Imaging the next great Cascadia earthquake: Optimal design for a seafloor acoustic-GNSS network

Up to 90% of the world?s primary plate boundaries are located in the ocean. This includes the Cascadia subduction zone offshore the Pacific Northwest of the United States of America, which is known to produce magnitude 9.0 earthquakes, and accompanying tsunamis, every ~600 years. An outstanding question in this region, as in most subduction zones, is the degree and spatial extent of strain accumulation (which will eventually release as an earthquake) on the subduction megathrust. Seafloor geodetic technology allows for measurements of crustal deformation at these otherwise inaccessible plate boundaries, but the ship time and logistics associated with these observations can be cost-prohibitive. The primary objectives of this work are to 1) quantify the relative information provided by expanding current Global Navigation Satellite Systems (GNSS) observation networks offshore northern California, Oregon and Washington, and 2) identify optimal locations for a network of seafloor geodetic instruments. Seafloor geodetic observations of the Cascadia subduction zone will address a previously unanswerable scientific question with direct hazard implications: Is the Cascadia megathrust fully locked? Because onshore observations are typically far from the seismogenic and tsunamigenic portions of subduction zone megathrusts, we have limited ability to image subduction zone locking, and there is significant uncertainty in locking distributions. We seek to quantify the relative importance of a well-placed seafloor geodetic observation for determining subduction zone locking using the theory of information entropy. Information entropy is a measure of the amount of missing information about a parameter (in this case, subduction zone locking) whose value is uncertain. Our goal is to identify new observation locations that minimize the information entropy. Or, equivalently, we want to maximize the information gain due to a new observation. By considering a suite of computational models that represent current understanding of the Cascadia subduction zone and its tectonic context, this optimization will incorporate expected uncertainties on model parameters, smoothing on triangular dislocation elements, and uncertainties on existing and new GNSS observations. This project will identify new observation locations that minimize the change in differential entropy (maximize the information gain), for networks that include up to 50 new stations. The results of this project will enable well-informed decisions about expanding current geodetic networks. For example, relative financial costs of seafloor geodetic technology may be compared in terms of information gain per dollar. Preliminary work for a single set of model assumptions suggests that a single seafloor observation may provide 5-10 times the information gain of the most optimal additional onshore observation. In other words, this suggests that deployment of seafloor geodetic observations will allow earthquake scientists to map strain accumulation on the Cascadia megathrust, prior to an inevitable future earthquake, in a way that is impossible with onshore observations alone.