CREST-PRF: Optimization of Joint Inversion of Geophysical Data to Improve 3-Dimensional Models of Earth Structures

优化地球物理数据的联合反演以改进地球结构的三维模型CREST计划内的科学和技术卓越研究中心-博士后研究奖学金(CREST-PRF)轨道支持CREST中心具有巨大潜力的初级研究人员，并为他们提供培训和研究经验，这些培训和研究经验将拓宽视野，促进跨学科互动，并使他们在科学界处于领导地位。这一CREST-PRF项目与CREST网络基础设施的研究重点相一致，以共享资源，以推进德克萨斯大学埃尔帕索分校的研究和教育(网络共享)卓越中心。地球科学的目标是通过使用广泛的方法来确定地球的内部组成，以了解其在数千年中的变化，并预测未来可能的行为。例如，地震活跃度相对较低的地区曾经历过地震，原因是地球动态内部的构造断层运动。此外，非典型的板块构造过程，如2017年9月墨西哥最近发生地震时发生的俯冲大洋板块破裂，使我们必须了解复杂的过程，以避免未来的灾难。这项研究项目将描述地震断层的位置和条件以及下地壳古应变的现状，并解决有关地球构造演化的基本问题。具体地说，该项目将通过优化互补地震(面波组速度和远震P波接收函数)和非地震(布格重力异常)数据集的联合反演，开发和改进目标区域的三维(3-D)地球模型。这些数据集的综合分辨率将使我们能够精确定位地下可能被地球物理调查忽视的关键特征。大学、高中和传统的K-12学生将参与这一项目。所提出的研究将开发和应用数值优化来提高三维地球结构模型的分辨率。尽管以前已经开发了重要的方法来优化针对日益复杂的地质环境的互补数据集的反演结果，但最近更强大的数学工具提供了证据，表明约束优化技术可以极大地改进地球内部最终的结构模型。利用原始-对偶内点方法进行地震和非地震数据集的联合反演，通过约束解空间和只关注可行的构造模型，可以进一步改进地球的三维成像。该项目的第一部分涉及开发计算算法，以优化从多个数据集的联合反演获得的结果。在第2部分中，将分析优化所涉及的进程，以确定在多核环境中同时执行代码(并行编程)的最佳和最具成本效益的工作分工。这些算法将应用于哥伦比亚的数据集，以获得对其复杂的子结构的新的构造理解--其特征是科科斯、加勒比海和南美板块交汇的独特地质环境。这一奖项反映了NSF的法定使命，并通过使用基金会的智力优势和更广泛的影响审查标准进行评估，被认为值得支持。

Optimization of Joint Inversion of Geophysical Data to Improve 3-Dimensional Models of Earth StructuresThe Centers of Research Excellence in Science and Technology-Postdoctoral Research Fellowship (CREST-PRF) track within the CREST program supports beginning CREST Center investigators with significant potential and provides them with training and research experiences that will broaden perspectives, facilitate interdisciplinary interactions and establish them in positions of leadership within the scientific community. This CREST-PRF project is aligned with the research focus of the CREST Cyberinfrastructure for Sharing resources to Advance Research and Education (Cyber-ShARE) Center of Excellence at the University of Texas at El Paso. Earth sciences aim to determine the interior composition of the Earth by using a wide array of methods to understand its changes throughout the millennia and predict possible future behavior. For example, regions with relatively low seismic activity have experienced earthquakes due to motion of tectonic faults within the Earth's dynamic interior. Moreover, atypical plate tectonic processes, such as the breaking of a subducting oceanic plate that occurred with recent Mexico earthquakes in September 2017 make understanding the complex processes imperative to avoid future catastrophes. This research project will characterize the location and condition of seismological faults and the current state of ancient strains in the lower crust and address fundamental questions about the tectonic evolution of the Earth. Specifically, the project will develop and improve 3-Dimensional (3-D) Earth models in targeted regions through the optimization of the joint inversion of complementary seismic (surface wave group velocities and teleseismic P-wave receiver functions) and non-seismic (Bouguer gravity anomalies) datasets. The combined resolution of these datasets will allow pinpointing of key features within the subsurface that may be overlooked by the geophysical surveys. Early college, high school and traditional K-12 students will be engaged in this project.The proposed research will develop and apply numerical optimization to improve the resolution of 3-D Earth structure models. Although important methods have been previously developed to optimize the results obtained from the inversion of complementary datasets for increasingly complex geological environments, recent more powerful mathematical tools have provided evidence that constrained optimization techniques can greatly improve final structural models for the interior of the Earth. The use of primal-dual interior point methods for the joint inversion of seismic and non-seismic datasets can provide further improvement to 3-D imaging of the Earth by constraining the solution space and focusing only on feasible structural models. Part 1 of this project involves the development of computational algorithms to optimize the results obtained from the joint inversion of multiple datasets. In Part 2, the processes involved in the optimization will be analyzed to determine the optimal and most cost-effective division of work to simultaneously execute the code (parallel programming) in a multi-core environment. These algorithms will be applied to datasets from Colombia to obtain a new tectonic understanding of its complex substructure-characterized by a unique geological setting where the Cocos, Caribbean, and South American plates converge.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.