Multichannel adaptive deconvolution based on streaming prediction-error filter

Field example

For the field data test, we use a 2D poststack section with time interval of 2 ms. The input is shown in figure 8a. For comparison, we apply the traditional predictive deconvolution (the filter length is 6) and the iterative method (the filter length is 6) to enhance the time resolution, as shown in figures 8b and 8c. Figure 8d shows a processing result using the proposed streaming PEF deconvolution method. The streaming PEF parameters are 6 ($N$ ), 0.032 ($b$ ), 25000 ( $\epsilon_t$ ), and 10000 ( $\epsilon_x$ ). The computation time of traditional, iterative method, and streaming PEF deconvolution methods are 0.024 s, 18.767 s, and 1.018 s, respectively, however, the traditional deconvolution method cannot enhance the time resolution at all time because of nonstationary of the field data. The proposed deconvolution and iterative methods can improve the vertical resolution at different times, so both methods are more suitable for processing nonstationary data. Moreover, compared with the traditional and iterative methods, the proposed method can better keep the continuity of events. Furthermore, we select a part of the data near to the reservoir layer from 3-3.5 s to calculate the average amplitude spectrum of the data before and after deconvolution, as shown in figure 9. Figure 9 confirms that the average amplitude spectrum of the seismic section after being processed by the streaming PEF deconvolution is broader than that of the traditional deconvolution result and slightly narrower than that of the iterative deconvolution result in the effective frequency range. However, according to the computation time of the different deconvolution methods, the computational efficiency of the proposed method is significantly improved compared with the iterative method. It further verifies the effectiveness and high efficiency of the streaming PEF deconvolution method in processing nonstationary seismic data.

data,tpef,apef,vlag-spef
Figure 8. Deconvolution results by using different methods. Poststack field data (a), traditional predictive deconvolution (b), iterative deconvolution (c), streaming PEF deconvolution (d).

zspec0,zspec1,zspec2,zspec3
Figure 9. Comparison of the average amplitude spectrum of the deconvolution results and original field data. Original field data (a), traditional predictive deconvolution (b), iterative deconvolution (c), streaming PEF deconvolution (d).

Multichannel adaptive deconvolution based on streaming prediction-error filter