Developing robust methodologies for predicting sandy shoreline changes over the next 10 to 50 years is a significant concern for coastal communities worldwide. Improving these shoreline models necessitates the refinement of the wave time series used to drive these models, as well as the accurate reproduction of the complete directional wave spectra, as opposed to relying on classic bulk parameters. The development of high-quality wave time series can be pursued through full dynamical downscaling of waves, establishing statistical relationships between offshore and nearshore waves, or employing hybrid methods. Hybrid models combine the computational efficiency of statistical models with the precision and detail of dynamical downscaling models, resulting in faster and more cost-effective wave predictions. While dynamical downscaling is the conventional approach in small geographical areas, hybrid methods are better suited for regional shoreline analyses and for examining long time series under various climate scenarios. In this study, we propose the utilization of BinWaves, an additive hybrid method, to downscale waves along the coastlines of New South Wales (NSW), Australia.

Satellite-derived shorelines (SDS) have been increasingly adopted in coastal research to analyze the migration of coastlines from regional to global scales (Luijendijk et al., 2018; Mao et al., 2021; Vos et al., 2023). However, most of the existing SDS algorithms rely on optical satellite images, which require cloud-free conditions to have an unobstructed view of the land surface. Compared to optical images, Synthetic Aperture Radar (SAR) images are less interpretable, so they are relatively unpopular for SDS. However, a key benefit of SAR images is that they are not impacted by cloud, which makes them ideal sources of information to reconstruct the cloud-contaminated optical images. Therefore, the fusion of SAR and optical images provides the unique opportunity to increase the data availability for earth observation (Schmitt et al., 2017), especially in cloudy regions. Although a variety of SAR- Optical fusion methods ranging from generic linear regressions to deep learning models have been developed for optical image reconstruction, their applicability to shoreline extraction has not been thoroughly investigated. In this study, we (1) develop a deep internal learning (Ebel et al., 2021; Zhang et al., 2019) model to reconstruct a sequence of cloud- contaminated Modified Normalized Difference Water Index (MNDWI) images with a sequence of SAR images; and (2) evaluate the applicability of the reconstructed MNDWI images for shoreline extraction.

Robust and reliable models are needed to understand how coastlines will evolve over the coming decades, driven by both natural variability and climate change. This study evaluated how accurately five popular ‘reduced-complexity’ models replicate multi-decadal shoreline change at Narrabeen-Collaroy Beach, a sandy embayment in Sydney, Australia. Measured shoreline positions derived from approximately monthly field surveys were used for 20-year calibration and 20-year validation periods. The models performed similarly on average but with large variability between transects. The set-up of several models was modified to compensate for their sensitivity to imperfect input wave data, and further site-specific improvements were identified. Capturing interannual to decadal-scale variability in cross-shore and longshore dynamics at this site was challenging for all five models. Models appeared to aggregate key processes at this timescale into parameter values rather than representing them directly. This suggests time-varying parameters or changes to model structure may be necessary for decadal-scale simulations.

Equilibrium-based models are a transparent method of modelling shoreline change, though often too simplistic to capture complex dynamics. Conversely, deep learning methodologies offer greater predictive power at the expense of transparency. In this research we scrutinize the internal workings of an LSTM shoreline model. A regression-based probe is used to show that cell state vectors, responsible for past-to-future information flow, autonomously generate equilibrium-like information akin to the physics-based equilibrium term of the ShoreFor model, Omega eq. The variation in probe skill throughout training is tracked to show that at 5 of 6 transects, the LSTM was able to meaningfully acquire equilibrium information (Sigma Delta R2 = 0.3-0.6). The results of this work offer evidence that an LSTM may model shoreline change with internal methods that are consistent with the current understanding of coastal shoreline dynamics. These physically meaningful representations emphasize the importance of co-evolution between machine learning and physics-based approaches moving forward.

Narrabeen Beach is 3.6 km-long embayed sandy beach located in southeast Australia on the northern beaches of Sydney. It is well known in the coastal research community for its long-term beach monitoring program, that was commenced in 1976 by Prof. Andy Short and has continued uninterrupted until the present day. This program has led to a number of groundbreaking research advances, including the Wright and Short morphodynamic beach state model, embayed beach rotation and links to climate cycles like the El Niño/Southern Oscillation. This presentation will present a significant extension of this monitoring program through the inclusion of advanced shoreline monitoring techniques. These techniques include: historical aerial photographs, airborne and fixed Lidar, UAV, satellite-derived shorelines (CoastSat), Argus coastal imaging and CoastSnap citizen science. This large dataset (comprising over 1 million data points) enables an unprecedented look at shoreline change in dynamic, wave-dominated environments on time scales from sub-daily to decadal. The talk will showcase new research derived from this dataset, including data-driven forecast models of shoreline erosion and insights on long-term coastal change. Finally, links to accessing this open-source dataset will be described.

