CGG: Improvements in Full Waveform Inversion
CGG case study demonstrates the benefits of full azimuths and long offsets for determining the velocities of a complex overburden in a deep-water region of the Gulf of Mexico.
Full Waveform Inversion (FWI) aims to estimate high-resolution velocity models by minimizing the difference between observed and modeled seismic waveforms. It goes beyond refraction and reflection tomography techniques, which use only the traveltime kinematics of the seismic data, and uses the additional information provided by the amplitude and phase of the seismic waveform.
Most FWI methods use transmitted energy for determining velocities. The major drawback of this approach is that the extent of the velocity update is limited by the offset range in the acquired data. If the water depth is greater than 1 km, near to middle offsets (less than 6 km) have reflections as the early arrivals, but these are more difficult to use for FWI because they contain density information. For most deep-water, wide-azimuth, towed-streamer surveys with a maximum offset of around 8.5 km, only the offsets greater than 6 km contain the refractions and diving wave energy that are useful for FWI. The penetration depth of the transmitted energy is restricted by this limited range of recorded offsets, producing velocity updates that are prone to acquisition-related artifacts.
CGG exploited transmitted energies to update the velocities in Keathley Canyon, Gulf of Mexico using long-offset and full-azimuth broadband data, acquired using its proprietary StagSeisTM acquisition geometry. The study area is in a deep-water area characterized by sedimentary basins with faults, carapaces, and complex salt structures. The StagSeis staggered configuration of two streamer boats and three additional source vessels allows the recording of ultra-long offsets of over 18 km in the inline direction. The study was configured with east-west and north-south sail lines to yield full-azimuth coverage up to 9 km (assuming source-receiver reciprocity). In this case study, data were acquired with BroadSeisTM variable-depth streamers with a maximum tow depth of 50 m. The proprietary curved shape of BroadSeis streamers creates ghost notch diversity, so that streamers can be towed deep for improved low-frequency content, without compromising the high frequencies. This is especially useful at far offsets where it can aid initial FWI iterations.
To examine the improvement in FWI velocity models gained from the long offsets, CGG compared inversion using only those offsets that would be recorded by conventional wide-azimuth acquisition with inversion using the full range of offsets recorded using StagSeis. Using only offsets up to 7.5 km inline and 4 km crossline produced FWI velocity updates with considerable oscillations in the deeper section and an incorrect, slow velocity, inversion of the carbonate section. Using the ultra-long offsets recorded by StagSeis resulted in velocity updates that were characteristic of the geology, produced flatter gathers and did not suffer from artificial oscillations.
To understand the benefits of the full-azimuth configuration, we compared the performance of FWI using only the north-south sail lines (wide-azimuth) with using all the sail lines (full-azimuth). The first inversion suffered from a north-south striping pattern, i.e. an acquisition footprint. The second inversion, using all the data, did not exhibit this pattern because of the improved illumination and transmission energy penetration from the transverse direction.
When the inversion was performed using all the StagSeis full-azimuth, long-offset data for frequencies up to 7 Hz, the velocity model clearly showed a good match with the geology (see figure). The faults and high-velocity condensed sections directly above the top of salt can easily be tracked by the FWI model, which also detects the carbonates and shale bodies. The results clearly demonstrate the imaging improvement that longer offsets and additional azimuths bring to Full Waveform Inversion. By illuminating more subsurface angles and deeper sections, StagSeis enables FWI to produce velocity models that faithfully characterize the different geological features in this complex region. For further details of this and other FWI datasets, visit CGG at SEG booth #2542.
