Improvements to Roller Dissipation Estimates from Optical Imagery
Abstract
As waves pass the break point and enter the surf-zone, they begin to break by plunging and spilling, creating a turbulent interaction at the water surface stirring up white foam often mathematically described as the front face of a rolling bore in a hydraulic jump. It is fairly easy for a human to identify the locations of the active breaking roller. However, methods to compare locations of wave breaking that can be identified in optical imagery with modeled roller dissipation have struggled due to the optical similarity between foam created during active breaking, and relic foam from previous rollers. This complicates the implementation of previous methods that attempt to estimate the magnitude of roller dissipation by relating phase-averaged model dissipation fields to the intensity of time-averaged optical imagery. To alleviate this issue, we propose an analysis framework that applies a machine learning algorithm for identifying active wave breaking to frame-by-frame video imagery (e.g. Saez et al., 2021, CoastEng), creating a full frame-by-frame catalog of wave breaking events. These wave breaking identifications can then be converted into a time-averaged map that quantifies locations of active roller dissipation. This approach effectively de-noises the optical imagery from relic foam and other environmental noise, allowing it to better correlate to phase-averaged coastal models than typical time-averaged optical imagery. This processing improves automated estimates of bathymetry in the surf-zone using the Beach Wizard algorithm (e.g. Aarninkhof et al., 2005, JGR:Oceans, VanDongeren et al., 2008, CoastEng.).
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2022
- Bibcode:
- 2022AGUFMOS42A..04C