Author
Thomas Bolton
Bio: Thomas Bolton is an academic researcher from University of Oxford. The author has contributed to research in topics: Parametrization & Deep learning. The author has an hindex of 4, co-authored 5 publications receiving 235 citations.
Papers
More filters
••
TL;DR: In this paper, a convolutional neural network (CNN) was trained on degraded data from a high-resolution quasi-geostrophic ocean model to predict turbulent processes and subsurface flow fields.
Abstract: Oceanographic observations are limited by sampling rates, while ocean models are limited by finite resolution and high viscosity and diffusion coefficients. Therefore, both data from observations and ocean models lack information at small and fast scales. Methods are needed to either extract information, extrapolate, or upscale existing oceanographic data sets, to account for or represent unresolved physical processes. Here we use machine learning to leverage observations and model data by predicting unresolved turbulent processes and subsurface flow fields. As a proof of concept, we train convolutional neural networks on degraded data from a high-resolution quasi-geostrophic ocean model. We demonstrate that convolutional neural networks successfully replicate the spatiotemporal variability of the subgrid eddy momentum forcing, are capable of generalizing to a range of dynamical behaviors, and can be forced to respect global momentum conservation. The training data of our convolutional neural networks can be subsampled to 10–20% of the original size without a significant decrease in accuracy. We also show that the subsurface flow field can be predicted using only information at the surface (e.g., using only satellite altimetry data). Our results indicate that data-driven approaches can be exploited to predict both subgrid and large-scale processes, while respecting physical principles, even when data are limited to a particular region or external forcing. Our in-depth study presents evidence for the successful design of ocean eddy parameterizations for implementation in coarse-resolution climate models.
236 citations
••
TL;DR: This work demonstrates the usefulness of the RVM algorithm to reveal closed‐form equations for eddy parameterizations with embedded conservation laws and shows the potential for new physics‐aware interpretable ML turbulence parameterizations for use in ocean climate models.
Abstract: The resolution of climate models is limited by computational cost. Therefore, we must rely on parameterizations to represent processes occurring below the scale resolved by the models. Here, we foc...
83 citations
••
TL;DR: In this paper, the authors proposed an eddy parametrization based on a non-Newtonian stress which depends on the partially resolved scales and their variability, and tested two versions of the parameter, one deterministic and one stochastic, at coarse and eddy-permitting resolutions in a double gyre quasi-geostrophic model.
80 citations
••
TL;DR: The best performing submitted models are described and compared to a Bayesian model previously developed by domain experts and an ensemble model is built that outperforms the individual component models in prediction accuracy.
8 citations
••
TL;DR: In this paper, a passive tracer driven by satellite-derived surface velocity fields is used to study the mixing of heat, carbon, and other climatically important tracers in geostrophic eddies.
Abstract: Geostrophic eddies contribute to the mixing of heat, carbon, and other climatically important tracers. A passive tracer driven by satellite-derived surface velocity fields is used to study ...
6 citations
Cited by
More filters
••
TL;DR: In this paper, a convolutional neural network (CNN) was trained on degraded data from a high-resolution quasi-geostrophic ocean model to predict turbulent processes and subsurface flow fields.
Abstract: Oceanographic observations are limited by sampling rates, while ocean models are limited by finite resolution and high viscosity and diffusion coefficients. Therefore, both data from observations and ocean models lack information at small and fast scales. Methods are needed to either extract information, extrapolate, or upscale existing oceanographic data sets, to account for or represent unresolved physical processes. Here we use machine learning to leverage observations and model data by predicting unresolved turbulent processes and subsurface flow fields. As a proof of concept, we train convolutional neural networks on degraded data from a high-resolution quasi-geostrophic ocean model. We demonstrate that convolutional neural networks successfully replicate the spatiotemporal variability of the subgrid eddy momentum forcing, are capable of generalizing to a range of dynamical behaviors, and can be forced to respect global momentum conservation. The training data of our convolutional neural networks can be subsampled to 10–20% of the original size without a significant decrease in accuracy. We also show that the subsurface flow field can be predicted using only information at the surface (e.g., using only satellite altimetry data). Our results indicate that data-driven approaches can be exploited to predict both subgrid and large-scale processes, while respecting physical principles, even when data are limited to a particular region or external forcing. Our in-depth study presents evidence for the successful design of ocean eddy parameterizations for implementation in coarse-resolution climate models.
236 citations
•
01 Jan 2020
TL;DR: An overview of techniques to integrate machine learning with physics-based modeling and classes of methodologies used to construct physics-guided machine learning models and hybrid physics-machine learning frameworks from a machine learning standpoint is provided.
