[OANNES Foro] Deep learning for multi-year ENSO forecasts

[Mario Cabrejos]

 <https://www.nature.com/nature> Nature volume 573, pages 568–572 (2019) 

 
<https://www.nature.com/articles/s41586-019-1559-7?WT.ec_id=NATURE-20190926&
utm_source=nature_etoc&utm_medium=email&utm_campaign=20190926&sap-outbound-i
d=65009000AFAE4CFBDC842073DE65D49CD7308762&utm_source=hybris-campaign&utm_me
dium=email&utm_campaign=000_SKN6563_0000015362_41586-Nature-20190926-EAlert&
utm_content=EN_internal_33925_20190926#article-info> Published: 18 September
2019

https://www.nature.com/articles/s41586-019-1559-7?WT.ec_id=NATURE-20190926&u
tm_source=nature_etoc&utm_medium=email&utm_campaign=20190926&sap-outbound-id
=65009000AFAE4CFBDC842073DE65D49CD7308762&utm_source=hybris-campaign&utm_med
ium=email&utm_campaign=000_SKN6563_0000015362_41586-Nature-20190926-EAlert&u
tm_content=EN_internal_33925_20190926

 

Deep learning for multi-year ENSO forecasts

 
<https://www.nature.com/articles/s41586-019-1559-7?WT.ec_id=NATURE-20190926&
utm_source=nature_etoc&utm_medium=email&utm_campaign=20190926&sap-outbound-i
d=65009000AFAE4CFBDC842073DE65D49CD7308762&utm_source=hybris-campaign&utm_me
dium=email&utm_campaign=000_SKN6563_0000015362_41586-Nature-20190926-EAlert&
utm_content=EN_internal_33925_20190926#auth-1> Yoo-Geun Ham,
<https://www.nature.com/articles/s41586-019-1559-7?WT.ec_id=NATURE-20190926&
utm_source=nature_etoc&utm_medium=email&utm_campaign=20190926&sap-outbound-i
d=65009000AFAE4CFBDC842073DE65D49CD7308762&utm_source=hybris-campaign&utm_me
dium=email&utm_campaign=000_SKN6563_0000015362_41586-Nature-20190926-EAlert&
utm_content=EN_internal_33925_20190926#auth-2> Jeong-Hwan Kim &
<https://www.nature.com/articles/s41586-019-1559-7?WT.ec_id=NATURE-20190926&
utm_source=nature_etoc&utm_medium=email&utm_campaign=20190926&sap-outbound-i
d=65009000AFAE4CFBDC842073DE65D49CD7308762&utm_source=hybris-campaign&utm_me
dium=email&utm_campaign=000_SKN6563_0000015362_41586-Nature-20190926-EAlert&
utm_content=EN_internal_33925_20190926#auth-3> Jing-Jia Luo 

 

Abstract

 

Variations in the El Niño/Southern Oscillation (ENSO) are associated with a
wide array of regional climate extremes and ecosystem impacts
<https://www.nature.com/articles/s41586-019-1559-7#ref-CR1> 1. Robust,
long-lead forecasts would therefore be valuable for managing policy
responses. But despite decades of effort, forecasting ENSO events at lead
times of more than one year remains problematic
<https://www.nature.com/articles/s41586-019-1559-7#ref-CR2> 2. Here we show
that a statistical forecast model employing a deep-learning approach
produces skilful ENSO forecasts for lead times of up to one and a half
years. To circumvent the limited amount of observation data, we use transfer
learning to train a convolutional neural network (CNN) first on historical
simulations <https://www.nature.com/articles/s41586-019-1559-7#ref-CR3> 3
and subsequently on reanalysis from 1871 to 1973. During the validation
period from 1984 to 2017, the all-season correlation skill of the Nino3.4
index of the CNN model is much higher than those of current state-of-the-art
dynamical forecast systems. The CNN model is also better at predicting the
detailed zonal distribution of sea surface temperatures, overcoming a
weakness of dynamical forecast models. A heat map analysis indicates that
the CNN model predicts ENSO events using physically reasonable precursors.
The CNN model is thus a powerful tool for both the prediction of ENSO events
and for the analysis of their associated complex mechanisms.

 

------------ próxima parte ------------
Se ha borrado un adjunto en formato HTML...
URL: <http://lista.oannes.org.pe/pipermail/oannes-oannes.org.pe/attachments/20191126/bf290567/attachment.html>