| When we are doing4D seismic monitoring in CCS projects, random noise has a direct influence on the CO2fluid identification and repeatability of4D seismic data. However, so far the quantitative research does not yet emerged. In my thesis, we analyze the effects of random noise on the CO2fluid identification and on repeatability analysis. During the analysis process, based on logging information and seismic data from Weyburn CCS project, Canada, we used two modeling methods. One is reflectivity method which can simulate the different wave fields, and another is convolution synthetic method which has been widely used and only simulates P-wave reflections. Then we combined with Gassmann fluid substitution theory to get4D seismic response caused by different fluid saturation and pore pressure. After studying these two forward models, we also analyze the real seismic data. We do repeatability analysis and S/N analysis during the processing flow in order to QC4D processing. Finally, we get the amplitude difference among vintages seismic data in three years, which was compared with the results of model.My study shows that:(1) In the model depending on reflectivity method, signal to noise ratio should arrive six and then the seismic response caused by5%CO2fluid can be seen. While in the convolution model, signal to noise ratio just required to be one. However, when the saturation of CO2increases continually, the amplitude difference will be smaller, and signal to noise ratio should be higher. But it is certainly sufficient to detect the distribution range of CO2using actual four-dimensional seismic, as well as whether there exists a leakage. If we want to confirm the CO2saturation, it maybe need more accurate models and calculation method.(2) As for the decreasing in pore pressure, P-wave amplitude seems change smaller, while S-wave seems more sensitive to pressure.(3) The NRMS values caused by5%CO2fluid are respectively around0.3and0.7, which is higher than the standard NRMS value (0.1-0.3) of ideally repeatable4D seismic data. Hence,4D information could be preserved on the premise of keeping good repeatability, but the effects every step works on4D information during the processing need further study.(4) Noise will decrease the repeatability among the different phase of seismic data, and it has fewer influences on strong reflector, relatively speaking weak reflect event. Then filtering processing can great improve its repeatability and make the amplitude differences more clearly, while filtering will have a drop on AVO effect.(5) In the AVO I&G crossplot, the rise of CO2saturation have a trend of moving along the direction to the third quadrant, while the increasing of pore pressure has a trend of moving to second quadrant; while a rise in noise makes the sub-point rotate to the third quadrant, which cause certain influences on the saturation of CO2. Filtering processing has a great effect on AVO results. In addition, the AVO effect is more obvious in reflectivity method modeling than in the convolution modeling.(6) During the process of Weyburn real four-dimensional seismic data, we used the NRMS and S/N profiles to track the processing flow. This can make the multi-phase seismic data repeated at the greatest extent, so that the effective4D information extracted can be more accurately. And generally, the more the S/N increases, the better the repeatability improves. |
PDF Full Text Request