地震波在地下层状介质中往复传播，形成层间多次波呈现于反射地震数据中，因其特征与有效波相接近，将会直接误导地质和储层解释。层间多次波基于波动理论可以表述为逆散射级数（ISS），本项目旨在系统研究完全数据驱动的ISS多次波压制方法，依序从下列四个方面开展攻关：1）研发压缩感知技术实现地震数据的插值、外推和重采样，满足ISS“全”数据要求；2）研发针对层间多次波的ISS预测算法，考虑实际数据的客观限制，使之能够应用于实际地震数据；3）研发L1/L2混合模的自适应相减方法，使之适用于非正交的多次波情形；4）研发GPU并行加速的多次波压制算法，提高多次波预测和相减两个过程的计算效率。综上所述，创新重点包括地震数据的空间域规则化、多次波预测和相减的计算效率和精度。本项研究将推动ISS方法在实际应用中的发展，发挥该方法的完全数据驱动优势，进而厘清和完善逆散射级数的地震波动理论。

Internal multiples in seismic data are formed by downward and upward propagation of seismic waves between different subsurfaces, internal multiples are able to mislead the geological interpretation because of their similar characterizations with primaries. Based on wave theories, internal multiple can be descried by inverse scattering series(ISS), the inverse scattering series algorithm is full data-driven, and will be the focal researching problem in this investigation. The follow four aspects will be studied in the proposed project respectively and successively: 1) compressed sensing technique is studied to complete seismic data interpolation, extrapolation and resample, and the final aim is to meet the requirements of full-data in ISS algorithm; 2)inverse scattering series internal multiple prediction algorithm is studied to ensure the proposed approach is applicable for all field data, because the application limitations of ISS for field data are taken consideration in data preprocessing; 3) hybrid L1/L2–norm internal multiple adaptive subtraction method is studied to eliminate the orthogonal hypothesis effectively in traditional subtraction algorithm; 4) The GPU/CPU parallel acceleration method is studied to solve the calculation efficiency problem effectively for internal multiple prediction and the subsequent adaptive subtraction. Above all, the innovations of this project are seismic regulation in spatial domain, accuracy and efficiency of internal multiple prediction and adaptive subtraction. The proposed project will promote the application developments of ISS algorithm in field data, exploit the full data-driven advantages of ISS algorithm, and also clarify and improve the ISS-based seismic wave theories.