基于机器学习的广东海域模式风暴潮数据订正及评估-海洋预报

基于机器学习的广东海域模式风暴潮数据订正及评估

单位：1. 河海大学海洋学院, 江苏南京 210098;
2. 国家海洋环境预报中心, 北京 100081

分类号：P731.23

出版年·卷·期（页码）：2026·43·第三期（31-41）

摘要：

研究基于欧洲中期天气预报中心的ERA5再分析风场与海平面气压数据，驱动ADCIRC模式构建了广东省2024年两次风暴潮过程的逐小时风暴潮数据集。针对ADCIRC模式在广东沿岸模拟存在的系统误差，发展了一种基于随机森林回归的机器学习订正方法。通过构建包含模拟增水、潮位、风增水、风速等17项物理特征及其衍生交互项的特征体系，建立了具有明确物理意义的误差订正模型。训练结果表明：模型在1 377个样本上表现出良好的学习能力，训练集与测试集的R²分别为0.72和0.65,RMSE分别为16.69和21.24，相关系数分别达到0.86和0.81。实例测试显示：模型对2024年9月过程的高风速个例的改进率为11.65%，对2024年11月过程的低风速个例的改进率达48.08%，平均改进率为29.87%。该机器学习订正方法能有效降低ADCIRC模式的系统误差，在低风速条件下订正效果尤为显著。

This study utilized the ERA5 reanalysis wind fields and sea level pressure data from the European Centre for Medium-Range Weather Forecasts（ECMWF） to drive the ADCIRC model, constructing an hourly storm surge dataset for two storm surge processes in Guangdong coastal waters in 2024. To correct the systematic errors in the storm surge simulations, by constructing a feature system comprising 17 physical features including simulated surge, tidal level, wind-induced surge, wind speed, and their derived interaction terms, a machine learning error correction model based on Random Forest regression with clear physical significance was established. Training results demonstrated the model's strong learning capability on 1 377 samples, with an R² of 0.71 and 0.65, an RMSE of 16.7 and 21.2, and correlation coefficients of 0.86 and 0.81 for the training and test sets, respectively. Case studies showed that the model achieved improvement rates of 11.65% for the high-windspeed case in September 2024 and 48.08% for the low-wind-speed case in November 2024, with an average improvement rate of 29.87%. The research suggests that this machine learning correction method effectively reduces systematic errors in the ADCIRC model, particularly demonstrating stronger correction capability under low-wind-speed conditions.

参考文献：

[1] 于福江,傅赐福,郭洪琳,等.现代风暴潮预报技术及应用[M].北京:科学出版社, 2020.YU F J, FU C F, GUO H L, et al. Modern technologies and application in storm surge forecasting[M]. Beijing:Science Press,2020.
[2] 冯士筰.风暴潮导论[M].北京:科学出版社, 1982.FENG S Z. Introduction to storm surge[M]. Beijing:Science Press,1982.
[3] 付翔,梁森栋,郭洪琳,等.中国沿海40年台风风暴潮特征研究[J].海洋预报, 2023, 40(6):1-11.FU X, LIANG S D, GUO H L, et al. Characteristic analysis of tropical storm surges affecting the coastal area of China in the past40 years[J]. Marine Forecasts, 2023, 40(6):1-11.
[4] 马永锋,祖子清,李涛,等.风暴潮增水长期模拟结果的订正方法研究[J].海洋预报, 2015, 32(5):35-44.MA Y F, ZU Z Q, LI T, et al. Study of the correction method on long-term simulations of storm surge elevation[J]. Marine Forecasts, 2015, 32(5):35-44.
[5] 熊杰,傅赐福,于福江,等.基于ECMWF细网格风场的台风“烟花”与飓风“艾达”风暴潮数值预报检验与分析[J].海洋预报,2022, 39(3):1-9.XIONG J, FU C F, YU F J, et al. Verification and analysis of numerical forecast of typhoon"In-Fa"and hurricane"Ida"storm surge based on ECMWF fine grid wind field[J]. Marine Forecasts,2022, 39(3):1-9.
[6] 侯琪,李晓敏,张杰,等.基于随机森林的台风风暴潮强度估计方法[J].信息技术, 2024(9):84-90.HOU Q, LI X M, ZHANG J, et al. Typhoon storm surge intensity estimation method based on Random Forest[J]. Information Technology, 2024(9):84-90.
[7] 尤成,于福江,原野.基于相空间重构的神经网络风暴潮增水预测方法[J].海洋预报, 2016, 33(1):59-64.YOU C, YU F J, YUAN Y. Storm surge prediction method of neural network based on phase space reconstruction[J]. Marine Forecasts, 2016, 33(1):59-64.
[8] 刘珍,莫冬雪,方泳,等.基于神经网络的台风风暴潮区域预报方法研究:以黄渤海为例[J].海洋与湖沼, 2025, 56(5):1058-1072.LIU Z, MO D X, FANG Y, et al. On regional forecasting method of typhoon storm surge based on neural network:a case study of the Yellow Sea and Bohai Sea[J]. Oceanologia et Limnologia Sinica,2025, 56(5):1058-1072.
[9] 张娟,周水华,黄宝霞,等.人工神经网络在台风风暴潮模拟中的解释应用[J].海洋预报, 2016, 33(2):60-65.ZHANG J, ZHOU S H, HUANG B X, et al. Interpretation of numerical storm surge model results using the artificial neural network[J]. Marine Forecasts, 2016, 33(2):60-65.
[10] LIU Y, ZHAO Q S, HU C C, et al. Prediction of storm surge water level based on machine learning methods[J]. Atmosphere, 2023,14(10):1568.
[11] AYYAD M, HAJJ M R, MARSOOLI R. Machine learning-based assessment of storm surge in the New York metropolitan area[J].Scientific Reports, 2022, 12(1):19215.
[12] CHEN K, KUANG C P, WANG L, et al. Storm surge prediction based on long short-term memory neural network in the East China Sea[J]. Applied Sciences, 2021, 12(1):181.
[13] DULLAART J C M, MUIS S, BLOEMENDAAL N, et al.Advancing global storm surge modelling using the new ERA5climate reanalysis[J]. Climate Dynamics, 2020, 54(1-2):1007-1021.
[14] WESTERINK J J, LUETTICH R A, FEYEN J C, et al. A basin-to channel-scale unstructured grid hurricane storm surge model applied to southern Louisiana[J]. Monthly Weather Review, 2008,136(3):833-864.
[15] LUETTICH JR R A, WESTERINK J J, SCHEFFNER N W.ADCIRC:an advanced three-dimensional circulation model for shelves, coasts, and estuaries. Report 1. Theory and methodology of ADCIRC-2DDI and ADCIRC-3DL[R]. 1992.
[16] DIETRICH J C, ZIJLEMA M, WESTERINK J J, et al. Modeling hurricane waves and storm surge using integrally-coupled,scalable computations[J]. Coastal Engineering, 2011, 58(1):45-65.

服务与反馈：

【文章下载】【发表评论】【查看评论】【加入收藏】