JAJack Atkinson

Jack Atkinson

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Jack Atkinson has a total of 24 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    TCTrack: Facilitating FAIR tropical cyclone tracking software and data

    by Jack Atkinson, Sam Avis, Alison Ming, Charles Powell

    Cyclone tracks are an important diagnostic in model evaluation when looking at long-term historical statistics and are also of interest in future scenario projections. In addition to modelling, generating tracks from observational or reanalysis data is important in aiding comparisons and assessing trends.Multiple approaches to generating cyclone tracks from model and observational data exist, each with their own opinion as to which variables are important, how candidate storms are identified, and how these candidates are stitched together to form tracks. Each method comes in its own codebase with varying degrees of user documentation meaning that those interested in tracking often select just one approach and stick to it. Further, all tracking codes produce different, often customised, output formats with minimal metadata that do not come close to meeting FAIR data standards.This makes any downstream data use or tracking intercomparison a challenge.We present TCTrack, an open-source Python-based package that provides a common interface to popular tropical cyclone tracking codes. Extensive documentation ensures accessible usage and manipulation of existing codes, as well as providing guidance on input requirements and preprocessing of data. Perhaps most notable is that all output data provided by TCTrack is in a common data format, regardless of the tracking algorithm used, and conforms to the Climate and Forecast (CF) metadata conventions (specifically H4:Trajectory data) preserving variable metadata information from the input files. This metadata-rich output aids in reproducibility and reusability and makes downstream analysis with a variety of other tools straightforward. Finally, TCTrack is built on an easily extendable framework meaning that addition of other tracking approaches from, and for use by, the community is both straightforward and encouraged.This poster showcases key aspects of the TCTrack software, a discussion of tracking data format using the CF-Conventions, and some results from deployment in an intercomparison study of different tracking methods applied to CMIP data.References:Atkinson, J.W. & Avis, S.J. (2025). TCTrack. https://github.com/Cambridge-ICCS/TCTrackEaton, B., et al. (2025). NetCDF Climate and Forecast (CF) Metadata Conventions v1.13. https://cfconventions.org/Hodges, K., Cobb, A., & Vidale, PL. (2017). How Well Are Tropical Cyclones Represented in Reanalysis Datasets? Journal of Climate 30, 14: 5243-5264Ullrich, P.A., et al. (2021) TempestExtremes v2.1: A Community Framework for Feature Detection, Tracking, and Analysis in Large Datasets. Geoscientific Model Development 14, no. 8: 5023–48.Vitart, F., & Stockdale T.N. (2001) Seasonal Forecasting of Tropical Storms Using Coupled GCM Integrations. Monthly Weather Review 129, 10: 2521-2537

  • Open Access

    Coupling a new convection parameterisation trained using high-resolution simulations to the Community Atmospheric Model

    by Jack Atkinson, Paul O'GOrman, Judith Berner, Marion Weinzierl

    A commonly observed issue in general circulation models is biases in the frequency distribution of precipitation, including too much weak rain (the drizzle problem) and either too much or too little heavy precipitation. High resolution models perform better on this front, but are restricted in the spatial and temporal scales they can simulate. Previous work (Yuval, O'Gorman, Hill (2021)) demonstrated that training a neural network parameterisation on high-resolution convection-resolving simulations and deploying it within the same model running at lower horizontal resolution can maintain a good representation of precipitation. Our work builds on this seeking to redeploy the parameterisation within a global atmospheric model, the Community Atmosphere Model (CAM), as a deep convection scheme, with the aim of running stable simulations with improved precipitation prediction. To do so requires interfacing the scheme to operate on a different vertical grid using a different system of variables to the original model in which it was trained. In this talk we will present this work discussing the objectives alongside the challenges faced moving the parameterisation from one model to another. We share the results from validation in single-column mode against field campaign observations, and of running the scheme globally in an aquaplanet configuration. We will also discuss software architecture and engineering considerations when seeking to develop and redeploy portable machine-learnt parameterisation schemes.

  • Open Access

    FTorch: a library for coupling PyTorch models to Fortran

    by Jack Atkinson, Athena Elafrou, Elliott Kasoar, Joseph G. Wallwork, Thomas Meltzer, Simon Clifford, Dominic Orchard, Chris Edsall

    In the last decade, machine learning (ML) and deep learning (DL) techniques have revolutionised many fields within science, industry, and beyond. Researchers across domains are increasingly seeking to combine ML with numerical modelling to advance research. This typically brings about the challenge of programming language interoperation. PyTorch (Paszke et al., 2019) is a popular framework for designing and training ML/DL models whilst Fortran remains a language of choice for many high-performance computing (HPC) scientific models. The FTorch library provides an easy-to-use, performant, cross-platform method for coupling the two, allowing users to call PyTorch models from Fortran. FTorch is open-source, open-development, and well-documented with minimal dependencies. A central tenet of its design, in contrast to other approaches, is that FTorch removes dependence on the Python runtime (and virtual environments). By building on the LibTorch backend (written in C++ and accessible via an API), it allows users to run ML models on both CPU and GPU architectures without needing to port code to device-specific languages.


