The application of deep learning is fast developing in climate prediction,in which El Ni?o–Southern Oscillation(ENSO),as the most dominant disaster-causing climate event,is a key target.Previous studies have shown th...The application of deep learning is fast developing in climate prediction,in which El Ni?o–Southern Oscillation(ENSO),as the most dominant disaster-causing climate event,is a key target.Previous studies have shown that deep learning methods possess a certain level of superiority in predicting ENSO indices.The present study develops a deep learning model for predicting the spatial pattern of sea surface temperature anomalies(SSTAs)in the equatorial Pacific by training a convolutional neural network(CNN)model with historical simulations from CMIP6 models.Compared with dynamical models,the CNN model has higher skill in predicting the SSTAs in the equatorial western-central Pacific,but not in the eastern Pacific.The CNN model can successfully capture the small-scale precursors in the initial SSTAs for the development of central Pacific ENSO to distinguish the spatial mode up to a lead time of seven months.A fusion model combining the predictions of the CNN model and the dynamical models achieves higher skill than each of them for both central and eastern Pacific ENSO.展开更多
Recent studies have shown that deep learning(DL)models can skillfully forecast El Niño–Southern Oscillation(ENSO)events more than 1.5 years in advance.However,concerns regarding the reliability of predictions ma...Recent studies have shown that deep learning(DL)models can skillfully forecast El Niño–Southern Oscillation(ENSO)events more than 1.5 years in advance.However,concerns regarding the reliability of predictions made by DL methods persist,including potential overfitting issues and lack of interpretability.Here,we propose ResoNet,a DL model that combines CNN(convolutional neural network)and transformer architectures.This hybrid architecture enables our model to adequately capture local sea surface temperature anomalies as well as long-range inter-basin interactions across oceans.We show that ResoNet can robustly predict ENSO at lead times of 19 months,thus outperforming existing approaches in terms of the forecast horizon.According to an explainability method applied to ResoNet predictions of El Niño and La Niña from 1-to 18-month leads,we find that it predicts the Niño-3.4 index based on multiple physically reasonable mechanisms,such as the recharge oscillator concept,seasonal footprint mechanism,and Indian Ocean capacitor effect.Moreover,we demonstrate for the first time that the asymmetry between El Niño and La Niña development can be captured by ResoNet.Our results could help to alleviate skepticism about applying DL models for ENSO prediction and encourage more attempts to discover and predict climate phenomena using AI methods.展开更多
模式分辨率对气候模式的模拟效果具有重要影响。然而,当前模式开发对于垂直分辨率的重视不够。以ENSO(厄尔尼诺-南方涛动)遥相关为例,利用CESM(Community Earth System Model)模式,探究不同模式垂直分辨率设置下模式模拟的ENSO对平流层...模式分辨率对气候模式的模拟效果具有重要影响。然而,当前模式开发对于垂直分辨率的重视不够。以ENSO(厄尔尼诺-南方涛动)遥相关为例,利用CESM(Community Earth System Model)模式,探究不同模式垂直分辨率设置下模式模拟的ENSO对平流层、对流层影响的差异,评估模式垂直分辨率在气候模拟中的重要性。结果表明,提高垂直分辨率可以显著改进模式对ENSO遥相关的模拟能力。以ECMWF(European Centre for Medium-Range Weather Forecasts)第五代再分析数据集(ERA5)为参照,ENSO对纬向平均温度的影响在北半球中高纬地区冬季呈现出“负正负”的三极子模态。CESM默认的垂直分辨率设置(L66)不能模拟出这一模态,而提高模式垂直分辨率(L103)后则可以较好地模拟出这个模态。对于水平分布而言,L66模拟的ENSO在对流层的信号与再分析资料相比明显偏强,L103则可以显著改善。同时,L103对ENSO影响平流层的模拟效果也比L66有所改善。进一步分析发现,L103模拟的行星波从对流层向平流层的传播更强,更接近再分析资料。提高垂直分辨率可以改善模式对大气波活动以及平流层-对流层动力耦合的模拟,重视模式的研发。展开更多
