Below is a list of DPhil (PhD) project areas for entry in 2022 in climate physics. If you are interested in any of the following research areas, please contact the relevant supervisor directly as they will be happy to have a dialogue with you.

Some projects may be filled as applications are reviewed, so we particularly encourage any candidates considering an application after the January deadline to contact prospective supervisors about available project options.

  • Earth and planetary climate dynamics: Raymond Pierrehumbert
  • Variability of atmospheres and oceans: Myles Allen, Tim Palmer, Tim Woollings, David Marshall, Lesley Gray, Antje Weisheimer and Hannah Christensen
  • Natural and anthropogenic drivers of climate change: Philip Stier, Don Grainger, Anu Dudhia, Raymond Pierrehumbert
  • Response to natural and anthropogenic drivers of climate change: Myles Allen, Lesley Gray, Antje Weisheimer
  • Physics of clouds and aerosols and their effects on climate: Philip Stier
  • Cryosphere and polar processes: Andrew Wells
  • Tropical climate variability: Hannah Christensen
  • Remote sensing and earth observations: Anu Dudhia, Philip Stier, Neil Bowles
  • Geophysical fluid dynamics: Peter Read, Andrew Wells, David Marshall, Tim Woollings, Lesley Gray, Raymond Pierrehumbert
  • Ocean physics: David Marshall, Andrew Wells
  • Predictability of weather and climate: Tim Palmer, Tim Woollings, Myles Allen, Lesley Gray, Antje Weisheimer Hannah Christensen
  • Stratosphere and climate: Lesley Gray
  • Weather and climate modelling: Hannah Christensen

Project examples

The following projects are posted to give overseas candidates an idea of possible options; candidates with home fee status who we ask to apply through the DTP will be encouraged to propose their own projects in liaison with supervisors during their first term in Oxford.

Machine Learning for Stochastic Parametrisation

Hannah Christensen

Atmospheric models lie at the heart of weather and climate predictions. The skill of the resultant forecasts is a substantial scientific achievement. However, these forecasts inevitably contain errors. We acknowledge this by producing probabilistic forecasts, which span the range of possible future states of the system. To produce these probabilistic forecasts, we must account for all sources of uncertainty in the forecast. A large source of uncertainty in weather and climate prediction arises from the simplifications made when developing the forecast model itself, particularly in how we represent small-scale processes. One approach to account for this source of uncertainty is the use of stochastic parametrisations, where stochasticity (random numbers) are introduced into the forecast model. This approach has become ubiquitous across operational weather forecasting centres, and is gaining traction in the climate community, even though current stochastic schemes are relatively simple.

In recent years there has been an explosion of activity in machine learning applied to weather and climate modelling. A key area of interest is replacing expensive parts of atmospheric models with cheaper, and potentially better, machine learnt emulators. In this project, we will explore the potential for machine learning to improve stochastic parametrisations. By building on existing close collaborations with modelling centres, the research undertaken in this DPhil has the potential to improve operational forecasts. In addition, by interpreting the resultant emulators, we seek to learn about the predictability of uncertain small-scale atmospheric processes.

The global impacts of El Niño

Hannah Christensen

One of the largest sources of information on seasonal timescales is the El Niño-Southern Oscillation (ENSO). ENSO consists of an irregular cycle in sea surface temperature in the Tropical Pacific, with associated changes in the atmospheric circulation. During a warm ‘El Niño’ event, SST anomalies in the central and eastern Tropical Pacific reach two to three degrees above average, while during a cold ‘La Niña’ event, SST anomalies reach two to three degrees below average. A strong El Niño or La Niña event substantially increases the likelihood of extreme weather events both near and far. For example, locally it leads to heavy rain and flooding in Peru and drought in Indonesia. Further afield, there is evidence to suggest it affects the intensity of precipitation in the Indian Monsoon, the likelihood of drought in the American mid-west, and even European weather patterns.

Quantifying these so-called ‘teleconnections’, whereby ENSO impacts extreme weather around the globe, is important to understand sources of skill in seasonal forecasts. If we know a strong El Niño is on the way, we would like to quantify the resultant chance of severe flood or drought events, months in advance. But it is difficult to quantify these changes in probability of extreme events due to the short observational record.

