Active Projects

Handling model-related uncertainty in weather and climate forecasts

People: Antje Weisheimer, Tim Palmer, Chris O'Reilly
This project is part of the National Centre for Atmospheric Science (NCAS).

PRIMAVERA: Process-based climate simulation: advances in high-resolution modelling and European climate risk assessment

PRIMAVERA is funded through the European Commission Horizon 2020 framework, and involves 19 partners across Europe. The aim of the project is to “develop a new generation of advanced and well-evaluated high-resolution global climate models, capable of simulating and predicting regional climate with unprecedented fidelity”.

The University of Oxford is involved in the work package “Frontiers of Climate Modelling”, and will investigate the use of stochastic parameterisations to represent the variability of unresolved sub-grid scale processes. Of particular interest is the extent to which stochastic physics can be used to represent small-scale components of the circulation that might otherwise be too expensive to resolve.

One of the major achievements so far in the project has been the development of novel stochastic schemes in both the atmosphere, land, ocean and sea-ice components of the EC-Earth climate model. These have been tested separately and jointly in a large ensemble of climate model simulations, generating around 80 Terabytes worth of processed data. In particular, a suite of simulations have been carried out with stochasticity added to every component: these form the first example of a `Probabilistic Earth-system Model’ (Palmer 2012).

Our work to date shows that stochastic schemes can greatly benefit the model in a manner similar to increased resolution, both in terms of smaller-scale structure (such as the number of tropical cyclones) and large-scale behaviour (such as global means of precipitation, cloud cover etc.). Further research is ongoing, with a focus on convective processes and key dynamical processes (such as ENSO).

Palmer, T. (2012) Towards the probabilistic Earth-system simulator: a vision for the future of climate and weather prediction. QJRMS

Key contacts:
Hannah Christensen, Kristian Strommen, Antje Weisheimer, Tim Palmer

ForPAc: Towards Forecast-based Preparedness Action

ForPAc is a project focusing on developing early warning systems for flooding and drought in Kenya and the Greater Horn of Africa. Working together with partners in the UK and Kenya, including the Red Cross/Red Crescent, Kenya Meteorological Department and the National Flood and Drought Monitoring Authority, we are:

  • identifing opportunities for forecast-based action in the region
  • quantifing the impact of rainfall shocks on society, at spatial scales from urban floods in Nairobi to sustained drought across the Greater Horn of Africa
  • diagnosing the predictability of anomalous rainfall on timescales ranging from 1 week to 1 year ahead

Work is being led in the group to assess the ability to anticipate flooding and drought events, from one week to several months ahead. Results are discussed with stakeholders in order to determine their potential to improved preparedness actions.

ForPAc is funded through the Science for Humanitarian Emergencies and Resilience (SHEAR) program, which is jointly funded by the Department for International Development and the Natural Environmental Research Council. See the project website, or contact Dave MacLeod for further details.

EUCP: European Climate Predictions


EUCP (EUropean Climate Prediction) will develop a regional climate prediction and projection system based on high-resolution climate models for Europe, to support climate adaptation and mitigation decisions for the coming decades. Where the term “climate prediction system” is used hereafter it covers both predictions and projections and the EUCP project will develop methods to stitch, or blend, these timescales seamlessly.
The project will also produce improved methods to characterise uncertainty in climate predictions and projections, regional downscaling, and evaluation against observations. The climate prediction system will produce consistent, authoritative and actionable climate information and the value of this system will be demonstrated through user relevant high impact extreme weather events in the near past and near future (1-40 years) drawing on convection permitting regional climate models translated into risk information. It will go beyond the outputs of, for instance, the IPCC assessments by turning raw model simulations into decision-relevant information, and will be designed to be sustainable and able to take account of new climate information, such as that produced for the next IPCC assessment, or the latest near-term decadal climate predictions as they become available.


EUCP is funded under the H2020 framework programme from Dec 2017 to Nov 2021. We lead the work package Towards a seamless near term European climate prediction system.

