BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Unsupervised Classification of Convective Organisation with Deep L
 earning - Leif Denby - University of Leeds
DTSTART:20191029T110000Z
DTEND:20191029T120000Z
UID:TALK130411@talks.cam.ac.uk
CONTACT:Jonathan Rosser
DESCRIPTION:The radiative properties of clouds\, and thus their impact on 
 Earth's climate\,\nare significantly affected by how clouds spatially orga
 nise. The precise\nmechanisms which drive different forms of convective or
 ganisation are however\ncurrently unknown. With a tool to automatically cl
 assify regions into distinct\nforms of convective organisation it is possi
 ble to produce a statistical\ndescription of the most likely large-scale a
 nd local environmental conditions\n(e.g. windshear\, horizontal convergenc
 e) present in differently organised\nstates. And further\, it will be poss
 ible to study their temporal evolution and\nquantify differences in behavi
 our of organisation in weather and climate models\nas compared to observat
 ions.\n\nUsing unsupervised learning of GOES-R imagery\, I have developed 
 machine\nlearning model to automatically identify regimes of convective or
 ganisation in\nsatellite imagery. I will show how this can be applied to s
 tudy their radiative\nproperties in the tropical Atlantic and the temporal
  evolution of marine\nstratocumulus.
LOCATION:Bullard Lab\, Seminar Room
END:VEVENT
END:VCALENDAR