Radon-Augmented Sentinel-2 Satellite Imagery to Derive Wave-Patterns and Regional Bathymetry

Abstract

Climatological changes occur globally but have local impacts. Increased storminess, sea level rise and more powerful waves are expected to batter the coastal zone more often and more intense. To understand climate change impacts, regional bathymetry information is paramount. A major issue is that the bathymetries are often non-existent or if they do exist, outdated. This sparsity can be overcome by space-borne satellite techniques to derive bathymetry. Sentinel-2 optical imagery is collected continuously and has a revisit-time around a few days depending on the orbital-position around the world. In this work, Sentinel-2 imagery derived wave patterns are extracted using a localized radon transform. A discrete fast-Fourier (DFT) procedure per direction in Radon space (sinogram) is then applied to derive wave spectra. Sentinel-2 time-lag between detector bands is employed to compute the spectral wave-phase shift and depth using the gravity wave linear dispersion. With this novel technique, regional bathymetries are derived at the test-site of Capbreton, France with an root mean squared (RMS)-error of 2.58 m and a correlation coefficient of 0.82 when compared to the survey for depths until 30 m. With the proposed method, the 10 m Sentinel-2 resolution is sufficient to adequately estimate bathymetries for a wave period of 6.5 s or greater. For shorter periods, the pixel resolution does not allow to detect a stable celerity. In addition to the wave-signature enhancement, the capability of the Radon Transform to augment Sentinel-2 20 m resolution imagery to 10 m is demonstrated, increasing the number of suitable bands for the depth inversion.

The Contribution of Wind-Generated Waves to Coastal Sea-Level Changes

“…Although constant efforts by the research community have been made toward either simple or complex paradigms of coastal evolution (e.g, static retreat by Bruun, 1962;Hanson et al 2010), the coastal response to perpetually changing ocean forcing conditions is still unclear (Cooper and Pilkey, 2004;Ranasinghe et al 2016). Recently, several methods using satellite imagery have been proven reliable (Poupardin et al 2016;Bergsma et al 2019;Raucoules et al 2019) in estimating large-sale coastal bathymetry at unprecedented resolution. One prominent approach is the use of multispectral imagery in estimating shallow bathymetry, until 20 m depth approximately in the case of clear waters.…”

Monitoring Beach Topography and Nearshore Bathymetry Using Spaceborne Remote Sensing: A Review

“…In recent years, robust depth estimation methods through wave celerity inversion have been developed for shore-based video systems and drones [92,100,102,103], optimizing the information gathered in space and time (generally 10-20 min). At present, there are some pioneering works on video from space to recover coastal bathymetry [104][105][106] that clearly show the potential to derive waves from a sequence of images.…”

“…Nevertheless, the method represents a real opportunity to estimate bathymetry where no other more conventional methods could be applied, while further developments are required to improve the accuracy of the method. In [100], a Radon transform-based wave-pattern extraction and depth inversion method is presented. Wave patterns are extracted by applying a Radon transform and subsequent angle filtering to Sentinel-2 imagery, such that the wave signal contains the most-dominant wave directions.…”

“…These results are beyond the expectations, considering the challenging environment, including a deep-water canyon and its effect on surrounding wave patterns. In addition to the wave pattern In [100], a Radon transform-based wave-pattern extraction and depth inversion method is presented. Wave patterns are extracted by applying a Radon transform and subsequent angle filtering to Sentinel-2 imagery, such that the wave signal contains the most-dominant wave directions.…”

Earth Observations for Monitoring Marine Coastal Hazards and Their Drivers

“…This allows depths of up to 40-50 m to be resolved. Bathymetric inversion code developed from the wave dispersion relationship in intermediate to shallow water make use of the temporal information contained in the spatial images and were applied to different satellite observations: IKONOS (Abileah 2006), WorldView-2 (McCarthy 2010), SPOT5/6 (Poupardin et al 2016), Sentinel-2 (Bergsma et al 2019) and Pleiades (Danilo and Binet 2013;Almar et al 2019). One of the great advantages of this approach is that it is autonomous and does not require additional wave information from observation or models besides those acquired by means of satellite remote sensing.…”