The complexity of sandy shoreline dynamics along storm-dominated coastlines is largely driven by the fundamentally distinct processes that govern individual storm events and post-storm recovery periods. Despite advancements in both physics-based and machine learning methods, accurately predicting both the rapid shift in shoreline response due to storms, and the subsequent recovery periods across multi-annual forecasting horizons remains a significant challenge. In this study, we introduce a Mixture of Experts (or 'Mixture') approach to shoreline modelling that augments a Long Short-Term Memory (LSTM) neural network with a specialized linear regression storm model. This state dependent approach, guided by a threshold gating mechanism, generates stable multi-year forecasts that effectively capture both storm impacts and longer-term shoreline trends. We apply the Mixture at two storm-dominated sites along the southeast Australian coastline and observe an improvement in NMSE of 0.26 at Narrabeen and 0.61 at the Gold Coast, relative to a baseline standalone LSTM model. The findings of this work emphasize that context-informed modelling decisions can significantly enhance machine learning methods, leading to more accessible and actionable forecasts while minimizing an increase in model complexity.

The present generation of widely-used dune erosion models (e.g., XBeach, Roelvink et al, 2009) do not explicitly consider geotechnical processes or attempt to model the internal sand-matrix stability of the eroding dune face during storms. However, several previous studies have shown that dunes fail due to a combination of several phenomena including shear, notching, and changing soil saturation (Erikson et al., 2007; Palmsten and Holman, 2010). This paper aims to improve the numerical modelling of shear failures in the absence of notching by including the presence of a phreatic surface and using realistic failure surfaces. Two slope stability methods that are commonly used in the field of geotechnics are tested, called the ‘Ordinary’ and ‘Bishop’ method of slices, respectively. These two so-called ’limit equilibrium’ (LE) methods are evaluated using a subset of the laboratory experimental data reported by Conti et al (2023).

Shorelines can be seen as representations of the constant interaction between hydrodynamic forces from the ocean and available sediment supply. The modelling of shoreline movement over short to medium timescales remains an active area of research owing to the difficulty of accounting for all the different factors influencing these dynamics. These include both cross-shore and alongshore processes acting at a range of timescales from individual waves to multi-decadal trends in waves and water levels. While many models to date have focused on either alongshore or cross-shore processes, the connectivity between adjacent locations requires a more inclusive approach. Along complex coastlines, coastal management requires flexible methodologies for pulling apart the signal considering a range of sources at any given location.

Understanding the drivers, as well as the ability to predict sandy shoreline change, is of primary interest to coastal engineers and managers, alike. As machine learning has increasingly gained prominence in the field of environmental science, there has also been a growing emphasis on the need for models that are more interpretable. Striking the right balance between model performance and interpretability has become increasingly nuanced. After all, establishing trust in a model necessitates a degree of comprehension regarding its underlying decision-making process. One path forward lies in the co-evolution of both methods to arrive at a modelling approach that retains interpretability and also captures the relevant dynamics to make accurate forecasts. In this study we develop several models combining multiple approaches designed to promote interpretability while retaining high predictive performance. Here we hypothesize that a smart mixture gate informed by prior knowledge will allow more flexibility in capturing the different modes of shoreline change which make up the overall response.

Predicting the extent of wave-driven dune erosion under wave impact and elevated water levels will improve our ability to safeguard the livelihood of ecosystems, communities and infrastructure living behind sandy beaches worldwide. However, the geophysical processes leading to time-dependent dune face failures are still not fully understood. Here, physical laboratory experiments are used to inform a coupled groundwater and limit equilibrium slope stability model to explore dune face stability. The model incorporates a spatially time-varying phreatic (or "water table") surface and the associated changes in pore water pressure due to wave runup to highlight three key physical processes leading to geotechnical dune face failure during wave-driven erosion. First, results from numerical modeling indicate that wave runup impacting the dune face has a destabilizing effect due to excess pore water pressure during downrush. Second, dune face instability during saturated pore water conditions occurs due to excess pore water pressure and the lack of apparent cohesion resulting from the super-elevated water table present inside the dune face. Instability is further exacerbated by wave runup reaching the dune and further elevating the phreatic surface. Third, an important feature in the timing and resiliency of the dune face under wave attack is the location and temporal evolution of the slumped sand post a failure event. The unconsolidated slumped sand acts to temporarily protect the base of the dune from direct wave attack until it is eroded away using swash processes.

