KlimaCampus Kolloquium: (5 Reasons For) The Probabilistic Earth-System Model

In this talk I give five reasons why weather and climate models should be stochastic. Amongst these reasons are: stochastic parametrisation is more consistent with scale-invariant structures in the atmosphere, and with coarse-grained parametrisation studies in high resolution weather forecast models; stochastic parametrisation improves the reliability and probabilistic skill of weather and climate forecasts on all timescales; stochastic parametrisation can reduce systematic errors substantially; the inherent stochasticity of the parametrisation problem opens new possibilities for the development of very high-resolution dynamical cores using fast energy-efficient approximate computing techniques; unlike multi-model ensembles as a means to represent model uncertainty, the stochastic parametrisation concept can help break down inter-institutional barriers and thus help promote international collaboration on model development.




15:15 h


Bundesstr. 53, room 022/023
Seminar Room 022/023, Ground Floor, Bundesstrasse 53, 20146 Hamburg, Hamburg


Tim Palmer, University of Oxford


Bjorn Stevens

Back to listing