ScienceDaily (Aug. 30, 2010) — Computational scientistsand geophysicists at the University of Texas at Austin and the CaliforniaInstitute of Technology (Caltech) have developed new computer algorithms thatfor the first time allow for the simultaneous modeling of Earth's mantle flow,large-scale tectonic plate motions, and the behavior of individual fault zones,to produce an unprecedented view of plate tectonics and the forces that driveit.
A paper describing the whole-earth model and itsunderlying algorithms will be published in the August 27 issue of the journalScience and also featured on the cover. 一篇作为封面将刊载于八月份第27期科学杂志的论文将描述这个全地球模型及其基层算法。 The work "illustrates the interplay between makingimportant advances in science and pushing the envelope of computationalscience," says Michael Gurnis, the John E. and Hazel S. Smits Professor ofGeophysics, director of the Caltech Seismological Laboratory, and a coauthor ofthe Science paper. 这项工作将“举例说明在利用科学给出重要建议和增强计算机科学方面的能力之间的相互影响”，麦克尔·格尼斯，地球物理学教授，加州理工地震学实验室主任，科学杂志论文的合著者约翰·E与黑兹尔·S·斯密斯说道。 To create the new model, computational scientists atTexas's Institute for Computational Engineering and Sciences (ICES) -- a teamthat included Omar Ghattas, the John A. and Katherine G. Jackson Chair inComputational Geosciences and professor of geological sciences and mechanical engineering,and research associates Georg Stadler and Carsten Burstedde -- pushed theenvelope of a computational technique known as Adaptive Mesh Refinement (AMR). 为建立这个新模型，德克萨斯计算机工程与科学学院（ICES)的计算机科学家们 —— 一个包括奥马尔·伽特斯，计算机地球科学系主任，地质学和机械工程学教授约翰·A同凯瑟琳·G·杰克森，以及研究合伙人格奥尔格·斯泰德勒和卡斯滕·布斯特德 —— 增强被称为自适应精细网格的计算机技术系统的运行能力 Partial differential equations such as those describingmantle flow are solved by subdividing the region of interest (such as themantle) into a computational grid. Ordinarily, the resolution is kept the samethroughout the grid. However, many problems feature small-scale dynamics thatare found only in limited regions. "AMR methods adaptively create finer resolutiononly where it's needed," explains Ghattas. "This leads to hugereductions in the number of grid points, making possible simulations that werepreviously out of reach." 像用于描述地幔流的偏微分方程就通过将感兴趣的区域（例如地幔）细分到计算机网格中被解答。一般，网格的解析率是均一的。但是，很多问题针对只建立在有限区域内的小规模动力学。“AMR方法仅在需要的地方自适应性的使用更高的解析率，”伽特斯解释道。“这导致格点数量大幅减少，使得原先超出研究范围的仿真成为可能。” "The complexity of managing adaptivity amongthousands of processors, however, has meant that current AMR algorithms havenot scaled well on modern petascale supercomputers," he adds. Petascalecomputers are capable of one million billion operations per second. To overcomethis long-standing problem, the group developed new algorithms that, Bursteddesays, "allows for adaptivity in a way that scales to the hundreds ofthousands of processor cores of the largest supercomputers availabletoday." “然而，管理数千处理器之间的自适应分配的复杂性，意味着目前的AMR算 法不能很好的在现代千万亿次超级计算机上按比例分配任务，”他补充道。千万亿次计算机能够每秒钟运算一千万亿次。为了解决这个长期存在的问题，这个团队开 发了新的算法，它“允许某种程度上使得在最大的超级计算机上自适应性的按比例分配任务给数十万处理器成为可能。”布斯特德说道。 With the new algorithms, the scientists were able tosimulate global mantle flow and how it manifests as plate tectonics and themotion of individual faults. According to Stadler, the AMR algorithms reducedthe size of the simulations by a factor of 5,000, permitting them to fit onfewer than 10,000 processors and run overnight on the Ranger supercomputer atthe National Science Foundation (NSF)-supported Texas Advanced Computing Center. 使用新算法，科学家们能够模拟全球地幔流动以及它是如何显明板块构造和断层移动的。据斯泰德勒说，AMR算法通过5千个要素减小仿真尺寸，允许它们在国家科学基金的空闲超级计算机—由德克萨斯先进计算机中心维护，上用少于1万个处理器通宵运行。 