基于 DSTFN(Deep Spatio-Temprral Fusion Network)模型的热带气旋轨迹预测方法

方巍; 杜娟; 齐媚涵; 胡鹏昱

doi:10.16032/j.issn.1004-4965.2024.077

基于 DSTFN(Deep Spatio-Temprral Fusion Network)模型的热带气旋轨迹预测方法

doi: 10.16032/j.issn.1004-4965.2024.077

1.
南京信息工程大学计算机学院/数字取证教育部工程研究中心，江苏南京 210044
2.
南京气象科技创新研究院中国气象局交通气象重点开放实验室，江苏南京 210041
3.
南京信息工程大学江苏省大气环境与装备技术协同创新中心，江苏南京 210044
4.
苏州大学江苏省计算机信息处理技术重点实验室，江苏苏州 215000
5.
上海农林职业技术学院，上海 201699

基金项目:

国家自然科学基金项目 42075007

苏州大学江苏省计算机信息处理技术重点实验室开放研究基金 KJS2275

中国气象局交通气象重点开放实验室开放研究基金 BJG202306

中国气象局流域强降水重点开放实验室开放研究基金 2023BHR-Y14

江苏省研究生科研与实践创新计划项目 SJCX24_0476

江苏省研究生科研与实践创新计划项目 SJCX24_0477

详细信息

通讯作者:
方巍，男，安徽省人，教授，博士研究生导师，主要从事台风预报等气象人工智能研究。E-mail:hsfangwei@sina. com

中图分类号: P444
计量
- 文章访问数: 10
- HTML全文浏览量: 4
- PDF下载量: 1
- 被引次数: 0
出版历程
- 收稿日期: 2023-10-19
- 修回日期: 2024-08-18
- 网络出版日期: 2025-03-28
- 刊出日期: 2024-12-20

Tropical Cyclone Track Prediction Method Based on DSTFN Model

1.
School of Computer Science/Engineering Research Center of Digital Forensics of Ministry of Education, Nanjing University of Information Science & Technology, Nanjing 210044, China
2.
Key Laboratory of Transportation Meteorology of China Meteorological Administration, Nanjing Joint Institute for Atmospheric Sciences, Nanjing 210041, China
3.
Jiangsu Collaborative Innovation Center of Atmospheric Environment and Equipment Technology (CICAEET), Nanjing University of Information Science & Technology, Nanjing 210044, China
4.
Provincial Key Laboratory for Computer Information Processing Technology, Soochow University, Suzhou, Jiangsu 215000, China
5.
Shanghai Vocational College of Agriculture and Forestry, Shanghai 201699, China)

摘要

摘要: 在全球气候变化背景下，越来越多的地区面临着热带气旋的威胁。因此，准确预测热带气旋的轨迹变化对于气象预警和灾害管理至关重要。然而，传统的基于深度学习的热带气旋预测方法在建模热带气旋的时空相关性方面存在局限。为此，提出了一种新的深度时空融合网络——DSTFN(Deep Spatio-Temporal Fusion Network)模型，以提高对热带气旋轨迹的预测精度和稳定性。构建了有效融合ConvNeXt(Convolutional Next) 模型和门控循环单元的 CaConvNeXt - GRU(Convolutional Block Attention Module Integrated with ConvNeXt and Gated Recurrent Unit)模型，以提取热带气旋三维时序数据中的复杂非线性时空特征。同时，引入了卷积块注意力模块，以自动聚焦不同等压面对热带气旋影响更大的特征。此外，设计了分阶段的训练策略，通过依次进行预训练、联合训练和整体训练实现了不同模块的有效融合。为了评估所设计的方法，在国际气候管理最佳路径档案和第五代大气再分析数据集上进行了大量实验。实验结果证明，在预测未来24 h的热带气旋轨迹时，相比于现有的基于深度学习的热带气旋轨迹预测模型，DSTFN模型的平均预测误差降低了约13.71 km。
- 热带气旋 /
- 路径预测 /
- DSTFN模型 /
- CaConvNeXt-GRU模型 /
- 时空序列预测
Abstract: In the context of global climate change, more and more regions are facing the threat of tropical cyclones. Therefore, accurate prediction of changes in the tracks of tropical cyclones is essential for meteorological warning and disaster reduction. However, existing tropical cyclone prediction methods based on deep learning have limitations in modeling the spatio-temporal correlation of tropical cyclones. In the present study, we proposed a new deep spatio-temporal fusion network (DSTFN) model to improve the prediction accuracy and stability of tropical cyclone tracks. We developed the CaConvNeXt-GRU model, which effectively integrated the ConvNeXt model and the gated recurrent unit, to extract complex nonlinear spatio-temporal features in the 3D time series data of tropical cyclones. Meanwhile, the convolutional block attention module was introduced to automatically focus on the features that were affected more heavily by different isobaric surfaces on tropical cyclones. Moreover, we designed a staged training strategy to realize the effective integration of different modules through pre-training, joint training, and ovERAll training. To evaluate the proposed model, we conducted extensive experiments on the International Best Track Archive for Climate Stewardship (IBTrACS) and the ERA5 dataset. OvERAll, in predicting tropical cyclone tracks for the next 24 hours, the DSTFN model reduced the avERAge prediction error by about 13.71 km compared to existing tropical cyclone track prediction models based on deep learning.
- tropical cyclone /
- path prediction /
- DSTFN model /
- CaConvNeXt-GRU model /
- spatio-temporal series prediction

