BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Artificial Intelligence Pathways from Weather to Climate - Tom  Be
 ucler (Université de Lausanne)
DTSTART:20260209T141500Z
DTEND:20260209T151500Z
UID:TALK242293@talks.cam.ac.uk
DESCRIPTION:Author list: Tom Beucler (UNIL)\, David Neelin (UCLA)\, Hui Su
  (HKUST)\, Ignacio Lopez-Gomez (Google Research)\, Chris Bretherton\, Oliv
 er Watt-Meyer (AI2)\, Tapio Schneider\, Costa Christopoulos (Caltech)\, Wi
 ll Chapman (UC Boulder)\, Laure Zanna (NYU)\, Aditya Grover (UCLA)\, Adam 
 Subel (NYU)\nDeep learning emulates atmospheric reanalyses with high fidel
 ity\, enabling increasingly well-calibrated ensemble weather forecasts at 
 progressively longer lead times. To extend these gains to climate-relevant
  horizons\, AI prediction systems must produce credible forced responses t
 o drivers of interest (e.g.\, greenhouse gases\, land-use change). We prop
 ose a minimal\, testable framework for AI climate modeling: (i) represent 
 external forcings explicitly and restrict them to physically appropriate s
 tate tendencies\; and (ii) stress-test robustness in out-of-distribution r
 egimes\, including extremes and counterfactual trajectories.\nUsing leadin
 g climate emulators and hybrid physics-AI models\, we identify coupling an
 d development challenges and compare scaling with resolution and effective
  complexity. AI models do not appear intrinsically more efficient than GPU
 -ported dynamical models once complexity is accounted for\, yet they can d
 irectly predict target variables at the desired grid without integrating t
 he full high-frequency\, multivariate state. Diverse ML downscaling strate
 gies can partially substitute for explicit fine-scale resolution when obse
 rvations are available\, paving the way towards inexpensive\, local risk a
 ssessment across prediction horizons.
LOCATION:Seminar Room 1\, Newton Institute
END:VEVENT
END:VCALENDAR