Search for: All records

Vegetation plays a crucial role in coastal dune building. Species‐specific plant characteristics can modulate sediment transport and dune shape, but this factor is absent in most dune building numerical models. Here, we develop a new approach to implement species‐specific vegetation characteristics into a process‐based aeolian sediment transport model. Using a three‐step approach, we incorporated the morphological differences of three dune grass species dominant in the US Pacific Northwest coast (European beachgrassAmmophila arenaria, American beachgrassA. breviligulata, and American dune grassLeymus mollis) into the model AeoLiS. First, we projected the tiller frontal area of each grass species onto a high resolution grid and then re‐scaled the grid to account for the associated vegetation cover for each species. Next, we calibrated the bed shear stress in the numerical model to replicate the actual sand capture efficiency of each species, as measured in a previously published wind tunnel experiment. Simulations were then performed to model sand bedform development within the grass canopies with the same shoot densities for all species and with more realistic average field densities. The species‐specific model shows a significant improvement over the standard model by (a) accurately simulating the sand capture efficiency from the wind tunnel experiment for the grass species and (b) simulating bedform morphology representative of each species' characteristic bedform morphology using realistic field vegetation density. This novel approach to dune modeling will improve spatial and temporal predictions of dune morphologic development and coastal vulnerability under local vegetation conditions and variations in sand delivery. 

Wind flow over coastal foredunes adapts to vegetation, resulting in spatial gradients in bed shear stresses that contribute to the formation of localized bedforms. Under- standing, and having the capability to numerically predict, the distribution of sedi- ment deposited within sparsely vegetated dune complexes is critical for quantifying the ecological, protective, and economic benefits of dune management activities. Data from wind tunnel experiments have indicated that there is a spatial lag from the canopy leading edge to a downwind location where sediment deposition first occurs. The length scale of this deposition lag is further quantified here using new field mea- surements of aeolian sediment transport across sparsely vegetated managed dune systems in Oregon, USA. We develop a deposition lag length scale parameter using both lab and this new field data and then incorporate this parameter into the process-based aeolian sediment transport model, Aeolis, which also includes a new far-field shear stress coupler. Results from numerical simulations suggest that the spatial deposition lag effect is significant for model skill in sparsely vegetated dunes. We observe with field and laboratory observations that, as canopy density increases, the length of the deposition lag decreases. As such, within the model framework the implementation of the deposition lag length does not affect the results of models of coastal dune geomorphological evolution within higher density canopies. Dune can- opy density can vary due to natural (e.g., storm overwash, burial, die-off) or anthro- pogenic (e.g., managed plantings, dune grading) processes. 

Coastal landscape change represents aggregated sediment transport gradients from spatially and temporally variable marine and aeolian forces. Numerous tools exist that independently simulate subaqueous and subaerial coastal profile change in response to these physical forces on a range of time scales. In this capacity, coastal foredunes have been treated primarily as wind-driven features. However, there are several marine controls on coastal foredune growth, such as sediment supply and moisture effects on aeolian processes. To improve understanding of interactions across the land-sea interface, here the development of the new Windsurf-coupled numerical modeling framework is presented. Windsurf couples standalone subaqueous and subaerial coastal change models to simulate the co-evolution of the coastal zone in response to both marine and aeolian processes. Windsurf is applied to a progradational, dissipative coastal system in Washington, USA, demonstrating the ability of the model framework to simulate sediment exchanges between the nearshore, beach, and dune for a one-year period. Windsurf simulations generally reproduce observed cycles of seasonal beach progradation and retreat, as well as dune growth, with reasonable skill. Exploratory model simulations are used to further explore the implications of environmental forcing variability on annual-scale coastal profile evolution. The findings of this work support the hypothesis that there are both direct and indirect oceanographic and meteorological controls on coastal foredune progradation, with this new modeling tool providing a new means of exploring complex morphodynamic feedback mechanisms. 

Abstract Classifying images using supervised machine learning (ML) relies on labeled training data—classes or text descriptions, for example, associated with each image. Data‐driven models are only as good as the data used for training, and this points to the importance of high‐quality labeled data for developing a ML model that has predictive skill. Labeling data is typically a time‐consuming, manual process. Here, we investigate the process of labeling data, with a specific focus on coastal aerial imagery captured in the wake of hurricanes that affected the Atlantic and Gulf Coasts of the United States. The imagery data set is a rich observational record of storm impacts and coastal change, but the imagery requires labeling to render that information accessible. We created an online interface that served labelers a stream of images and a fixed set of questions. A total of 1,600 images were labeled by at least two or as many as seven coastal scientists. We used the resulting data set to investigate interrater agreement: the extent to which labelers labeled each image similarly. Interrater agreement scores, assessed with percent agreement and Krippendorff's alpha, are higher when the questions posed to labelers are relatively simple, when the labelers are provided with a user manual, and when images are smaller. Experiments in interrater agreement point toward the benefit of multiple labelers for understanding the uncertainty in labeling data for machine learning research.