S2S reboot: An argument for greater inclusion of machine learning in subseasonal to seasonal forecasts

Judah Cohen, Dim Coumou, Jessica Hwang, Lester Mackey, Paulo Orenstein, Sonja Totz, Eli Tziperman

Research output: Contribution to JournalReview articleAcademicpeer-review

Abstract

The discipline of seasonal climate prediction began as an exercise in simple statistical techniques. However, today the large government forecast centers almost exclusively rely on complex fully coupled dynamical forecast systems for their subseasonal to seasonal (S2S) predictions while statistical techniques are mostly neglected and those techniques still in use have not been updated in decades. In this Opinion Article, we argue that new statistical techniques mostly developed outside the field of climate science, collectively referred to as machine learning, can be adopted by climate forecasters to increase the accuracy of S2S predictions. We present an example of where unsupervised learning demonstrates higher accuracy in a seasonal prediction than the state-of-the-art dynamical systems. We also summarize some relevant machine learning methods that are most applicable to climate prediction. Finally, we show by comparing real-time dynamical model forecasts with observations from winter 2017/2018 that dynamical model forecasts are almost entirely insensitive to polar vortex (PV) variability and the impact on sensible weather. Instead, statistical forecasts more accurately predicted the resultant sensible weather from a mid-winter PV disruption than the dynamical forecasts. The important implication from the poor dynamical forecasts is that if Arctic change influences mid-latitude weather through PV variability, then the ability of dynamical models to demonstrate the existence of such a pathway is compromised. We conclude by suggesting that S2S prediction will be most beneficial to the public by incorporating mixed or a hybrid of dynamical forecasts and updated statistical techniques such as machine learning. This article is categorized under: Climate Models and Modeling > Knowledge Generation with Models.

Original languageEnglish
Article numbere00567
Pages (from-to)1-15
Number of pages15
JournalWiley Interdisciplinary Reviews. Climate Change
Volume10
Issue number2
Early online date18 Dec 2018
DOIs
Publication statusPublished - Mar 2019

Keywords

  • climate prediction
  • machine learning
  • polar vortex
  • unsupervised learning

Cite this

Cohen, Judah ; Coumou, Dim ; Hwang, Jessica ; Mackey, Lester ; Orenstein, Paulo ; Totz, Sonja ; Tziperman, Eli. / S2S reboot : An argument for greater inclusion of machine learning in subseasonal to seasonal forecasts. In: Wiley Interdisciplinary Reviews. Climate Change. 2019 ; Vol. 10, No. 2. pp. 1-15.
@article{7d4fc8050bf94985a2d78f6fbd5bc6a3,
title = "S2S reboot: An argument for greater inclusion of machine learning in subseasonal to seasonal forecasts",
abstract = "The discipline of seasonal climate prediction began as an exercise in simple statistical techniques. However, today the large government forecast centers almost exclusively rely on complex fully coupled dynamical forecast systems for their subseasonal to seasonal (S2S) predictions while statistical techniques are mostly neglected and those techniques still in use have not been updated in decades. In this Opinion Article, we argue that new statistical techniques mostly developed outside the field of climate science, collectively referred to as machine learning, can be adopted by climate forecasters to increase the accuracy of S2S predictions. We present an example of where unsupervised learning demonstrates higher accuracy in a seasonal prediction than the state-of-the-art dynamical systems. We also summarize some relevant machine learning methods that are most applicable to climate prediction. Finally, we show by comparing real-time dynamical model forecasts with observations from winter 2017/2018 that dynamical model forecasts are almost entirely insensitive to polar vortex (PV) variability and the impact on sensible weather. Instead, statistical forecasts more accurately predicted the resultant sensible weather from a mid-winter PV disruption than the dynamical forecasts. The important implication from the poor dynamical forecasts is that if Arctic change influences mid-latitude weather through PV variability, then the ability of dynamical models to demonstrate the existence of such a pathway is compromised. We conclude by suggesting that S2S prediction will be most beneficial to the public by incorporating mixed or a hybrid of dynamical forecasts and updated statistical techniques such as machine learning. This article is categorized under: Climate Models and Modeling > Knowledge Generation with Models.",
keywords = "climate prediction, machine learning, polar vortex, unsupervised learning",
author = "Judah Cohen and Dim Coumou and Jessica Hwang and Lester Mackey and Paulo Orenstein and Sonja Totz and Eli Tziperman",
year = "2019",
month = "3",
doi = "10.1002/wcc.567",
language = "English",
volume = "10",
pages = "1--15",
journal = "Wiley Interdisciplinary Reviews. Climate Change",
issn = "1757-7799",
publisher = "John Wiley and Sons Inc.",
number = "2",

}

S2S reboot : An argument for greater inclusion of machine learning in subseasonal to seasonal forecasts. / Cohen, Judah; Coumou, Dim; Hwang, Jessica; Mackey, Lester; Orenstein, Paulo; Totz, Sonja; Tziperman, Eli.