HTML全文

1 CaConvNeXt-GRU结构

下载: 全尺寸图片幻灯片

2 空间特征提取模块结构

下载: 全尺寸图片幻灯片

3 下采样模块结构

下载: 全尺寸图片幻灯片

4 DSTFN模型结构

下载: 全尺寸图片幻灯片

5 一个气压层上的热带气旋二维数据结构

下载: 全尺寸图片幻灯片

6 热带气旋及其周边环境数据的三维时序结构

下载: 全尺寸图片幻灯片

7 台风“天鸽”预测误差可视化

下载: 全尺寸图片幻灯片

8 台风“山竹”预测误差可视化

下载: 全尺寸图片幻灯片

9 台风“莫拉菲”预测误差可视化

下载: 全尺寸图片幻灯片

1 热带气旋特征数据

字段记号	字段	备注
sid	热带气旋编号	表示热带气旋的唯一标识符
iso_time	热带气旋发生时间	使用世界时坐标(utc)，格式为yyyy-mm-dd hh: mm: ss，时间点以6 h为间隔
usa_lat	热带气旋中心纬度	热带气旋中心点的地理纬度，以度为单位
usa_lon	热带气旋中心经度	热带气旋中心点的地理经度，以度为单位
usa_pres	热带气旋中心经度	热带气旋中心点的地理经度，以度为单位
usa_wind	热带气旋中心风速	表示热带气旋中心点的最大持续风速
dist2land	热带气旋中心位置与陆地距离	表示热带气旋中心点到最近陆地的距离

下载: 导出CSV

2 卷积网络分支模块24 h预测距离误差

方法	与真实点的误差距离/km
XGBoost	205.54
LSTM	200.64
GRU	190.54
Dlinear	193.99
卷积网络分支模块	180.57

下载: 导出CSV

3 CaConvNeXt-GRU分支模块24 h预测距离误差

方法	与真实点的误差距离/km
2DCNN	176.77
Inception	171.07
ConvLSTM	168.84
Multi-ConvGRU	165.54
CaConvNeXt-GRU分支模块(无卷积块注意力模块)	164.87
CaConvNeXt-GRU分支模块	162.63

下载: 导出CSV

4 DSTFN模型24 h预测距离误差

方法	与真实点的误差距离/km
CaConvNeXt-GRU分支模块(1)	183.49
卷积网络分支模块	180.57
CaConvNeXt-GRU分支模块(2)	162.63
GBRNN	145.38
1DCNN+2DCNN	140.45
1DCNN+ Multi-ConvGRU + CBAM	134.34
DSTFN模型(无预训练)	134.89
DSTFN模型	131.67
CMA	74.69

下载: 导出CSV

参考文献(40)

