Deriving nearshore bathymetry and waves characteristics from a single UAV video

Abstract

Nearshore bathymetry and water surface elevation are estimated using the temporal correlation method and the sky glint method, respectively, from a 14 min UAV video captured at Torrey Pines State Beach, CA, USA, under relatively energetic wave conditions ($H_0=1.96$ m) generating breaking waves. The estimated bathymetry shows strong agreement with in-situ measurements, with an RMSE of 0.74 m, a bias of 0.52 m, and $R^2=0.98$, comparable to other established methods. More importantly, the resulting depth estimation remains unaffected by the abrupt transitions caused by wave breaking and the presence of wave-induced foam in the surf zone. At the breaking point, the temporal correlation method consistently estimates wave celerities at this point, providing accurate depth measurements - a persistent challenge in nearshore bathymetry studies. A methodology to assess the correlation significance of the estimated bathymetry is presented. These results highlight the method’s potential for monitoring nearshore morphological changes such as induced by storms. The application of the sky glint method shows a tendency to overestimate individual wave heights (RMSE $=0.59$ m and Bias $=0.31$ m) compared to in-situ pressure sensor measurements. Surface elevations time series derived from the video are phase-coherent with in-situ pressure sensor measurements but overestimations are attributed to the use of a linear Modulation Transform Function (MTF). While further study is required, this proof of concept could constitutes a non-labour-intensive, remotely sensed method for generating water surface elevation datasets for sea state assessments and hydrodynamic model forcing. This study demonstrates the abundance of information that can be extracted from a 14 min UAV flight, offering a cost-effective, high-frequency sampling and high-resolution approach for coastal managers and researchers to monitor and study their nearshore environments, with potentially short - daily - revisit time.