Tobias Thornes

profile image

Tobias Thornes

Graduate student (DTP)

I am a DPhil student studying under Tim Palmer and Peter Duben in the Predictability of Weather and Climate Group. I am part of the first cohort of Oxford's cross-disciplinary Doctoral Training Partnership (DTP) in environmental research, and am interested in a variety of environmental and other topics outside my main research. I am currently a student of Oriel college, of which I am greatly fond, having graduated with a Master of Physics degree from University College, Durham in 2014.

My motivation for studying weather and climate is two-fold. The primary motivation, as with all good science, is curiosity: Earth's atmosphere is a complex and fascinating phenomenon to study in its own right, and though we can never understand it fully or predict the weather and climate with complete accuracy, it is a delight for us as a species to be able to describe it more accurately and simulate it with our computer models. Secondly, though, it is vital that we improve our understanding of how weather and climate will be affected by the most serious problem that we face - that of human-induced climate change - for which such models are essential.

Outside my physics research, I enjoy reading, writing and debating on diverse themes such as philosophy, history and politics, and am politically active. I regularly publish articles for online magazines such as 'Bang! Science' and my college's publication the 'Poor Print'. I am originally from the Worcestershire countryside and regularly return there to appreciate the natural world. I greatly value the service provided by my college chapel, where I am one of the Bible clerks.

I am tutor in Second-year Statistical Physics at Oriel College (Hilary and Trinity terms 2016).

My main research is into the possibility of making our models of weather and climate more efficient by, somewhat counterintuitively, reducing the precision or reliability with which computers carry out calculations relating to variables that are resolved on small spatial scales. Such variables are not very well-observed, so it may be a waste of computer resources if they are represented 'exactly', using the standard 64 'bits' of memory, on a computer. It may be more efficient to represent them with only 32 or 16 bits, for example, which would save a lot of computer power using the appropriate hardware. This would then allow us to re-use the computer power that we save to increase the resolution of the models instead, improving the accuracy of the forecast. I test this hypothesis using idealised models such as Lorenz' 1996 model and one based on Surface Quasi-Geostrophic Equations.

I have also published work relating to the Heat Island Effect in Birmingham, and on the impact of climate change on animals (please see the attached publications for more information).