In this project, we will produce simulations using world-leading climate models to quantify the impact that ENSO has on extreme weather around the globe. This will provide an important source of information on seasonal timescales, enabling industry, policy makers, and the humanitarian sector to quantify the risks of natural hazards, and act accordingly.

Required skills: good mathematical knowledge and basic programming skills

Observing the origins of tropical cyclones

Anu Dudhia and Tony McNally (Reading)

Tropical cyclones are potentially devastating weather systems for which accurate early warning is critical. However, their origins and behaviour often depend on rather small scale atmospheric structures which are not well-resolved in global numerical weather prediction models. The aim of this project is to explore if key features signalling the genesis, or sudden intensification, of tropical cyclones can be observed in infrared spectra from a 12km footprint measured by the IASI instrument on the MetOp satellites. Such features include warm Sea Surface Temperatures (SSTs), high levels of mid-tropospheric humidity (fuel) and cloud top temperatures. The work would combine the IASI expertise in AOPP with the forecast modelling and data assimilation at ECMWF.

The conclusions from this project will be relevant for improving the use of data from current satellite instruments in operational tropical cyclone prediction. In addition this study will help pave the way for the successful exploitation of future satellite instruments. The next generation geostationary Meteosat sounders (2023 onwards) will observe the genesis of Atlantic tropical cyclones with high temporal resolution (up to every 15 minutes) and IASI-NG instruments (2021 onwards) will provide enhanced information about the vertical structure of these storms.

Relevant skills: computing, atmospheric radiative transfer and thermodynamics.

Full details can be found here.

Satellite observations of air quality

Anu Dudhia and Brian Kerridge (Remote Sensing Group, RAL Space)

Ammonia (NH3) and carbon monoxide (CO) are important atmospheric pollutants, contributing to the production of ozone and particulates, which are important to air quality and also to climate change. Ammonia is generated from agricultural processes and carbon monoxide from fossil fuel combustion, but both are also products of biomass burning.
Concentrations of these (and many other molecules) can be retrieved from their infrared signatures in the spectra taken by the current generation of polar-orbiting interferometers, such as the IASI instrument on the MetOp satellites.

Both RAL (CO) and Oxford (NH3) have developed retrieval algorithms for IASI and the first part of the project will be to compare these results with other sources, eg other IASI retrievals, other satellite instruments, models, and surface measurements.

The second part of the project involves the adaptation of these algorithms for two new satellite instruments:
IASI-NG, the next generation IASI instruments, the first of which will be launched in 2021 and should provide data in the time-frame of the DPhil. This will have improved spectral resolution, allowing more species to be retrieved and with greater accuracy.

MTG-S, the third generation of the Meteosat geostationary satellite which will carry an interferometer viewing the entire earth's disk every hour. The first of these is due for launch in 2023. Unlike the IASI instruments which are on polar-orbiters and hence view each location only every 12 hours, the MTG-S will allow the full diurnal cycle to be observed.

Relevant skills: ability to write computer code in Fortran, C, IDL or Python, basic knowledge of radiative transfer and inverse methods.

Full details can be found here.

Retrieval of aerosol from single-view imagery

Don Grainger and Adam Povey

Atmospheric aerosols and cloud influence the climate by scattering and absorbing light. The most recent IPCC report labelled these interactions the most uncertain aspect of our understanding of the Earth's radiation budget. An accurate and long-term record of aerosol properties is essential to understanding the mechanics of our atmosphere, monitoring the effects of wildfires, and adapting to the growing problem of air pollution. As part of a European Space Agency project, researchers at Oxford helped produce a 30+ year record of cloud properties from satellite observations. A successful DPhil candidate will join that team and devise a means to extend our aerosol record to cover the same period. The method could be adapted to the next generation of geostationary imagery, providing near-real-time observations of aerosol and cloud of unrivaled detail and duration.

