BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Causal networks for process-oriented climate model evaluation - Pe
 er Nowack | Imperial College London/University of East Anglia
DTSTART:20210504T100000Z
DTEND:20210504T113000Z
UID:TALK159052@talks.cam.ac.uk
CONTACT:87364
DESCRIPTION:Global climate models are central tools for understanding past
  and future climate change. The assessment of model skill\, in turn\, can 
 benefit from modern data science approaches. Here I will present recent wo
 rk on causal discovery algorithms as a novel approach for process-oriented
  climate model evaluation [1\,2].\n\nFollowing an introduction to the conc
 ept of causal discovery\, I will move on to key scientific implications of
  this new approach when applied to global sea level pressure datasets. Usi
 ng causal networks learned from meteorological reanalysis data (as a proxy
  for observations) and from CMIP5 climate model output\, I demonstrate tha
 t climate models which better reproduce the observed causal information fl
 ow also better reproduce important precipitation patterns over highly popu
 lated areas such as the Indian subcontinent\, Africa\, East Asia\, Europe\
 , and North America. In addition\, the method identifies expected model in
 terdependencies due to shared development backgrounds of many climate mode
 ls. Finally\, I find that causal network metrics provide stronger relation
 ships for constraining precipitation projections under climate change than
  traditional model evaluation metrics. Such emergent relationships highlig
 ht the potential of causal discovery approaches to constrain longstanding 
 uncertainties in climate change projections.\n\nTime allowing\, I will als
 o briefly touch on a few other projects of my group at UEA/Imperial Colleg
 e London. These include recent work on statistical learning approaches to 
 constrain the uncertain role of clouds in global warming\, machine learnin
 g parameterizations for ozone in Earth system models [3\,4]\, low-cost air
  pollution sensor calibrations using machine learning [5]\, and a new bloc
 king detection algorithm using self-organizing maps [6].\n\nReferences:\n\
 n[1] Nowack P\, Runge J\, Eyring V\, Haigh JD. Causal networks for climate
  model evaluation and constrained projections. Nature Communications 11\, 
 1415 (2020).\n[2] Runge J\, Nowack P\, Kretschmer M\, Flaxman S\, Sejdinov
 ic D. Detecting and quantifying causal associations in large nonlinear tim
 e series datasets. Science Advances 5\, eaau4996 (2019).\n[3] Nowack P\, B
 raesicke P\, Haigh J\, Abraham NL\, Pyle JA\, Voulgarakis A. Using machine
  learning to build temperature-based ozone parameterizations for climate s
 ensitivity simulations. Environmental Research Letters 13\, 104016 (2018).
 \n[4] Nowack P\, Ong QYE\, Braesicke P\, Haigh J\, Abraham NL\, Pyle J\, V
 oulgarakis A. Machine learning parameterizations for ozone: climate model 
 transferability. Proceedings of the 9th International Workshop on Climate 
 Informatics 9\, 263-268 (2019).\n[5] Nowack P\, Konstantinovskiy L\, Gardi
 ner H\, Cant J. Towards low-cost and high-performance air pollution measur
 ements using machine learning calibration techniques. Atmospheric Measurem
 ent Techniques Discussions (2020).\n[6] Thomas C\, Voulgarakis A\, Lim G\,
  Haigh J\, Nowack P. An unsupervised learning approach to identifying bloc
 king events: the case of European summer. Weather and Climate Dynamics Dis
 cussions (2021).
LOCATION:https://zoom.us/j/6708259482?pwd=Qk03U3hxZWNJZUZpT2pVZnFtU2RRUT09
END:VEVENT
END:VCALENDAR