[1]	顾小丽, 钱燕珍, 周伟军, 等 . 台风 “利奇马” 灾害风险及气象服务效益评估[j]. 热带气象学报, 2022, 38(1): 35-42.
[2]	李艳, 符彩芳, 金茹. 西北太平洋近赤道热带气旋生成的特征分析[j]. 大气科学学报, 2019, 42(5): 695-704.
[3]	Global increase in major tropical cyclone exceedance probability over the past four decades[J]. Proceedings of the National Academy of Sciences, 2020, 117(22): 11 975-11 980.
[4]	ROBERTS M J, CAMP J, SEDDON J, et al. Impact of model resolution on tropical cyclone simulation using the HighResMIPPRIMAVERA multimodel ensemble[J]. Journal of Climate, 2020, 33(7): 2 557-2 583.
[5]	麻素红, 张进, 沈学顺, 等. 2016年grapes_tym改进及对台风预报影响[j]. 应用气象学报, 2018, 29(3): 257-269.
[6]	PALMER T, STEVENS B. The scientific challenge of understanding and estimating climate change[J]. Proceedings of the National Academy of Sciences, 2019, 116(49): 24 390-24 395.
[7]	王相军, 李秋胜. enso事件影响下登陆中国热带气旋年际变化特征的分位数回归分析[j]. 热带气象学报, 2022, 38(1): 11-22.
[8]	QIAN Y, HSU P C, MURAKAMI H, et al. A hybrid dynamical‐statistical model for advancing subseasonal tropical cyclone prediction overthe western North Pacific[J]. Geophysical Research Letters, 2020, 47(20): e2020GL090095.
[9]	CHEN R, ZHANG W, WANG X. Machine learning in tropical cyclone forecast modeling: A review[J]. Atmosphere, 2020, 11(7): 676
[10]	CHEN R, WANG X, ZHANG W, et al. A hybrid CNN-LSTM model for typhoon formation forecasting[J]. GeoInformatica, 2019, 23(3): 375-396.
[11]	LIU Z, MAO H, WU C Y, et al. A convnet for the 2020s[C]//Proceedings of the IEEE /CVF conference on computer vision and pattern recognition. 2022: 11 976-11 986.
[12]	BELLO I, FEDUS W, DU X, et al. Revisiting resnets: Improved training and scaling strategies[J]. Advances in Neural Information Processing Systems, 2021, 34: 22 614-22 627.
[13]	LIU Z, LIN Y, CAO Y, et al. Swin transformer: Hierarchical vision transformer using shifted windows[C]//Proceedings of the IEEE/CVF international conference on computer vision. 2021: 10 012-10 022.
[14]	WANG P, WANG P, WANG C, et al. Using a 3D convolutional neural network and gated recurrent unit for tropical cyclone track forecasting [J]. Atmospheric Research, 2022, 269: 106053.
[15]	FANG W, LU W, LI J, et al. A Novel Tropical Cyclone Track Forecast Model Based on Attention Mechanism[J]. Atmosphere, 2022, 13(10): 1607.
[16]	GIFFARD-ROISIN S, YANG M, CHARPIAT G, et al. Tropical cyclone track forecasting using fused deep learning from aligned reanalysis data[J]. Frontiers in big Data, 2020, 3: 1.
[17]	XU G, XIAN D, FOURNIER-VIGER P, et al. AM-ConvGRU: a spatio-temporal model for typhoon path prediction[J]. Neural Computing and Applications, 2022, 34(8): 5 905-5 921.
[18]	ZHANG Y, LI K, LI K, et al. Image super-resolution using very deep residual channel attention networks[C]//Proceedings of the European Conference on Computer Vision (ECCV). 2018: 286-301.
[19]	HERSBACH H, BELL B, BERRISFORD P, et al. The ERA5 global reanalysis[J]. Quarterly Journal of the Royal Meteorological Society, 2020, 146(730): 1 999-2 049.
[20]	KNAPP K R, KRUK M C, LEVINSON D H, et al. The international best track archive for climate stewardship (IBTrACS) unifying tropical cyclone data[J]. Bulletin of the American Meteorological Society, 2010, 91(3): 363-376.
[21]	WOO S, PARK J, LEE J Y, et al. Cbam: Convolutional block attention module[C]//Proceedings of the European conference on computer vision (ECCV). 2018: 3-19.
[22]	CHEN J H, LIN S J, MAGNUSSON L, et al. Advancements in hurricane prediction with NOAA's next ‐ generation forecast system[J]. Geophysical Research Letters, 2019, 46(8): 4 495-4 501
