地质学家报告了首项证据，显示 2 个古老、面积相当于大陆尺度的地幔构造，称为大型低剪切波速省（large low-shear-velocity provinces, LLSVPs），已影响地球磁场约 265 million years。每个 LLSVP 的估计面积可与非洲大陆相比，位于地核-地幔边界附近、约 2,900 kilometers 深度，处在形状不规则的地幔区域之中；这些区域更热、更致密、化学性质也不同，边缘则被较冷的「环带」包围，该处地震波传播更快。
在一项由利物浦大学领导、发表于 Nature Geoscience 的研究中，研究人员主张，这些极高温的 LLSVPs 与周围较冷地幔之间的温度反差，会改变地核内液态铁的环流方式，使不同区域的流动加速或减速，并在地球发电机（geodynamo）中产生不对称性，从而塑造今日非理想的磁场几何（不规则、倾斜与各种模式）。团队将地幔观测与超级电脑模拟结合，比较均一地幔模型与包含 2 个 LLSVPs 的非均质模型，并把两者与真实磁场资料核对；只有纳入 LLSVPs 的模拟能重现观测到的特征。
这项工作为先前的侦测努力加入了可量化的时间尺度：异常在 1970s late 被怀疑，约 2 decades later 得到确认，之后又经过另 10 years 的研究才提出这项连结主张。模拟显示，某些磁场成分可在 hundreds of millions of years 内相对稳定，而其他成分则会大幅改变，挑战长久以来的假设，即长期平均的磁场行为是一个完全对齐的棒磁铁。作者认为，这些发现可能改进与盘古大陆（Pangaea）聚合与解体相关的重建，并帮助约束古气候、古生物学与自然资源形成中的不确定性，但其依据仍是模型与资料的一致性，而非在 2,900 kilometers 深度进行直接取样。
Geologists report first evidence that 2 ancient, continent-sized mantle structures called large low-shear-velocity provinces (LLSVPs) have influenced Earth’s magnetic field for about 265 million years. Each LLSVP is estimated to be comparable in area to the African continent and sits near the core-mantle boundary at roughly 2,900 kilometers depth, within irregular mantle regions that are hotter, denser, and chemically distinct and bordered by a cooler “ring” where seismic waves travel faster.
In a Nature Geoscience study led by the University of Liverpool, researchers argue that temperature contrasts between these ultrahot LLSVPs and surrounding cooler mantle modify how liquid iron circulates in the core, speeding or slowing flow by region and producing asymmetry in the geodynamo that shapes today’s non-ideal magnetic geometry (irregularities, tilts, and patterns). The team combined mantle observations with supercomputer simulations, comparing a uniform-mantle model versus heterogeneous models that include the 2 LLSVPs, then checked both against real magnetic-field data; only the LLSVP-inclusive runs reproduced observed features.
The work adds a quantitative timescale to earlier detection efforts: anomalies were suspected in the late 1970s, confirmed about 2 decades later, and followed by another 10 years of research leading to this linkage claim. Simulations indicate some magnetic-field components can remain relatively stable for hundreds of millions of years while others change substantially, challenging the long-held assumption that long-term averaged field behavior is a perfectly aligned bar magnet. Authors suggest the findings could improve reconstructions tied to Pangaea’s assembly and breakup and help constrain uncertainties in ancient climate, paleobiology, and natural-resource formation, while still resting on model-data agreement rather than direct sampling at 2,900 kilometers depth.