中文摘要：

为实现复杂起伏地表条件下基于单程波动方程的高效叠前深度偏移，本文采用“逐步一累加”思想，应用傅里叶有限差分法（FFD）算法实现复杂起伏地表条件下的波动方程叠前深度偏移．为提高效率，本文提出在整个成像实现过程中，将整个成像区域分两块区别对待，在地表高程最低点所在的水平面到基准面的区域内，应用常规精细延拓步长进行波场延拓及成像；在以下区域内，应用大延拓步长进行波场延拓，计算延拓至各层的上下行波场的互相关值，并利用延拓步长顶底两层已经得到的互相关值，结合本文所提的插值算法来实现延拓步长内部层位的互相关成像．模型试算表明本文中提出的方法可以实现对复杂起伏地表条件下的偏移成像，而且可以在不影响精度的同时提高成像效率．

英文摘要：

To realize pre-stack depth migration of complex surface based on one-way wave-equation with high efficiency, the paper recommends wave field downward continuation based on accumulating step by step, using FFD continuation operator. To improve its migration efficiency, this paper proposes the application of different imaging strategies in different blocks. That is, for the depths above the minimum elevation, we do wave field downward continuation based on accumulating step by step using delicate extrapolation step and with cross-correlation imaging condition to get its imaging result, to solve the effect of complex surface; for the area below the minimum elevation, we do the wave field extrapolation on a coarse grid, for the depths right on the coarse extrapolation steps, we perform cross-correlation of extrapolated upgoing and downgoing wave in frequency domain, summing the correlation result on frequency to obtain the imaging result of the depths, note that the correlation result without summing on frequency on the coarse extrapolation steps are reserved for the imaging of depths within the coarse extrapolation steps. And for the depths within the coarse extrapolation step, we do interpolation using the correlation results of nearest upper and lower levels, to get the correlation value and then summing the correlation values by frequency to get the imaging results. Consequently, the correlation value of depths within coarse extrapolation steps are get with interpolation, its effect is equivalent to result by cross-correlation of extrapolated upgoing and downgoing wave field but save the timeconsuming extrapolation process, the efficiency can be improved. Numerical example shows that the method can realize pre-stack depth migration of complex surface with high efficiency and no significant loss of image quality.