Collaborative Research: EarthCube Capabilities: Open Polar Radar (OPoRa) Software and Service

地球的极地冰盖在地质时期的海平面形成中发挥着关键作用，但冰盖对当代气候变化的反应仍然高度不确定。通过测量海拔、重力和冰流变化的变化，四十年的星载观测揭示了最近冰盖质量损失的加速。然而，利用这些卫星观测来预测未来的行为仍然很困难。机载雷达测深仪测量提供了约束冰盖表面动态的潜力，因为它们可以在整个冰盖范围内绘制出地下参数。然而，50年来，多个组织部署了一系列系统，数据分布在不一致的数据策略和处理方法下，这些系统的雷达数据收集导致了孤立的研究，从而降低了效率，增加了科学研究的时间。开放极地雷达小组掌握了83%的南极洲雷达测深数据，几乎覆盖了整个格陵兰岛和极地海冰。这些数据集将以通用格式放置，并通过最终用户驱动的过程，通过具有一套通用工具的通用接口提供给科学家。开放极地雷达有潜力在多个尺度上极大地改进冰盖模型。这可以从根本上提高对冰盖崩塌过程的认识，修正对过去冰盖动力学的认识，并迅速提高对海平面上升的估计。改进海平面预测将导致更好的减灾战略，从而减少沿海洪水的危险和成本。雷达软件和服务将用于其他探测雷达问题，包括行星应用。共享的工具将有助于并利用EarthCube生态系统，为数据共享和图像分析研究中的其他挑战提供解决方案。开放极地雷达将有助于本科生、研究生和博士后的多学科培训。主要研究人员将与科学家进行年度培训研讨会、黑客马拉松和特定学科会议活动，以收集反馈并推进冰冻圈科学界的雷达能力。该项目还将在夏季研究经验中纳入代表性不足的社区。开放极地雷达项目将建立一个软件生态系统，以整合极地雷达软件和服务，并使相关数据集标准化和可搜索-所有这些都在社区驱动的过程中。目标受众是放射冰川学领域的科学家、人工智能数据科学家，以及在冰盖、高山冰川、海冰和行星冰体上收集各种卫星、机载和地面冰穿透雷达数据的软件和数据工程师。极地雷达测深数据在很大程度上是一种未开发的资源，因为分散处理和人工解释带来的挑战阻碍了雷达数据在全冰盖范围内的先进应用。这项工作开启了一个不断增长的地球物理图像人工智能训练数据数据库，使传统摄影中常见的计算机视觉技术能够进入一个直到最近还依赖人类从地球地下图像中提取意义的领域。为实现科学创新数据共享，本项目提出：(1)将来自每个机构的数据处理链合并到一个开源软件套件中；(2)升级当前的Open Polar服务器，以满足可查找可访问互操作和可重用（FAIR）原则和多机构需求，并纳入一些新的数据层；(3)在科学应用和人工智能模型需求的驱动下，采用共享的去噪和图像增强工具，提高所有数据产品的质量。(4)展示几个用例，包括应用人工智能和物理模型相结合的方法来跟踪冰川特征，以验证两个区域的结果，这两个区域是多个数据提供商覆盖的交叉点。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Earth’s polar ice sheets play a critical role in shaping sea level over geological time, yet ice-sheet response to contemporary climate change remains highly uncertain. Forty years of spaceborne observations reveal the recent acceleration in mass loss of the ice sheets through measured change in elevation, gravity, and ice-flow variation. It remains difficult, however, to use these satellite observations to predict future behavior. Airborne radar sounder measurements offer the potential to constrain ice-sheet surface dynamics because they can map out subsurface parameters on an ice-sheet-wide scale. However, five decades of radar data collection by multiple organizations deploying a range of systems with data distributed under inconsistent data policies and processing methods leads to siloed research resulting in lower efficiency and increased time to science. The Open Polar Radar team accounts for 83% of Antarctica radar sounder data and nearly complete Greenland and polar sea-ice coverage. These datasets will be placed in common formats and made available through a common interface with a common set of tools for scientists via an end-user driven process. Open Polar Radar has the potential to vastly improve ice-sheet models at multiple scales. This could radically improve the understanding of the ice-sheet collapse processes, revise the understanding of past ice-sheet dynamics, and rapidly improve sea-level rise estimates. Improved sea-level projections will lead to better mitigation strategies that could reduce the dangers and costs of coastal flooding. The radar software and services will be common to other sounding radar problems, including planetary applications. The shared tools will contribute to and leverage the EarthCube ecosystem leading to solutions to other challenges in data sharing and image analytics research. Open Polar Radar will contribute to the multidisciplinary training of undergraduate and graduate students and postdocs. The principal investigators will engage scientists with annual training workshops and hackathons and activities at discipline-specific meetings to gather feedback and to advance the radar capabilities of the cryospheric sciences community. The project will also include under-represented communities in summer research experiences.The Open Polar Radar project will establish a software ecosystem to consolidate polar radar software and services and make the associated datasets standardized and searchable – all in a community driven process. The target audiences are radioglaciology domain scientists, AI data scientists, and the software and data engineers behind the various satellite, airborne, and ground-based ice penetrating radar data collected over ice sheets, mountain glaciers, sea ice, and planetary icy bodies. Polar radar sounder data are a largely untapped resource, because the challenges associated with decentralized processing and manual interpretation prevent advanced applications of radar data on an ice-sheet-wide scale. This work starts a growing database of AI training data for geophysical imagery, enabling computer vision techniques common for traditional photography to a field that has, until very recently, relied on humans to extract meaning from images of Earth’s subsurface. To enable sharing data for scientific innovation, this project proposes to: (1) Merge data processing chains from each institution into one open-source software suite, (2) Upgrade the current Open Polar Server to meet Findable Accessible Interoperable and Reusable (FAIR) principles and multi-institution needs as well as incorporate a number of new data layers, (3) Enhance the quality of all data products by employing the shared tools for noise removal and image enhancement driven by science applications and AI model needs, (4) Demonstrate several use cases including the application of combined AI and physical models for tracking englacial features to validate results in two regions that are intersections of multiple data providers’ coverage.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.