The Research Of Deconvolution Based On Signal Fidelity

With the continuous progress in the field of Petroleum Exploration and Production, a great deal of attention has been paid to the finding of hydrocarbons hidden in the complex and lithological traps.Presently,people are seeking new processing methods with the target of high resolution,high S/N ratio and high fidelity. To improve resolution, all kinds of deconvolutions have been applied. But some of them reduce the S/N ratio of seismic data. Some reduce the fidelity of seismic data. Some have too many hypothesises or their hypothesises are a little excessive. Accordingly our targets are to improve the resolution, S/N ratio and fidelity of seismic data on one hand and, on the other hand, improve the resolution of seismic data.After the theory analysis, this paper in-depth studied deconvolutions on basis of signal fidelity.First, Aimed at the circum stances of the resolution is improved meanwhile the signal-to-noise ratio has decreased significantly after deconvolution.The paper introduced a method of spectrum-constrained deconvolution, which designed auto-matically spectrum -constrained operator according to character of signal-to-noise ration on seismic data, limited the degree of deconvolution at frequency band with low signal-to-noise ratio and achieved object of while improving resolution, some signal-to-noise ratio can be p- reserved after deconvolution prcessing. At the same time, this method used space and time variations of operator maintained at the input data of the signal to noise ratio without lowering, then improve high-frequency components. So that the resolution and continuity of seismic data are improved. Used broadband Ricker wavelet as the desired output, because of broadband Ricker wavelet with a high-resolution, high-fidelity and high-frequency noise suppression characteristics, thus making the results this method has a higher resolution, fidelity and the ability to suppress noise.The second part discusses some approaches on spectrum fitting deconvolution and puts forward, after analyzing the irrationality of the deconvolution hypothesis, a new assumption that the amplitude spectrum of seismic wavelet is smooth so as to replace the one that the reflection coefficients are white random noises. The resolution and fidelity of processing results have been improved by spectrum fitting deconvolution.Taking into account the wavelet space and time-varying characteristics, adoptting an improved method, achieved a better space and time variations deconvolution processing in the frequency domain.Model spreadsheet and the actual data processing test proved the validity of the method.