The extratropical circulation's variabilities and extremes: old problems, new advances
March 11, 2019
Pedram Hassanzadeh, Visiting from Rice University
Hosted by Dave Thompson
Abstract
The variabilities of the extratropical atmospheric circulation and the resulting extreme weather events have significant societal impacts. However, due to complexities such as eddy-mean flow interaction, important aspects of the extratropical variability such as annular modes and blocking events are still not well understood.In this talk I will first discuss how a novel tool, the linear response function (LRF), can be computed and used to accurately quantify eddy-mean flow interaction in climate models, thus providing new insight into the dynamics of the annular modes and similar modes of climate variability. Focusing on an idealized dry GCM and the barotropic annular mode (the leading empirical orthogonal function of zonal-mean zonal wind), I show that the response of eddy momentum and eddy heat fluxes to the annular mode anomaly at the quasi-steady limit can be quantified using the LRF. Using these results, the existence of a positive eddy-jet feedback is unequivocally confirmed and the magnitude of the feedback, its dynamics, and its contribution to the persistence of the annular mode are quantified. Using a new statistical method designed based on these findings, an improved estimate of the feedback in the reanalysis data of the Southern Hemisphere is presented.
I will also briefly discuss how deep learning techniques can be used for data-driven forecasting of extreme-casting weather patterns such as blocking events. After introducing a new framework that casts extreme weather prediction as a pattern recognition problem (analog forecasting) and uses convolutional neural networks, I show the skills of the framework in forecasting heat waves and cold spells over North America 1 to 5 days ahead in data from NCAR's Large Ensemble simulations.