The project consists of developing a machine learning method for finding reliable and non-spurious emergent constraints. Emergent constraints is a technique to reduce the uncertainty of future climate predictions. This is achieved by finding a linear relationship in an ensemble of climate models, between a measure of an observable or trend and the sensitivity in climate projections. Finding non-spurious and reliable emergent constraints is a hard task for conventional techniques, due to the complexity of the Earth system dynamics and the high-dimensionality of climate data.
Xavier Andoni Tibau Alberdi
Veronika Eyring (DLR IPA, Uni Bremen)
Joachim Denzler (FSU Jena)
Markus Reichstein (MPI Biogeochemistry)