University of Reading
The University is home to more than 23,000 students from over 150 countries and 4,000 staff.
A world leader in climate science and winner of the Queen’s Anniversary Prize. The University’s School of Mathematical, Physical and Computational Science was awarded the Athena Swan Silver in 2023, in recognition of good employment practices in relation to women working in science, engineering and technology.
The Times and Sunday Times, Good University Guide 2025, Sustainable University of the Year.
Winner of Outstanding Contribution to Sustainability Leadership, London Higher Award 2024.
Research is divided into four broad themes: Agriculture, Food and Health; Environment; Heritage and Creativity; and Prosperity and Resilience.
Ranked 1st in the People & Planet University League 2023/24 and 4th in 2024/25.
Ranked 13% in the QS World University Rankings in 2025.
In excess of 70 Global partners.
First ever winner of the Times Higher Education Outstanding Contribution to Environmental Leadership award 2023.
Potential PhD projects – 2027
- Forecasting the spatiotemporal distribution of Celtic Sea cod using satellite ocean colour and machine learning
This project will develop a novel, data-driven machine-learning-based modelling framework to provide near-real time predictions of the distributions of the highly mobile Celtic Sea cod.
This is sought after by UK and EU fisheries managers and policy makers, as well as by the International Council for Exploration of the Seas.
Partner: Centre for Environment, Fisheries and Aquaculture Science.
- Accounting for model error in Earth system digital twins
Earth system digital twins fuse machine learning models with real-world observations to predict the impact of policy choices on the environment and society under different scenarios.
This project involves working at the intersection of applied mathematics, machine learning and climate science to develop next-generation digital twins that capture uncertainties in complex physical models, advancing these powerful tools for better-informed decision-making in a changing climate.
Partner: National Centre for Earth Observation.
- Constructing plausible worst-case scenarios for UK hydrological drought
UK water companies have a regulatory requirement to protect against severe drought, defined as a 1-in-200-year event.
This project will develop defensible methods of testing this requirement using physical climate storylines (physically self-consistent, causal explanations) to construct plausible worst cases, anchored in downward counterfactuals of historical events.
Partner: Anglian Water.
- Stability and evolution of the background state of the atmosphere
The atmosphere is changing in response to global warming. This project will develop a model for the evolution of the global climate state in response to forcing using the mathematical theory of optimal transport and investigate the key question of stability of those states using Hamiltonian dynamics.
- The future of sting-jet storms: an AI-powered climate exploration
Sting jets are airstreams within intense winter storms that can lead to damaging surface wind gusts. Detecting them in gridded numerical output, for example from climate models, is currently computationally expensive.
This project will investigate climate change effects in sting-jet occurrence through the development of AI-based sting-jet detection methods.
Partner: National Centre for Atmospheric Science.
- Statistical reproducibility in climate forecast evaluation
Climate forecasts underpin critical decisions, yet evaluating their performance is statistically challenging because climate systems are non-stationary and analytical choices can strongly influence conclusions.
This project will develop methods to assess the robustness and reproducibility of forecast evaluations, distinguishing unavoidable forecasting uncertainty from variability introduced by alternative, defensible statistical analyses.
- Numerical methods for assimilating satellite altimetry data
Use mathematics to better understand the ocean. In this project, you will develop numerical methods to combine satellite measurements of sea surface height with dynamical ocean models, improving estimates of the ocean state.
After initially working on idealised systems, in collaboration with European Centre for Medium-Range Weather Forecasts (ECMWF) scientists, you will work on real data and contribute to important advances in weather prediction and climate monitoring.
Partners: European Centre for Medium-Range Weather Forecasts; National Centre for Earth Observation.
- Faster and greener data assimilation using mixed-precision algorithms
This PhD tackles one of the biggest computational challenges in climate science: how to combine vast datasets and models efficiently.
By developing new mixed-precision algorithms, you’ll explore how reduced numerical accuracy can unlock faster, greener weather and climate prediction – working at the frontier of mathematics, computing, and Earth system science.
Partner: National Centre for Earth Observation.
- Heat networks in support of a net-zero carbon national power system
Help shape the future of low-carbon energy. This PhD will explore how district heat networks can support the transition to decarbonised heating by easing pressure on power systems.
You will develop statistical models of heat demand and dynamic network models, using sensitivity and uncertainty analysis to identify smart, time-dependent demand-side management strategies.
Partners: Reading Borough Council; Royal Berkshire Hospital Estates Team.
- Physical models of climate with machine learnt turbulent fluxes
Machine learning (ML) is a transformative force in weather prediction, but the same methods cannot be used for climate because they cannot learn from a climate that hasn’t happened yet.
This project will create a new hybrid ML/physical model, learning the detail that is too expensive for a physical model, while keeping the large-scale laws of the physical model, such as energy conservation.
Partners: UK Met Office; European Centre for Medium-Range Weather Forecasts.
- The effects of long-term heat shocks on UK fruit and vegetable supply chain resilience
Climate change is having adverse effects on our ability to supply food. In research at the forefront of supply chain dynamics we will use mathematical modelling to improve the resilience of UK fresh food supply chains in the context of future heat shocks in the UK and Europe.
- Physics-informed machine learning model for global tropical cyclone prediction
Tropical cyclones are among the costliest weather systems. This PhD aims to build physics-informed AI to improve the physical consistency of global tropical cyclone prediction from monthly to decadal timescales.
Explore new drivers of predictability, develop probabilistic forecasts with quantified uncertainty, and engage national forecasting agencies and industry on societal and economic impacts.
- On the information content of ocean observations: inverse methods for circulation and turbulent mixing
Develop new rigorous inverse ocean models, deriving absolute circulation, eddy-induced transports and turbulent mixing coefficients, from observations.
Using dynamic and thermodynamic conservation laws, you will investigate identifiability, conditioning and well-posedness, seeking five-parameter descriptions of ocean stirring and mixing that can be used to improve understanding of the current ocean circulation state.
- Mathematical modelling of fish bioenergetics and environmental resilience in aquaculture
This PhD will develop an individual-based bioenergetic model to predict how environmental variability affects Asian seabass growth, energy allocation and resilience in integrated multitrophic aquaculture.
Combining selectively bred fish and environmental data from Thailand with mechanistic modelling, the project will provide a predictive framework for optimising production under changing environmental conditions.
Partner: Centre for Environment, Fisheries and Aquaculture Science.
- The Charney conjecture of data assimilation in (relatively) simple dynamic ocean models
Charney’s conjecture in data assimilation states that for suitable geo-physical models, temperature observations determine all other state variables such as wind or ocean current velocities, or pressure fields.
This project investigates the Charney conjecture in the context of a relatively simple ocean and temperature model. The research combines partial differential equations, dynamical systems, and data assimilation and has applications to weather, ocean, and climate prediction.
- Learning to learn: optimal machine learning training objectives for weather prediction
Machine learning is transforming weather and climate prediction, but how should forecasting models be trained?
This PhD will find the best ways to train probabilistic forecasting models, so that models learn as much as possible from limited data, using probability, statistics, optimisation, and dynamical systems to improve forecast uncertainty.
Do you have more questions?
Contact us
University of Reading
How to apply
Step 1
Send Your Application
Step 2
Provide Supporting Documents
Step 3
Assessment & Interviews
Step 4
Get An Offer
Apply now
Imperial College London
University of Reading
University of Southampton
Fully Funded Studentships
Study while receiving a full stipend (with enchanced stipend), PhD fees paid for 4 years, and a generous allowance for research-related travel.