Advancing correlation-based Bayesian modeling of the geomagnetic field to include paleomagnetic sediment records

在千年时间尺度上全球重建地磁核心场具有广泛的应用，例如用于研究地球发电机过程和地球核心动力学的特征；用于与数值发电机模拟进行比较；作为沉积岩芯、火山物质和考古文物的测年工具；或用于研究宇宙成因的同位素产生率。这些模型是从两类数据中推断出来的：一类是来自具有剩余磁化作用的材料的数据，如来自考古遗址的火山岩、砖块或烧粘土碎片，另一类是来自嵌入磁性颗粒的海洋或湖泊沉积物的数据。在一个正在进行的项目中，开发了一种新的基于贝叶斯和相关性的古地磁核场建模方法。这种方法提供了稳健和现实的不确定性估计，以及改进的场模型。然而，到目前为止，该方法仅基于前一类数据。这是由于与模拟沉积物记录有关的挑战：1.在沉积物记录中，磁场的强度和偏角仅相对于岩心的其余部分已知。有必要进行校准。沿单一岩心的数据不能被认为是独立的，它们的时间误差是相关的。沉积物记录不可避免地呈现出磁场变化的平滑版本。拟议项目的目的是克服这些挑战，并将开发的基于相关性的建模方法扩展到同样适用于沉积物记录。这包括对测年不确定性和与沉积相关的过程进行贝叶斯建模，例如平滑。扩展的框架不仅将允许以一致的方式利用全新世的所有现有数据，而且还将在时间上更早地应用该方法，届时可获得的剩余记录较少，数据库以沉积物为主。

Global reconstructions of the geomagnetic core-field on millennial timescales have broad applications e.g. for studying characteristics of the geodynamo process and Earth's core dynamics; for comparison to numerical dynamo simulations; as a dating tool for sediment cores, volcanic material and archaeological artifacts; or to study cosmogenic isotope production rates. These models are inferred from two classes of data: Data from materials with thermoremanent magnetization, such as volcanic rocks, bricks or burnt clay fragments from archaeological sites, and data from marine or lacustrine sediments with embedded magnetic particles. In an ongoing project, a novel Bayesian and correlation-based modeling method for the archeomagnetic core-field was developed. This method provides robust and realistic uncertainty estimates, together with an improved field model. However, the method is based only on the former class of data so far. This is due to challenges associated with modeling sediment records:1. In sediment records of the magnetic field, intensity and declination are only known relative to the rest of the core. A calibration is necessary.2. Data along a single core cannot be considered independent and their temporal errors are correlated.3. Sediment records inevitably present smoothed versions of magnetic field variations.The aim of the proposed project is to overcome these challenges and to extend the developed correlation-based modeling method to be applicable to sediment records as well. This includes the Bayesian modeling of dating uncertainties and sedimentation related processes, such as smoothing. The extended framework will not only allow the exploitation of all available data for the Holocene in a consistent manner, but also to apply the method further back in time, when fewer thermoremanent records are available and the database is dominated by sediments.