Estimating Water Depth Remotely




Our ongoing work uses both video and LiDAR to estimate water depth, with goal of improving understanding of runup and overtopping at vulnerable sites. Water depth controls wave breaking across the surfzone and shoreline wave runup but is difficult to measure directly during storms. However, water depth can be estimated by measuring the speed of wave crests as they propagate across the surfzone and using wave theory to relate speed to depth. Video-based algorithms for depth estimation (e.g. cBATHY, Holman et al., 2018) are publicly available, and under continuous development by many researchers worldwide (e.g., Argus).


 

During the 2019-20 SCaRP (Storm CoAstal Response Project) at Torrey Pines State Beach, surfzone observations were obtained with a LiDAR on a stationary truck overlooking the beach, LiDAR mounted on a drone hovering above the beach, and with swash zone pressure sensors. With small instabilities in the drone data suppressed with a single ground control point, the hovering LiDAR is about as accurate the stationary LiDAR, with errors of a few cm. However, the inversion from the observed crest speed to the desired water depth has sometimes large errors. Ongoing work aims to quantify errors in existing depth-estimation from video methods, and to explore inversions using LiDAR, which provides more information than crest speed. Inversions from LiDAR crest and sea-surface gradient tracking are promising (Figure1).


Image
Figure 1. Subsurface bathymetry estimates from the drone LiDAR (Dec 14, 2019 Hover 4) are compared with ground truth ('jumbo' GPS jetski) surveys 3 days before and 3 days after the LiDAR (before and after surveys typically differ by 20cm, dots). During the hover, the tidal water level (yellow curve) was about 1.6m, and LiDAR depth-inversion estimates extended from the shoreline to about the seaward edge of the surfzone (depth about 2m, incident wave height 1.4m.) Depth inversions are two flavors of crest tracking (green and dark red) and gradient-derived (cyan). Including bore height detectably improves the fit to ground-truth data.