[23]	Bechdtold P. Challenges in tropical numerical weather prediction at ECMWF[C]//Current trends in the representation of physical processes in weather and climate models. Singapore: Springer, 2019: 29-50.
[24]	CHEN N, SUN H, ZHANG Q, et al. A Short-Term Wind Speed Forecasting Model Based on EMD/CEEMD and ARIMA-SVM Algorithms [J]. Applied Sciences, 2022, 12(12): 6085.
[25]	陈春, 陶丽. 热带气旋潜在生成指数的对比分析及其在西北太平洋的改进[j]. 大气科学学报, 2023, 46(4): 615-629.
[26]	HSAN T Z, SEIN M M. Combining Support Vector Machine and Polynomial Regressing to Predict Tropical Cyclone Track[C]//2021 IEEE 3rd Global Conference on Life Sciences and Technologies (LifeTech). IEEE, 2021: 220-221.
[27]	TAN J, CHEN S, WANG J. Western North Pacific tropical cyclone track forecasts by a machine learning model[J]. StochasticEnvironmental Research and Risk Assessment, 2021, 35(6): 1 113-1 126.
[28]	WANG C, XU Q, LI X, et al. CNN-based tropical cyclone track forecasting from satellite infrared images[C]//IGARSS 2020-2020 IEEE International Geoscience and Remote Sensing Symposium. IEEE, 2020: 5 811-5 814.
[29]	Alemany S, Beltran J, Perez A, et al. Predicting hurricane trajectories using a recurrent neural network[C]//Proceedings of the AAAI Conference on Artificial Intelligence. 2019, 33(1): 468-475.
[30]	GAO S, ZHAO P, PAN B, et al. A nowcasting model for the prediction of typhoon tracks based on a long short term memory neural network [J]. Acta Oceanologica Sinica, 2018, 37(5): 8-12.
[31]	LIAN J, DONG P, ZHANG Y, et al. A novel deep learning approach for tropical cyclone track prediction based on auto-encoder and gated recurrent unit networks[J]. Applied Sciences, 2020, 10(11): 3965.
[32]	SONG T, LI Y, MENG F, et al. A novel deep learning model by Bigru with attention mechanism for tropical cyclone track prediction in the Northwest Pacific[J]. Journal of Applied Meteorology and Climatology, 2022, 61(1): 3-12.
[33]	TONG B, WANG X, FU J Y, et al. Short-term prediction of the intensity and track of tropical cyclone via ConvLSTM model[J]. Journal of Wind Engineering and Industrial Aerodynamics, 2022, 226: 105026.
[34]	SU X. Using deep learning model for meteorological satellite cloud image prediction[C]//AGU Fall Meeting Abstracts. 2017, 2017: IN13B- 0064
[35]	LUI Y S, TSE L K S, TAM C Y, et al. Performance of MPAS-A and WRF in predicting and simulating western North Pacific tropical cyclone tracks and intensities[J]. Theoretical and Applied Climatology, 2021, 143(1): 505-520.
[36]	TIPPETT M K, CAMARGO S J, SOBEL A H. A Poisson regression index for tropical cyclone genesis and the role of large-scale vorticity in genesis[J]. Journal of Climate, 2011, 24(9): 2 335-2 357.
[37]	NATH S, KOTAL S D. Seasonal prediction of tropical cyclone activity over the North Indian Ocean using the neural network model[J]. Atmósfera, 2015, 28(4): 271-281.
[38]	CHEN T, GUESTRIN C. Xgboost: A scalable tree boosting system[C]//Proceedings of the 22nd acm sigkdd international conference on knowledge discovery and data mining. 2016: 785-794.
[39]	ZENG A, CHEN M, ZHANG L, et al. Are transformers effective for time series forecasting?[C]//Proceedings of the AAAI conference on artificial intelligence. 2023, 37(9): 11 121-11 128.
[40]	LU X, YU H, YING M, et al. Western North Pacific tropical cyclone database created by the China Meteorological Administration[J]. Advances in Atmospheric Sciences, 2021, 38(4): 690-699.