Two hyperparameters, the mean Moho depth and the Moho density contrast, must be specified before the gravity inversion of Moho. Incorrect estimation will impact inversed Moho morphology. The purpose of this study is to present a new gravimetric Moho inversion method. The key improvement of the new method is to accurately estimate the Moho density contrast, based on a linear relationship between the depth of known points and gravity observations. The method is illustrated by a synthetic experiment where the estimated density contrast differs from the true value by only 0.0011 , showing a 93% improvement compared to the initial estimate. The results of processing the noise data show that, our method's accuracy is minimally affected by noise, but is sensitive to the number of known points. Nevertheless, when using only 10 known points, there is still a 50% probability of obtaining a solution with a root mean square (RMS) less than 1 km. When the number of points is greater than 64, the effect of the uniformity of the point distribution is almost negligible. In the real case, we employed the proposed method to invert the South China Sea (SCS) Moho depth. The Moho model reveals that, there is a distinct zoning feature at Moho depth in the SCS, and the 13.5 km isodepth line indicates the continent-ocean boundary. Furthermore, the RMS of the difference between the gravimetric Moho model and the seismological data is 1.64 km, which is primarily attributed to the lateral variation of the density contrast.

中文翻译：

---

重力扰动莫霍面深度恢复改进方法及其在南海的应用

两个超参数，平均莫霍面深度和莫霍面密度对比，必须在莫霍面重力反演之前指定。不正确的估计会影响反莫霍面形态。本研究的目的是提出一种新的重量莫霍面反演方法。新方法的关键改进是基于已知点深度与重力观测之间的线性关系，准确估计莫霍面密度对比度。该方法通过一个综合实验来说明，其中估计的密度对比度与真实值仅相差 0.0011，与最初的估计相比，提高了 93%。对噪声数据的处理结果表明，我们的方法的准确性受噪声影响最小，但对已知点的数量敏感。尽管如此，当仅使用 10 个已知点时，仍有 50% 的概率获得均方根 (RMS) 小于 1 km 的解。当点数大于 64 时，点分布均匀性的影响几乎可以忽略不计。在实际案例中，我们采用所提出的方法来反演南海（SCS）莫霍面深度。莫霍面模型表明，南海莫霍面深度具有明显的分带特征，13.5 km等深线为大陆-海洋边界。此外，重力莫霍面模型与地震数据之间的差异的 RMS 为 1。