This article proposes a new methodological approach to measure and map the size of coarse clasts on a land surface from photographs. This method is based on the use of the Mask Regional Convolutional Neural Network (R-CNN) deep learning algorithm, which allows the instance segmentation of objects after an initial training on manually labeled data. The algorithm is capable of identifying and classifying objects present in an image at the pixel scale, without human intervention, in a matter of seconds. This work demonstrates that it is possible to train the model to detect non-overlapping coarse sediments on scaled images, in order to extract their individual size and morphological characteristics with high efficiency (R² = 0.98; Root Mean Square Error (RMSE) = 3.9 mm). It is then possible to measure element size profiles over a sedimentary body, as it was done on the pebble beach of Etretat (Normandy, France) in order to monitor the granulometric spatial variability before and after a storm. Applied at a larger scale using Unmanned Aerial Vehicle (UAV) derived ortho-images, the method allows the accurate characterization and high-resolution mapping of the surface coarse sediment size, as it was performed on the two pebble beaches of Etretat (D50 = 5.99 cm) and Hautot-sur-Mer (D50 = 7.44 cm) (Normandy, France). Validation results show a very satisfying overall representativity (R² = 0.45 and 0.75; RMSE = 6.8 mm and 9.3 mm at Etretat and Hautot-sur-Mer, respectively), while the method remains fast, easy to apply and low-cost, although the method remains limited by the image resolution (objects need to be longer than 4 cm), and could still be improved in several ways, for instance by adding more manually labeled data to the training dataset, and by considering more accurate methods than the ellipse fitting for measuring the particle sizes.

This research was carried out in the framework of the Surface Water and Ocean Topography (SWOT) program of the French National Centre of Space Studies (CNES). In the context of global climate change, increases in frequency and intensity of extreme events are considered potential threats. Hence, many efforts have been devoted to acquiring a better understanding of the natural processes driving the hydrodynamics’ multiscale variability in order to produce a more accurate estimation of their fluctuations. In this situation, and in the framework of Normandy’s coastal risk management, the dynamics of extreme events were investigated from regional scales (the English Channel basin) to local scales (the Normandy beaches in Etretat and Hautot-sur-Mer). Offshore wave data were obtained and hourly wave data were simulated for two years using the SWAN (Simulating WAves till Nearshore) model, including ten highly energetic stormy events selected by their different physical characteristics. Spectral analysis was conducted to compare the transformation in frequencies of the wave spectra from the Channel to shallow waters. A special focus was given to the hydrodynamics close to Normandy during these extreme events depending on their duration, their strength, and the persistence of the maximum wave height.

This study delves into the morphodynamic changes of pebble beaches in response to storm events, employing a combination of observational and numerical approaches. This research focuses on three extreme events, meticulously examining morhological changes in intertidal topography on the beach of Etretat (Normandy, France). A robust dataset of daily beach topography, derived from video monitoring systems, validates a set of numerical simulations of cross-shore dynamics performed by the process-based model XBeach-G. Our study evaluates the model’s efficacy in estimating beach profile evolution under high-energy conditions and explores its sensitivity to the physical properties of pebbles, including permeability. The results underscore the significance of considering spatial and temporal variations in permeability during storms to enhance the numerical model’s accuracy in predicting pebble beach dynamics. Furthermore, this study advocates for the incorporation of grain size mapping techniques to refine numerical model implementations.