BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20190719T085744Z
LOCATION:HG D 1.1
DTSTART;TZID=Europe/Stockholm:20190613T181500
DTEND;TZID=Europe/Stockholm:20190613T184500
UID:submissions.pasc-conference.org_PASC19_sess118_msa121@linklings.com
SUMMARY:Deep Learning Predicts ENSO
DESCRIPTION:Minisymposium\nComputer Science and Applied Mathematics, Emerg
 ing Application Domains, Climate and Weather, Physics\n\nDeep Learning Pre
 dicts ENSO\n\nHuang, Chen, Wang\n\nThe El Niño-Southern Oscillation
  (ENSO)is one of the most influential coupled air-sea phenomena in earth s
 ystem. Here we explore a novel method for ENSO prediction based on deep le
 arning technology. The network is deep enough with 36 ResNet layers in tot
 al and can be fed with single or multiple physical fields as precursors at
  the same time. Our model is promising to catch spatial dependence of phys
 ical fields and dig the relative importance of the different predictors on
  the evolution of ENSO as well. It outperforms the general performance of 
 conventional prediction models, with higher accuracy and better predictabi
 lity even at longer leading times. On the 6-month limitation, correlation 
 coefficient between predictions and observations can reach 0.85, while RMS
 E be no more than 0.62°C. On the 12-month limitation, correlation coef
 ficient can reach 0.72, while RMSE can be no more than 0.7°C. The deep
  learning model is also effective in exploring the spatio-temporal connect
 ions between precursors and ENSO events, which helps for the understanding
  on ENSO physics.
END:VEVENT
END:VCALENDAR