基于波动方程三维表面多次波预测方法研究

与传统的二维表面多次波预测算法相比,基于波动方程的全三维表面多次波预测方法无需对地下介质做简单近似,其更符合地震波在地下介质中传播的真实状况,是地震资料处理中解决多次波预测问题的强有力工具.本文从三维多次波预测的基本理论出发,给出了全三维多次波预测算法的预测矩阵表示、计算方法以及实现条件,采用GPU(图形处理器)加速全三维表面多次波预测,较传统的CPU串行计算,GPU并行预测表面多次波的计算效率约提高165倍.文中分别利用二维和三维表面多次波预测算法对理论模拟的含表面多次波的三维地震数据进行多次波预测计算,对比分析结果表明,相比于二维算法,文中所述的基于波动方程的全三维表面多次波预测效果明显改善,其计算精度更高,辅以合理有效的自适应相减算法,可获得高精度的地震勘探资料表面多次波压制数据.

3D surface-related multiple prediction approach investigation based on wave equation

Compare with the traditional 2D surface-related multiple prediction algorithm, 3D surface-related multiple prediction approach based on wave equation needs no assumption for underground medium. It is a powerful tool to deal with multiple prediction problem in seismic data processing, which is more conform to the actual situation that seismic wave propagate through the underground medium. Based on the fundamentals of multiple prediction, the matrix representation, calculation approach and realization condition for full 3D surface-related multiple prediction are proposed in this paper. Also using GPU (Graphic Processing Unit) to accelerate full 3D surface-related multiple prediction algorithm, compare to traditional CPU computation, the speed is 165 times faster. Multiple prediction tests are done on synthetic 3D seismic data with surface-related multiples, 2D and 3D surface-related multiple prediction effect from 2D and 3D algorithms are compared and analyzed, and the result demonstrates that full 3D surface-related multiple prediction approach based on wave equation has higher computation accuracy and better prediction effect than 2D algorithm. With the help of reasonable adaptive subtraction, we can obtain seismic prospecting data with high precision after surface-related multiple suppression.