Aim: this project aims to produce an accurate retrieval of aerosol properties from single-view satellite imagery. The Optimal Retrieval of Aerosol and Cloud (ORAC) algorithm is open-source implementation of optimal estimation retrieval developed by EODG in conjunction with scientists at the Rutherford Appleton Laboratory. The code is used to infer aerosol (and cloud) properties from radiometer imagers, such as MODIS, the Sea and Land Surface Temperature Radiometer (SLSTR) and the Spinning Enhanced Visible and InfraRed Imager (SEVIRI). Currently, ORAC requires two views of the atmosphere to produce an aerosol retrieval. This limits us to observations since 1995 despite acceptable satellite records beginning in 1978. This also excludes observations from the SEVIRI geostationary imagers, which provide planetary-scale imagery every 15-minutes. Aerosol retrievals at such resolution are necessary to evaluate some of the most uncertain processes.
This project will determine the best means to produce aerosol retrievals from single-view imagery. This is expected to involve constraining the surface reflectance, through some combination of theoretical modelling and independent observations. Various single-view aerosol retrievals exist, such as NASA's Deep Blue, that are expected to provide a starting point.
Depending on the student's interest, there is scope to manage and operate remote sensing instruments in Oxford to generate data that can be compared to the new retrievals for validation. Products can also be intercompared to trace gas and volcanic products generated within the group.

Skills that would be helpful: computing, statistics, atmospheric radiative transfer, remote sensing (especially of the surface).

Full details can be found here.

Satellite remote sensing of volcanic plumes

Don Grainger, Isabelle Taylor and Tamsin Mather (Earth Sciences)

Plumes of ash and gas are one of the far-reaching hazards associated with volcanic eruptions. Monitoring them helps to mitigate their effects, understand the impacts they have on the environment and climate, and interpret volcanic activity. Satellite remote sensing offers a cost-effective solution to some of the limitations of ground-based monitoring and can be used to study the propagation of these plumes as they move across the globe and interact with different weather systems. The IASI instrument is an infrared hyperspectral sensor on-board three of EUMETSAT’s meteorological satellites, with each obtaining global coverage twice a day. IASI can be used to obtain the mass and height of both ash and SO2 and previous studies have demonstrated its capability for studying the plumes from large explosive eruptions and also continuous degassing.

Aim: to explore some of the ways in which the IASI instruments (and future generations of this instrument) can be used to study volcanic plumes of ash and gas. This project will build on the previous research in Oxford using IASI to study volcanic plumes. The focus of the PhD can be tailored to the student’s strengths and interests. A student from a physics or more technical background may wish to work on the further development of retrievals for IASI or other hyperspectral satellites; while a student with an Earth Sciences or Geography background may wish to explore how this data may be used to understand volcanic activity. Students may be interested to explore the impacts of these plumes on atmospheric dynamics. Opportunities may arise during the PhD to study recent eruptions.

Skills that would be helpful: programming (IDL, python), experience with remote sensing, experienced with multi-disciplinary work.

Using trace gas measurements to infer aerosol type

Don Grainger and Anu Dudhia

The effect of atmospheric aerosols is the most poorly quantified element of the Earth's radiative budget (see Figure 1). Aerosols act directly by reflecting or absorbing solar radiation, or indirectly by altering cloud properties. This project focuses on understanding the direct effect of aerosols. Ground and aircraft based instruments can be used to measure aerosol properties locally whereas satellite measurements provide a global perspective. Progress has been made in using satellite instruments to quantify aerosol optical depth and particle size (using, for example, NASA's Moderate Resolution Imaging Spectrometer, MODIS, or ESA's Advanced Along Track Scanning Radiometer, ATSR) but much uncertainty remains in the type of aerosol being observed. Understanding the type of aerosol is critical as measurements from a thin layer of highly reflective aerosol can be indistinguishable from a thicker layer of strongly absorbing aerosol. Without knowledge of aerosol type it is not possible to tell if solar radiation has been reflected harmlessly back to space or absorbed into the atmosphere, further heating the Earth-atmosphere system.

The aim of this project is to quantify aerosol absorption (as opposed to scattering) over an annual cycle, globally. The Optimal Retrieval of Aerosol and Cloud (ORAC) algorithm is open source software developed by EODG in conjunction with scientists at the Rutherford Appleton Laboratory. The code is used to infer aerosol (and cloud) properties from radiometer imagers such as MODIS, the Sea and Land Surface Temperature Radiometer (SLSTR) and the Spinning Enhanced Visible and InfraRed Imager (SEVIRI). Currently ORAC uses a combination of spectral signature augmented by geographic location to select a most probable aerosol type. This is not always successful. In this project additional measurements of aerosol-forming gases will be used to construct a prior probability of aerosol type. This will be used within ORAC to create a best guess of aerosol type.

