Drift time estimation by dynamic time warping
Tianci Cui, Gary F. Margrave
The drift time is the difference between the event time at seismic frequencies and at sonic logging frequencies predicted by the attenuation theory. The synthetic seismogram needs drift time correction to tie the seismic trace. In this report, the drift time is estimated by dynamic time warping (DTW) method, which is based on the constrained optimization algorithm. By matching the stationary and nonstationary seismograms, DTW can estimate the drift time automatically without knowledge of Q or check-shot records. The estimated drift time is caused by apparent Q attenuation and may contain extra time shift due to the phase error in the wavelet used to construct the stationary seismogram. After applying drift time correction to the stationary seismogram, the residual phase between it and the nonstationary seismogram is almost constant in both time and frequencies. Then time-variant amplitude balancing and time-variant or time-invariant constant-phase rotation are applied to the nonstationary seismogram to perfect the matching.