Runup is the wave-driven component of the total water level and is a critical metric for characterizing exposure to coastal hazards (Sallenger, 2000; Stockdon et al., 2006). The vertical extent of wave runup is typically parameterized as a function of offshore wave characteristics and beach slope (e.g., Stockdon et al., 2006). Recent research has considered the role of nearshore bars in regulating wave breaking and found that the depth of the bar regulates the elevation of runup at the shoreline (e.g., Cohn et al., 2014). However, the relationship between the morphology of the bar and the various hydrodynamic components related to runup (i.e., swash and setup) has not been explored. Further, the relationship between the bar morphology and the style of dune erosion has not been analyzed. In this study we use the numerical model XBeach (XB; Roelvink et al., 2009) to reproduce a flume experiment performed at the University of New South Wales (UNSW) Sydney Water Research Laboratory (Conti et al., 2023). We then synthesize and modify the bar in the profile to explore how the bar crest depth, bar trough depth and bar distance from the shoreline can affect runup and resulting dune erosion. Note that we use the Conti et al. (2023) study to define initial conditions for XBeach and validate the model setup but not to compare results as that study was ultimately focused on the role of moisture content in the dune on erosion.

An increased availability of long-term coastal imaging datasets has opened the door to the use of data-driven modelling approaches to predict shoreline evolution in response to wave and water level conditions. In this study an autoregressive neural network approach has been applied to predict shoreline change over daily to yearly timescales. A dataset comprising two embayed beaches (Narrabeen Beach, Australia and Tairua Beach, New Zealand) has been used, spanning 10 years of daily shoreline position observation at each site. The model shows good cross-validation performance, predicting the shoreline position with an average 4.64 m RMSE (0.78 NMSE) at Tairua and 5.73 m RMSE (0.46 NMSE) at Narrabeen over approximately 2-year test periods.The autoregressive component of the model involved the use of the last predicted shoreline position in the prediction of shoreline change over the next timestep. This “memory” of past conditions was found to be crucial to maintaining model stability and prediction accuracy over timescales of weeks to years. Model outputs were interrogated to show the structure of the equilibrium response to previous shoreline position which was more prevalent at Tairua. The model is quite robust to changes in the quantity and temporal resolution of the training data, though training data of more than 2-years was desirable, particularly at Narrabeen.

As the coastal population continues to expand, the risk of experiencing social and economic losses due to the effects of a changing climate also increases. Although considerable advances have been achieved in terms of developing numerical process-based models for shoreline change, ensuring reliable predictions remains a formidable task. We propose an innovative approach by implementing Convolutional Neural Networks (CNNs) to predict the evolution of shorelines in response to wave forcings. While the use of these data-driven model frameworks in coastal research is still in its early stages, it holds promise in capturing the strong autocorrelation and memory/storage effects involved in shoreline evolution and may therefore offer a viable alternative to process-based models, at least in some locations. Our models were put to the test at two distinct beach locations: Ocean Beach in California, characterized by its seasonal patterns, and Duck in North Carolina, which lacks a strong seasonal signal. We offer an assessment of the models' performance through the use of absolute- value error metrics. The findings presented shed light on the potential of Deep Learning in forecasting shoreline change.

Yes. Equilibrium shoreline models, which simulate wave-driven cross-shore erosion and accretion, are mathematically equivalent to a discrete convolution (i.e., a weighted, moving average) of a time series of wave-forcing conditions with a parameterized memory-decay kernel function. The direct equivalence between equilibrium shoreline models and convolutions reveals key theoretical aspects of equilibrium behavior. Convolutions (representing quasi-low-pass filter operations) provide an intuitive theoretical description of shoreline erosion and accretion behavior in response to waves: that is, shoreline position often mirrors the weighted moving average of wave time series. Model-convolution equivalence also provides a conceptual basis to interpret, evaluate, and construct data-driven Machine-Learning/Deep-Learning (ML/DL) models that use convolutions to extract features from data and then apply them for prediction (e.g., Convolutional Neural Networks (CNNs)). Finally, our findings provide a methodological pathway (based on Fourier transforms) for future understanding of wave-driven shoreline change, which can be used to interpret the coherence between the frequency spectrum of the processes of waves and shoreline change and construct more computationally efficient and effective shoreline-modeling approaches.

