Weak Signal Detection And Denoising Method Research Based On High-density Seismic Data

With the faster development of seismic exploration instrument in recent years, single-point high-density technology has become one of the geophysical prospecting technologies rapidly at home and abroad. The characters of High-density seismic data have the signal bandwidth, wavefield information-rich and true reflection information about reservoir. But it can also record the mass noise, which leads the low signal-to-noise ratio of seismic data. So, it is the great significance for research of weak signal detection and denoising which can fully play the advantage of high-density seismic data.At first, the paper introduces the research presented for the high-density seismic technology and related with processing technology at home and abroad. Then, it also analyzes the technology characters and work principle about single-point detector, which has the following advantages: the dynamic range, signal no distortion. For the character of high-density acquisition technology, it should meet following three basic requirements: time space sampling theorem, the size of the Bin scale and offset. Later, it also analyzed the impact of alias with group interval factor.The third chapter mainly discussed three methods for weak signal detection research including autocorrelation, chaos theory and stochastic resonance. The weak multiple was detected making use of autocorrelation method, which is major of internal multiple. It has archived better effect in the single-frequency signal model detection test, which is based on chaos oscillator by the way of Lyapunov exponent. By thin interbed model, I extended it to seismic signal detection with certain bandwidth. Later, it is successful to deal with the problem of weak signal detection that traditional spectrum failed.For the high-density seismic data, three research methods were selected including singular value decomposition, vector filtering and curvelet denoising. By energy ratio method is used to automatically selecting SV number which has been proposed. It can suppress regular noises by advantage of combined with local window SVD and wavelet transform; furthermore its effect is better than FK method. The effect is very better with suppressing random noises for the data after NMO using SVD method. Vector flitering is based on multi-channel similar principle, which can be applied to the data of post-stack and reflector with little curvature, but it is inability to suppress coherent noises. Therefore, it need further study. Curvelet denoising can process the data with signal-to-ration less than 1 and its result has well preserved-amplitude. By real data processing, there is the best effect when the threshold is selected 6-7 percent for the maximum amplitude with real data. The data can be analyzed and processed by choosing different thresholds. Curvelet transform is a new denoising method and well preserved-amplitude, which is a worth further studying approach. Finally, a software development based GUI environment, its name is weak seismic signal detection and denoising applications.