中文摘要：

目前对多次波的有效利用仅围绕多次波成像技术展开,通过成像多次波试图获取更丰富的地下构造信息.不同于该思路,本文另辟蹊径,从利用多次波提高地震数据分辨率角度出发,对多次波的有效利用进行了深入挖掘.首先基于聚焦变换思想在聚焦域内实现多次波的降阶,通过理论推导得出聚焦域内多次波表现为原始数据的多维子波反褶积这一重要结论,从理论上证明了本文方法提高地震数据分辨率的可行性;然后采用引入整形正则化的非稳态回归自适应匹配滤波方法将聚焦域内由多次波构建的高分辨率数据分离出来,实现原始数据的高分辨率转换.与常规反褶积模型不同,该方法基于波动理论推导得出,可以适用于任意复杂情况;每一道输出结果中所有炮记录都参与了运算,从空间上加以约束,在提高纵向分辨率的同时可以改善数据的横向分辨率.最后通过模型试算和实际资料处理对本文方法的有效性、适应性和实用性进行了验证.

英文摘要：

Multiples can be migrated to obtain much broader illumination range and richer structural information under the surface, and the present studies are all carried out based on this idea. In this paper, we propose a totally different way, according to the relationship of the propagation path between multiples and primaries, and from the perspective of improving the resolution of seismic data by using multiples, we study deeply on the use of multiples. We first focus on the theory derivation of focal transform, by choosing the primaries as the focus operator, do focal transform to the data containing both primaries and multiples, the result in focal domain can be obtained, in which the primaries focus on the area around zero time, while a total new response like the original data has been recovered from the multiples. During the derivation, we obtain the important conclusion that the multiples in focal domain can be regarded as a multidimensional wavelet deeonvolution process of the original data. So the effectiveness of this method on improving resolution by using multiples can be verified in theory. By introducing the non-stationary regression adaptive match filtering with sharping regularization to separate thehigh resolution data recovered from the multiples in the focal domain, the high resolution transform of the original data will be completed finally. Then we can use the processing methods for primaries to deal with the new high resolution data so as to get a high resolution processing result. Be totally different from the conventional deconvolution model, this method is derived based on wave-equation theory, it can be applied to any complex situations. Besides, all the shot records are involved in the computation of each output trace, so by the constraints in the space, the lateral resolution of data can be improved at the same time of improving vertical resolution. Examples from synthetic data and real field data verify the effectiveness, adaptability and practicability of this method.