CONCLUSIONS

We compared two methods of determining the near seafloor parameters from data recorded by OBS and OBH. The first method is based on AVO from pressure and vertical particle velocity of the seafloor signals. The second method uses AVO of the vertical and radial particle velocity. Using simple synthetic seismograms, we evaluated the effect of layering and poroelasticity on the parameter inversion.

We can conclude that the method using Vz-Vr seems to be more robust and stable. Reasonable inversion results can be obtained for angular coverage considerably smaller than the critical angle. Furthermore, the two components need not to be calibrated prior to the inversion. In case of layering, the P-Vz method does not result in good estimates for density, P-wave and S-wave velocity. The Vz-Vr method, on the other hand, can still yield a P-wave velocity with an error of 1.9% and an S-wave velocity of 6% when only the near offsets are used in the inversion. Since the effect of layering was simulated with a hard layer, these error bounds are probably upper limits. Layers with parameters closer to the ones of the near seafloor should result in better constrained inversion results.

Poroelasticity has a distinct effect on both reflection and AVO coefficients. The Vz-P method gives good estimates of P-wave velocity ($\le$ 1% error) and S-wave velocity ($\le$ 2% error). The density cannot be estimated within small error bounds. The error is about 20%. The Vz-Vr method produces only a good constraint on the P-wave velocity ($\le$ 1% error). The fluid appears to have a significant effect on the radial component, resulting in strong deviations between the true and inverted S-wave velocity ($\ge$ 30% error). M ore work is required in order to fully understand the effect of poroelasticity on the inversion results.

Finally, we can conclude that the Vz-Vr method seems to be superior to the P-Vz method for the determination of the near seafloor parameters. It appears to be more stable and robust even in the case of layering.

We would like to thank Schlumberger Cambridge Research for a great and productive research environment with excellent personal and computer resources. The paper was the result of a summer internship of Christine Ecker with SCR this summer. We would also like to thank Chris Chapman for his helpful discussions about Hankel transforms.