Enhanced Coastal Shoreline Modeling Using an Ensemble Kalman Filter to Include Nonstationarity in Future Wave Climates

Abstract

A novel approach to improve seasonal to interannual sandy shoreline predictions is presented, whereby model-free parameters can vary in time, adjusting to potential nonstationarity in the underlying model forcing. This is achieved by adopting a suitable data assimilation technique (dual state-parameter ensemble Kalman filter) within the established shoreline evolution model ShoreFor. The method is first tested and evaluated using synthetic scenarios, specifically designed to emulate a broad range of natural sandy shoreline behavior. This approach is then applied to a real-world shoreline data set, revealing that time-varying model-free parameters are linked through physical processes to changing characteristics of the wave forcing. Greater accuracy of shoreline predictions is achieved, compared to existing stationary modeling approaches. It is anticipated that the wider application of this method can improve our understanding and prediction of future beach erosion patterns and trends in a changing wave climate. Plain Language Summary Understanding and predicting future changes along sandy coastlines worldwide is highly relevant for coastal management in the context of climate change. In the future, the changing occurrence of storms-and over longer time scales, rising sea levels-are expected to result in new patterns of shoreline erosion. It is very common for shoreline change models to use past records of measured shorelines and waves to match mathematical equations to these existing observations. However, the validity of these types of shoreline models to predict the future is questionable, when waves and storm patterns around the world in coming decades are expected to be different to those observed in the past. A new methodology is presented to address this issue by exploring how a mathematical shoreline model can self-adjust to wave climates that vary through time. The proposed methodology is shown to be successful at improving shoreline predictions.

Uncertainties in Shoreline Projections to 2100 at Truc Vert Beach (France): Role of Sea‐Level Rise and Equilibrium Model Assumptions

“…This was also observed in previous studies, which showed that changes in wave regime can alter the model parameters and the functional relations between them (Ibaceta et al, 2020;Splinter et al, 2017). As a perspective of future work, one way to reduce the effects of model free parameters' uncertainties on modeled shoreline may be to employ non-stationary parameters that can adjust to changes in wave-climate regimes (Ibaceta et al, 2020). The use of non-stationary parameters would also imply a dynamic value of the r parameter, reducing uncertainties associated to the assumption of a linear relationship, between SF's response rate parameters.…”

Anthropogenic Impact on Beach Heterogeneity within a Littoral Cell (Northern Tuscany, Italy)

“…More recently, aerial and camera photogrammetry and drone-derived datasets have been widely used e.g., [26][27][28][29]. Many approaches use modelling as a coastal monitoring tool e.g., [30,31]. Models can also be applied to evaluate areas prone to risk related to coastal processes e.g., [32,33].…”

Modelling cross-shore shoreline change on multiple timescales and their interactions

“…However, due to this single memory decay factor, model skill deteriorates considerably if multiple dominant forcing and beach response timescales are present (Vitousek et al, 2017;Almar et al, 2017;. Recent work by Splinter et al (2017) and Ibaceta et al (2020) also showed that timescales of beach change and forcing may be temporally dependent, with beaches undergoing rapid adjustment to the changes in the dominant wave forcing over time and where single memory decay models can fail to capture the observed shoreline signal.…”