GEO OSE Track 1: Enhancing the accessibility of novel geostatistical inversion workflows for cryosphere research

许多地球科学数据集需要空间内插才能用于科学研究。通常重要的是，这些内插是以满足某些物理和统计特性的方式进行的。这些“地统计学”方法被广泛应用于采矿和石油行业，但没有免费软件使学生、学者和环境研究人员可以使用这些方法。该项目通过创建免费提供的地质统计学软件来解决这一无障碍问题。该软件将包括将物理信息和统计信息相结合的新方法。这些算法将足够快，以便它们可以用于非常大的数据集。为了方便不同学科的其他研究人员使用这些工具，该软件将安装在几个不同的在线科学计算中心。该软件将附带在线教育材料。该软件将用于结合各种不同的测量来源，创建强大的南极冰川下地形图。这些地形图将有助于提高冰盖和海平面上升模型的精确度。这项工作将建立在现有的GStatSim Python地统计学软件和教育Jupyter Book的基础上，以执行高速物理信息的地统计学模拟。研究小组将开发一种新的马尔可夫链蒙特卡罗方法来执行地球物理反演，其中地质统计模拟被迭代扰动，直到地球物理正演模型的输出与测量结果收敛。将使用并行处理技术来提高该方法的可扩展性。这将使用户能够高速生成大规模地质统计模拟的集合，同时考虑到空间和物理限制。该方法将被测试并应用于两个案例研究：1)利用重力观测模拟冰架下地形；2)模拟具有质量守恒约束的冰下地形。这些案例研究将为冰盖模型提供关键参数。为了促进GStatSim在冰冻圈社区的使用，这些工具将托管在GHub和CryoCloud计算平台上。该软件还将链接到地球科学信息合作伙伴工具箱页面，以便使更广泛的地球科学受众能够访问该包。该奖项反映了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。

Many Earth science data sets require spatial interpolation in order to be used in scientific studies. It is often important that these interpolations are done in a way that satisfies certain physical and statistical properties. These “geostatistical” methods are widely used in the mining and petroleum industries, but there is no free software that makes these methods available to students, academics, and environmental researchers. This project addresses this accessibility barrier by creating freely available geostatistics software. This software will include novel methods for combining physical and statistical information. These algorithms will be fast enough so that they can be used for very large data sets. To make it easy for other researchers in different disciplines to use these tools, the software will be installed in several different online scientific computing hubs. The software will be accompanied by online educational materials. The software will be used to create robust maps of the topography underneath Antarctic glaciers by combining a variety of different measurement sources. These topographic maps will help improve the rigor of ice-sheet and sea level rise models.This work will build upon the existing GStatSim Python geostatistics software and educational Jupyter Book to perform high-speed physics-informed geostatistical simulations. The research team will develop a novel Markov Chain Monte Carlo approach for performing geophysical inversions where geostatistical simulations are iteratively perturbed until the outputs of a geophysical forward model converge with measurements. Parallel processing techniques will be used to improve the scalability of this method. This will enable users to generate ensembles of large-scale geostatistical simulations at high speeds while accounting for both spatial and physical constraints. This method will be tested and applied to two case studies: 1) the simulation of sub-ice-shelf topography using gravity observations, and 2) the simulation of subglacial topography with mass conservation constraints. These case studies will provide critical parameters for ice-sheet models. To facilitate the use of GStatSim in the cryosphere community, these tools will be hosted on the GHub and CryoCloud computing platforms. This software will also be linked to the Earth Science Information Partners toolbox page in order to make this package accessible to broader geosciences audiences.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.