of the sign of the contrast (positive or negative) can be quite hard. Assuming small lateral changes in the subsurface structure, averaging of neighboring CMP gathers should, however, improve this analysis. A subsequent 2-D seismic migration/inversion of a velocity model varying laterally only in S-wave velocity shows that even in the presence of noise a reasonable good result can be obtained.