The Infrared Atmospheric Sounding Interferometer (IASI) is a nadir-viewing Fourier transform spectrometer carried on the MetOp series of satellites (October 2006 - ~2030). Work within EODG has shown IASI can provide information on gases that form aerosol (e.g. SO2 or NH3) or gases associated with industrial pollution or biomass burning (e.g. CO or HCN). The goal of this project is to use the IASI gas measurements to improve ORAC's prediction of aerosol type. For example an enhanced value in the linear flag of SO2 will be used to increase the likelihood of the ORAC algorithm selecting sulphate aerosol. Much of the work will involve optimising the choice of aerosol given the values of the gas measurements. Once the method is finalised the algorithm will be applied to at least a year of data and the results analysed to give a measure of absorbing versus non-absorbing aerosol. If time permits further investigations could include fire emission indices, aerosol-formation mass budgets or aerosol composition climatologies.

Skills that would be helpful: computing, atmospheric radiative transfer, remote sensing, chemistry.

Full details can be found here.

Idealised models of polar climate in warm and cold climates

Raymond Pierrehumbert

Recent research has shown that polar climate is strongly influenced by incursions of outbreaks of warm, moist air from lower latitudes. In cold climates this process affects the growth and decay of sea ice, and in warm climates such as the Eocene this determines the ability of polar continents (e.g. Antarctica) to avoid hard freezes in the winter. This project will explore idealized models of the response of polar climates to stochastic intrusions, with a particular eye to understanding the seasonal cycle, the extent of fluctuations about it, the influence of clouds and the way the seasonal cycle is affected by precessional variations in insolation.

This project requires a thorough understanding of fundamental physics, including thermodynamics, mechanics and electromagnetic radiation, as well as a facility with analysis of mathematical models. Familiarity with physical chemistry is also desirable. Hence, a first degree in Physics, Mathematics or a related discipline is required. The project involves considerable use of computational techniques, so basic familiarity with numerical analysis and familiarity with programming techniques in some computer language is required. The main programming languages used are Python and Fortran, but prior experience with these specific languages, while desirable, is not required.

Nonlinearity, climate sensitivity and bifurcations in the climate system

Raymond Pierrehumbert

Much current analysis of the problem of climate sensitivity is based on linearization of the Earth's energy balance about its pre-industrial state. However, our work has shown that the high-sensitivity "fat tail" of the distribution, which poses the greatest risk, is generically indicative of the climate system being near a bifurcation or tipping point. Analyses of climate sensitivity that ignore the presence of a nearby bifurcation risk missing important features of climate change. We are working on a variety of means to detect and explore tipping points in multiphysics model ensembles. We are also examining idealized mathematical models which have a dense set of bifurcations, so that climate is nowhere differentiable with regard to its parameters. Why is such exotic behavior not seen in general circulation models? Or is it just that we haven't looked hard enough?

What ended the boring billion?

Raymond Pierrehumbert

The “boring billion” is the period of Earth history roughly bookended at one end by the Great Oxygenation Event and Makganyene Snowball at the dawn of the Proterozoic, and the Cryogeneian glaciations of the Neoproterozoic (about 700 million years ago) at the other. During this period, there are few major carbon isotope excursions, and there is no evidence of major glaciations. The termination of this period of relative stasis is a key event in Earth history, as it was followed not long afterwards by the first multicellular life (the Ediacarans) and somewhat later the Cambrian Explosion which marks the dawn of the modern world of the Phanerozoic. The question of what terminated the Boring Billion is closely associated with the more specific question of how Snowball Earth events (global glaciations) are initiated, and involves consideration of the physical climate system as well as biogeochemistry. This project involves wide-ranging research aimed at understanding Proterozoic climate evolution, from a standpoint of climate dynamics, atmospheric chemistry and biogeochemistry. General circulation models, as well as a hierarchy of simpler process models, are all employed in the work. A typical DPhil research project in this area would not involve the problem as a whole, but rather some specific aspect such as the effect of ocean/atmosphere dynamics on the greenhouse gas threshold for global glaciation, or the nature of carbon cycle fluctuations that could draw down CO2 while being compatible with the proxy record of carbon isotope excursions.

