Deep learning for multi-year ENSO forecasts

doi:10.1038/s41586-019-1559-7

IEECAS OpenIR > 黄土与第四纪地质国家重点实验室（2010~）

	Deep learning for multi-year ENSO forecasts
	Ham, Yoo-Geun 1; Kim, Jeong-Hwan 1; Luo, Jing-Jia 2,3
通讯作者	Ham, Yoo-Geun(ygham@jnu.ac.kr)
	2019-09-26
发表期刊	NATURE
ISSN	0028-0836
卷号	573 期号:7775 页码:568-+
摘要	Variations in the El Nino/Southern Oscillation (ENSO) are associated with a wide array of regional climate extremes and ecosystem impacts(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(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(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.
DOI	10.1038/s41586-019-1559-7
关键词[WOS]	EL-NINO ; INDIAN-OCEAN ; PREDICTION ; FLAVORS
收录类别	SCI ; SCI
语种	英语
资助项目	Korea Meteorological Administration Research and Development Program[KMI2018-03214] ; Basic Science Research Program through the National Research Foundation of Korea (NRF) - Ministry of Education[NRF-2016R1A6A1A03012647] ; 'The Startup Foundation for Introducing Talent' of NUIST
WOS研究方向	Science & Technology - Other Topics
项目资助者	Korea Meteorological Administration Research and Development Program ; Basic Science Research Program through the National Research Foundation of Korea (NRF) - Ministry of Education ; 'The Startup Foundation for Introducing Talent' of NUIST
WOS类目	Multidisciplinary Sciences
WOS记录号	WOS:000488247600055
出版者	NATURE PUBLISHING GROUP
引用统计	被引频次：517[WOS] [WOS记录] [WOS相关记录]
文献类型	期刊论文
条目标识符	http://ir.ieecas.cn/handle/361006/13361
专题	黄土与第四纪地质国家重点实验室（2010~）
通讯作者	Ham, Yoo-Geun
作者单位	1.Chonnam Natl Univ, Dept Oceanog, Gwangju, South Korea 2.Nanjing Univ Informat Sci & Technol, Inst Climate & Applicat Res ICAR CICFEM KLME ILCE, Nanjing, Jiangsu, Peoples R China 3.Chinese Acad Sci, Inst Earth Environm, SKLLQG, Xian, Shaanxi, Peoples R China
推荐引用方式 GB/T 7714	Ham, Yoo-Geun,Kim, Jeong-Hwan,Luo, Jing-Jia. Deep learning for multi-year ENSO forecasts[J]. NATURE,2019,573(7775):568-+.
APA	Ham, Yoo-Geun,Kim, Jeong-Hwan,&Luo, Jing-Jia.(2019).Deep learning for multi-year ENSO forecasts.NATURE,573(7775),568-+.
MLA	Ham, Yoo-Geun,et al."Deep learning for multi-year ENSO forecasts".NATURE 573.7775(2019):568-+.