Joint Inversion Of P-wave NMO Velocities And VSP Traveltimes In 3D TI Media

The main goal of this thesis is to make inversion for parameters of anisotropy and interfacesin 3D multilayer TI media with 3D dipping interfaces. The inversion id based on jointly using 3DP-wave NMO velocities, corresponding zero-offset reflection times t 0, interface depths from aborehole, and multiazimuthal VSP downward P-wave traveltimes. Firstly, in order to carry out thetarget, forward-modelling computations are realized for traveltimes of waves and NMO velocitiesin TI media. Then, by means of the genetic algorithm, inversion is performed for parameters ofanisotropy and interfaces.In the part of forward-modelling, I put forward the coordinate transform ray tracing methodfor TI media with 3D dipping interfaces on the basis of existing work (Yao, 2004). It can be usedto compute traveltimes of reflection waves on CSP(Common Shot Point)and CMP(CommonMiddle Point)gathers, and of VSP downward wave concisely and effectively. The accuracy ofcomputation is analyzed, and the paths of phase and group propagations are compared. Theinfluences of model conditions on time-distance of P-wave reflection on CMP gathers are studied.Moreover, I derive the analytic solution, which is in accordance with the existing NMO-velocitysolutions when limited to some special cases, to NMO-velocity surface for a single layer of TImedia with a 3D dipping interface, and unify the NMO-velocity solutions to pure mode waves(not converted waves) in TI media (including isotropic media). The computational method ofeffective NMO velocities for multilayer TI media with dipping interfaces is given using thesolution to NMO-velocity surface, and the numerical computation is performed. The mainconclusions are drawn through numerical computations:(1) The coordinate transform 3D ray tracing method presented in the thesis is stable, and thetraveltime computations have good accuracy on condition that the controlling parameters areproperly chosen.(2) When the media attributes are not known, the time-distance relation of P-wave reflectionon CMP gathers can be described reliably by a hyperbolic curve if the ratio of the maximum offsetto the depth of the reflector doesn't exceed 1.0.(3)The analytic solution presented in the thesis to NMO-velocity surface can be used toconveniently compute effective NMO velocities. The errors of NMO velocities determined by theactual survey data relative to NMO velocities computed by the analytic solution increase with theenlargement of the maximum offset. The maximum offset used shouldn't exceed the depth of thereflector as observational NMO velocities are determined.In the part of inversion for parameters, all of 7 parameters of anisotropy and the interface arefor the first timer recovered in a 3D TI model with a dipping interface by means of GA jointlyusing multiazimuthal (no less than 2) NMO velocities and corresponding zero-offset reflectiontimes t 0 at multi-CMP (no less than 3), the interface depths presented by a borehole and thedownward P-wave traveltimes on 2 VSP sections. The inversion for interval parameters of amultilayer model is also accomplished. The influences of factors of used data, medium attributesand structures on inversion results are analyzed. The main conclusions are as follows:(1) The stable inversion results for anisotropy parameters can not be obtained, but the stableinversion results for parameters of the dipping interface can be obtained, using onlymultiazimuthal NMO velocities and corresponding zero-offset reflection times t 0 at multi-CMPand the borehole data, or using multiazimuthal NMO velocities and corresponding zero-offsetreflection times t 0 at multi-CMP, the borehole data and the downward P-wave traveltimes on aVSP section. All of the parameters of anisotropy and the dipping interface for TI media can bestably obtained (the dip of the symmetry axis for TI is comparatively large), using multiazimuthalNMO velocities and corresponding zero-offset reflection times t 0 at multi-CMP, the boreholedata and the downward P-wave traveltimes on 2 VSP sections, or using zero-offset reflection timest 0 at multi-CMP, the borehole data and the downward P-wave traveltimes on 5 VSP sections.(2) Increase of the offsets for VSP benefits to the inversion stability, and the accuracy ofinversion results depends sensitively on the errors of the input data.(3) Joint inversion of multiazimuthal NMO velocities and corresponding zero-offsetreflection times t 0 at multi-CMP, the borehole data and the downward P-wave traveltimes on 2VSP sections can stably acquire all of the model parameters for TI media with a comparativelylarge dipping angle of the symmetry axis, but can not stably acquire the azimuth of the symmetryaxis and can stably acquire the other parameters of anisotropy and the dipping interface for TImedia with a comparatively small dipping angle of the symmetry axis. The dip angle of theinterface doesn't much affect the stability of inversion results, and the enlargement of anisotropyfor TI media benefits to the stable inversion for all the model parameters.(4) Even if the input data theoretically satisfy the constraint condition for stable inversion, theinversion results may still bring about non-uniqueness which makes a feature of obviousdifference between the inverted values and the actual values for a few parameters. Thisnon-uniqueness is basically in accordance with actual model in describing the kinematicsignatures of P-wave, and may be used as stratum parameters with certain reliability.