Yiran Xue ¹ Jian Zhang ^1* Gan Zhang ^2* Dongchang Zhao ^2*

• ¹ Southwest Jiaotong University, Chengdu, China
• ² Sichuan Water Development Investigation, Design & Research Co., Ltd, Chengdu, Sichuan Province, China

Seismic inversion is one of the key techniques for reservoir characterization. Depth-domain seismic inversion eliminates the cumulative errors associated with depth-to-time and time-to-depth conversions, thus providing geologists and reservoir engineers with an intuitive basis for geological interpretation. The method has received increasing attention in the field of reservoir characterization. Extracting accurate depth-domain seismic wavelets is a prerequisite for successful depth-domain seismic inversion. However, the depth-domain wavelet is velocity-dependent and exhibits significant nonstationarity, which leads to the failure of seismic wavelet estimation methods based on the stationary convolutional model. To this end, we propose a modified wavenumber-domain unscaled S transform (MWUST) method to accomplish accurate estimation of depth-domain seismic wavelets. The proposed method enhances the accuracy of wavenumber components by removing the linear-wavenumberdependent term from the S transform. Furthermore, it introduces slope and intercept parameters to improve the depth resolution at low wavenumbers, thereby yielding a more reliable depth-wavenumber spectrum. Subsequently, the relationship between the non-stationary depth-domain seismic wavelet and the depth-wavenumber spectrum is established, allowing for the accurate extraction of nonstationary wavelets under the assumption that the depth-domain reflectivity is a random sequence. The synthetic and real data applications have been used to verify the effectiveness of the proposed method.