Matching Study Of Seismic Data And Well Data In Time Domain And Depth Domain

Presently, it seems impossible to study and solve problems by mean of single data along with the increasing exploration difficulty. In geophysics, it need to research synthetically with other data for reservoir information. This is realized through the tie of seismic and well logging. The well logging have precedence over the seismic, because the vertical resolution of the well logging is higher than the seismic and the abundant frequency information in the well logging just make up the seismic. In the same way, the seismic supplement the transverse information of the well logging by large-scale overlay and continuous horizons tracing.In this thesis, the sonic curve is converted from depth to time. After getting serial reflection coefficients, we resample those reflection coefficients by multi-resolution and compare with the result of conventional method. In the same time the reflection coefficients are denoised by wavelet transform. Accordingly the disposed reflection coefficients not only keep the trend of original data as a whole but also reserve some local details by comparison. To realize the automatic segmentation, we segment the sonic curve under different scales using the multi-resolution properties of wavelet transform. In this way some thin horizons are holded preferably and the matching precision between well logging and seismic is increased. In the course of making synthetic seismogram, an extracted wavelet is selected in several different time windows which is based on the fractal property of well logging and is restricted with the well by the side of trace. The extracted wavelet take advantage of other preestablished wavelet without the hypothesis of phase. In the end, the synthetic seismogram and seismic traces near the borehole is decomposed under different scales using the multi-resolution properties of wavelet transform. Then the Least Square Algorithm is used to get the optimal matching filters in different scale space and the minimum entropy filtering is used to correct the matching operators. Finally the synthetic seismogram will be reconstructed. In the process, the iteration is proposed to improve the effect of contrast. The method can match seismic traces near the borehole with synthetic seismogram in time, phase, frequency and amplitude to some extent. In the thesis, some theoretic models are maked and several real data are disposed. It is accurate and effective for the theory and technique in the paper by comparison with conventional methods.