In: Wiley Interdisciplinary Reviews. Climate Change, Vol. 10, No. 2, e00567, 03.2019, p. 1-15.

Research output: Contribution to JournalReview articleAcademicpeer-review

TY - JOUR

T1 - S2S reboot

T2 - An argument for greater inclusion of machine learning in subseasonal to seasonal forecasts

AU - Cohen, Judah

AU - Coumou, Dim

AU - Hwang, Jessica

AU - Mackey, Lester

AU - Orenstein, Paulo

AU - Totz, Sonja

AU - Tziperman, Eli

PY - 2019/3

Y1 - 2019/3

N2 - The discipline of seasonal climate prediction began as an exercise in simple statistical techniques. However, today the large government forecast centers almost exclusively rely on complex fully coupled dynamical forecast systems for their subseasonal to seasonal (S2S) predictions while statistical techniques are mostly neglected and those techniques still in use have not been updated in decades. In this Opinion Article, we argue that new statistical techniques mostly developed outside the field of climate science, collectively referred to as machine learning, can be adopted by climate forecasters to increase the accuracy of S2S predictions. We present an example of where unsupervised learning demonstrates higher accuracy in a seasonal prediction than the state-of-the-art dynamical systems. We also summarize some relevant machine learning methods that are most applicable to climate prediction. Finally, we show by comparing real-time dynamical model forecasts with observations from winter 2017/2018 that dynamical model forecasts are almost entirely insensitive to polar vortex (PV) variability and the impact on sensible weather. Instead, statistical forecasts more accurately predicted the resultant sensible weather from a mid-winter PV disruption than the dynamical forecasts. The important implication from the poor dynamical forecasts is that if Arctic change influences mid-latitude weather through PV variability, then the ability of dynamical models to demonstrate the existence of such a pathway is compromised. We conclude by suggesting that S2S prediction will be most beneficial to the public by incorporating mixed or a hybrid of dynamical forecasts and updated statistical techniques such as machine learning. This article is categorized under: Climate Models and Modeling > Knowledge Generation with Models.

AB - The discipline of seasonal climate prediction began as an exercise in simple statistical techniques. However, today the large government forecast centers almost exclusively rely on complex fully coupled dynamical forecast systems for their subseasonal to seasonal (S2S) predictions while statistical techniques are mostly neglected and those techniques still in use have not been updated in decades. In this Opinion Article, we argue that new statistical techniques mostly developed outside the field of climate science, collectively referred to as machine learning, can be adopted by climate forecasters to increase the accuracy of S2S predictions. We present an example of where unsupervised learning demonstrates higher accuracy in a seasonal prediction than the state-of-the-art dynamical systems. We also summarize some relevant machine learning methods that are most applicable to climate prediction. Finally, we show by comparing real-time dynamical model forecasts with observations from winter 2017/2018 that dynamical model forecasts are almost entirely insensitive to polar vortex (PV) variability and the impact on sensible weather. Instead, statistical forecasts more accurately predicted the resultant sensible weather from a mid-winter PV disruption than the dynamical forecasts. The important implication from the poor dynamical forecasts is that if Arctic change influences mid-latitude weather through PV variability, then the ability of dynamical models to demonstrate the existence of such a pathway is compromised. We conclude by suggesting that S2S prediction will be most beneficial to the public by incorporating mixed or a hybrid of dynamical forecasts and updated statistical techniques such as machine learning. This article is categorized under: Climate Models and Modeling > Knowledge Generation with Models.

KW - climate prediction

KW - machine learning

KW - polar vortex

KW - unsupervised learning

UR - http://www.scopus.com/inward/record.url?scp=85058864589&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85058864589&partnerID=8YFLogxK

U2 - 10.1002/wcc.567

DO - 10.1002/wcc.567

M3 - Review article

VL - 10

SP - 1

EP - 15

JO - Wiley Interdisciplinary Reviews. Climate Change

JF - Wiley Interdisciplinary Reviews. Climate Change

SN - 1757-7799

IS - 2

M1 - e00567

ER -