Is model complexity holding back our understanding of the aerosol-climate email@example.com
This project will use a number of advanced international models and statistical methods to understand the how increasing the number of atmospheric processes in a climate model affects the uncertainty in the aerosol-climate effect. The PhD will exploit recent major advances in the understanding of computer model uncertainty and the student will have the opportunity to visit and collaborate with international climate modellers. The research will involve the Met Office climate and Earth system model (HadGEM) and other international climate models, combined with new approaches that enable sources of model uncertainty to be quantified efficiently. By combining the model uncertainty information from a number of climate model ensembles the aim will be to understand how model complexity is holding back our understanding of the difference between the models in the aerosol-climate effect.
Background and rationale
Aerosol particles from pollution sources have a net cooling effect on climate that may have offset a large fraction of the warming caused by greenhouse gas emissions. Through all Intergovernmental Panel for Climate Change (IPCC) assessments since 2001 aerosol forcing has remained the largest uncertainty – the climate models don’t agree on the magnitude of the aerosol effect on the climate. The uncertainty in this dominant forcing term limits our ability to determine the ‘climate sensitivity’ (how much the temperature will rise in response to future forcing).
One reason for the slow progress in aerosol modelling is that global climate models are highly complex and computationally demanding. They are therefore typically run only a few times, which means it is not possible to understand whether additional complexity truly leads to ‘better’ model performance. The Leeds aerosol group has developed some unique approaches to overcome this limitation. By building statistical emulators it has been possible to run a complex global aerosol model essentially many thousands of times. This enables the effect of all the aerosol uncertainties to be quantified and separated. These approaches are described in a series of papers (see below), and the aerosol group now has two statisticians working alongside the aerosol and climate modellers. The new statistical approaches have enabled us to quantify how model uncertainty affects the prediction of aerosols around the world. We extend this by analyzing how the complexity of the model can be reduced without increasing its uncertainty. “Everything should be made as simple as possible, but not simpler”, Albert Einstein.
The big challenge for this PhD is to understand why climate models may agree on the global mean climate in the present-day the way even though the way in which they simulate that climate is known to differ and how this may lead to differences in the predictions of future climate. The student will have access to modeled aerosol data from various international research groups designed by Leeds specifically to investigate model uncertainty. The student will use statistical tools to analyse this data evaluating and understanding complexity in the aerosol predictions of each of the international models, and thereby help to reduce the uncertainty in the aerosol-climate relationship.
Key research questions
- What are the major differences in how global models simulate aerosols?
- How does the complexity of a climate model affect estimates of aerosol radiative forcing of climate and the response of the climate system to these forcings?
- How can the research community make aerosol models that are as simple as possible, but not simpler? What is gained by adding complexity, making the model slower, and reducing the number of times it can be run?
The project will use several advanced modeling tools and datasets:
- A number of international climate models used in previous IPCC assessments including advanced simulation of global aerosols.
- New statistical approaches to quantify and understand sources of uncertainty in complex models. These approaches have been developed by statisticians in the aerosol group. The PhD student could extend these methodologies or use them ‘out of the box’ to understand the model complexities.
Undergraduate training in any physical/chemical science, computing, mathematics or applied statistics would be appropriate. In this sort of project it is possible to explore a particular line of research depending on experience and interest. For example, someone with a mathematical or statistics background may want to explore more advanced techniques for the statistical evaluation of the models. Other students may be more interested to exploit existing model information to explore in more detail the causes and consequences of model uncertainty and error.
Students will receive training in running and visualizing global model results. Dr Kirsty Pringle in the aerosol group is a dedicated research and support scientist who leads the technical aspects of the model development. Specific training in statistical methods and computing will be provided as necessary.
The PhD student will be a member of the aerosol modeling research group in the Institute for Climate and Atmospheric Science (http://goo.gl/pzvwVw). This group has a strong track record of producing excellent postgraduates, several of whom have won prizes and other commendations for their theses (see http://goo.gl/S52pfT). Students are strongly encouraged, and helped, to produce publications during their PhD time (examples for past students are given at http://goo.gl/017jQ0). There is a high level of cooperation in the group, and students quickly learn to use and extend the advanced models. The student will also have the opportunity to collaborate with a wide range of international collaborators associated with the group. Come and visit us and talk to our existing and past PhD students.
Further reading / bibliography
A full list of the group’s papers is available at http://www.see.leeds.ac.uk/people/k.carslaw. Some relevant papers include:
Lee LA; Reddington CL; Carslaw KS (2016) On the relationship between aerosol model uncertainty and radiative forcing uncertainty, Proceedings of the National Academy of Sciences of the United States of America, 113, pp.5820-5827. doi: 10.1073/pnas.1507050113
Lee LA; Pringle KJ; Reddington CL; Mann GW; Spracklen DV; Carslaw KS; Stier P; Pierce JR (2013) The magnitude and causes of uncertainty in global model simulations of cloud condensation nuclei, Atmospheric Chemistry and Physics, 13, pp.8879-8914. doi: 10.5194/acp-13-8879-2013
Lee LA; Carslaw KS; Pringle KJ; Mann GW (2012) Mapping the uncertainty in global CCN using emulation, Atmospheric Chemistry and Physics, 12, pp.9739-9751. doi: 10.5194/acp-12-9739-2012
Lee LA; Carslaw KS; Pringle KJ; Mann GW; Spracklen DV (2011) Emulation of a complex global aerosol model to quantify sensitivity to uncertain parameters, Atmospheric Chemistry and Physics, 11, pp.12253-12273. doi: 10.5194/acp-11-12253-2011
Carslaw KS; Lee LA; Reddington CL; Pringle KJ; Rap A; Forster PM; Mann GW; Spracklen DV; Woodhouse MT; Regayre LA; Pierce JR (2013) Large contribution of natural aerosols to uncertainty in indirect forcing, Nature, 503, pp.67-+. doi: 10.1038/nature12674
The aerosol group: http://www.see.leeds.ac.uk/aerosol
The Global Aerosol Synthesis and Science Project: http://gassp.org.uk
Related undergraduate subjects:
- Applied mathematics
- Environmental science