基于LSTM构建海岸侵蚀风险预测模型的研究及应用
Research and application of constructing a coastal erosion risk prediction model based on LSTM
梁东 1高娜 2英晓明 1周曾 3舒勰俊 2徐婉明 2赵明利1
作者信息
- 1. 自然资源部南海发展研究院,广东 广州 510300;自然资源部海洋环境探测技术与应用重点实验室,广东 广州 510300
- 2. 自然资源部南海发展研究院,广东 广州 510300
- 3. 河海大学 江苏省海岸海洋资源开发与环境安全重点实验室,江苏 南京 210098
- 折叠
摘要
岸线侵蚀预测是海岸动力地貌学研究的热点问题之一.本文基于长短期神经记忆网络LSTM,采用 1985-2023年江苏射阳县附近海域收集到的岸线、水深和潮间带宽度数据,结合ERA5数据反演的波浪和潮流数据,构建海岸侵蚀风险预测模型.该预测模型可以准确反映海岸线的加速侵蚀、稳定侵蚀、淤积的非线性变化或线性变化趋势.预测结果表明,在沙源减少条件下,波浪和潮流增强是近20年来射阳海域海岸侵蚀的主要因素.此外,利用预测模型开展了海岸加固、消浪和弱流对海岸的防护效果试验,试验结果表明,海岸加固防护效果最佳,消浪防护效果较弱流防护效果好.预测模型设置运行过程高效,具有较高的应用价值和开发潜力.
Abstract
Shoreline erosion prediction is one of the hot issues in coastal dynamic geomorphology research.Based on the long short term memory(LSTM),the data of shoreline,water depth,intertidal zone width,and wave and tid-al current for ERA5 inversion clollected from 1985 to 2023 near Sheyang County of Jiangsu Province were used to construct a coastal erosion risk prediction model in this study.The prediction model could accurately predict the nonlinear/linear change trend of accelerated erosion,stable erosion or coastline sedimentation.The results showed that the increasing of wave and tidal currents was the main factor of coastal erosion in Sheyang area in recent 20 years under the condition of sand source reduction.Besides,an ideal experiment of coastal protection activities was conducted by using the prediction model,and the protection effects of coastal reinforcement,wave dissipation and weak current engineering were discussed.The results showed that the protection effect of coastal reinforcement is the best,and wave dissipation is better than weak current.The prediction model is reasonable,and has great applica-tion value and development potential.
关键词
海岸侵蚀预测/江苏射阳/LSTM/非线性变化Key words
coastal erosion prediction/Sheyang,Jiangsu/LSTM/Nonlinear variation引用本文复制引用
基金项目
国家重点研发计划(2022YFC3106203)
自然资源部海洋环境探测技术与应用重点实验室自主设立课题(MESTA-2022-C005)
出版年
2024
NETL
NSTL
万方数据