Abstract: In this manuscript, we provide a structured and comprehensive overview of techniques to integrate machine learning with physics-based modeling. First, we provide a summary of application areas for which these approaches have been applied. Then, we describe classes of methodologies used to construct physics-guided machine learning models and hybrid physics-machine learning frameworks from a machine learning standpoint. With this foundation, we then provide a systematic organization of these existing techniques and discuss ideas for future research.
230 citations
••
Commonwealth Scientific and Industrial Research Organisation1, University of Western Australia2, Imperial College London3, Harper Adams University4, Scotland's Rural College5, Bavarian Forest National Park6, University of Würzburg7, York University8, University of New England (Australia)9, University of Sussex10, University of Louisville11, Smithsonian Institution12, University of Leeds13
TL;DR: In this paper, the authors identify seven key challenges in drawing robust inference about insect population declines: establishment of the historical baseline, representativeness of site selection, robustness of time series trend estimation, mitigation of detection bias effects, and ability to account for potential artefacts of density dependence, phenological shifts and scale-dependence in extrapolation from sample abundance to population level inference.
Abstract: 1. Many insect species are under threat from the anthropogenic drivers of global change. There have been numerous well‐documented examples of insect population declines and extinctions in the scientific literature, but recent weaker studies making extreme claims of a global crisis have drawn widespread media coverage and brought unprecedented public attention. This spotlight might be a double‐edged sword if the veracity of alarmist insect decline statements do not stand up to close scrutiny.
2. We identify seven key challenges in drawing robust inference about insect population declines: establishment of the historical baseline, representativeness of site selection, robustness of time series trend estimation, mitigation of detection bias effects, and ability to account for potential artefacts of density dependence, phenological shifts and scale‐dependence in extrapolation from sample abundance to population‐level inference.
3. Insect population fluctuations are complex. Greater care is needed when evaluating evidence for population trends and in identifying drivers of those trends. We present guidelines for best‐practise approaches that avoid methodological errors, mitigate potential biases and produce more robust analyses of time series trends.
4. Despite many existing challenges and pitfalls, we present a forward‐looking prospectus for the future of insect population monitoring, highlighting opportunities for more creative exploitation of existing baseline data, technological advances in sampling and novel computational approaches. Entomologists cannot tackle these challenges alone, and it is only through collaboration with citizen scientists, other research scientists in many disciplines, and data analysts that the next generation of researchers will bridge the gap between little bugs and big data.
220 citations
••
TL;DR: This work introduces a systematic way of enforcing nonlinear analytic constraints in neural networks via constraints in the architecture or the loss function, which reduces errors in the subsets of the outputs most impacted by the constraints.
Abstract: Neural networks can emulate nonlinear physical systems with high accuracy, yet they may produce physically inconsistent results when violating fundamental constraints. Here, we introduce a systematic way of enforcing nonlinear analytic constraints in neural networks via constraints in the architecture or the loss function. Applied to convective processes for climate modeling, architectural constraints enforce conservation laws to within machine precision without degrading performance. Enforcing constraints also reduces errors in the subsets of the outputs most impacted by the constraints.
187 citations
••
TL;DR: It is shown that RC–ESN substantially outperforms ANN and RNN–LSTM for short-term predictions, e.g., accurately forecasting the chaotic trajectories for hundreds of numerical solver's time steps equivalent to several Lyapunov timescales.
Abstract: . In this paper, the performance of three machine-learning methods for predicting short-term
evolution and for reproducing the long-term statistics of a multiscale spatiotemporal Lorenz 96
system is examined. The methods are an echo state network (ESN, which is a type of reservoir computing; hereafter RC–ESN), a deep feed-forward artificial neural network (ANN), and a recurrent neural network (RNN) with long
short-term memory (LSTM; hereafter RNN–LSTM). This Lorenz 96 system has three tiers of nonlinearly interacting
variables representing slow/large-scale ( X ), intermediate ( Y ), and fast/small-scale ( Z )
processes. For training or testing, only X is available; Y and Z are never known or used. We
show that RC–ESN substantially outperforms ANN and RNN–LSTM for short-term predictions, e.g.,
accurately forecasting the chaotic trajectories for hundreds of numerical solver's time steps
equivalent to several Lyapunov timescales. The RNN–LSTM outperforms ANN, and both methods show
some prediction skills too. Furthermore, even after losing the trajectory, data predicted by
RC–ESN and RNN–LSTM have probability density functions (pdf's) that closely match the true pdf – even at the tails. The pdf of the data predicted using ANN, however, deviates from the true
pdf. Implications, caveats, and applications to data-driven and data-assisted surrogate modeling
of complex nonlinear dynamical systems, such as weather and climate, are discussed.
142 citations