海洋热浪(简称MHWs)是在一定海域内发生的海表温度异常偏高的现象,本文利用在再分析数据资料,分析了ENSO对中国近海40年间(1982~2023年)冬季MHWs事件的影响。长期趋势上,El Niño年冬季中国近海各海域MHWs分布较平均,La Niña...海洋热浪(简称MHWs)是在一定海域内发生的海表温度异常偏高的现象,本文利用在再分析数据资料,分析了ENSO对中国近海40年间(1982~2023年)冬季MHWs事件的影响。长期趋势上,El Niño年冬季中国近海各海域MHWs分布较平均,La Niña年不同海域发生MHWs差异较大。El Niño、La Niña年总发生天数和平均持续时间在空间分布上几乎相反。发生El Niño事件后,中国近海都发生了大量的MHWs,La Niña事件后,MHWs发生很少甚至没有发生。同时,ENSO影响西太副高的加强西伸,对冬季MHWs的发生密切相关,环流异常会增大南海北部潜热通量与南部的短波辐射通量,促进MHWs发生。The phenomenon of marine heatwaves (MHWs) refers to abnormally high sea surface temperatures in specific marine areas. This paper utilizes reanalysis data to investigates the influence of El Niño-Southern Oscillation (ENSO) on winter MHWs in offshore China over the past 40 years (1982~2023). In terms of long-term trends, the distribution of MHWs in offshore China during El Niño winters exhibits relative uniformity, while significant variations are observed among different regions during La Niña winters. Spatially, there exists an inverse relationship between the total occurrence days and average duration of MHWs during El Niño and La Niña events. Subsequent to an El Niño event, a substantial number of MHWs occur in China’s coastal waters;however, following a La Niña event, occurrences of MHWs are infrequent or even absent. Furthermore, ENSO influences the westward extension and intensification of the West Pacific subtropical high pressure system and demonstrates close associations with winter occurrences of MHWs. Circulation anomalies enhance latent heat fluxes in northern South China Sea regions and short-wave radiation fluxes in southern South China Sea regions, thereby promoting the development of MHWs.展开更多
文章利用中国气象局发布的热带气旋(TC)最佳路径数据资料,以及NCEP/NCAR大气再分析数据,分析了盛季(7~9月份)、后季(10~11月份)西北太平洋(WNP)TC累积气旋能量(ACE)与ENSO事件相关性的年代际变化。发现WNP TC盛季ACE与ENSO的相关性在198...文章利用中国气象局发布的热带气旋(TC)最佳路径数据资料,以及NCEP/NCAR大气再分析数据,分析了盛季(7~9月份)、后季(10~11月份)西北太平洋(WNP)TC累积气旋能量(ACE)与ENSO事件相关性的年代际变化。发现WNP TC盛季ACE与ENSO的相关性在1980年发生了年代际的突变,二者相关性由不显著变为显著的正相关。TC后季ACE与ENSO的相关性在1990年前后发生了年代际的突变,相关性由不相关变为显著的正相关。盛季、后季二者相关性年代际转变发生的事件不同,影响机制也不同。盛季ACE与ENSO相关性发生转变的主要原因是连续型ENSO在1980年之前发生频次较高,减弱了ENSO对ACE的影响。后季ACE与ENSO相关性发生转变的原因主要是1990年之前,El Niño多为东太平洋型,在WNP激发的环流异常为偶极子型分布,不能影响ACE的总量,1990年之后,El Niño发生时异常对流的位置偏西,WNP大部分区域被气旋式异常环流控制,有利于TC的生成和加强,因此TC ACE与ENSO有较好的相关性。This paper utilizes the best track data of tropical cyclones from the China Meteorological Administration, along with NCEP/NCAR reanalysis atmospheric data, to analyze the decadal shift in the relationship between the Accumulated Cyclone Energy (ACE) of tropical cyclones (TC) in the Northwest Pacific (WNP) during the peak (July-September) and late (October-November) seasons and ENSO events. It was found that the correlation between WNP TC ACE during the peak season and ENSO underwent a decadal shift in 1980, changing from non-significant to a significant positive correlation. The correlation between late-season TC ACE and ENSO also experienced a decadal shift around 1990, changing from no correlation to a significant positive correlation. The decadal shifts in correlation during the peak and late seasons occurred at different times and were driven by different mechanisms. The shift in the correlation between peak season TC ACE and ENSO is primarily due to the high frequency of continuous-type ENSO events before 1980, which weakened the influence of ENSO on ACE. The reason for the shift in the correlation between late-season TC ACE and ENSO is mainly because before 1990, El Niño was predominantly of the Eastern Pacific type, and the circulation anomalies it triggered in the WNP were of a dipole distribution, which did not affect the total ACE. After 1990, the position of anomalous convection during El Niño events shifted westward, and most of the WNP was controlled by cyclonic anomaly circulation, which was conducive to the generation and strengthening of TCs, hence the better correlation between TC ACE and ENSO.展开更多
