Remotely sensed monitoring of coastal geomorphology: coupling satellite-derived vegetation edges with other proxy metrics

Building an understanding of how coastal processes interact is key for modelling future change, particularly in a rapidly changing climate with rising sea levels. Coastal change indicators can offer a wealth of information on past and potential future shoreface trends but have been traditionally difficult and costly for users to acquire. Novel, open-source, coastal monitoring tools using satellite imagery now present an exciting opportunity for generating diverse time series to fuel coastal digital twins, with previously unseen frequency. Using the novel Python toolkits VedgeSat and CoastSat powered by cloud-based data processing, neural network pixel classifiers, and dynamic thresholding, we automatically extracted 467,142 measurements of coastal vegetation edge and instantaneous waterline positions from Sentinel-2 imagery. Near-weekly changes across almost a decade and storm response and seasonal variability are quantified for waterlines and (for the first time) vegetation edges along ~1,400 cross-shore transects at an environmentally varied Scottish coast, revealing strong seasonal signals where vegetation accraces. We define here the vegetation transition zone, which represents the overlapping region between vegetation and bare sediment pixels and is narrower across steeper dune slopes and eroding vegetation edges and waterlines. Copernicus Marine Service wave hindcasts and public topography data were also incorporated to allow rapid investigations of hydrodynamic forcing on vegetation and sediment, confirming a satellite-derived link between sediment transport and vegetation. The value of remote sensing for gathering large and varied coastal datasets near-global scales is demonstrated with a multivariate analysis, which is scalable and applicable to other complex environmental systems via the open-source toolkit CoasTrack. These relationships fueled by Big Data can inform environmental model domains for real-time coastal change predictions.