基于贝叶斯方法的地壳厚度定量估算

大陆地壳一直以来都扮演着记录过去40亿年地球演化历史的重要角色。现代板块汇聚边界形成的岩浆岩地球化学特征与岩浆活动时的地壳厚度高度相关，因此，一系列地球化学指标可作为地壳厚度的优秀示踪剂。然而，由于岩浆岩地球化学组成的复杂性，准确量化过去地质时期地壳厚度一直是一项具有挑战性的任务。本文基于地球化学大数据，运用贝叶斯方法，建立了一个利用多种地化指标（CaO、K₂O、MnO、Dy、Ho、Lu、Y、Sr/Y、Ce/Y、La/Yb、Sm/Yb、Dy/Yb）定量估算地壳厚度的贝叶斯模型。利用中新世以来（＜15 Ma）的全球数据验证表明，与传统的单指标方法相比，贝叶斯模型对现今地壳厚度提供了更准确的估计。利用该模型重建了新生代拉萨地块地壳厚度变化，重建结果表明，拉萨地块在50~30 Ma经历了多阶段地壳增厚，最终形成如今的巨厚地壳。

Quantitative estimation of crustal thickness based on Bayesian method

The continental crust has played an important role in recording the Earth's evolution over the past four billion years. The geochemical characteristics of magmatic rocks formed by the modern plate convergence boundary are highly correlated with the crust thickness during magmatic activity, so a series of geochemical indicators can be used as excellent tracers of crust thickness. However, due to the complexity of the geochemical composition of magmatic rocks, accurately quantifying crustal thickness in past geological periods has been a challenging task. Based on large geochemical databases, a Bayesian model for the quantitative estimation of crustal thickness using various geochemical indices (CaO, K2O, MnO, Dy, Ho, Lu, Y, Sr/Y, Ce/Y, La/Yb, Sm/Yb, Dy/Yb) is established in this paper. Validation results using global data since the Miocene (< 15 Ma) show that the Bayesian model provides a more accurate estimate of present-day crustal thickness than traditional single-indicator methods. This model is used to reconstruct the crustal thickness changes of the Lhasa block in the Cenozoic. The reconstruction results show that the Lhasa block experienced multiple stages of crustal thickening from 50 Ma to 30 Ma, and finally formed today's extremely thick crust.