Pushing the frontiers in climate modelling and analysis with machine learning

Nature Climate Change, Journal Year: 2024, Volume and Issue: 14(9), P. 916 - 928

Published: Aug. 23, 2024

Language: Английский

---

Closing the Loops on Southern Ocean Dynamics: From the Circumpolar Current to Ice Shelves and From Bottom Mixing to Surface Waves

Reviews of Geophysics, Journal Year: 2024, Volume and Issue: 62(3)

Published: July 30, 2024

Abstract A holistic review is given of the Southern Ocean dynamic system, in context crucial role it plays global climate and profound changes experiencing. The focuses on connections between different components drawing together contemporary perspectives from research communities, with objective closing loops our understanding complex network feedbacks overall system. targeted at researchers physical science ambition broadening their knowledge beyond specific field, aims facilitating better‐informed interdisciplinary collaborations. For purposes this review, system divided into four main components: large‐scale circulation; cryosphere; turbulence; gravity waves. Overviews are key dynamical phenomena for each component, before describing linkages components. reviews complemented by an overview observed trends future projections. Priority areas identified to close remaining

Language: Английский

---

Turbulence Closure With Small, Local Neural Networks: Forced Two‐Dimensional and β‐Plane Flows

Journal of Advances in Modeling Earth Systems, Journal Year: 2024, Volume and Issue: 16(4)

Published: April 1, 2024

Abstract We parameterize sub‐grid scale (SGS) fluxes in sinusoidally forced two‐dimensional turbulence on the β ‐plane at high Reynolds numbers (Re ∼25,000) using simple 2‐layer convolutional neural networks (CNN) having only O(1000) parameters, two orders of magnitude smaller than recent studies employing deeper CNNs with 8–10 layers; we obtain stable, accurate, and long‐term online or a posteriori solutions 16× downscaling factors. Our methodology significantly improves training efficiency speed large eddy simulations runs, while offering insights into physics closure such turbulent flows. approach benefits from extensive hyperparameter searching learning rate weight decay coefficient space, as well use cyclical annealing, which leads to more robust accurate compared fixed rates. either coarse velocity vorticity strain fields inputs, output components deviatoric stress tensor, S d . minimize loss between SGS flux divergence (computed high‐resolution solver) that obtained CNN‐modeled , without requiring energy enstrophy preserving constraints. The success shallow accurately parameterizing this class flows implies stresses have weak non‐local dependence fields; it also aligns our physical conception small‐scales are locally controlled by larger scales vortices their strained filaments. Furthermore, CNN‐parameterizations likely be interpretable.

Language: Английский

---

Parameterizing Vertical Mixing Coefficients in the Ocean Surface Boundary Layer Using Neural Networks

Journal of Advances in Modeling Earth Systems, Journal Year: 2023, Volume and Issue: 15(10)

Published: Oct. 1, 2023

Abstract Vertical mixing parameterizations in ocean models are formulated on the basis of physical principles that govern turbulent mixing. However, many include ad hoc components not well constrained by theory or data. One such component is eddy diffusivity model, where vertical fluxes a quantity parameterized from variable diffusion coefficient and mean gradient quantity. In this work, we improve parameterization surface boundary layer enhancing its model using data‐driven methods, specifically neural networks. The networks designed to take extrinsic intrinsic forcing parameters as input predict profile trained output data second moment closure scheme. modified scheme predicts through online inference maintains conservation standard equations, which particularly important for targeted use climate simulations. We describe development stable implementation an general circulation demonstrate enhanced outperforms predecessor reducing biases mixed‐layer depth upper stratification. Our results potential physics‐aware global models.

Language: Английский

---

Mesoscale Eddy‐Induced Sharpening of Oceanic Tracer Fronts

Journal of Advances in Modeling Earth Systems, Journal Year: 2025, Volume and Issue: 17(3)

Published: March 1, 2025

Abstract Oceanic fronts are ubiquitous and important features that form evolve due to multiscale oceanic atmospheric processes. Large‐scale temperature tracer fronts, such as those found along the eastward extensions of Gulf Stream Kuroshio currents, crucial components regional ocean environment climate. This numerical study examines relative importance large‐scale currents mesoscale (“eddies”) in front formation evolution. Using an idealized model double‐gyre system on both eddy‐resolving coarse‐resolution grids, we demonstrate effect eddies is sharpen front, whereas current counteracts this acts create a broader front. The eddy‐driven frontogenesis further described terms recently proposed framework generalized eddy‐induced advection, which represents all eddy effects tracers not mass fluxes traditionally parameterized by isopycnal diffusion. In advection formulated using effective velocity (EEIV), speed at move contours. advantage formulation frontal sharpening can be readily reproduced EEIVs. A functional EEIV variables effectively simulation. shows promise for advective parameterize models eddy‐resolving.

