A 2D Fully Convolutional Neural Network for nearshore and surf-zone bathymetry inversion from synthetic imagery of the surf-zone using the model Celeris

Adam Collins1, Katherine L Brodie2, Spicer Bak2, Tyler Hesser3, Matthew Farthing3, Douglas W Gamble4 and Joseph W Long5, (1)University of North Carolina at Wilmington, Earth and Ocean Sciences, Wilmington, United States, (2)U.S. Army Engineer Research and Development Center, Coastal and Hydraulics Laboratory, Duck, NC, United States, (3)U.S. Army Engineer Research and Development Center, Coastal and Hydraulics Laboratory, Vicksburg, United States, (4)University of North Carolina at Wilmington, Earth and Ocean Sciences, Wilmington, NC, United States, (5)University of North Carolina at Wilmington, Physics and Physical Oceanography, Wilmington, United States
Abstract:
Up-to-date observations of nearshore and surf-zone bathymetry have a variety of important uses, such as locating and warning beach-goers about rip currents due to sandbar gaps, informing regional sediment management, or quantifying storm impacts. Typical survey techniques are expensive and time-consuming, and are not feasible in a variety of situations (e.g. manpower, on-site access). However, the emergence of nearshore remote sensing platforms (e.g. UAS, towers, and satellites) from which high-resolution imagery of the sea-surface can be collected at frequent intervals, has unleashed the potential for accurate bathymetric estimation from wave-inversion techniques without in-situ measurements or a physical presence in the study area. While a variety of physics-based algorithms have been applied to nearshore and surf-zone bathymetric inversion, they have difficulty accounting for non-linear effects and can have biases during breaking waves. Fully convolutional neural networks (FCNs) are a branch of artificial intelligence algorithms that have proven effective at computer vision tasks in semantic segmentation and regression problems. In this work, we consider the use of FCNs for inferring bathymetry from video-derived imagery. The model presented shows the feasibility of using an AI system to perform bathymetric inversion on time-lapsed images (timex) of realistic-looking, synthetically generated surf-zone imagery from the wave model Celeris[1]. Development continues on the FCN with additional inputs of synthetic video frames, and eventually real tower and satellite imagery. Funded by the Deputy Assistant Secretary of the Army for Research and Technology under ERDC’s Military Engineering research program titled “Force Projection Entry Operations”.

[1] Tavakkol, Sasan, and Patrick Lynett. "Celeris: A GPU-accelerated open source software with a Boussinesq-type wave solver for real-time interactive simulation and visualization." Computer Physics Communications 217 (2017): 117-127.

Back to: Accelerating Ocean Science: Using Artificial Intelligence and Machine Learning to Gain New Insights I