Multi-trace Seismic Deconvolution Method With Lateral Constraint Regularization

The sparse constraint inversion can break through the limitation of the original frequency band of seismic data,and obtain higher resolution than the conventional deconvolution method.However,sparse inversion often assumes that the reflection coefficient satisfies a certain sparse distribution,so the mathematical significance of the high-frequency components obtained by this model-based prior distribution is obviously greater than the physical meaning,and the high-frequency components of the inversion result lack accuracy.Besides,conventional methods perform deconvolution on seismic data trace by trace,and does not consider the constraint relationship between the trace and the trace.Therefore,the seismic profile obtained by the trace-by-trace inversion is discontinuous in the lateral direction and has a discontinuity phenomenon.This phenomenon is particularly serious when the input seismic traces have low signal-to-noise ratios(SNRs)or complex structures.In order to solve the above problems,this paper studies the multi-channel seismic inversion lateral constraint regularization method.Multi-channel inversion processing two-or three-dimensional data simultaneously,so lateral regularization constraints can be introduced.We first extract dip and the reflection structure information of time-space domain and frequency-space domain from the seismic data,construct the lateral constraint regularization term,and then introduce it into the multi-channel inversion system based on the Bayesian framework.The advantage of multi-channel inversion is that it can balance the vertical resolution and spatial continuity of seismic data.While improving the resolution,the spatial structure of the original data can be guaranteed and the weak reflection can be effectively restored.