双变网格正演模拟方法因其低存储、高效率等优势,在刻画地下小尺度异常体,尤其是超深部微结构方面发挥了越来越重要的作用。变网格正演模拟方法在实现过程中需要考虑两个问题:网格步长变化处的虚假反射和长时间采样下的不稳定性。从理论上推导了虚假反射误差的数学表达式,分析了虚假反射影响因素,并通过引入Lanczos滤波算子,实现了一种基于交错网格的稳定且高精度的双变网格正演模拟算法。模型正演试算结果表明,在超长采样时间迭代测试下本算法依然稳定,且虚假反射也得到明显压制。随后,将双变网格波场延拓算子应用于偏移成像过程,实现了双变网格逆时偏移算法。裂缝储层模型的成像结果表明,双变网格逆时偏移方法能够以较小的内存,较为准确、高效地成像地下微小构造。
Owing to its low memory and high efficiency, dual-variable grid forward modeling method has played an important role in characterizing the seismic response of small-scale anomalies, especially for ultra-deep microstructures. In the implementation process of variable grid forward modeling method, it needs to consider two issues. Firstly, the fake reflection occurred where the grid interval changes. The second is this method prone to be unstable under a long sampling time. in this paper, we derived the mathematical expressions of fake reflection error theoretically, and analyzed its influencing factors. Through the introduction of LanczoS filtering operator, we proposed a stable and high-precision dual-variable grid forward modeling algorithm based on staggered grid. Numerical results show that our algorithm not only remains stable at more than 30 thousand times iterations, but also suppresses fake reflections significantly. Subsequently,by applying our algorithm to reverse time migration,we developed a dual-variable grid reverse time migration algorithm. Numerical results of fractured reservoirs proved that our method can characterize underground microstructures accurately and efficiently.