51³Ô¹ÏºÚÁÏ

PI: Prof. Nathan Mayne
Collaborators: Met Office / Momentum Partnership, National Centre for Atmospheric Science (NCAS)

Weather and climate models are some of the most computationally demanding scientific software systems in use. As supercomputers become more powerful and their architectures more diverse, the software used to understand our atmosphere must evolve alongside them. is the Met Office’s next-generation modelling system, designed to take advantage of advances in high-performance computing and support the future of weather prediction and climate projections.

Working with the Met Office Momentum Partnership and the National Centre for Atmospheric Science (NCAS), and led at 51³Ô¹ÏºÚÁÏ by Prof. Nathan Mayne, the University of 51³Ô¹ÏºÚÁÏ has contributed to the development, review and testing of the new course. The practical, self-paced materials provide a route into the LFRic ecosystem for researchers and operational users, helping them progress from understanding the concepts behind the model to configuring and running LFRic Atmosphere simulations.

The Research Software and Analytics Group’s contribution brings together expertise in research software engineering, high-performance computing and technical training. Our work has included reviewing and testing training materials from the perspective of an external research partner, improving practical exercises and technical guidance, and considering prerequisites, accessibility and the overall learner experience. Our wider work with the Met Office has also involved testing LFRic and its associated workflows on partner HPC infrastructure, helping identify the practical challenges researchers encounter when moving scientific software and training between computing environments.

The course introduces users to LFRic and the wider Momentum framework, unstructured grids and data, the LFRic software architecture and supporting tools such as PSyclone. Learners gain practical experience building applications, configuring and running simulations on high-performance computing systems, and analysing model output using Python and Jupyter notebooks.

Importantly, these are open and evolving training materials rather than a static course. The is openly available on GitHub, allowing the wider community to report issues, provide feedback and contribute improvements as LFRic and the computing environments supporting it continue to develop.

Developing the next generation of weather and climate models is not only a software challenge: researchers also need the skills and training to use them effectively. By combining expertise from the Met Office, NCAS and partner organisations such as the University of 51³Ô¹ÏºÚÁÏ, the project is helping turn a highly sophisticated modelling system into something that a growing research community can learn, use and contribute to.

The University of 51³Ô¹ÏºÚÁÏ’s contribution to this work was highlighted by the Met Office in its blog, , which explores how advances in atmospheric science, software engineering, supercomputing and training are coming together to support the next generation of weather and climate modelling.

PIs: Dr Luke Pilling, Prof. Jack Bowden 

The research team sought support from the Research Software Engineering (RSE) team to improve the reproducibility, maintainability and shareability of their analytical workflows, particularly within the context of working in Trusted Research Environments (TREs). Scoping sessions highlighted several key challenges, included enabling researchers to collaborate on and modify analyses created by others, managing versions of code over time, reproducing research outputs such as figures and models and working efficiently within TRE constraints.

The RSEs embedded within the research group, working alongside researchers to understand their day-to-day workflows and identify practical solutions. They developed a standardised repository structure and workflow template to bring consistency to analysis projects and make code easier to navigate, maintain and share. (A copy of this template is available at ). This was supported through targeted training workshops, including sessions on adopting the new workflow and improving programming practices.  Self-directed learning materials on topics such as function writing and unit testing in R were also developed.

Discussions with researchers also highlighted a specific need to streamline the process of querying UK Biobank and CPRD medication data. The RSEs therefore also created a bespoke R package for querying the prescription data based on a simple, declarative file format, enabling researchers to perform the queries more reproducibly and supporting collaboration. 

"Having Ruxandra and Tom from the Research and Software and Analytics group embedded in our epidemiology team made an enormous difference to several projects in our group. The most tangible were the development of specific software tools to robustly document medication codes in our epidemiological work, but also the delivery of specific, tailored training sessions to the group on reproducible, collaborative, organised, and open coding. Together, they promoted better coding practice and programmatic thinking in the wider team, which has had lasting benefits beyond the end of the specific project. Thank you!" 

Dr. Luke Pilling

Collaborators: ECEHH, Cornwall County Council

Working with the  and , the RSE group have developed LCAT, the Local Climate Adaptation Tool, an open-source tool which supports local decision makers across the UK to plan and adapt to climate change. The live tool can be accessed .

LCAT brings together complex climate models, adaptation options and health impact evidence all in one place, and importantly, generates recommendations for appropriate adaptation approaches, to support the health and wellbeing of local people. These recommendations are based on the best available peer-reviewed evidence. LCAT has been co-designed with over 100 different local decision-makers across the UK and is one of the outputs of the EU  project.

The RSE group have been responsible for everything from climate data analysis, to the app's deployment, to front and back-end development, and more. We developed scripts in PostgreSQL and Python to process large NetCDF climate data files, improving previous processing times by >95%. Working with the project team, we have developed new front-end features, iteratively improving the UX/UI of the tool based on stakeholder feedback. Finally, we worked alongside  to develop the LCAT deployment pipeline, deploy the first version, and maintain and manage new releases of LCAT going forward.

