Skip to main content

Events

CGAFD Seminar: Machine learning for climate simulations

Machine learning for climate simulations


Event details

Abstract

In recent years, the progress of artificial intelligence in generating weather forecasts has been incredibly impressive.  How artificial intelligence will be useful on climate time scales is still unclear.  In this talk we present two potential applications for climate developed at the Met Office.  The first demonstrates use of a neural network to mimic the behaviour of one of the sub-grid parameterization schemes used in global climate models: the non-orographic gravity wave scheme.  A climate model simulation, using the neural network in place of the existing parameterization scheme, is found to accurately generate a quasi-biennial oscillation of the tropical stratospheric winds, and correctly simulate the non-orographic gravity wave variability associated with the El Niño–Southern Oscillation and stratospheric polar vortex variability. These internal sources of variability are essential for providing seasonal forecast skill, and the gravity wave forcing associated with them is reproduced without explicit training for these patterns.  The second application concerns the flow dependent bias correction of future seasonal forecasts, with the aim of improving forecast skill.  A neural network is trained to learn appropriate bias corrections to atmospheric wind and temperature fields in the Met Office seasonal forecast system.  This work is ongoing, and the current results of simulations run with climatological bias correction, and flow dependent bias correction using the neural network, will be presented.

 

Location:

Amory C417