Geology topics

Nicholas Cohn

Publications and source records attributed to Nicholas Cohn.

4 recordsLinked to original sources

USGS2021

Labeling poststorm coastal imagery for machine learning: Measurement of interrater agreement

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.

Earth and Space Science↗

USGS2021

Hotspot dune erosion on an intermediate beach

A large, low pressure Nor’easter storm and Hurricane Joaquin contributed to multiple weeks of sustained, elevated wave and water level conditions along the southeastern Atlantic coast of the United States in Fall 2015. Sea level anomalies in excess of 1 m and offshore wave heights of up to 4 m were recorded during these storms, as observed at the U.S. Army Corps of Engineers’ Field Research Facility in Duck, NC, USA. In response to these energetic oceanographic conditions, there were highly variable morphologic changes to the dune over short spatial scales (<km) which included a range of responses from vertical dune scarping to no measureable response. The portion of the study area with the largest dune erosion occurred at a location fronted by an abnormally deep nearshore bathymetric feature, which altered surf-zone waves and hydrodynamics. The pre-storm beach and dune topography also varied throughout the study area, additionally influencing the frequency of dune collision and contributing to the spatially variable erosion patterns. This work uses field datasets and numerical modeling tools to investigate the causation of hotspot dune erosion at the Field Research Facility. Three different numerical models were tested against the available data in order to assess model skill at resolving complex spatial dune erosion patterns. The three models successfully reproduce the general spatial trends in alongshore variable responses, although not necessarily the details of profile response or net erosion magnitude. Analysis of the model outputs, in conjunction with the available field data, suggests that the observed hotspot dune erosion is related to a complex combination of both topographic and bathymetric controls on the processes driving dune erosion. Therefore, the most simplistic model tested, which only accounts for alongshore variations in topographic profile details, can only predict hotspot dune erosion in locations where steep beach and/or dune topography is the primary control on collisional dune impacts. The higher fidelity models , which account for feedback effects from subaqueous morphology, are similarly able to predict the locations of maximum hotspot erosion, but are sensitive to beach over-steepening and/or errors in wave runup calculations that can lead to over-prediction of simulated dune erosion. This work highlights that numerous existing tools are capable of identifying the foredune regions at most risk from hotspot erosion, as well as the need for continued research to improve representation of all relevant intra-storm morphodynamic processes.

Coastal Engineering↗

USGS2017

Extreme oceanographic forcing and coastal response due to the 2015–2016 El Niño

The El Niño-Southern Oscillation is the dominant mode of interannual climate variability across the Pacific Ocean basin, with influence on the global climate. The two end members of the cycle, El Niño and La Niña, force anomalous oceanographic conditions and coastal response along the Pacific margin, exposing many heavily populated regions to increased coastal flooding and erosion hazards. However, a quantitative record of coastal impacts is spatially limited and temporally restricted to only the most recent events. Here we report on the oceanographic forcing and coastal response of the 2015–2016 El Niño, one of the strongest of the last 145 years. We show that winter wave energy equalled or exceeded measured historical maxima across the US West Coast, corresponding to anomalously large beach erosion across the region. Shorelines in many areas retreated beyond previously measured landward extremes, particularly along the sediment-starved California coast.

Pacific coast↗

USGS2016

Morphodynamics of prograding beaches: A synthesis of seasonal- to century-scale observations of the Columbia River littoral cell

Findings from nearly two decades of research focused on the Columbia River littoral cell (CRLC), a set of rapidly prograding coastal barriers and strand-plains in the U.S. Pacific Northwest, are synthesized to investigate the morphodynamics associated with prograding beaches. Due to a large sediment supply from the Columbia River, the CRLC is the only extensive stretch of shoreline on the U.S. west coast to have advanced significantly seaward during the late Holocene. Since the last Cascadia Subduction Zone (CSZ) earthquake in 1700, with associated co-seismic subsidence and tsunami, much of the CRLC has prograded hundreds of meters. However, the rates of progradation, and the processes most responsible for sediment accumulation, vary depending on time scale and the morphological unit in question. Remarkably, the 20th and early 21st century shoreline change rates were more than double the late prehistoric rates that include recovery from the last major CSZ event, most likely due to an increase in sediment supply resulting from inlet jetty construction. In some locations detailed beach morphology monitoring reveals that at interannual- to decadal-scale the upper shoreface aggraded about 2 cm/yr, subtidal sandbars migrated offshore and decayed while intertidal bars migrated onshore and welded to the shoreline, the shoreline prograded about 4 m/yr, and 1 to 2 new foredune ridges were generated. A detailed meso-scale sediment budget analysis in one location within the littoral cell shows that approximately 100 m 3 /m/yr accumulated between − 12 m (seaward limit of data) and + 9 m (crest of landward-most foredune). Gradients in alongshore sediment transport, net onshore-directed cross-shore sediment transport within the surf zone, and cross-shore feeding from a shoreface out of equilibrium with forcing conditions are each partially responsible for the significant rates of sediment supplied to the beaches and dunes of the CRLC during the observational period. Direct observations of beach progradation at seasonal- to decadal-scale are put in context of measured or inferred changes over time scales of decades to centuries.

Oregon, Washington↗