论文详情
最小二乘偏移理论上能够消除采集照明不佳的影响,减少成像假象、均衡成像振幅、提高成像分辨率,然而,受实际观测炮集数据与模拟炮集数据不易匹配、迭代反演计算量过大等因素影响,三维数据域最小二乘偏移的规模化应用受到制约。为此,提出了基于双程波傅里叶有限差分反偏移方法、共轭梯度反演步长三点抛物搜索寻优方法和GPU加速偏移/反偏移核函数计算方法的三维数据域最小二乘逆时偏移技术。通过有效压制反偏移正演模拟的数值频散,提高反偏移模拟数据与观测数据匹配的一致性,采用计算量更少的迭代寻优方法和GPU算法,实现了迭代反演效率的大幅提升,在此基础上形成了一套具有实际处理能力的三维数据域最小二乘逆时偏移技术方案。将该技术应用于西部和南方两个工区的实际地震资料处理,提高了地震成像分辨率,改善了同相轴的连续性,整体成像效果优于逆时偏移,验证了该技术方案的有效性。
In theory,least-squares migration can remove the influence of poor illumination pattern,reduce image artifacts,balance amplitude,and improve resolution.However,because observed shot gathers do not match simulated shot gathers well,and the cost of iterative inversion is high,large-scale application of least-squares reverse time migration is very difficult in the three-dimensional data domain.To address the key problems of data matching and computational efficiency,we propose a fourier finite-difference demigration method based on two-way wave equation,a three-point parabolic search step optimization for conjugate gradient inversion,and an efficient calculation method based on GPU for kernel of migration and demigration.By effectively suppressing the numerical dispersion of demigration,the consistency between the simulated and observed data was improved.By using less computational iterative optimization method and GPU algorithm,the efficiency of iterative inversion was improved.In summary,a technological scheme for least-squares reverse time migration in the three-dimensional data domain,with the ability to be used in practical processing was established.Two practical applications showed that this technique improves the resolution of seismic imaging and the continuity of seismic events,demonstrating the advantage and efficacy of least-squares reverse time migration.
国家重点研发计划(2017YFC0602804-02)、国家科技重大专项(2016ZX05014-001-002)共同资助。