In recent years,deep learning methods have gradually been applied to prediction tasks related to Arctic sea ice concentration,but relatively little research has been conducted for larger spatial and temporal scales,ma...In recent years,deep learning methods have gradually been applied to prediction tasks related to Arctic sea ice concentration,but relatively little research has been conducted for larger spatial and temporal scales,mainly due to the limited time coverage of observations and reanalysis data.Meanwhile,deep learning predictions of sea ice thickness(SIT)have yet to receive ample attention.In this study,two data-driven deep learning(DL)models are built based on the ConvLSTM and fully convolutional U-net(FC-Unet)algorithms and trained using CMIP6 historical simulations for transfer learning and fine-tuned using reanalysis/observations.These models enable monthly predictions of Arctic SIT without considering the complex physical processes involved.Through comprehensive assessments of prediction skills by season and region,the results suggest that using a broader set of CMIP6 data for transfer learning,as well as incorporating multiple climate variables as predictors,contribute to better prediction results,although both DL models can effectively predict the spatiotemporal features of SIT anomalies.Regarding the predicted SIT anomalies of the FC-Unet model,the spatial correlations with reanalysis reach an average level of 89%over all months,while the temporal anomaly correlation coefficients are close to unity in most cases.The models also demonstrate robust performances in predicting SIT and SIE during extreme events.The effectiveness and reliability of the proposed deep transfer learning models in predicting Arctic SIT can facilitate more accurate pan-Arctic predictions,aiding climate change research and real-time business applications.展开更多
Utilizing the Community Atmosphere Model,version 4,the influence of Arctic sea-ice concentration(SIC)on the extended-range prediction of three simulated cold events(CEs)in East Asia is investigated.Numerical results s...Utilizing the Community Atmosphere Model,version 4,the influence of Arctic sea-ice concentration(SIC)on the extended-range prediction of three simulated cold events(CEs)in East Asia is investigated.Numerical results show that the Arctic SIC is crucial for the extended-range prediction of CEs in East Asia.The conditional nonlinear optimal perturbation approach is adopted to identify the optimal Arctic SIC perturbations with the largest influence on CE prediction on the extended-range time scale.It shows that the optimal SIC perturbations are more inclined to weaken the CEs and cause large prediction errors in the fourth pentad,as compared with random SIC perturbations under the same constraint.Further diagnosis reveals that the optimal SIC perturbations first modulate the local temperature through the diabatic process,and then influence the remote temperature by horizontal advection and vertical convection terms.Consequently,the optimal SIC perturbations trigger a warming center in East Asia through the propagation of Rossby wave trains,leading to the largest prediction uncertainty of the CEs in the fourth pentad.These results may provide scientific support for targeted observation of Arctic SIC to improve the extended-range CE prediction skill.展开更多
文摘随着全球气候变暖加剧,北极地区的大气海洋环境剧烈变化,导致海冰变化更加不稳定,使得海冰预测的难度增大。本研究选择海表温度、2 m平均气温、二氧化碳浓度为大气海洋变量,海冰范围距平为时序特征参数,将上述参量作为北极海冰范围(Sea Ice Extent,SIE)的预测要素,建立了面向SIE的多变量长短期记忆(Long Short Term Memory,LSTM)神经网络模型,对比分析了2015-2021年不同时间序列预测模型的预测结果。结果显示:本研究所构建模型的RMSE、MAE、MAPE分别为0.353×106 km2、0.261×106 km2和3.191%。相比于其他预测模型,结合大气海洋变量和时序特征参数后的LSTM模型预测结果误差更小,拟合效果更好,可以消除夏季海冰剧烈变化对预测效果的影响,提高海冰范围的预测精度,对北极航道的通航安全保障工作具有重要的研究与应用价值。
基金supported by the National Natural Science Foundation of China(Grant Nos.41976193 and 42176243).
文摘In recent years,deep learning methods have gradually been applied to prediction tasks related to Arctic sea ice concentration,but relatively little research has been conducted for larger spatial and temporal scales,mainly due to the limited time coverage of observations and reanalysis data.Meanwhile,deep learning predictions of sea ice thickness(SIT)have yet to receive ample attention.In this study,two data-driven deep learning(DL)models are built based on the ConvLSTM and fully convolutional U-net(FC-Unet)algorithms and trained using CMIP6 historical simulations for transfer learning and fine-tuned using reanalysis/observations.These models enable monthly predictions of Arctic SIT without considering the complex physical processes involved.Through comprehensive assessments of prediction skills by season and region,the results suggest that using a broader set of CMIP6 data for transfer learning,as well as incorporating multiple climate variables as predictors,contribute to better prediction results,although both DL models can effectively predict the spatiotemporal features of SIT anomalies.Regarding the predicted SIT anomalies of the FC-Unet model,the spatial correlations with reanalysis reach an average level of 89%over all months,while the temporal anomaly correlation coefficients are close to unity in most cases.The models also demonstrate robust performances in predicting SIT and SIE during extreme events.The effectiveness and reliability of the proposed deep transfer learning models in predicting Arctic SIT can facilitate more accurate pan-Arctic predictions,aiding climate change research and real-time business applications.
基金the National Natural Science Foundation of China(Grant Nos.42288101,41790475,42175051,and 42005046)the State Key Laboratory of Tropical Oceanography(South China Sea Institute of Oceanology,Chinese Academy of Sciences+1 种基金Grant No.LTO2109)the Guangdong Basic and Applied Basic Research Foundation(Grant No.2021A1515011868).
文摘Utilizing the Community Atmosphere Model,version 4,the influence of Arctic sea-ice concentration(SIC)on the extended-range prediction of three simulated cold events(CEs)in East Asia is investigated.Numerical results show that the Arctic SIC is crucial for the extended-range prediction of CEs in East Asia.The conditional nonlinear optimal perturbation approach is adopted to identify the optimal Arctic SIC perturbations with the largest influence on CE prediction on the extended-range time scale.It shows that the optimal SIC perturbations are more inclined to weaken the CEs and cause large prediction errors in the fourth pentad,as compared with random SIC perturbations under the same constraint.Further diagnosis reveals that the optimal SIC perturbations first modulate the local temperature through the diabatic process,and then influence the remote temperature by horizontal advection and vertical convection terms.Consequently,the optimal SIC perturbations trigger a warming center in East Asia through the propagation of Rossby wave trains,leading to the largest prediction uncertainty of the CEs in the fourth pentad.These results may provide scientific support for targeted observation of Arctic SIC to improve the extended-range CE prediction skill.
