海洋学研究 ›› 2022, Vol. 40 ›› Issue (2): 19-31.DOI: 10.3969-j.issn.1001-909X.2022.02.003

• • 上一篇    下一篇

基于GHSOM网络的南海风场时空变化特征分析

  

  1. 河海大学海洋学院,江苏 南京 210013
  • 出版日期:2022-06-15 发布日期:2022-06-15
  • 基金资助:
    国家重点研发计划(2017YFA0604104);国家自然科学基金(42076015)

Spatio-temporal variation characteristics of wind field in South China Sea based on Growing Hierarchical Self-Organizing Map analysis

  1. College of Oceanography, Hohai University, Nanjing 210013, China
  • Online:2022-06-15 Published:2022-06-15

摘要:

基于1979—2018年欧洲中期天气预报中心(ECMWF)近海面10 m风场资料,采用增长型分层自组织映射(GHSOM)神经网络方法,对南海海表面风场(SSW)的季节变化和年际异常变化进行了分析,结果表明:(1)GHSOM网络训练原始风场数据第一层结果揭示了4个特征模态,高度概括了南海近海面风场的季节变化特征;第二层结果提取了风场的月变化特征。(2)GHSOM网络训练异常风场数据第一层结果揭示了4类异常风场特征模态:反气旋式异常、气旋式异常、西南风异常和东北风异常模态。其中反气旋式异常和气旋式异常模态呈现出不对称现象,即反气旋式异常风场的振幅大于气旋式异常风场;且这两个模态与ENSO事件密切相关,它们的时间序列与Niño 3.4指数序列存在显著的延迟相关。同时,东北风异常风场模态的发生频率大于西南风异常模态。向下扩展的第二层结果揭露了异常风场模态更多的细节特征。


关键词: 海面风, 增长型分层自组织映射, 季节变化, 年际异常变化, 南海

Abstract:

Based on the sea surface wind data at 10 m during 1979 to 2018 from European Center for MediumRange Weather Forecasts (ECMWF), the Growing Hierarchical Self-Organizing Map (GHSOM) method were used to analyze the seasonal variation and interannual anomaly variation characteristics of near-surface wind field over the South China Sea (SCS). Four feature patterns are extracted in the first-layer GHSOM from original wind field data, which highly summarize the seasonal variation characteristics, and the second-layer results reveal the monthly variation characteristics. Four anomaly feature patterns also are extracted in the first-layer GHSOM network and they are anticyclonic wind anomaly, cyclonic wind anomaly, southwest wind anomaly and northeast wind anomaly patterns, respectively. Anticyclonic and cyclonic wind anomaly patterns are closely related to ENSO events with time lags by three months and five months comparing with Niño3.4 index. Anticyclonic and cyclonic wind anomalies also show asymmetry, that is, the amplitude of anticyclonic wind anomaly is obviously larger than that of cyclonic wind anomaly. The occurrence frequency of the northeast wind anomaly pattern is greater than that of the southwest wind anomaly pattern. The more SOM patterns in the second layer expose particulars of anomaly wind.


Key words: sea surface wind, Growing Hierarchical Self-Organizing Map, season variability, inter-annual anomaly variability, South China Sea

中图分类号: