锆石微量元素机器学习示踪Jack Hills锆石岩浆源区

澳大利亚的Jack Hills锆石是地球早期地质历史研究的最重要研究对象。然而，其岩浆源区是否具有沉积物加入这一关键科学问题尚存分歧，这也引发了一系列关于早期地球地壳组成、板块俯冲起始等重大地球科学问题的争论。因此，厘定Jack Hills锆石的岩浆源区具有重要的科学价值。.锆石微量元素组成可以用于示踪岩浆源区。目前研究者往往采用I型花岗岩锆石（源区无沉积物）和S型花岗岩锆石（源区有沉积物）二分法开展源区示踪研究，往往通过构建各类图解，在2维平面上示踪锆石的源区类型。本研究尝试采用机器学习方法，将研究维度从2维提升至17维，即利用锆石中17种主要微量元素特征开展锆石的分类研究。这种基于17种锆石微量元素的机器学习分类方案，将比传统2维图解更加可靠、准确地揭示锆石的岩浆源区，有望准确回答Jack Hills锆石的岩浆源区是否具有沉积物加入这一关键科学问题。

Jack Hills zircons from Australia, are the most important samples to study the earliest geological history of the Earth. However, whether sediment is involved in the magma source of Jack Hills zircons is still in debate, which further raises a series of debates in many aspects, such as in the continental crust composition of the early Earth and the time for the earliest subduction. Thus, determining the magma sources of Jack Hills zircons is of great importance..Trace element characteristics could be used to trace the magma source of Jack Hills zircons. To trace the magma source, researchers usually use 2-dimensional diagrams to classify zircon into I-type granitic zircon, whose magma source is mainly consisted of igneous rocks, and S-type granitic zircon, whose magma source is sediment. Here we apply machine learning methods into the classification of zircons by using 17 key trace element of zircon. This new 17-dimensional classifier would be more reliable than traditional 2-dimentional diagram in tracing the magma source of Jack Hills zircons, would be a promising method in precisely answering the key problems in whether sediment is involved in the magma source of Jack Hills zircons.