  • Open Access

    FTorch - lowering the technical barrier of incorporating ML into Fortran models

    by Dominic Orchard, Elliott Kasoar, Jack Atkinson, Thomas Meltzer, Simon Clifford, Athena Elafrou

    Across geoscience, numerical models are used for understanding, experimentation, and prediction of complex systems. Many of these models are computationally intensive and involve sub-models for certain processes, often known as parameterisations. Such parameterisations may capture unresolved sub-grid processes, such as turbulence, or represent fast-moving dynamics, such as gravity waves, or provide a combination of the two, such as microphysics schemes.Recently there has been significant interest in incorporating machine learning (ML) methodsinto these parameterisations. Two of the main drivers are the emulation of computationally intensive processes, thereby reducing computational resources required, and the development of data-driven parameterisation schemes that could improve accuracy through capturing ‘additional physics’.Integrating ML sub-models in the context of numerical modelling brings a number of challenges, some of which are scientific, others computational. For example, many numerical models are written in Fortran, whilst the majority of machine learning is conducted using Python-based frameworks such as PyTorch that provide advanced ML modelling capabilities. As such there is a need to leverage ML models developed externally to Fortran, rather than the error-prone approach of writing neural networks directly in Fortran, missing the benefits of highly-developed libraries.Interoperation of the two languages requires care, and increases the burden on researchers and developers. To reduce these barriers we have developed the open-source FTorch library [1] for coupling PyTorch models to Fortran. The library is designed to streamline the development process, offering a Fortran interface mimicking the style of the Python library whilst abstracting away the complex details of interoperability to provide a computationally efficient interface.A significant benefit of this approach is that it enables inference to be performed on either CPU or GPU, enabling deployment on a variety of architectures with low programmer effort. We will report on the performance characteristics of our approach, both in the CPU and GPU settings and include a comparison with alternative approaches.This approach has been deployed on two relevant case studies in the geoscience context: a gravity-wave parameterisation in an intermediate complexity atmospheric model (MiMA) based on Espinosa et al. [2], and a convection parameterisation in a GCM (CAM/CESM) based on Yuval et al. [3]. We will report on these applications and lessons learned from their development. [1] FTorch https://github.com/Cambridge-ICCS/FTorch[2] Espinosa et al., Machine Learning Gravity Wave Parameterization Generalizes to Capture the QBO and Response to Increased CO2, GRL 2022 https://doi.org/10.1029/2022GL098174[3] Yuval et al., Use of Neural Networks for Stable, Accurate and Physically Consistent Parameterization of Subgrid Atmospheric Processes With Good Performance at Reduced Precision, GRL 2021 https://doi.org/10.1029/2020GL091363

  • Open Access

    Tools and techniques for modular, portable (Machine Learning) parameterisations

    by Jack Atkinson, Dominic Orchard, Elliott Kasoar, Thomas Meltzer

    Numerical models across the geophysical sciences make extensive use of physical parameterisations to represent sub-grid processes such as eddies, convection, gravity waves, and microphysics. With the advent of machine learning (ML), big data, and artificial intelligence there are increasing efforts to develop data-driven parameterisation schemes or emulate existing, but computationally-intensive, parameterisations using ML techniques.Despite their name, the irony is that traditional design approaches often lead to parameterisations which are not themselves parameters, but rather tightly integrated, enmeshed pieces of code in a larger model. Such parameterisations and practices pose a threat to the reproducibility, accuracy, ease-of-(re)-use, and general FAIRness (Findable, Accessible, Interoperable, Reusable) [1] of the schemes being developed.In contrast, a modular approach to parameterisations (and their receivers, e.g., GCMs), would enable them to be more easily (1) interchangeable and Interoperable, to compare different schemes and assess uncertainty due to their inherent approximate behaviour, and (2) portable and Reusable, between models, to reduce engineering effort, and (3) understandable and Accessible, by being clear about dependencies and the physical meaning of program variables, and (4) testable, to aid verification and correctness testing. Achieving this goal in the context of numerical modelling brings a number of scientific, engineering, and computational challenges. In this talk we aim to set out some best-practice principles for achieving modular parameterisation design for geoscience modellers. We then focus on the particular challenges around modem ML parameterisations. To this end we have developed a library for easily interoperating ML-based parameterisations in PyTorch with Fortran-based numerical models, called FTorch [2]. By reducing the Fortran-PyTorch Interoperability burden on researchers this framework should reduce errors that arise and increase the speed of development when compared to other approaches such as re-coding models in Fortran. FTorch aims to make emerging ML parameterisation research more Accessible to those who may not have deep expertise of ML, Fortran, and/or computer science. It also means that models developed using PyTorch can leverage its feature-rich libraries and be shared in their native format, maximising Reusability. We discuss the design principles behind FTorch in the context of modular parameterisations and demonstrate our principles and approach by coupling ML parameterisations to atmospheric and climate models.In general, we present a number of considerations that could be used to make all parameterisation schemes more easily Interoperable and Re-useable by their developers. [1] Barker, M. et al, Introducing the FAIR Principles for research software, Sci Data 9, 622 (2022) https://doi.org/10.1038/s41597-022-01710-x[2] FTorch https://github.com/Cambridge-ICCS/FTorch

