Multichannel adaptive deconvolution based on streaming prediction-error filter

Conclusion

We have improved the theory of streaming PEF with temporal and spatial constraints and proposed a multichannel adaptive deconvolution based on streaming PEF in the time-space domain. Our approach uses a time-varying prediction step to guarantees the reasonable amplitude relationship of deconvolution results. The proposed method updates the filter coefficients simultaneously when each new data point arrives, which effectively represents the nonstationary characteristics of the seismic data. The efficient computational feature of streaming computation is useful for efficient adaptive deconvolution in high-dimensional data processing. Synthetic examples and field data test confirm that the proposed deconvolution method is successful in enhancing the temporal resolution of seismic data while preserving the relative amplitude relationship and structural continuity.