In the equivalent source method used for continuation of potential field from an arbitrary surface, the kernel function of Fredholm equation shows self‐similarity and redundancy. According to these characteristics, we adopted the wavelet compression algorithm with non‐linear cosine threshold value to accomplish the order depression of the Fredholm equation, thus making the equivalent source method, put forward by Bhattacharyya in 1977, can deal with data from large area and with mass volume, which is hard to achieve before. Theoretical modeling indicates that, the hard threshold method obtains great accuracy when the compression ratio is 40.5%, while its accuracy decreases when the compression ratio is 81.1%. At this time, the nonlinear cosine threshold can improve the continuation accuracy obviously. At last, the hard threshold, soft threshold, and nonlinear threshold are used in the compression of line MT‐1 data of northeast Sichuan gas field. Results indicate that, when the compression ratio is 79%, all the three methods can acquire good results. But the nonlinear threshold has the highest fidelity, which can provide important evidence in searching for local structures related with petroleum reservoir.

Common Reflection Surface (CRS) stack can improve signal-to-noise ratio of seismic signals. However, the key issue to improve resolution is how to set CRS stack aperture. In this paper, we discuss the characteristics of Fresnel zone during the reflection of seismic waves and suggest that the Fresnel zone under undulate layerinterface condition can be approximated with ellipse. Then we bring forward a method to determine the aperture size of CRS stack based on Fresnel zone and use it to deal with seismic section of the steep slope belt in Miyang Depression. The results show that Fresnel zone based CRS stack improves both the signal-to-noise ratio and resolution because the stacking aperture determined by Fresnel zone maximizes the superposition energy of seismic signals. In addition, deep weak seismic information is preserved and displayed, while normal CRS stack can only improve the signal-to-noise ratio and the shallow seismic information.