基金supported by the National Key R&D Program of China(Grant No.2019YFA0606703)the National Natural Science Foundation of China(Grant No.41975116)the Youth Innovation Promotion Association of the Chinese Academy of Sciences(Grant No.Y202025)。
文摘The application of deep learning is fast developing in climate prediction,in which El Ni?o–Southern Oscillation(ENSO),as the most dominant disaster-causing climate event,is a key target.Previous studies have shown that deep learning methods possess a certain level of superiority in predicting ENSO indices.The present study develops a deep learning model for predicting the spatial pattern of sea surface temperature anomalies(SSTAs)in the equatorial Pacific by training a convolutional neural network(CNN)model with historical simulations from CMIP6 models.Compared with dynamical models,the CNN model has higher skill in predicting the SSTAs in the equatorial western-central Pacific,but not in the eastern Pacific.The CNN model can successfully capture the small-scale precursors in the initial SSTAs for the development of central Pacific ENSO to distinguish the spatial mode up to a lead time of seven months.A fusion model combining the predictions of the CNN model and the dynamical models achieves higher skill than each of them for both central and eastern Pacific ENSO.
基金supported by the Shanghai Artificial Intelligence Laboratory and National Natural Science Foundation of China(Grant No.42088101 and 42030605).
文摘Recent studies have shown that deep learning(DL)models can skillfully forecast El Niño–Southern Oscillation(ENSO)events more than 1.5 years in advance.However,concerns regarding the reliability of predictions made by DL methods persist,including potential overfitting issues and lack of interpretability.Here,we propose ResoNet,a DL model that combines CNN(convolutional neural network)and transformer architectures.This hybrid architecture enables our model to adequately capture local sea surface temperature anomalies as well as long-range inter-basin interactions across oceans.We show that ResoNet can robustly predict ENSO at lead times of 19 months,thus outperforming existing approaches in terms of the forecast horizon.According to an explainability method applied to ResoNet predictions of El Niño and La Niña from 1-to 18-month leads,we find that it predicts the Niño-3.4 index based on multiple physically reasonable mechanisms,such as the recharge oscillator concept,seasonal footprint mechanism,and Indian Ocean capacitor effect.Moreover,we demonstrate for the first time that the asymmetry between El Niño and La Niña development can be captured by ResoNet.Our results could help to alleviate skepticism about applying DL models for ENSO prediction and encourage more attempts to discover and predict climate phenomena using AI methods.
文摘模式分辨率对气候模式的模拟效果具有重要影响。然而,当前模式开发对于垂直分辨率的重视不够。以ENSO(厄尔尼诺-南方涛动)遥相关为例,利用CESM(Community Earth System Model)模式,探究不同模式垂直分辨率设置下模式模拟的ENSO对平流层、对流层影响的差异,评估模式垂直分辨率在气候模拟中的重要性。结果表明,提高垂直分辨率可以显著改进模式对ENSO遥相关的模拟能力。以ECMWF(European Centre for Medium-Range Weather Forecasts)第五代再分析数据集(ERA5)为参照,ENSO对纬向平均温度的影响在北半球中高纬地区冬季呈现出“负正负”的三极子模态。CESM默认的垂直分辨率设置(L66)不能模拟出这一模态,而提高模式垂直分辨率(L103)后则可以较好地模拟出这个模态。对于水平分布而言,L66模拟的ENSO在对流层的信号与再分析资料相比明显偏强,L103则可以显著改善。同时,L103对ENSO影响平流层的模拟效果也比L66有所改善。进一步分析发现,L103模拟的行星波从对流层向平流层的传播更强,更接近再分析资料。提高垂直分辨率可以改善模式对大气波活动以及平流层-对流层动力耦合的模拟,重视模式的研发。