We've had some excellent press around LCAT as we have rolled it out, from  and the . And we are very proud to say that LCAT won the

“Working with the Research and Software Analytics group has been game-changing for the LCAT project. Their knowledge, expertise and engagement with the project has allowed us to be responsive to user feedback and need, and iteratively develop an award-winning decision-support tool on climate change adaptation. This development would not have been possible without them.”

Prof. Emma Bland, ECEHH

Collaborator: The GW4 Alliance

X-disciplinary Challenges from Industry for Technical Expert Development () is an EPSRC-funded programme that will develop the skills of Research Technical Professionals (RTPs), supporting them to work with industrial partners in addressing real-world challenges. 

You can read more about the project here, and read an interview with Nic Whippey, senior RSE and X-CITED lead, .

Collaborators: Dr Steven Rieder, Professor Clare Dobbs, Dr Thomas Guillet, Department of Physics and Astronomy

AMUSE (Astrophysical Multipurpose Software Environment) allows different astrophysical codes to run and communicate with a common interface. The RSE Group worked to integrate basic functionality of Arepo, a moving-mesh gravity and magnetohydrodynamics code, into AMUSE. This will lead to follow-on work expanding this integration, for example testing Arepo in conjunction with stellar evolution and gravitational dynamics codes which are part of AMUSE. This will make it ready for use by researchers.

Collaborators: Departments of Computer Science and Engineering, The Alan Turing Institute and NATS
PI: Professor Richard Everson‌â¶ÄŒ

The 51³Ô¹ÏºÚÁÏ RSE Group have a longstanding commitment to project Bluebird, a Prosperity Partnership with the National Air Traffic Service (NATS) and The Alan Turing Institute. This project aims to deliver the world's first AI-supported system that can control UK airspace.

The RSE Group has aided in the development of a data-driven, high-fidelity model of UK airspace that can be controlled via machine learning and other AI algorithms. Other tasks have included the implementation of algorithms used for aircraft trajectory prediction and atmospheric modelling, the creation of a web-app based human-machine interface to display, monitor and control simulations, and the development of a data pipeline, which feeds the model with data from thousands of aircraft from across the UK.

We ended last summer with a highly successful full stack evaluation, working with NATS to evaluate our first generation of air traffic control algorithms. Currently, the group is working to improve the aircraft trajectory modelling used in the simulations, supporting researchers by implementing continuous testing and evaluation of their algorithms, and developing a gamified simulation, where the general public can test their air traffic control skills against an AI controller.

Collaborators: Wellcome Trust, Dr Chloe Onoufriou

The University of 51³Ô¹ÏºÚÁÏ is developing an open-source and widely adaptable tool for allowing researchers to 'cost' the carbon impact of their research project. This tool has completed phase 1 of development and been through initial testing at UoE to develop a proof-of concept model. You can read more about the project here.

Collaborator: Prof Lars Johanning, Department of Engineering 

The Cornwall FLOW Accelerator project developed a simulator to compare strategies for floating offshore wind (FLOW) farm operations, looking to help reduce the associated carbon footprint. The RSE Group provided the much-needed technical expertise, with the RSEs involved giving guidance on the project's development from a software engineering perspective, writing software to combine the core components of the simulator into a unified whole, and developing a graphical user interface. 

Collaborator: Prof Pierre Friedlingstein FRS, Department of Mathematics and Statistics

The RSE Group are working with the Global Carbon Budget (GCB) Office, which publishes authoritative research tracking the trends in global carbon emissions. The RSE Group are developing a new website that provides users with access to GCB data and enables them to explore it through an interactive data visualisation dashboard. This will increase the impact and influence of the GCB with policy makers and non-academic organisations. The RSE Group have also improved the codebase and processes used for creating analyses for the annual GCB report, making it easier to maintain and more accommodating to dataset updates. This is helping to streamline the development effort required to produce each year's report.

Collaborator: Dr Oleksandr Kyriienko, Department of Physics and Astronomy

The aim of this project was to scope the potential of quantum machine learning methods for fraud detection to inform the quantum effort at HSBC. The RSEs involved in the project provided classical machine learning expertise, prompted valuable discussion, helped benchmarking quantum machine learning algorithms against their classical counterparts. The results of this project gave rise to  and fed into HSBC strategy.

Collaborator: Professor Katharine Tyler, Department of Social and Political Sciences, Philosophy and Anthropology

This project involved the creation of an interactive app that forms part of an entitled ‘Red, Amber, Green Britain’, which makes research findings on inequality in the context of Brexit and COVID-19 accessible to those outside of the academic community. The RSE Group provided the much-needed technical expertise to develop the app, which was featured at the Science Gallery in Detroit in their recent exhibition ‘Tracked and Traced’.