Reconstruction of Three-dimensional Temperature Field in South China Sea Based on Generative Adversarial Networks
-
摘要: 针对当前南海高分辨率海洋次表层温度场数据稀缺问题，根据海洋表层遥感观测和次表层海水温度的时空关联性，开展基于生成对抗网络的南海高分辨率三维温度场重构方法研究。基于2013−2017年的海面温度、海面盐度、海面高度异常以及海面风场等多源海洋表层遥感数据，构建了基于生成对抗网络的南海三维温度场重构模型。利用训练完成的模型和海表多源遥感数据，开展2018年南海541 m以浅19个深度层的海洋三维温度场重构实验，并与GLORYS12V1再分析数据和Argo剖面数据进行比较评估，检验所提出方法的可行性。实验结果表明，三维温度场重构结果各个深度层的空间分布特征与GLORYS12V1再分析数据保持良好的一致性，能够反应出南海中部海域典型垂向断面的季节变化特征。通过对比南海不同海域三个位置点海水温度时间序列，验证了模型重构性能的稳定性。与Argo温度剖面的对比验证了重构结果能够较好地反应出真实海水温度的垂向变化，证明所提出的方法具有一定的实际应用价值。此外，重构的2018年南海三维温度场平均RMSE为0.704℃，优于CNN（0.952℃）和U-net（0.863℃）模型。Abstract: To address the scarcity of high-resolution ocean subsurface temperature field data in the South China Sea (SCS), this study proposes a Generative Adversarial Network (GAN) for reconstructing high-resolution three-dimensional ocean temperature field based on the spatiotemporal correlation between ocean surface remote sensing observations and subsurface ocean temperature. The proposed GAN model is trained by multi-source ocean surface remote sensing data from 2013 to 2017, including sea surface temperature, sea surface salinity, sea level anomaly, and sea surface wind. The three-dimensional ocean temperature fields for 19 depth layers shallower than 541 meters in SCS in 2018 are reconstructed using the trained model and ocean surface multi-source remote sensing data. The ocean temperature fields reconstruction results are compared with GLORYS12V1 reanalysis data and Argo profile data to assess the feasibility of the proposed model. The results of experiments show that the spatial distribution characteristics of the reconstructed temperature field at different depth layers are in good agreement with the GLORYS12V1 reanalysis data, and can reflect the seasonal variation features of typical vertical cross-sections in the central SCS. The comparison of ocean temperature time series at three different locations in the SCS verifies the stability and accuracy of the proposed model. The evaluation experiments based on Argo in-situ observations show that the model can accurately reconstruct the vertical variation of real ocean temperature, demonstrating the practical application value of the proposed method. The average RMSE of the reconstructed three-dimensional temperature field in the South China Sea for 2018 is 0.704℃, which outperforms the CNN (0.952℃) and U-net (0.863℃) models.
-
图 5 三维温度场重构结果与GLORYS12V1海洋再分析数据的在不同深度层水平分布特征比较
Fig. 5 Comparison of horizontal distribution characteristics of 3d temperature field reconstruction results and GLORYS12V1 ocean reanalysis data at different depth layers
图 6 经向断面海水温度垂向分布
Fig. 6 Vertical distribution of ocean temperature in the meridional section
图 7 纬向断面海水温度垂向分布
Fig. 7 Vertical distribution of ocean temperature in latitudinal section
图 8 海水温度重构结果与GLORYS12V1在不同深度层的比较散点图
Fig. 8 Comparison of scatter plot between reconstructed ocean temperature and GLORYS12V1 at different depth layers
图 9 不同验证点海水温度时间序列
Fig. 9 Time series of ocean temperature at different verification point
图 10 每日三维温度场重构结果与GLORYS12V1数据在不同深度层平均RMSE
Fig. 10 The daily averaged RMSE between 3D ocean temperature field reconstruction results and GLORYS12V1 data at different depth layer
图 11 三维温度场重构结果与GLORYS12V1数据在不同深度层月均平均RMSE
Fig. 11 The month averaged RMSE between 3D ocean temperature field reconstruction results and GLORYS12V1 data at different depth layer
图 13 不同输入参数组合下的温度场重构精度
Fig. 13 Reconstruction accuracy under different sea surface input variables
图 14 2018年各模型在南海19个深度层的平均RMSE
Fig. 14 Average RMSE for each model in 2018 for 19 depth layers in the South China Sea
表 1 不同验证点不同深度层2018年平均RMSE
