Coal seam strong reflection separation with matching pursuit
Li Haishan1, Yang Wuyang1, Tian Jun2, Wu Guochen2
1. PetroChina Research Institute of Petroleum Exploration & Development—NorthWest, Lanzhou, Gansu 730020, China;
2. School of Geosciences, China University of Petroleum (East China), Qingdao, Shandong 266580, China
Abstract:Strong reflection events occur in seismic section if coal seams exist in the study area. In this case, the reflection information from targets above or below the coal seam will be submerged in the strong reflection events. And then the reservoir prediction can hardly be carried out. Based on sparse representation theory, a method for separation of strong coal seam reflection via matching pursuit is put forward in order to deal with this problem. The key to remove coal seam reflection by matching pursuit is the choice of sparse dictionary. Through the analysis of seismic data sparsity in different one-dimensional dictionaries, and taking into account of low-frequency characteristics of strong reflection, the one-dimensional undecimated discrete wavelet transform (UDWT1D) dictionary composed of low-frequency atoms is adopted to sparsely represent the strong reflection information. The matching pursuit decomposition is carried out with the iterative algorithm proposed by Blumensath et al., although this algorithm is a greedy algorithm, the decomposition is carried out only one time for each seismic trace, so the computational efficiency is high. Test results of theoretical model and real data verified the validity of the proposed method.