Fig 1.
Architecture of SCNN for HR seismic processing.
Fig 2.
Comparison of different wavelets: (a) Time domain (b) Frequency Domain.
Fig 3.
Generated seismic reflection coefficient sequence.
Fig 4.
Synthetic seismic traces and amplitude spectra by different wavelets: (a) synthetic seismic traces (b) amplitude spectra.
Fig 5.
Seismic synthetics by convolution from 14 wells: (a) log reflectivity (b) synthetic traces with a 30Hz Richer wavelet (c) synthetic traces with a wide-band Ricker wavelet.
Fig 6.
Synthetic LR and HR seismic traces of a certain well.
Fig 7.
P wave impedance of Marmousi2 model.
Fig 8.
Low-resolution synthetic seismic of Marmousi2 model convoluted with a 30Hz Richer wavelet.
Fig 9.
High-resolution synthetic seismic of Marmousi2 model convoluted with a wide-band B-spline wavelet.
Fig 10.
Training performance of high-resolution seismic processing SCNN model.
Fig 11.
Validation of the well-trained SCNN model by well synthetic data.
Fig 12.
Validation of the well-trained SCNN model by Marmousi2 synthetic data.
Fig 13.
High-resolution processing for Marmousi2 low-resolution synthetic by SCNN model.
Fig 14.
Single trace comparison for high-resolution processing of Marmousi2 model.
Fig 15.
Processing to a poststack seismic data by the well-trained SCNN model: (a) raw seismic section (b) HR processed seismic section.
Fig 16.
Comparison of near-well raw trace and its corresponding HR processed trace from the poststack seismic data.
Fig 17.
Spectra comparison of near-well raw trace and its corresponding HR processed trace from the poststack seismic data.
Fig 18.
Raw seismic gathers.
Fig 19.
HR processed seismic gathers by the well-trained SCNN model.
Fig 20.
Comparison of near-well raw trace and its corresponding HR processed trace from the prestack seismic data.
Fig 21.
Spectra comparison of near-well raw trace and its corresponding HR processed trace from the prestack seismic data.