Research On Seismic Wavelet Estimation In Depth Domain

Over the past decades,the conventional methods of seismic inversion are almost implemented in time domain.In the wake of developments in computer technology,depth migration processing is all the rage at the moment.Compared with seismic profile in time domain,seismic depth migration profile can reflect the underground structure more superiorly and make it easier to carry out the geologic interpretation.The depth-domain synthetic seismogram is necessary to perform the horizon calibration on the seismic gathers in depth domain.In general,we can generate the time domain synthetic seismogram by convoluting the reflectivity with the seismic wavelet.When the well log data are converted from depth domain into time domain,the logging curve inevitably exists tension and compression,which causes the time shift and introduces errors.At the same time,many details of interest on the well log data are lost after depth-to-time conversion.The seismic wave field in the depth domain does not satisfy the linear time invariant condition,thus there is no way to use convolution model directly.But convolution model can be implemented in the constant-velocity depth domain via constant velocity depth transformation.In this kind of depth domain,reflectivity can be derived from well log.However,the seismic wavelet in depth domain is not estimated as easily as in time domain.Thus,the crux of the problem is how to obtain a precise depth-domain seismic wavelet.The reliability of depth-domain seismogram synthesis would be higher than time-domain one if we can solve this problem.In this dissertation,we first present a convolution model based on C matrix that can reduced computation time and computation overhead.In addition,we propose to use Fibonacci optimization algorithm to determine the number of sampling points for seismic wavelet.In view of these aspects,ridge regression and Gibbs sampling algorithm are presented for depth-domain seismic wavelet estimation.The result shows that these two methods can be effectively applied to seismic wavelet estimation in depth domain.