Climate change is having lasting effects such as changing ocean acidity, temperatures, and heights, increasing the chances of natural hazards such as forest fires and hurricanes, and leading to the disappearance of water sources supplied by glaciers and snow. Carbon dioxide (CO2) produced by humans is a primary contributor to man-made climate change thus finding storage mechanisms for human produced CO2 is necessary to mitigate climate change. Mantle rocks, like peridotite, offer a storage option for CO2, in the presence of water, peridotite will alter into new kinds of rocks. If CO2 is present in the water this CO2 will be stored as a solid in the rocks, keeping it trapped underground forever. Currently this chemical alteration process is well understood however the physical process that allows water to reach unaltered peridotite is not. SerpRateAI will discover the physical processes that allow water to be pumped underground to understand how peridotite alters. Ultimately we will produce natural alteration rates, i.e., how fast the rock changes over time. This can be used to estimate how much CO2 we can store in a region. Given that mantle rocks like peridotite are common, this could feasibly be used to store a large fraction of human produced CO2.
Peridotite rocks are found in oceanic environments and outcrop on continents around the world. The alteration of peridotite, i.e., serpentinization, is a major geodynamic process that controls the density of crustal rocks and the global cycles of carbon dioxide and hydrogen. The physical mechanisms that allow peridotite to fully serpentinize remain elusive. The mechanism of reaction driven cracking has been suggested to promote serpentinization because the increase in volume creates stress on the surrounding rock, promoting fracture development in unaltered rock. This process also acts as a pumping mechanism, delivering fluids from the recently altered rock into new areas. Understanding and detecting this process in natural environments is important for estimating serpentinization rates. SerpRateAI aims to both constrain and explain serpentinization rates using dozens of terabytes of data collected in the Oman Drilling Program Multi-Borehole Observatory, in a region undergoing active serpentinization. The data include seismic data from hydrophone and geophone networks, X-ray tomography imaging of the entire borehole cores, down-borehole video, core logging data, hydrological data such as flow rates and fluide levels in the borehole, and ambient atmospheric and borehole conditions. SerpRateAI will produce an ensemble of machine learning models that will 1) explore the relationship between serpentinization rates and cracking events and degassing events detected in the seismic data, 2) explore the development of serpentine vein networks using the X-ray tomography data and use these networks to identify areas of active serpentinization, and 3) predict and explain the observed serpentinization rates in the context of the observed seismicity, degassing, vein developments, and ambient conditions. The project builds on the strong international collaboration between the University of Oslo, Utrecht University, Woods Hole Oceanographic Institution, and Columbia University.