在地形不规则的地区所采集的地震资料,通常在偏移前要校正到一个平直的基准面上。用于处理高程变化的基于旅行时的技术是在偏移应用之前对地震资料做时移。这种简单的时移或者说高程静校正不能恰当地把广角反射或倾角反射表示成为它们本来应该在基准面上记录的那样。结果,当高程变化明显时,由偏移和诸如倾角时差校正(DMO)之类的其他波动方程处理方法都可能会损害同相轴定位的精度。传统的方法是采用增大或减小偏移速度来对付这种偏移过量和偏移不足的假象。可是,简单地调整偏移速度并不能消除在高程变化的地区所采集的地震资料的波场畸变。而像波动方程基准面重建这样更完善更精确的解决办法作为常规应用又计算量太大。这里,我们提出了一种使用常规偏移算法进行不规则地表偏移的有效、精确的方法。像高程静校正一样,把地面记录到的数据时移到一个水平基准面上。根据我们的方法,我们把基准面选在探区的最高高程上或高于最高高程。在偏移之后,通过整体时移总是可以将基准面高程调整到其他任意高度。在偏移阶段,地表和基准面之间的地层速度置为零(或是很小的值),在原来地表下方,层速度代表了地下地质情况的最佳估计。通过增加一个零速度层的办法,对平基准面的数据应用偏移算法,直到在记录面遇到非零速度时才允许横向传播。野外数据的实例说明,使用“零速度层”与平坦基准面的常规偏移相比明显地改善了成象精度。此外,人们不需要调节地质上求得的偏移速度场来补偿基准面静校正中的不足。这种方法可以扩展到像DMO、炮点和检波点道集向下外推和偏移这类叠前处理中,从而提出了一种处理不规则地表数据的统一方法。 Seismic data collected in irregularly shaped terrains are usually corrected to a flat datum before they are offset. Travel-based techniques for handling elevation changes are time-shifting seismic data before offsetting the application. This simple time-shift or elevation static correction does not properly represent wide-angle or dip reflections as they should have been recorded on the datum. As a result, offsetting and other wave equation processing methods such as dipole moment correction (DMO) may compromise the accuracy of the in-phase positioning as the elevation changes significantly. The traditional method is to increase or decrease the offset speed to deal with this illusion of over-migration and under-migration. However, simply adjusting the migration velocity does not eliminate the wavefield distortions of seismic data collected in areas of varying elevation. And a more complete and accurate solution like a wave equation datum reconstruction is too computationally expensive as a routine application. Here, we propose an efficient and accurate method of irregular surface offsets using the conventional offset algorithm. Like elevation static correction, the data recorded on the ground is shifted to a horizontal plane. According to our method, we choose the datum as the highest elevation or higher than the highest altitude. After offsetting, it is always possible to adjust the datum elevation to any other height with the overall time shift. During the offset phase, the velocity of the formation between the surface and the datum is set to zero (or a small value) below which the layer velocity represents the best estimate of the subsurface geological conditions. By adding a zero velocity layer, the offset algorithm is applied to the datum of the flat datum, allowing lateral propagation only when the recording surface encounters a nonzero velocity. The field data examples show that using the “zero velocity layer” significantly improves the imaging accuracy compared to the normal offset of a flat datum. In addition, one does not need to adjust the geostatically-derived offset velocity field to compensate for deficiencies in datum static correction. This method can be extended to such prestack processing as DMO, shot point and receiver drop extrapolation and offset, and a unified method to deal with irregular surface data is proposed.