The Research Of Noise Elimination Methods For Three-component Seismic Pre-stack Data

Not only single component seismic data but also multi-component seismic data have noises. Eliminate and suppressing the noise of seismic data to improve the signal-to-noise ratio(SNR) of data is an important research content. To achieve these goals, so far, people have developed a lot of advanced digital processing technologies, but most of these technologies were developed for the z-component seismic data. Compared with p-wave data, converted wave data has its own complexity. Surface wave in seismic data is very strong. So compared with p-wave data, the SNR of converted wave data is much lower. At the same time, dominant frequency of converted wave is lower so that frequency band of effective wave and surface wave has great overlap, traditional noise attenuation methods for p-wave data can't obtain ideal effect.First the author suppresses the noises in 3c data using conventional p-wave noise elimination methods , then analyzes the applicability of conventional p-wave noise elimination methods for 3c data and existing problems. After this, the paper researches the characteristics of 3c seismic data and makes full use of the information from these three components to study noise elimination methods for three-component seismic pre-stack data . In this paper, the methods what we study are adaptive vector filter method and vector decomposition method. Through the study, we get the following conclusions: (1) Using conventional p-wave noise elimination methods based on frequency domain filtering methods to supress noises in 3c data will greatly destroy effective wave. (2) Adaptive vector filter method: for the simulated data, the application effect of adaptive filter method was good, but, due to the complexity of the noises in real three components data, this method has great limitations, and its results rely too much on referenced noise. (3) Vector decomposition method can be used to remove the random noises in conventional seismic data, the effect will be influenced by the length of time window and relevant trace numbers. When applied in 2D2C data, the vector decomposition method not only can suppress the converted wave and random noises in z-component; but also it can suppress the p-wave and random noise in x-components.