An unbiased atmospheric climate model
To correct the large biases in state-of-the-art climate models with a ML model of the sub-grid tendencies.
Projects
Samples of published and ongoing projects by the Climate Modelling Research Group. These are also intended to give prospective students project ideas, particularly have a look at those tagged “OPEN” but feel free to reach out with your own ideas too!
Atmospheric modelling, GPU computing, Software engineering, Data compression, Information theory, Scientific computing, Data visualisation, Feels-like temperature, Forecast communication, Machine learning, Numerical weather prediction, Generalization, Automatic differentiation, Exoplanets, OPEN, GPU programming, Stochastic
StochasticRounding.jl: Up or down, maybe both?
Stochastic rounding introduces random numbers in every little +, -, *, /, … with big impacts on large scales of a simulated system
Fast yet flexible: Atmospheric modelling and high-perfomance computing on the GPU
High-resolution weather and climate models are a massive high-performance computing endeavour
Differentiable atmospheric modelling for Earth and exoplanets
Atmospheric general circulation models are the backbone of climate models, being used to understand and predict climate change on Earth and understand climat...
Machine learning in atmospheric models with generalization in space and time
Learning the physics not a hardcoded representation of continents with their specific climates
Heat forecasts
Forecasting feels-like temperatures as a strategy to reduce heat illnesses during sport events
Data compression and bitinformation
Compressing atmospheric data into its real bitwise information content
SpeedyWeather.jl
Play climate modelling like it’s LEGO!