BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:From Machine Learning Parameterization to Full Model Emulation - D
 avid John Gagne - Head of Machine Integration and Learning for Earth Syste
 ms\, NCAR
DTSTART:20240503T100000Z
DTEND:20240503T110000Z
UID:TALK216202@talks.cam.ac.uk
CONTACT:Jack Atkinson
DESCRIPTION:The Machine Integration and Learning for Earth Systems (MILES)
  group at the US NSF National Center for Atmospheric Research collaborates
  with groups across the weather/climate spectrum to develop physics-inform
 ed machine learning systems that integrate closely with existing community
  physics-based models. One of our first emulation projects focused on the 
 development of a spectral bin warm-rain microphysics emulator for the Comm
 unity Atmospheric Model. In recent work\, we have reduced our original 7 n
 eural network model to 1 and 3 neural network versions that we have integr
 ated into CAM with a fortran-based neural network inference framework. We 
 have also developed a machine learning surface layer parameterization base
 d on observed near surface atmospheric conditions and surface fluxes. We h
 ave tested this model within WRF and the FastEddy GPU LES model to underst
 and some of the sensitivities and important variables. Finally\, I will di
 scuss some preliminary results from our new full atmospheric model emulato
 r\, the Community Runnable Earth Digital Intelligence Twin (CREDIT). The m
 odel is trained on ERA5 model level output and can perform stable 1-hour r
 ollouts to 10 days with a single model. We intend the framework to support
  community training and inference of AI weather models.\n\nHybrid particip
 ants can join using the following zoom link: https://cam-ac-uk.zoom.us/j/8
 8550555680?pwd=VHgwMW8zRXJ4UTBnUExSSW10NXNjUT09
LOCATION:MR5 DAMTP and online
END:VEVENT
END:VCALENDAR