This project requires a thorough understanding of fundamental physics, including thermodynamics, mechanics and electromagnetic radiation, as well as a facility with analysis of mathematical models. Familiarity with physical chemistry is also desirable. Hence, a first degree in Physics, Mathematics or a related discipline is required. The project involves considerable use of computational techniques, so basic familiarity with numerical analysis and familiarity with programming techniques in some computer language is required. The main programming languages used are Python and Fortran, but prior experience with these specific languages, while desirable, is not required.

Nonlinear dynamics of the Quasi-Biennial Oscillation in laboratory and theoretical models

Peter Read (co-supervision by Scott Osprey and Alfonso Castrejon-Pita from Engineering Sciences (and potentially Neil Butchart from the Met Office)

The Quasi-Biennial Oscillation (QBO) dominates the climate of the tropical stratosphere, influencing the long range transport of momentum, heat and chemical constituents. It is also thought to play an important role in influencing the predictability of various features, such as the Madden-Julian Oscillation (MJO) and other phenomena in the troposphere. So understanding what determines its variability in space and time is important for a range of problems in seasonal climate prediction. Although the basic nonlinear wave-driven mechanisms that drive the QBO are reasonably well understood, its detailed variability is complex, chaotic and much less well understood. The phenomenon is notoriously difficult to capture realistically in global climate models. The likely impact of future global climate change on the QBO is also quite controversial and uncertain.

In this project, we propose to study a number of mechanisms that might influence the behavior of the QBO, using a combination of numerical models and a laboratory analogue of the QBO, in which factors such as the wave forcing and other parameters and feedbacks can be closely controlled and varied. In the laboratory experiment, internal waves are launched into a salt-stratified fluid in an annular channel by oscillating flexible membranes in the bottom of the tank. Each segment of the membrane in the new experiment can be separately controlled by computer (a uniquely novel aspect) to enable varying spectra of internal waves to be excited and for the amplitude of the waves to be varied in time (thereby emulating the seasonal cycle and other modulations). The response of the fluid to this forcing in the form of time varying velocity fields will then be measured by optical particle imaging techniques, while conductivity probes will measure the stratification. The experiments will be complemented by a series of numerical model simulations (a) to achieve direct numerical simulation of some of the laboratory flows themselves for comparison with and validation against experimental measurements, and (b) to explore idealized simulations of QBO-like phenomena in global atmospheric circulation models. The numerical models will make use of Met Office codes such as ENDGAME to maximize opportunities to transfer benefits of this research directly to Met Office researchers.

This is a blue skies project that will suit a student with an interest in fluid dynamics, numerical modelling and with strong physical/mathematical skills. The project will involve collaboration with the Department of Engineering Science on experimental aspects, and with the Met Office on numerical modeling (for which additional funding may be available).

Constraining aerosol-cloud interactions through multi-sensor fusion of satellite-based Earth observations using machine learning

Philip Stier and Yarin Gal Department of Computer Science (co-supervisor)

Aerosol-cloud interactions remain the largest uncertainty in anthropogenic perturbations to the Earth’s radiation balance underlying climate change. In this project we will develop new machine learning based approaches to combine the strengths and information content of multiple heterogeneous satellite observations to provide novel observational constraints on aerosol-cloud interactions.

The Earth is being continuously observed by a wide array of satellite sensors, some of which are flown in synchronised constellations observing the same scenes only seconds apart. Yet, retrievals of atmospheric properties are almost exclusively done independently for each instrument. This is because traditional methods make it difficult to combine heterogeneous satellite observations (e.g. along track aerosol-backscatter curtains from lidars with wide-swath aerosol optical depth retrievals from spectral imagers or along-track cloud radar reflectivity curtains with wide-swath spectral imager cloud property retrievals) in an optimal way. Therefore, significant opportunities to strengthen our observational constraints on aerosol-cloud interactions, in particular in terms of the vertical distribution of clouds and aerosol properties, remain currently unexplored.

This project will develop novel machine learning methods for multi-sensor fusion of heterogeneous satellite observations based on modern deep convolutional neural network architectures, such as Conditional Generative Adversarial Networks (CGANs), to provide constraints on the vertical distribution of clouds and aerosols, which will also provide more robust constraints on the retrieved physical cloud and aerosol properties.