Key contacts:
PI: Antje Weisheimer
Chris O'Reilly
Daniel Befort
Tim Palmer


ITHACA: An Information Theoretic Approach to Improving the Reliability of Weather and Climate Simulations

Weather and climate prediction models take a large amount of time and energy to run. However, it is well known that current models are not able to make use of the full capacity of existing supercomputer technology. This problem is expected to become even worse on future supercomputers. Therefore, the aim of the project is to enable better predictions of weather and climate behaviour by exploring new ways to use supercomputers most efficiently. Any improvements in prediction skill will lead to major benefits for society, the economy and the environment.
When weather and climate predictions are made, they use computer-based mathematical models. For a single prediction, a model is typically run many times using slightly different input parameters and even wilfully including some random elements. This approach for making a prediction is used for two reasons. Firstly, we cannot be entirely certain about the true state of the atmosphere and the oceans because we only have limited measurements. Secondly, we know that the mathematical model can never be absolutely perfect. This means there will always be some ‘uncertainty’ left in each of our predictions.
The underlying question of this project is now:

If we are left with an uncertainty in our prediction anyway, do we really need to do all the calculations of our mathematical model as precisely as we currently do?

To answer this question, we will use approaches from information theory and emulators of precision to investigate and understand how much precision in the calculations is really needed, without losing predictive power. Doing calculations with less precision will free lots of computing capacity on our supercomputers. This spare capacity can then in turn be used to improve the predictions by reinvesting it into more important aspects of the model.

Disruptive Computing - Enhancing weather and climate simulations using Quantum Technologies

Funded by The Royal Society

The field of Quantum Information Processing (QIP) has seen rapid development over the past decade in both theory and experimental implementations. A number of quantum computation algorithms have been demonstrated to outperform the best known classical algorithms. As a result, QIP has gained considerable attention from various other scientific disciplines. In particular, a range of basic linear algebra operations has been shown to exhibit (in parts exponential) quantum speed-ups. This indicates that future Quantum Computers might be beneficial to numerical computation tasks. Modern weather and climate science heavily relies on efficient numerical simulations of non-linear dynamics. This project aims to investigate and evaluate the potential usefulness of QIP for such simulations. The major challenge hereby is to find a way to simulate a non-linear system on a quantum computer whose state transitions are fundamentally linear. Both possible ways to overcome this hurdle, post-measurement selection and effective non-linearities in many-body systems, are closely investigated.

Expired Projects

Towards the Prototype Probabilistic Earth-System Model for Climate Prediction

Summary of results
People: Tim Palmer
This project is funded by an ERC Advanced Investigator Grant 2011.


People: Antje Weisheimer, Chris O'Reilly, Tim Woollings
We are part of the NERC funded project SummerTIME (Summer: Testing Influences and Mechanisms for Europe).


People: Antje Weisheimer, Tim Palmer, David MacLeod
We are part of the FP7 project SPECS (Seasonal-to-decadal climate Prediction for the improvement of European Climate Services) and lead the workpackage 4.4. on Addressing Model Inadequacy. SPECS intends to develop the new generation of European operational seasonal-to-decadal climate forecast systems for the production of reliable, local climate information at the global scale. SPECS is a collaborative project with 20 partners from Europe and Brazil and is part of a cluster of European projects that will provide a coordinated response to the societal need for climate services.


People: Antje Weisheimer, Nathalie Schaller
We are part of the FP7 project EUCLEIA (EUropean CLimate and weather Events: Interpretation and Attribution).


People: Antje Weisheimer, Tim Palmer, Tim Woollings
We are part of the NERC funded project IMPETUS (Improved Drought Forecasts for User Decision-Making).


People: Andrew Dawson, Tim Palmer
We are part of the TEMPEST Project. Our involvement in TEMPEST covers "Investigating the response of intense extratropical cyclones to climate change in very high-resolution global atmospheric model experiments capable of capturing mesoscale structures."