Full waveform seismic monitoring of CO2 injection and storage with broadband and fiberoptic sensors

加拿大地球科学的一个关键优先事项是实现碳捕集、利用和封存（CCUS），即温室气体永久封存在地质结构中，减少其对气候变化的影响。应用地震学将在这方面发挥关键的监测作用。我和我的团队将寻求在我们使用地震方法监测二氧化碳储存的能力方面取得突破，充分利用数据，可靠且成本低。这项研究将得到与弹性波勘探地震学研究联合会（一个有能力开展定制陆地地震实验的小组）以及通过其实地研究站（CaMI.FRS）与遏制和监测研究所合作的支持。我的HQP和我将利用这种协作环境来创建新的、基于全波形反演的地震方法，以确保安全稳定的存储。在建立储层模型所需的FWI变体中，存在基本的数学和数值问题，这使得绘制多种岩石物性及其随时间变化的图变得更加复杂。我们将通过算法构建和数值分析来解决这个问题，并使用CaMI.FRS获得的自定义现场数据验证结果。我和我的小组也将针对基于fwi的监控的计算费用。近年来，硬件/软件的进步使FWI的计算工具更容易获得，而且，关键的是，它们还提供了两种新技术，这将是低成本、可靠监测问题的核心。第一种是分布式声学传感（DAS），其中光纤电缆被用作灵活、易于部署的采集系统。与DAS数据相关的成本降低使其成为新监测技术的关键，但数据的不寻常属性意味着需要特殊方法来发挥其潜力。我和我的团队将把DAS与高频FWI合并，作为我们提出的核心研究目标之一。第二项技术是与深度学习相关的计算平台套件，以及它们为FWI的扩展和计算效率优化带来的可能性。我的团队曾开发过用于地震反演的深度学习工具，但数据丰富的二氧化碳监测问题尚未被提出。在拟议的计划中，我和我的团队将开发定制设计的神经网络，在涉及波动理论和岩石物理学的物理指导下，输出密封和/或储存故障的指标。在这个研究计划中，科学需要创造工具来监测二氧化碳的地质储存，将在帮助加拿大实现其气候目标方面发挥作用。重要的是，通过指导、推广和有针对性的招聘，这项研究将在我们在一个包容的环境中培训多样化的HQP群体的同时完成。

A critical geoscientific priority in Canada is to enable Carbon Capture, Utilization, and Storage (CCUS), in which greenhouse gases are permanently sequestered in geological structures, reducing their impact on climate change.  Applied seismology stands to play a critical monitoring role in this. My group and I will seek to make breakthroughs in our ability to use seismic methods to monitor CO2 storage, making full use of data, reliably and at low cost. The research will be supported by collaboration with the Consortium for Research in Elastic Wave Exploration Seismology, a group with the capacity to carry out custom land seismic experiments, and with the Containment and Monitoring Institute through their Field Research Station (CaMI.FRS). My HQP and I will leverage this collaborative environment to create new, full waveform inversion-based seismic methods for ensuring safe and stable storage. In the FWI variants needed for reservoir model-building, basic mathematical and numerical issues exist, that make mapping of multiple petrophysical properties, and their changes with time, more complicated to produce. We will address this through algorithm-building and numerical analysis, and validate the results with custom field data acquired at the CaMI.FRS.  My group and I will also target the computational expense of FWI-based monitoring. In recent years, hardware/software advances have made computational tools for FWI more easily available, and, critically, they have also made available two new technologies which will be central to the problem of low-cost, reliable monitoring. The first is Distributed Acoustic Sensing (DAS), in which fiberoptic cables are used as flexible, easily-deployable acquisition systems. The reduction in cost associated with DAS data makes it a key to new monitoring technology, but the unusual properties of the data mean that special methods will be needed to leverage its potential. My group and I will merge DAS with high-frequency FWI as one of our core proposed research objectives. The second technology is the suite of computational platforms associated with deep learning, and the possibilities they introduce for extension, and computationally efficient optimization, of FWI. My group has a record of developing deep learning tools for seismic inversion, but the data-rich problem of CO2 monitoring has not yet been broached. In the proposed plan, my group and I will develop custom-designed neural networks, constrained with physics-guiding involving wave theory and petrophysics, to output indicators of containment and/or storage failures. In this research plan, the science needed to create tools for monitoring of geological storage of CO2 will be carried out playing a role in helping Canada achieve its climate objectives. Importantly, by means of mentoring, outreach, and targeted recruitment, this research will be accomplished at the same time as we train a diverse cohort of HQP, in an inclusive environment.