Tab. 1 Average RMSE of different validation point and depth layer in 2018
RMSE (℃) 10 m 77 m 155 m 266 m 541 m 平均值 T1 0.42 0.85 0.69 0.38 0.25 0.51 T2 0.48 0.70 0.56 0.41 0.36 0.50 T3 0.61 0.77 0.55 0.36 0.20 0.49 
下载: 导出CSV
-
[1] Oliver E C J, Donat M G, Burrows M T, et al. Longer and more frequent marine heatwaves over the past century[J]. Nature Communications, 2018, 9(1): 1324. doi: 10.1038/s41467-018-03732-9 [2] 林明森, 何贤强, 贾永君, 等. 中国海洋卫星遥感技术进展[J]. 海洋学报, 2019, 41(10): 99−112. doi: 10.3969/j.issn.0253-4193.2019.10.007Lin Mingsen, He Xianqiang, Jia Yongjun, et al. Advances in marine satellite remote sensing technology in China[J]. Haiyang Xuebao, 2019, 41(10): 99−112. doi: 10.3969/j.issn.0253-4193.2019.10.007 [3] Johnson G C, Hosoda S, Jayne S R, et al. Argo—two decades: global oceanography, revolutionized[J]. Annual Review of Marine Science, 2022, 14: 379−403. doi: 10.1146/annurev-marine-022521-102008 [4] 王喜冬, 韩桂军, 李威, 等. 利用卫星观测海面信息反演三维温度场[J]. 热带海洋学报, 2011, 30(6): 10−17. doi: 10.3969/j.issn.1009-5470.2011.06.002Wang Xidong, Han Guijun, Li Wei, et al. Reconstruction of ocean temperature profile using satellite observations[J]. Journal of Tropical Oceanography, 2011, 30(6): 10−17. doi: 10.3969/j.issn.1009-5470.2011.06.002 [5] Wu Xiangbai, Yan Xiaohui, Jo Y H, et al. Estimation of subsurface temperature anomaly in the North Atlantic using a self-organizing map neural network[J]. Journal of Atmospheric and Oceanic Technology, 2012, 29(11): 1675−1688. doi: 10.1175/JTECH-D-12-00013.1 [6] Khedouri E, Szczechowski C, Cheney R. Potential oceanographic applications of satellite altimetry for inferring subsurface thermal structure[C]//Proceedings OCEANS '83. San Francisco: IEEE, 1983: 274-280. [7] Liu Lei, Peng Shiqiu, Huang Ruixin. Reconstruction of ocean's interior from observed sea surface information[J]. Journal of Geophysical Research: Oceans, 2017, 122(2): 1042−1056. doi: 10.1002/2016JC011927 [8] 周超杰, 张杰, 杨俊钢, 等. 基于ROMS模式的南海SST与SSH四维变分同化研究[J]. 海洋学报, 2019, 41(1): 32−40. doi: 10.3969/j.issn.0253-4193.2019.01.004Zhou Chaojie, Zhang Jie, Yang Jungang, et al. 4DVAR assimilation of SST and SSH data in South China Sea based on ROMS[J]. Haiyang Xuebao, 2019, 41(1): 32−40. doi: 10.3969/j.issn.0253-4193.2019.01.004 [9] Thacker W C, Long R B. Fitting dynamics to data[J]. Journal of Geophysical Research: Oceans, 1988, 93(C2): 1227−1240. doi: 10.1029/JC093iC02p01227 [10] Lapeyre G, Klein P. Dynamics of the upper oceanic layers in terms of surface quasigeostrophy theory[J]. Journal of Physical Oceanography, 2006, 36(2): 165−176. doi: 10.1175/JPO2840.1 [11] Nardelli B B, Santoleri R. Reconstructing synthetic profiles from surface data[J]. Journal of Atmospheric and Oceanic Technology, 2004, 21(4): 693−703. doi: 10.1175/1520-0426(2004)021<0693:RSPFSD>2.0.CO;2 [12] Fox D N, Teague W J, Barron C N, et al. The modular ocean data assimilation system (MODAS)[J]. Journal of Atmospheric and Oceanic Technology, 2002, 19(2): 240−252. doi: 10.1175/1520-0426(2002)019<0240:TMODAS>2.0.CO;2 [13] Ali M M, Swain D, Weller R A. Estimation of ocean subsurface thermal structure from surface parameters: a neural network approach[J]. Geophysical Research Letters, 2004, 31(20): L20308. [14] Su Hua, Zhang Tianyi, Lin Mengjing, et al. Predicting subsurface thermohaline structure from remote sensing data based on long short-term memory neural networks[J]. Remote Sensing of Environment, 2021, 260: 112465. doi: 10.1016/j.rse.2021.112465 [15] Zhang Shuyu, Deng Yuesen, Niu Qianru, et al. Multivariate temporal self-attention network for subsurface thermohaline structure reconstruction[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61: 