|
| 引用本文: | 赵健,蔡瑞阳,孙伟富,杨俊钢.基于海洋气候数据集的区域海平面变化非线性预测[J].海洋科学,2023,47(4):69-78. |
| |
|
| |
|
|
| 本文已被：浏览 631次 下载 1524次 |
码上扫一扫！ |
|
|
| 基于海洋气候数据集的区域海平面变化非线性预测 |
|
赵健1, 蔡瑞阳1, 孙伟富2, 杨俊钢2
|
|
1.中国石油大学(华东) 海洋与空间信息学院, 山东 青岛 266580;2.自然资源部 第一海洋研究所, 山东 青岛 266061
|
|
| 摘要: |
| 本文基于中国首套长时间序列、高精度、高时空一致性的全球海洋气候数据集产品, 利用1993年1月至2015年12月的山东半岛近海海平面异常数据, 构建了基于集合经验模式分解(EEMD)和长短期记忆神经网络(LSTM)的海平面非线性变化组合预测模型。EEMD可以得到海平面异常的各周期项、线性趋势及残差部分, LSTM模型可对其进行逐个预测并重构得到最终的海平面异常预测结果。EEMD-LSTM组合模型海平面异常预测的均方根误差仅为25.87 mm, 取得了令人满意的效果。基于该组合模型预测2016-2025年山东半岛近海海平面上升速率将达到3.54 mm·a-1。 |
| 关键词: 海平面变化 气候数据集 集合经验模式分解 长短期记忆神经网络 预测 |
| DOI：10.11759/hykx20210128002 |
| 分类号:P228 |
| 基金项目:国家重点研发计划项目(2016YFA0600102) |
|
| Prediction of regional sea-level nonlinear change based on ocean climate data records |
|
ZHAO Jian1, CAI Rui-yang1, SUN Wei-fu2, YANG Jun-gang2
|
|
1.College of Oceanography and Space Informatics, China University of Petroleum, Qingdao 266580, China;2.First Institute of Oceanography, MNR, Qingdao 266061, China
|
| Abstract: |
| China’s first global ocean climate data records exhibiting long time-series data, high accuracy, and spatiotemporal consistency were used to study regional nonlinear sea-level changes. Using the sea-level anomaly (SLA) data around the Shandong Peninsula from January 1993 to December 2015, we established a combined model based on ensemble empirical mode decomposition (EEMD) and the long and short-term memory (LSTM) approach to forecast nonlinear sea-level trends around the Shandong Peninsula. Herein, period terms, noise terms and residuals (trend item) are individually obtained from EEMD, forecasted using an LSTM neural network, and then reconstructed to obtain the trends in sea-level change. The EEMD-LSTM combined model may be a valuable approach in the prediction of sea-level change as indicated by the minimum 25.59 mm root mean square error values in the SLA prediction obtained during the testing period. The model predicts that, for the period 2016-2025, the rate of sea-level rise around the Shandong Peninsula will increase to 3.54 mm·a-1. |
| Key words: sea level change climate data records ensemble empirical mode decomposition long short-term memory prediction |
|
|
|
|
|
|