基于GPU交错网格三维正演模拟的上覆地层对目的层成像振幅的影响研究

发布时间：2018-02-15 03:07

本文关键词： 交错网格 GPU 三维模型正演振幅分析　出处：《中国地质大学(北京)》2017年博士论文　论文类型：学位论文

【摘要】：在油气勘探中,随着对地震勘探精度要求的不断提高,对地震成像方法的要求已不仅仅是满足于得到地下构造成像,更希望能够得到地下目标岩性体的振幅信息。人们通过对地震数据进行偏移成像得到反射界面的成像振幅,并认为成像振幅能够反映出反射界面的岩性变化,进而通过沿层属性切片技术对地下储层空间展布进行刻画。但是受上覆地层复杂地质构造的影响,偏移得到的目标层位的振幅信息无法正确反映岩性信息的变化。保幅偏移成像方法是从反演角度来消除上覆地层介质的影响。但是对于上覆地层和目的层成像之间的关系依然鲜有研究。针对这一问题,从正演的角度出发,借助GPU并行加速技术,使用高阶交错网格有限差分方法对不同地质模型进行正演模拟,研究上覆地层和目的层成像振幅信息之间的关系,为后续实际地震资料处理过程中进行岩性识别、储层预测提供理论帮助。在交错网格有限差分公式的推导过程中,将PML衰减系数整合到差分公式中,将核心计算区域和PML边界吸收区域统一起来,建立了含PML衰减系数的交错网格有限差分公式。结合交错网格和GPU的特点,使用GPU对交错网格有限差分算法进行加速,充分利用GPU芯片内的高速共享存储器,实现了基于GPU的三维高阶交错网格有限差分正演模拟算法,提高了算法的计算效率并取得了满意的结果。通过三维高阶交错网格有限差分正演模拟GPU算法对不同的模型进行正演模拟,研究了目的层成像振幅随上覆倾斜界面倾角变化的规律。从上覆界面的深度、高差和上下层速度等几个方面,研究了水平界面、背斜界面和向斜界面对目的层成像振幅的影响。分析了上覆岩性异常体的速度、半径的变化和目的层成像振幅之间的关系。基于一系列正演模拟结果,分析了上覆地层介质变化与目标层振幅之间的量化关系,对于分析偏移成像后地震振幅的保真性具有理论意义和实用价值。
[Abstract]:In oil and gas exploration, with the increasing requirement of seismic exploration precision, the requirement of seismic imaging method is not only satisfied with the imaging of underground structures. It is more desirable to get amplitude information of underground target lithologic body. The imaging amplitude of reflection interface is obtained by migration imaging of seismic data, and it is considered that imaging amplitude can reflect the lithologic change of reflection interface. Furthermore, the spatial distribution of underground reservoir is characterized by the technique of stratigraphic slicing. However, influenced by the complex geological structure of overlying strata, The amplitude information of the target layer obtained by migration can not reflect the change of lithologic information correctly. The method of amplitude preserving migration imaging is to eliminate the influence of overlying stratum media from the angle of inversion. There is still little research on the relationship. From the forward modeling point of view, with the help of GPU parallel acceleration technique, the forward modeling of different geological models is carried out by using the high-order staggered grid finite difference method, and the relationship between the imaging amplitude information of the overlying strata and the target layer is studied. It provides theoretical help for lithology identification and reservoir prediction in the course of seismic data processing. In the derivation of staggered grid finite difference formula, the PML attenuation coefficient is integrated into the difference formula. By unifying the core computing region and the absorbing region of PML boundary, a finite difference formula of staggered grid with PML attenuation coefficient is established. Combined with the characteristics of staggered grid and GPU, GPU is used to accelerate the staggered grid finite difference algorithm. Taking full advantage of the high-speed shared memory in GPU chip, a 3D high order staggered grid finite difference forward modeling algorithm based on GPU is realized. The computational efficiency of the algorithm is improved and satisfactory results are obtained. The forward simulation of different models is carried out by using the GPU algorithm of three-dimensional high-order staggered grid finite difference forward modeling. The variation of the imaging amplitude of the target layer with the dip angle of the overlying inclined interface is studied. The horizontal interface is studied from the depth of the overlying interface, the height difference and the velocity of the upper and lower layers. The influence of anticline interface and syncline interface on imaging amplitude of target layer is analyzed. The relationship between velocity, radius variation of overlying lithologic anomaly and imaging amplitude of target layer is analyzed. Based on a series of forward simulation results, The quantitative relationship between the variation of overlying media and the amplitude of the target layer is analyzed, which is of theoretical significance and practical value for the analysis of the fidelity of seismic amplitude after migration imaging.
【学位授予单位】：中国地质大学(北京)
【学位级别】：博士
【学位授予年份】：2017
【分类号】：P631.4

【参考文献】