A key to the model was the incorporation of data on amultitude of scales. "Many natural processes display a multitude ofphenomena on a wide range of scales, from small to large," Gurnisexplains. For example, at the largest scale -- that of the whole earth -- themovement of the surface tectonic plates is a manifestation of a giant heatengine, driven by the convection of the mantle below. The boundaries betweenthe plates, however, are composed of many hundreds to thousands of individualfaults, which together constitute active fault zones. "The individualfault zones play a critical role in how the whole planet works," he says, "andif you can't simulate the fault zones, you can't simulate plate movement"-- and, in turn, you can't simulate the dynamics of the whole planet. 模型的关键在于超大尺寸上的数据结合。“很多自然过程在一个宽泛的尺寸上显示大量的现象，从微观到宏观，”格尼斯解释道。例如，在最大的尺寸上—整个地球—表面构造板块的移动表现为一个由地幔流的热循环驱动的巨型热力发动机。然而，板块间的边界由无数独立断层组成，它们一起组成断层区域。“独立断层区域在整个板块如何工作上扮演着重要角色，”他说，“如果你不能模拟断层区域，你就不能模拟板块运动”--其次，你就不能模拟整个板块的动态。 In the new model, the researchers were able to resolvethe largest fault zones, creating a mesh with a resolution of about onekilometer near the plate boundaries. Included in the simulation wereseismological data as well as data pertaining to the temperature of the rocks,their density, and their viscosity -- or how strong or weak the rocks are,which affects how easily they deform. That deformation is nonlinear -- withsimple changes producing unexpected and complex effects. 在新的模型里，研究人员能够分解最大的断层区域，在接近板块边缘处创建一个分辨率大约一公里的网络。模拟包括地震学数据还有岩石的温度，密度，粘性--或者这些岩石有多坚硬或柔软这类影响它们有多容易变形的固有数据。这些变形是非线性的--简单的改变也会产生出乎预料和复杂的影响。 "Normally, when you hit a baseball with a bat, theproperties of the bat don't change -- it won't turn to Silly Putty. In theearth, the properties do change, which creates an exciting computationalproblem," says Gurnis. "If the system is too nonlinear, the earthbecomes too mushy; if it's not nonlinear enough, plates won't move. We need tohit the 'sweet spot.'" “平常，当你用球棒击球时，球棒的属性不会发生变化—它不会变成橡皮泥。在地球模型上，属性会发生变化，这种变化产生一个令人激动的计算问题，”格尼斯说。“如果系统过于非线性，这个地球会变得太软；如果不够非线性，板块将不会运动。我们需要正好打倒‘击球点’上。 After crunching through the data for 100,000 hours ofprocessing time per run, the model returned an estimate of the motion of bothlarge tectonic plates and smaller microplates -- including their speed anddirection. The results were remarkably close to observed plate movements. 在以每次运行用十万小时处理时间对数据进行处理后，模型反馈一个大型结构板块与较小的微型板块两者的运动的估计值。计算结果与观测到的板块运动非常接近。 In fact, the investigators discovered that anomalousrapid motion of microplates emerged from the global simulations. "In thewestern Pacific," Gurnis says, "we have some of the most rapidtectonic motions seen anywhere on Earth, in a process called 'trench rollback.'For the first time, we found that these small-scale tectonic motions emergedfrom the global models, opening a new frontier in geophysics." 事实上，调查人员从地球模拟器中发现微型板块不寻常的快速移动。“在西太平洋，”格尼斯说，“我们获得了在地球某些地方发现的大量快速结构移动中的几处，这一过程被称为‘海沟反转’。我们首次发现这样一些小规模结构运动从地球模型中浮现出来，它们在地球物理领域开创了一个新前沿。 One surprising result from the model relates to theenergy released from plates in earthquake zones. "It had been thought thatthe majority of energy associated with plate tectonics is released when platesbend, but it turns out that's much less important than previouslythought," Gurnis says. "Instead, we found that much of the energydissipation occurs in the earth's deep interior. We never saw this when welooked on smaller scales." 一 个来自地球模型的令人意外的成果涉及到地震区域里自板块释放的能量。“能量的大部分曾被认为与板块弯曲时板块结构被释放的过程有关，然而结果显示其并没有 原先想象得那么重要。“在这之外，我们发现多数能量耗散发生在地球内部深处。这是在我们观察更小规模的模型时无法看到的。” |
|
|
科学日报（
