Data-Driven Methods for Ocean Surface Wave Measurements in the Coastal Arctic Using Submarine Fiber Optic Cables

Distributed Acoustic Sensing (DAS) provides a novel method to measure ocean surface waves in the coastal ocean with unprecedented persistence and spatial coverage. DAS uses existing submarine fiber optic cables to observe seafloor strain, a proxy for seafloor pressure, at “channels” spaced every several meters along a many kilometer-long cable. A key challenge is converting measured strain rate to seafloor pressure, since necessary parameters such as seabed composition and cable burial depth are often unknown or difficult to characterize over the full extent of the fiber. We use a 3-month dataset of colocated DAS and ground truth measurements (seafloor pressure moorings) offshore of Oliktok Point, Alaska, to explore the use of sequential machine learning models for the robust conversion of DAS strain rate to seafloor pressure. We focus on short time series with stationary statistics and include other external information such as water depth and temperature. Learning the transformation from strain rate to seafloor pressure in the time domain leverages the phase information of incident waves measured across neighboring spatial channels, thus helping to distinguish ocean waves from other noise common to DAS signals. An accurate and stable DAS calibration will enable the study of coastal processes using spatiotemporal methods including wave-by-wave analysis and frequency-wavenumber (f-k) analysis. The development of these methods will be powerful for characterizing highly variable wave environments such as the surf zone and seasonally ice-covered coastal environments in the Arctic.