Seismic data interpolation without iteration using $t$ -$x$ -$y$ streaming prediction filter with varying smoothness

conclusions

In this study, we proposed a fast approach based on SPF for simultaneously reconstructing irregular and regular missing traces in the $t$ -$x$ -$y$ domain. With the help of local smoothness constraints, we defined a streaming computation manner to calculate nonstationary PF without multiple iterations. The invertible characteristics of SPF has direct applications, such as for seismic data interpolation. The non-causal in space filter structure and the similarity matrix guaranteed the accuracy of the interpolation results. Compared with the Fourier POCS method, the $t$ -$x$ -$y$ SPF can characterize reasonably nonstationary signal while avoiding artifacts that occur in the frequency domain method. Moreover, the proposed method is superior in terms of its computational cost. Experiments with synthetic examples and field data demonstrated that the $t$ -$x$ -$y$ SPF is effective at efficiently recovering irregular and regular missing traces in nonstationary seismic data.