Language: Английский

---

Application of machine learning and convex limiting to subgrid flux modeling in the shallow-water equations

Mathematics and Computers in Simulation, Journal Year: 2025, Volume and Issue: unknown

Published: May 1, 2025

Language: Английский

---

Machine Learning for Online Sea Ice Bias Correction Within Global Ice‐Ocean Simulations

Geophysical Research Letters, Journal Year: 2024, Volume and Issue: 51(3)

Published: Jan. 30, 2024

Abstract In this study, we perform online sea ice bias correction within a Geophysical Fluid Dynamics Laboratory global ice‐ocean model. For this, use convolutional neural network (CNN) which was developed in previous study (Gregory et al., 2023, https://doi.org/10.1029/2023ms003757 ) for the purpose of predicting concentration (SIC) data assimilation (DA) increments. An initial implementation CNN shows systematic improvements SIC biases relative to free‐running model, however large summertime errors remain. We show that these residual can be significantly improved with novel augmentation approach. This approach applies sequential and DA corrections new simulation over training period, then provides set refine weights network. propose machine‐learned scheme could utilized generating conditions, also real‐time seasonal‐to‐subseasonal forecasts.

Language: Английский

---

Bringing it all together: Science and modelling priorities to support international climate policy

Published: Feb. 19, 2024

Abstract. We review how the international modelling community, encompassing Integrated Assessment models, global and regional Earth system climate impact have worked together over past few decades, to advance understanding of change its impacts on society environment, support policy. then recommend a number priority research areas for coming ~6 years (i.e. until ~2030), timescale that matches newly starting activities encompasses IPCC 7th Report (AR7) 2nd UNFCCC Global Stocktake. Progress in these will significantly our increase quality utility science emphasize need continued improvement of, ability simulate, coupled change. There is an urgent investigate plausible pathways emission scenarios realize Paris Climate Targets, including overshoot 1.5 °C 2 targets, before later returning them. System models (ESMs) be capable thoroughly assessing such warming overshoots, particular, efficacy negative CO2 actions reducing atmospheric driving cooling. An improved assessment long-term consequences stabilizing at or above pre-industrial temperatures also required. ESMs run CO2-emission mode, more fully represent - carbon cycle feedbacks. Regional downscaling should use forcing data from simulations, so projections are as realistic possible. accurate simulation observed record remains key requirement does metrics, Effective Sensitivity. For adaptation, guidance potential changes extremes modes variability develop in, demand. Such improvements most likely realized through combination increased model resolution parameterizations. propose deeper collaboration across efforts targeting process realism coupling, enhanced resolution, parameterization improvement, data-driven Machine Learning methods. With respect sampling future uncertainty, between approaches large ensembles those focussed statistical emulation attention paid High Impact Low Likelihood (HILL) outcomes. In risk exceeding critical tipping points during overshoot. comprehensive change, arising directly specific mitigation actions, it important detailed, disaggregated information Models (IAMs) used generate available models. Conversely, methods developed incorporate societal responses into scenario development. Finally, new data, scientific advances, proposed this article not possible without development maintenance robust, globally connected infrastructure ecosystem. This must easily accessible useable all communities world, allowing community engaged developing delivering knowledge

Language: Английский

---

Closing the loops on Southern Ocean dynamics: From the circumpolar current to ice shelves and from bottom mixing to surface waves

Authorea (Authorea), Journal Year: 2023, Volume and Issue: unknown

Published: July 8, 2023

A holistic review is given of the Southern Ocean dynamic system, in context crucial role it plays global climate and profound changes experiencing. The focuses on connections between different components drawing together contemporary perspectives from research communities, with objective 'closing loops' our understanding complex network feedbacks overall system. For purposes this review, system divided into four main components: large-scale circulation; cryosphere; turbulence; gravity waves. Overviews are key dynamical phenomena for each component, before describing linkages components. reviews complemented by an overview observed trends future projections. Priority areas required to improve identified.

Language: Английский

---

Data Imbalance, Uncertainty Quantification, and Transfer Learning in Data‐Driven Parameterizations: Lessons From the Emulation of Gravity Wave Momentum Transport in WACCM

Journal of Advances in Modeling Earth Systems, Journal Year: 2024, Volume and Issue: 16(7)

Published: July 1, 2024

Abstract Neural networks (NNs) are increasingly used for data‐driven subgrid‐scale parameterizations in weather and climate models. While NNs powerful tools learning complex non‐linear relationships from data, there several challenges using them parameterizations. Three of these (a) data imbalance related to rare, often large‐amplitude, samples; (b) uncertainty quantification (UQ) the predictions provide an accuracy indicator; (c) generalization other climates, example, those with different radiative forcings. Here, we examine performance methods addressing NN‐based emulators Whole Atmosphere Community Climate Model (WACCM) physics‐based gravity wave (GW) as a test case. WACCM has complex, state‐of‐the‐art orography‐, convection‐, front‐driven GWs. Convection‐ orography‐driven GWs have significant due absence convection or orography most grid points. We address resampling and/or weighted loss functions, enabling successful emulation all three sources. demonstrate that UQ (Bayesian NNs, variational auto‐encoders, dropouts) ensemble spreads correspond during testing, offering criteria identifying when NN gives inaccurate predictions. Finally, show decreases warmer (4 × CO 2 ). However, their is significantly improved by applying transfer learning, re‐training only one layer ∼1% new climate. The findings this study offer insights developing reliable generalizable various processes, including (but not limited to)

Language: Английский

---