| Popis: |
The main focus of our work is application of visco-elastic Full Waveform Inversion (eFWI) to estimation of elastic parameters directly from seismic shot data. In particular, we aim to recover the ratio between pressure wave velocity Vp and shear wave velocity Vs, which can be indicative of the type of fluid present in a potential subsurface reservoir. The workflow we describe does not directly use well ties for wavelet control, as is typical in seismic inversion. We provide intuitive explanations for choices behind the proposed multi-stage hierarchical algorithm, outline its key steps, and present field data examples from a structurally simple clastic setting and a complex sub-salt environment. The latter, in particular, makes differentiating net versus non-net in pre-salt carbonate reservoirs a challenging problem when using narrow azimuth streamer data. The challenges include significant illumination variability caused by complex geometries of the evaporites, complex wave modes and scattering caused by strong property contrasts in the evaporites and carbonates, and layering of anhydrite and other salts within the evaporites causing complex transmission losses. The thick carbonate reservoir units beneath the salt further complicate the estimation process due to lack of low frequency signal recorded in streamer data. Our methodology shows that it is possible to directly invert shot data to obtain geologically meaningful elastic properties that can be useful in exploration and early development phases. However, challenges remain. We demonstrate that inverted acoustic impedance (Ip) from eFWI has higher fidelity and more reliable magnitude compared to the ratio between compressional and shear wave propagation speeds (Vp/Vs). In fact, the former is sufficiently accurate to be reliably used for porosity prediction. In turn, eFWI Vp/Vs inversion results are in qualitative agreement with well information (as a blind test) for the clastic example and able to discriminate the net versus non-net for the pre-salt example without well information. However, the quantitative match may be insufficient to determine the type of fluid via rock property inversion without any well control. Further research is needed to investigate the sensitivity of Vp/Vs and determine if the quality of the data is key a factor, in addition to stabilizing simultaneous extraction of several elastic parameters from seismic data via constrained inversion framework. |