文摘海洋热浪(简称MHWs)是在一定海域内发生的海表温度异常偏高的现象,本文利用在再分析数据资料,分析了ENSO对中国近海40年间(1982~2023年)冬季MHWs事件的影响。长期趋势上,El Niño年冬季中国近海各海域MHWs分布较平均,La Niña年不同海域发生MHWs差异较大。El Niño、La Niña年总发生天数和平均持续时间在空间分布上几乎相反。发生El Niño事件后,中国近海都发生了大量的MHWs,La Niña事件后,MHWs发生很少甚至没有发生。同时,ENSO影响西太副高的加强西伸,对冬季MHWs的发生密切相关,环流异常会增大南海北部潜热通量与南部的短波辐射通量,促进MHWs发生。The phenomenon of marine heatwaves (MHWs) refers to abnormally high sea surface temperatures in specific marine areas. This paper utilizes reanalysis data to investigates the influence of El Niño-Southern Oscillation (ENSO) on winter MHWs in offshore China over the past 40 years (1982~2023). In terms of long-term trends, the distribution of MHWs in offshore China during El Niño winters exhibits relative uniformity, while significant variations are observed among different regions during La Niña winters. Spatially, there exists an inverse relationship between the total occurrence days and average duration of MHWs during El Niño and La Niña events. Subsequent to an El Niño event, a substantial number of MHWs occur in China’s coastal waters;however, following a La Niña event, occurrences of MHWs are infrequent or even absent. Furthermore, ENSO influences the westward extension and intensification of the West Pacific subtropical high pressure system and demonstrates close associations with winter occurrences of MHWs. Circulation anomalies enhance latent heat fluxes in northern South China Sea regions and short-wave radiation fluxes in southern South China Sea regions, thereby promoting the development of MHWs.
文摘文章利用中国气象局发布的热带气旋(TC)最佳路径数据资料,以及NCEP/NCAR大气再分析数据,分析了盛季(7~9月份)、后季(10~11月份)西北太平洋(WNP)TC累积气旋能量(ACE)与ENSO事件相关性的年代际变化。发现WNP TC盛季ACE与ENSO的相关性在1980年发生了年代际的突变,二者相关性由不显著变为显著的正相关。TC后季ACE与ENSO的相关性在1990年前后发生了年代际的突变,相关性由不相关变为显著的正相关。盛季、后季二者相关性年代际转变发生的事件不同,影响机制也不同。盛季ACE与ENSO相关性发生转变的主要原因是连续型ENSO在1980年之前发生频次较高,减弱了ENSO对ACE的影响。后季ACE与ENSO相关性发生转变的原因主要是1990年之前,El Niño多为东太平洋型,在WNP激发的环流异常为偶极子型分布,不能影响ACE的总量,1990年之后,El Niño发生时异常对流的位置偏西,WNP大部分区域被气旋式异常环流控制,有利于TC的生成和加强,因此TC ACE与ENSO有较好的相关性。This paper utilizes the best track data of tropical cyclones from the China Meteorological Administration, along with NCEP/NCAR reanalysis atmospheric data, to analyze the decadal shift in the relationship between the Accumulated Cyclone Energy (ACE) of tropical cyclones (TC) in the Northwest Pacific (WNP) during the peak (July-September) and late (October-November) seasons and ENSO events. It was found that the correlation between WNP TC ACE during the peak season and ENSO underwent a decadal shift in 1980, changing from non-significant to a significant positive correlation. The correlation between late-season TC ACE and ENSO also experienced a decadal shift around 1990, changing from no correlation to a significant positive correlation. The decadal shifts in correlation during the peak and late seasons occurred at different times and were driven by different mechanisms. The shift in the correlation between peak season TC ACE and ENSO is primarily due to the high frequency of continuous-type ENSO events before 1980, which weakened the influence of ENSO on ACE. The reason for the shift in the correlation between late-season TC ACE and ENSO is mainly because before 1990, El Niño was predominantly of the Eastern Pacific type, and the circulation anomalies it triggered in the WNP were of a dipole distribution, which did not affect the total ACE. After 1990, the position of anomalous convection during El Niño events shifted westward, and most of the WNP was controlled by cyclonic anomaly circulation, which was conducive to the generation and strengthening of TCs, hence the better correlation between TC ACE and ENSO.