The derived products will be applied as novel observational constraint on the highly uncertain representation of aerosol-cloud interactions in cloud-resolving and climate models, reducing the overall uncertainty in anthropogenic perturbations to the climate system.

For further information on eligibility criteria and how to apply to this project please see here. Please note that the application deadline is Sunday 16th January 2022. Applications should be made directly to the iMIRACLI website.

Clouds in a changing climate

Philip Stier

Clouds play a key role in the climate system via modulation of the Earth’s energy balance and their role in the hydrological cycle. It is therefore vital to understand the uncertain response of clouds to anthropogenic perturbations in the form of greenhouse-gas induced warming (cloud feedbacks) and the emission of air pollutants (aerosol effects). You will join the dynamical climate processes group tackling these and related questions employing advanced computer models of the atmosphere and Earth’s climate, in synergy with measurements from satellites, aircrafts and ground-based instruments, theory and increasingly machine learning techniques.

Specific DPhil projects will be developed jointly with the student and often combine multiple methodologies and external partners or supervisors. I would be interested to explore joint supervision with our new visiting professors Sue van den Heever (CSU/Oxford) and/or Graeme Stephens (NASA JPL / Oxford) as well as with our machine learning collaborators in the departments of statistics and computer science.

Climate variability and predictability during the 20th Century

Antje Weisheimer

This project will look into the variability of the coupled atmosphere-ocean-sea-ice climate system during the 20th Century which was characterised by complex variations linked to natural variability (internal and external) as well as anthropogenic forcings. A key aim of the project will be to link the multi-decadal climate variability as seen in observational datasets with the fluctuations that state-of-the-art climate forecast models used in operational seasonal forecasts exhibit.

Forecasts of seasonal climate anomalies using physically based global circulation models are routinely made at operational meteorological centres around the world. A crucial component of any seasonal forecast system is the set of retrospective forecasts, or hindcasts, from past years that are used to estimate skill and to calibrate the forecasts. You will be working with a new hindcast data set called Coupled Seasonal Forecasts of the 20th Century (CSF-20C) to improve our understanding of the physical mechanisms in the atmosphere, ocean and sea-ice responsible for the low-frequency modulations of forecast skill that have been found in these hindcasts.

Further reading: Weisheimer et al., 2020 and references therein 

The project requires strong analytical skills in Physics, Mathematics, Meteorology, Environmental sciences or related disciplines and a high level of curiosity to explore the marvels of nature. In return, it offers a wide range of opportunities and applications in the broad field of weather and climate predictions.

Revisiting atmospheric timescales

Tim Woollings

Weather patterns are generally unpredictable beyond a week or so due to the chaotic nature of the atmospheric fluid dynamics. This is the well-known butterfly effect. However, this is an average picture; some events are clearly more predictable and others less so.

This project will revisit the issue of predictability and timescales in the atmosphere using long modern datasets. New methods will be developed to characterise the memory of the atmosphere and these will be used to determine not only the average predictability but also the range of behaviour.

We rely on complex numerical models for predictions of the weather and climate next week, next summer and the next century. A key aim of this project is to assess the memory timescales in current models when compared to observations. Are models under- or over-confident and how much should we trust their projections of future change?

This project requires a strong first degree in either Physics, Mathematics, Engineering, or Earth Science, with a particular emphasis on skills in mathematical methods or computational modelling.

The dynamics of polar weather and its impact on lower latitudes

Tim Woollings

The rapidly warming Arctic is often referred to as the canary in the coal mine, as the impacts of climate change are already striking hard. In addition to the strong local effects of this warming, weather patterns over and surrounding the Arctic are expected to be impacted, including potential effects on the mid-latitude jet streams and storm tracks.

Despite considerable scientific, social and political interest in these changes, we still lack fundamental understanding of the atmospheric fluid dynamics which controls polar weather patterns and their interaction with lower latitudes. This project will investigate this dynamics, using a range of tools including analytical theory, detailed dynamical analysis of observed data and experiments performed with idealised or full complexity climate models. The central questions are what physical processes control the weather and climate variability of the polar regions, and what role will this physics play in the evolving behaviour of our climate system.

This project requires a strong first degree in either Physics, Mathematics, Engineering, or Earth Science, with a particular emphasis on skills in mathematical methods or computational modelling.