The current level of seismic exploration and the expectations of geophysical customers require an increase in the seismic resolution. This is necessary to obtain More detailed depth images of a fragment of the earth's crust are to be obtained, as well as the accuracy of inversion and migration transformations is to be improved. For this purpose, the frequency range of the signal should be extended, and the complications of the seismic wavelet shape should be corrected by means of inverse filtering. The paper describes the signature deconvolution optimization method. The inverse filtering operator is calculated from a known signal, and then adaptive filtering of the input data is performed. The algorithm is robust in the sense that it does not lead to an increase of the random and the regular noise. Adaptation is applied in a sliding rectangular window.
Materials and methods
The source materials are the seismic traces recorded during seismic surveys and the source signature. The processing method is non-stationary adaptive deconvolution in a sliding window.
An adaptive non-stationary robust signature deconvolution method is proposed. The efficiency of the algorithm is proved by the results of a marine real dataset processing. The new method will be an alternative data processing scheme that replenish the arsenal of deconvolution tools for a geophysicist. The parameters of the procedure are such that a simple choice of a deliberately large threshold value transforms the algorithm into the conventional and well known signature deconvolution.
Instability of the signature deconvolution can be considered one of its main problems. The signature deconvolution operator is designed using only the estimated source signature. It is then applied to the seismic gathers that may be contaminated by noise. As the operator does not take into account the existence of noise, the noise may be amplify as a result of such deconvolution. To deal with that problem, we introduce a data-driven component into the commonly deterministic signature deconvolution algorithm. To ensure the signature deconvolution is stable, we propose a self-tuning filter that is capable of detecting noise for a given frequency in a sliding window and modifying the amplitude spectrum of a deconvolution operator for that frequency. This approach makes it possible to obtain the sought-for deconvolution result for noiseless data windows and limit noise amplification for noisy parts of the data, which we prove using a real marine dataset.