  • Open Access

    Practical machine learning with PyTorch

    by Jack Atkinson, Jim Denholm


  • Open Access

    A New Model of Electron Pitch Angle Distributions and Loss Timescales in the Earth's Radiation Belts

    by S. A. Glauert, J. W. Atkinson, J. P. Ross, R. B. Horne

    As the number of satellites on orbit grows it is increasingly important to understand their operating environment. Physics‐based models can simulate the behavior of the Earth's radiation belts by solving a Fokker‐Planck equation. Three‐dimensional models use diffusion coefficients to represent the interactions between electromagnetic waves and the electrons. One‐dimensional radial diffusion models neglect the effects of energy diffusion and represent the losses due to the waves with a loss timescale. Both approaches may use pitch angle distributions (PADs) to create boundary conditions, to map observations from low to high equatorial pitch angles and to calculate phase‐space density from observations. We present a comprehensive set of consistent PADs and loss timescales for 2 ≤ L* ≤ 7, 100 keV ≤ E ≤ 5 MeV and all levels of geomagnetic activity determined by the Kp index. These are calculated from drift‐averaged diffusion coefficients that represent all the VLF waves that typically interact with radiation belt electrons and show good agreement with data. The contribution of individual waves is demonstrated; magnetosonic waves have little effect on loss timescales when lightning‐generated whistlers are present, and chorus waves contribute to loss even in low levels of geomagnetic activity. The PADs vary in shape depending on the dominant waves. When chorus is dominant the distributions have little activity dependence, unlike the corresponding loss timescales. Distributions peaked near 90° are formed by plasmaspheric hiss for L* ≤ 3 and E < 1 MeV, and by EMIC waves for L* > 3 and E > 1 MeV. When hiss dominates, increasing activity broadens the distribution but when EMIC waves dominate increasing activity narrows the distribution.


  • Magnetically driven flow in a liquid-metal battery

    by P. A. Davidson, O. Wong, J. W. Atkinson, A. Ranjan

    We investigate the flow within a liquid-metal battery induced by an externally imposed magnetic field, B0. An analytical model for laminar flow is proposed and this is found to be in excellent agreement with numerical simulations, not only for weakly forced steady flow, but also for the time-averaged velocity in more strongly forced flows where the motion is either unsteady (but laminar) or else weakly turbulent. Our primary conclusion is that surprisingly weak magnetic fields are capable of destabilizing the flow and inducing turbulence, with turbulence first appearing at B0≈1.0G. By comparison, the earth's magnetic field is ∼0.5 G. This extreme sensitivity of liquid-metal pools carrying current to an external magnetic field has long been known in the context of other industrial processes.


  • The evolution of laminar thermals

    by J. W. Atkinson, P. A. Davidson

    We consider the life cycle of an axisymmetric laminar thermal starting from the initial condition of a Gaussian buoyant blob. We find that, as time progresses, the thermal transitions through a number of distinct stages, undergoing several morphological changes before ending up as a vortex ring. Whilst each stage is interesting in its own right, one objective of this study is to set out a consistent mathematical framework under which the entire life cycle can be studied. This allows examination of the transition between the different stages, as well as shedding light on some unsolved questions from previous works. We find that the early stages of formation are key in determining the properties of the final buoyant vortex ring and that, since they occur on a time scale where viscosity has little effect, the final properties of the ring display an independence above a critical Reynolds number. We also find that rings consistently contain the same proportion of the initial heat and have a consistent vorticity flux. By considering the effect of Prandtl number, we show that thermal diffusion can have a significant impact on development, smoothing out the temperature field and inhibiting the generation of vorticity. Finally, by considering the wake left behind as well as the vortex ring that is generated, we observe that the wake can itself roll up to form a second mushroom cap and subsequently a secondary vortex ring that follows the first.


  • Dynamics of a trapped vortex in rotating convection

    by J. W. Atkinson, P. A. Davidson, J. E. G. Perry

    We consider axisymmetric rotating convection in a cylindrical domain, focusing on the eye that can form at the center of a cyclonic vortex. Upon increasing the thermal forcing we observe that the system undergoes a Hopf bifurcation from a state with a steady eye to one in which the eye oscillates. For an aspect ratio, Ekman number, and Prandtl number of 0.1 we find that the critical Reynolds number at which this transition occurs is 398.