ABSTRACT High‐frequency multichannel seismic systems provide detailed images of the shallow marine subsurface. In order to exploit the redundancy inherent in such data optimally, traveltime corrections need to account for normal moveout and static effects due to vertical source and receiver variations. Misalignment of reflections in common‐midpoint gathers will significantly lower the frequency content in the final stack, making this correction particularly important for very high‐frequency seismic data. Traditionally, normal moveout correction involves labour‐intensive picking of stacking velocities, while static corrections can be, by some techniques, performed automatically. In this paper, we present a high‐frequency seismic case study from the Baltic Sea, using seismic image matching as a novel, fully automated technique to perform joint moveout and static corrections. Our multichannel test profiles were acquired offshore Rugen island for wind farm development. Owing to the regular passage of up to 1.5 m high ocean waves during data acquisition, these boomer profiles suffer from strong static effects. We perform joint normal moveout and static corrections by defining the nearest common offset section as a fixed reference frame and minimizing its difference in traveltime with respect to all available common offset sections. Time shifts are computed independent of a pre‐defined traveltime curve, using the normalized cross‐correlation as a measure of data similarity while penalizing irregular displacements by a regularization term. Time shifts are converted to stacking velocities based on the traditional hyperbolic traveltime equation. Our results are compared with those derived by conventional manual velocity analysis and subsequent trim static corrections. We find that image matching produces stacks of similar quality and stacking velocity models of similar to slightly better quality compared with the conventionally derived ones, revealing the potential of this technique to automatize and significantly speed up this first part of the seismic processing chain.

中文翻译：

---

波罗的海高分辨率地震数据的自动静态和时差校正

摘要 高频多道地震系统提供了浅海海底的详细图像。为了最佳地利用此类数据中固有的冗余，旅行时校正需要考虑由于垂直源和接收器变化引起的正常时差和静态效应。共中点道集的反射错位将显着降低最终叠加中的频率成分，这使得这种校正对于非常高频的地震数据尤为重要。传统上，正常时差校正涉及劳动密集型选择堆叠速度，而静态校正可以通过某些技术自动执行。在本文中，我们介绍了波罗的海的高频地震案例研究，使用地震图像匹配作为一种新颖的方法，全自动技术来执行关节移动和静态校正。我们的多通道测试配置文件是为风电场开发获得的离岸鲁根岛。由于数据采集期间高达 1.5 m 高的海浪有规律地通过，这些婴儿潮一代剖面会受到强烈的静态影响。我们通过将最近的公共偏移部分定义为固定参考系并最小化其相对于所有可用公共偏移部分的旅行时间差异来执行联合法向时差和静态校正。时间偏移的计算独立于预定义的旅行时间曲线，使用归一化互相关作为数据相似性的度量，同时通过正则化项惩罚不规则位移。基于传统的双曲走时方程，时移被转换为叠加速度。我们的结果与通过传统手动速度分析和随后的微调静校正得出的结果进行了比较。我们发现，与传统派生的模型相比，图像匹配产生的堆栈质量相似，叠加速度模型的质量略好一些，这揭示了该技术在自动化和显着加快地震处理链的第一部分方面的潜力。