[1] A model for wave and wind stress prediction is constructed. The source functions that drive the space-time evolution of the energy spectra are developed in form based on theory and laboratory and field experiments. The calibration factors (proportionality constants of the source functions) are determined from a comparison of modeled and observed significant height and mean period. The observations are for the month of January 2005 and are derived from an array of laser range finders mounted on a bridge between two platforms in the Ekofisk oil field in the North Sea. The model calculates the form stress on the waves and adds it vectorially to the sheltering-modified skin stress. The resulting drag coefficient versus wind speed is shown to have the observed structure: low in light winds, increasing in moderate winds, and increasing more slowly in very strong winds. Modeled spectral shapes in the four quadrants of Hurricane Bonnie (1998) match the Scanning Radar Altimeter measurements. Modeled spectral properties in Hurricane Ike (2008) are compared against NDBC buoy estimates with good results. Drag coefficients in the mixed seas produced by hurricanes show dependence on wave age of the wind sea, swell propagation direction, and water depth. The need for wave and stress modeling for atmosphere-ocean coupling is emphasized. The new wave model has all the necessary attributes to be the basis for such a coupler.
Implementing artificial neural networks is commonly achieved via high-level programming languages such as Python and easy-to-use deep learning libraries such as Keras. These software libraries come preloaded with a variety of network architectures, provide autodifferentiation, and support GPUs for fast and efficient computation. As a result, a deep learning practitioner will favor training a neural network model in Python, where these tools are readily available. However, many large-scale scientific computation projects are written in Fortran, making it difficult to integrate with modern deep learning methods. To alleviate this problem, we introduce a software library, the Fortran-Keras Bridge (FKB). This two-way bridge connects environments where deep learning resources are plentiful with those where they are scarce. The paper describes several unique features offered by FKB, such as customizable layers, loss functions, and network ensembles. The paper concludes with a case study that applies FKB to address open questions about the robustness of an experimental approach to global climate simulation, in which subgrid physics are outsourced to deep neural network emulators. In this context, FKB enables a hyperparameter search of one hundred plus candidate models of subgrid cloud and radiation physics, initially implemented in Keras, to be transferred and used in Fortran. Such a process allows the model’s emergent behavior to be assessed, i.e., when fit imperfections are coupled to explicit planetary-scale fluid dynamics. The results reveal a previously unrecognized strong relationship between offline validation error and online performance, in which the choice of the optimizer proves unexpectedly critical. This in turn reveals many new neural network architectures that produce considerable improvements in climate model stability including some with reduced error, for an especially challenging training dataset.
Hurricane Isaac induced large surface waves and a significant change in upper ocean circulation in the Gulf of Mexico before making landfall at the Louisiana coast on 29 August 2012. Isaac was observed by 194 surface drifters during the Grand Lagrangian Deployment (GLAD). A coupled atmosphere‐wave‐ocean model was used to forecast hurricane impacts during GLAD. The coupled model and drifter observations provide an unprecedented opportunity to study the impacts of hurricane‐induced Stokes drift on ocean surface currents. The Stokes drift induced a cyclonic (anticyclonic) rotational flow on the left (right) side of the hurricane and accounted for up to 20% of the average Lagrangian velocity. In a significant deviation from drifter measurements prior to Isaac, the scale‐dependent relative diffusivity is estimated to be 6 times larger during the hurricane, which represents a deviation from Okubo's (1971) canonical results for lateral dispersion in nonhurricane conditions at the ocean surface.
Forecasting hurricane impacts of extreme winds and flooding requires accurate prediction of hurricane structure and storm-induced ocean surface waves days in advance. The waves are complex, especially near landfall when the hurricane winds and water depth varies significantly and the surface waves refract, shoal and dissipate. In this study, we examine the spatial structure, magnitude, and directional spectrum of hurricane-induced ocean waves using a high resolution, fully coupled atmosphere-wave-ocean model and observations. The coupled model predictions of ocean surface waves in Hurricane Ike (2008) over the Gulf of Mexico and Superstorm Sandy (2012) in the northeastern Atlantic and coastal region are evaluated with the NDBC buoy and satellite altimeter observations. Although there are characteristics that are general to ocean waves in both hurricanes as documented in previous studies, wave fields in Ike and Sandy possess unique properties due mostly to the distinct wind fields and coastal bathymetry in the two storms. Several processes are found to significantly modulate hurricane surface waves near landfall. First, the phase speed and group velocities decrease as the waves become shorter and steeper in shallow water, effectively increasing surface roughness and wind stress. Second, the bottom-induced refraction acts to turn the waves toward the coast, increasing the misalignment between the wind and waves. Third, as the hurricane translates over land, the left side of the storm center is characterized by offshore winds over very short fetch, which opposes incoming swell. Landfalling hurricanes produce broader wave spectra overall than that of the open ocean. The front-left quadrant is most complex, where the combination of windsea, swell propagating against the wind, increasing wind-wave stress, and interaction with the coastal topography requires a fully coupled model to meet these challenges in hurricane wave and surge prediction.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.