Robust predictions of shoreline change are critical for sustainable coastal management. Despite advancements in shoreline models, objective benchmarking remains limited. Here we present results from ShoreShop2.0, an international collaborative benchmarking workshop, where 34 groups submitted shoreline change predictions in a blind competition. Subsets of shoreline observations at an undisclosed site (BeachX) over short (5-year) and medium (50-year) periods were withheld from modelers and used for model benchmarking. Using satellite-derived shoreline datasets for calibration and evaluation, the best performing models achieved prediction accuracies on the order of 10 m, comparable to the accuracy of the satellite shoreline data, indicating that certain beaches can be modelled nearly as well as they can be remotely observed. The outcomes from this collaborative benchmarking competition critically review the present state-of-the-art in shoreline change prediction as well as reveal model limitations, facilitate improvements, and offer insights for advancing shoreline-prediction capabilities.

Understanding and predicting dune erosion is crucial for coastal hazards mitigation, ecosystem preservation, as well as the protection of human settlements and infrastructure along sandy coastlines. However, our knowledge regarding the influence and potential importance of internal sand moisture content on wave-driven dune face erosion processes remains limited. This paper presents the findings from an extensive series of controlled wave flume experiments of eroding, unvegetated dunes, combining a range of wave conditions and water levels. A further variable that is studied directly for the first time is the internal moisture content of the dune. The initial moisture content and its evolving dynamics within the eroding dune face are quantified, revealing this to be a determining factor of the observed erosion rates, the final horizontal erosion distance, and the dune face failure type. Importantly, infiltration into the dune with each wave was not a driving mechanism of observed failures, but rather the rapid increase and then decrease of the phreatic surface within the dune, resulting in excess pore pressures and destabilization of the sediment matrix. Based on these observations two distinct mechanisms of shear failure and resulting dune face slumping are identified for unsaturated dunes corresponding to 'minimum ' and 'field capacity ' internal moisture content: Type 1 is associated with a circular failure surface in the absence of notching at the base of the dune, while for Type 2, notching is present and the failure surface is vertical. Under fully saturated conditions, the phreatic surface is observed to decouple from the wave runup, with excess water continuously exiting the dune face near the base. This results in rapid shear failure with no notching (Type 3). Significantly, dunes with saturated initial pore moisture content above the capillary fringe are observed to have up to 35 % greater erosion potential, consistently receding further landward than the two unsaturated counterparts as well as undergoing slumping in the absence of wave impact. A new conceptual framework is presented, comprising of a three-phase erosion sequence that directly links dune face failure mechanisms to wave runup and prevailing groundwater conditions.

This paper introduces a comprehensive protocol leveraging open-access techniques to create small- to medium-scale 3D representations of the environment by using iPhone and iPad light detection and ranging (LiDAR). The protocol focuses on two capabilities of the iPhone LiDAR. The first capability is 3D modeling: iPhone LiDAR rapidly generates detailed indoor and outdoor 3D models, providing insights into object size, volume and geometry. The second capability is change detection: the 3D models created by the LiDAR sensor can be used for precise measurement of changes over time. Compared to other 3D topographic surveying methods, this method is rapid, high resolution, low cost and easy to use. The protocol outlines iPhone LiDAR scanning practices, model export and change detection. The expected results after executing the protocol are (i) a detailed 3D model of a small- to medium-sized object or area of interest and (ii) a distance point cloud revealing change between two point clouds of the same object or area between different times. The entire protocol can be conducted within 2 h by anyone with an iPhone with the LiDAR sensor and a computer. This protocol empowers scientists, students and community members conducting research with a cheap, easy-to-use method for addressing a range of questions and challenges, thus benefiting experts and the broader community.

Download This Paper Open PDF in Browser Add Paper to My Library Share: Permalink Using these links will ensure access to this page indefinitely Copy URL Copy DOI

Coastal change is a complex combination of multi-scale processes (e.g., wave-driven cross-shore and longshore transport; dune, bluff, and cliff erosion; overwash; fluvial and inlet sediment supply; and sea-level-driven recession). Historical sea-level-driven coastal recession on open ocean coasts is often outpaced by wave-driven change. However, future sea-level-driven coastal recession is expected to increase significantly in tandem with accelerating rates of global sea-level rise. Few models of coastal sediment transport can resolve the multitude of coastal-change processes at a given beach, and fewer still are computationally efficient enough to achieve large-scale, long-term simulations, while accounting for historical behavior and uncertainties in future climate. Here, we show that a scalable, data-assimilated shoreline-change model can achieve realistic simulations of long-term coastal change and uncertainty across large coastal regions. As part of the modeling case study of the U.S. South Atlantic Coast (Miami, Florida to Delaware Bay) presented here, we apply historical, satellite-derived observations of shoreline position combined with daily hindcasted and projected wave and sea-level conditions to estimate long-term coastal change by 2100. We find that 63 to 94% of the shorelines on the U.S. South Atlantic Coast are projected to retreat past the present-day extent of sandy beach under 1.0 to 2.0 m of sea-level rise, respectively, without large-scale interventions.