Radial-Trace Time-Frequency Peak Filtering Based On Correlation Integral Suppress Seismic Random Noise

Seismic exploration is the main method in searching for oil and natural gas. Earthquake experts can identify geological structure underground rock by the seismic data, and then located the oil and gas fields and get an effective evidence for the exploitation of oil and gas. There are three parts of seismic exploration:seismic data acquisition, seismic data processing and seismic data interpretation. Seismic data processing is the most important part in seismic exploration, a well processed s provide a legible seismic data for the interpretation of seismic data. If we do not process it well at this part, it effect the seismic data interpretation directly. Thus, how to obtain a high signal-to-noise ratio of seismic data is particularly important. Seismic data denoising methods for seismic exploration has became a hot research topic today.Time frequency peak filtering technology is based on time frequency analysis theory to suppress the random noise. It can suppress the random noise in nonstationary deterministic band-limited signals. In recent years, TFPF has been applied to seismic random noise attenuation by the modern signal processing laboratory of Jilin University in China successfully. TFPF does not depend on the signal prior information, it has been proven feasible to recover the desired signal in the case of a low SNR. TFPF can effectively remove the random noise of seismic exploration and improve the signal to noise ratio greatly compared to other traditional de-noising method. However, a fixed WL in conventional TFPF fails to take both the random noise suppression and the valid signal amplitude preservation into account, it processes the signal only along the temporal direction, is an one-dimensional frequency filtering algorithm, which ignores the information space of the golden mean between the seismic data and the road, does not take the correlation information of seismic data road and road into account. A radial-trace TFPF is developed to reduce the sensitivity to WL compared to TFPF. It transforms the seismic data through radial trace transform, then TFPF is used to filter the random noise in the radial domain, the filtered record inverse to the original domain at the last. The radial-trace transform can stretch the valid signal to make it as linear as possible within the window, so that we can use a longer WL in radial-trace TFPF. The longer WL is more conducive to long filter random noise suppression without affecting the maintain of the signal amplitude. In this way, we can get a more satisfactory result than the TFPF on both the random noise suppression and the valid signal amplitude preservation. However, the radial frequency peak filtering uses a fixed angle of sampling methods, if the intersection angle between the seismic events and the radial trace is large, the radial-trace transform cannot stretch the signal effectively. It will cause serious distortion if we still use a longer WL in radial-trace TFPF.In this letter, by introducing the definition of the correlation integral, we propose the radial-Trace Time-Frequency Peak Filtering Based on Correlation Integral to suppress seismic random noise in the seismic data. We discuss the principle of distinguishing the noise segment and the signal segment through the correlation integral values in the radial-trace domain, so that we can apply different WLs in the radial-trace TFPF. Then a longer WL according to the noise intensity is used to remove the random noise and a shorter WL according to the frequency characteristics of signal is used to preserve the details of signal to improve the deficiencies of the radial-trace TFPF. The experiment results on both the synthetic model and the field seismic data show that this method can effectively remove noise from seismic record and maintain the amplitude of the valid signal and improve the signal to noise ratio of seismic exploration data effectly.