Faculty Candidate Seminar - Michael Pritchard

03/30/2016 - 2:30pm
Engineering Bldg Unit 2, Room 479 (Main Campus)
Event Description: 

DATE:     March 30th

LOCATION:  2:30pm Engineering Bldg Unit 2, room 479 (main campus)
SPEAKER: Michael Pritchard, Ph.D.
UC Irivine

TITLE:   Confronting root problems of climate hazard prediction with cloud-permitting algorithms amidst a paradigm shift in modern supercomputer hardware

I will argue that the critical path to improving climate hazard prediction is fixing root problems in representing clouds and turbulence in global climate models, emphasizing the “cloud superparameterization” paradigm as a promising strategy for progress. In this approach, thousands of convection-permitting models are embedded in a planetary model of the atmosphere. The algorithm scales well and takes better advantage of large distributed supercomputers than typical methods.  I will focus on what this approach has revealed about the emerging hazard of an amplified Madden-Julian Oscillation (MJO) with global warming. The MJO is a planetary-scale pattern of rainfall in the tropical atmosphere that modulates damaging tropical cyclones and atmospheric rivers. The problem is the MJO is poorly understood and has proved difficult to simulate in standard climate models. I will show how hypothesis-driven interventions in a superparameterized climate model can help reveal the dynamics of the MJO towards understanding the plausibility of its future amplification.  I will conclude with a vision for how to improve predictions of two other especially uncertain climate hazards –the climate response of marine stratocumulus clouds, which dominates uncertainty of overall climate sensitivity, and the future response of tropical cyclone intensity. I will argue that these problems can now be brought into the range of explicit simulation by vastly expanding the superparameterization paradigm, using new accelerator algorithms and software engineering to exploit current trends in high performance computing. 
Faculty Hosts:    Dan Cayan (dcayan@ucsd.edu) & Keiko Nomura (knomura@ucsd.edu)
For more information on this event, contact: 
Leslie Costi
Event Calendar: