Superplumes, superpiles or superpuddings? Understanding the thermochemical dynamics of the mantle with waveform seismology

我们了解板块构造。我们仍然不明白的是地幔和板块是如何共同作用产生这种独特的行为的。这是一个突出的根本性问题。特别是，我们不了解地幔对流的动力学，包括成分扮演的角色。地震学研究告诉我们，地幔底部有两个巨大的神秘结构，一个在非洲之下，另一个在太平洋之下。我们不了解它们在地幔对流中的作用。这些结构有三种假说，每一种都有不同的动力学含义。首先，它们是热对流的结果，例如，由一簇羽流（超级羽流）造成的;第二，它们代表密集的碎屑堆（超级堆）;第三，它们是热成分，由洋壳的再循环造成，导致地幔中的分布不均匀性（有人将其等同于布丁中的李子），可能更集中在这些地区（超级布丁）。我们将用电脑模拟地幔环流来研究每一类假说。我们将把板块运动的历史应用到这些模型中，以产生可以与真实的地球相比较的模型。我们团队的早期工作表明，初步的超级羽流和超级堆模型产生的结构与地震层析成像的两个大型结构相似。不同的假设，虽然将有不同的内部和顶部结构，这不能解决与目前的地震层析成像。它们可以从地震学上加以区分，但这需要更先进的方法。本项目将利用这些更先进的方法。预测地震结构的模型将从所产生的地幔环流模型的现阶段产生。这将使用来自数百个实验室实验的矿物学及其弹性特性的世界级热力学数据库来完成。这些模型将使用地震探测器进行测试，这些探测器可以观察结构内部，另一组则集中在结构的上边缘。探测器对不同假设的预测将通过精确地直接模拟传播的地震波来产生。这将使用光谱有限元代码SPECFEM3D_GLOBE在国家超级计算机HECToR（不久将是ARCHER）上完成。一组探测器是所谓的“ScS”地震波。这是一种从核幔边界反射的波-这提供了一种工具，可以以高横向分辨率观察结构内部。第二组探测器将是围绕结构顶部底部的体波。如果这些结构延伸到核幔边界之上，那么只对它们进行采样的波就会受到影响。然后将不同模型的独特预测地震特征与现在可用的大型数据集进行比较，以便准确地测试假设。数据中的微弱特征将通过使用叠加波形的观测地震学先进技术（添加多个地震图）得到放大，最好使用地震仪阵列。它们还将为目前用于地幔结构成像和建模的近似方法提供一个很好的测试。（获得高性能计算）、工具（模拟地幔环流（TERRA，Fluidity）和地震波传播的代码（SPECFEM3D_GLOBE））、数据、跟踪记录和专业知识（包括板块运动历史、矿物物理学、建模和地震数据分析方面的合作伙伴）来承担这一雄心勃勃的项目。

We understand plate tectonics. What we still do not understand is how the mantle and plates work together to generate this unique behaviour. This is an outstanding fundamental problem. In particular we do not understand the dynamics of mantle convection including what role composition plays. Seismological studies have shown us that the mantle has two giant mysterious structures at its base, one beneath Africa the other beneath the Pacific. We do not understand their role in mantle convection. There are three hypotheses for these structures, each with different dynamical implications. First that they are the result of thermal convection e.g. result from a cluster of plumes (super-plumes); second that they represent dense detrital pile (super-piles); and third that they are thermo-compositional, resulting from recycling of oceanic crust, leading to distributed heterogeneity (equated by some to plums in a pudding) through the mantle, which maybe more concentrated in these regions (super-puddings). We will produce computer simulations of mantle circulation to investigate each class of hypothesis. We will apply plate motion history to these to produce models that can be compared with the real Earth. Earlier work of our team has shown that both preliminary superplume and superpile models produce structures similar to the two large structures imaged with seismic tomography. The different hypotheses though will have different internal and top structures, which cannot be resolved with current seismic tomography. They can be differentiated seismically, but it requires more advanced methods.This project will bring these more advanced methods to bear. Models of the predicted seismic structure will be produced from the present-day stage of the resulting mantle circulation models. This will be done using a world-class thermodynamic database of the mineralogy and its elastic properties, derived from hundreds of laboratory experiments. The models will be tested using seismic probes that can look inside the structures, and another set that focus at their upper edge. The predictions of the probes for the different hypotheses will be produced by accurately directly simulating the propagating seismic waves. This will be done using the spectral finite-element code SPECFEM3D_GLOBE on the National Supercomputer, HECToR (and soon ARCHER). One set of probes is the so called 'ScS' seismic wave. This is a wave that reflects off the core mantle boundary - this provides a tool to look inside the structure with high lateral resolution. The second set of probes will be body-waves that bottom around the top of the structures. If the structures extend high above the core mantle boundary then waves that only sample them will be affected. The distinctive predicted seismic signatures of the different models will then be compared to the large datasets now available allowing the hypotheses to be exactingly tested. The weak signatures in the data will be amplified by using the advanced techniques of observational seismology of stacking waveforms (adding multiple seismograms), which are best done using arrays of seismometers. They will also provide a good test for current approximate methods used to image and model the mantle structure.The research team has the resources (access to high performance computing), tools (code to model mantle circulation (TERRA, Fluidity) and seismic wave propagation (SPECFEM3D_GLOBE)), data, track record and expertise (including partners for plate motion histories, mineral physics, modelling, and seismic data analysis) in place to undertake this ambitious project.