4507116. [16] Meng Lingsheng, Yan Chi, Zhuang Wei, et al. Reconstructing high-resolution ocean subsurface and interior temperature and salinity anomalies from satellite observations[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 4104114. [17] Xiao Changjiang, Tong Xiaohua, Li Dandan, et al. Prediction of long lead monthly three-dimensional ocean temperature using time series gridded Argo data and a deep learning method[J]. International Journal of Applied Earth Observation and Geoinformation, 2022, 112: 102971. doi: 10.1016/j.jag.2022.102971 [18] Yan Hengqian, Zhang Ren, Wang Huizan, et al. Estimating daily subsurface thermohaline structure from satellite data: a deep network with embedded empirical orthogonal functions[J]. Deep Sea Research Part I: Oceanographic Research Papers, 2024, 206: 104257. doi: 10.1016/j.dsr.2024.104257 [19] Xie Huarong, Xu Qing, Cheng Yongcun, et al. Reconstruction of subsurface temperature field in the South China Sea from satellite observations based on an attention U-net model[J]. IEEE Transactions on Geoscience and Remote Sensing, 2022, 60: 4209319. [20] Wang Guihua, Su Jilan, Chu P C. Mesoscale eddies in the South China Sea observed with altimeter data[J]. Geophysical Research Letters, 2003, 30(21): 2121. [21] Zheng Quanan, Susanto R D, Ho C R, et al. Statistical and dynamical analyses of generation mechanisms of solitary internal waves in the northern South China Sea[J]. Journal of Geophysical Research: Oceans, 2007, 112(C3): C03021. [22] Nan Feng, He Zhigang, Zhou Hui, et al. Three long-lived anticyclonic eddies in the northern South China Sea[J]. Journal of Geophysical Research: Oceans, 2011, 116(C5): C05002. [23] Zheng Quanan, Hu Jianyu, Zhu Benlu, et al. Standing wave modes observed in the South China Sea deep basin[J]. Journal of Geophysical Research: Oceans, 2014, 119(7): 4185−4199. doi: 10.1002/2014JC009957 [24] He Qingyou, Zhan Haigang, Cai Shuqun, et al. A new assessment of mesoscale eddies in the South China Sea: surface features, three‐dimensional structures, and thermohaline transports[J]. Journal of Geophysical Research: Oceans, 2018, 123(7): 4906−4929. doi: 10.1029/2018JC014054 [25] Xie Huarong, Xu Qing, Zheng Quanan, et al. Assessment of theoretical approaches to derivation of internal solitary wave parameters from multi-satellite images near the Dongsha Atoll of the South China Sea[J]. Acta Oceanologica Sinica, 2022, 41(6): 137−145. doi: 10.1007/s13131-022-2015-3 [26] Wang Weiqiang, Wang Chunzai. Formation and decay of the spring warm pool in the South China Sea[J]. Geophysical Research Letters, 2006, 33(2): L02615. [27] 王凡, 赵永平, 冯志纲, 等. 1998年春夏南海温盐结构及其变化特征[J]. 海洋学报, 2001, 23(5): 1−13. doi: 10.3321/j.issn:0253-4193.2001.05.001Wang Fan, Zhao Yongping, Feng Zhigang, et al. Thermohaline structure and variation of the South China Sea in the spring and summer of 1998[J]. Haiyang Xuebao, 2001, 23(5): 1−13. doi: 10.3321/j.issn:0253-4193.2001.05.001 [28] Tan Hongjian, Cai Rongshuo, Wu Renguang. Summer marine heatwaves in the South China Sea: trend, variability and possible causes[J]. Advances in Climate Change Research, 2022, 13(3): 323−332. doi: 10.1016/j.accre.2022.04.003 [29] Goodfellow I J, Pouget-Abadie J, Mirza M, et al. Generative adversarial nets[C]// Proceedings of the 28th International Conference on Neural Information Processing Systems. Montreal: NIPS, 2014. [30] Perezhogin P, Zanna L, Fernandez‐Granda C. Generative data‐driven approaches for stochastic subgrid parameterizations in an idealized ocean model[J]. Journal of Advances in Modeling Earth Systems, 2023, 15(10): e2023MS003681. doi: 10.1029/2023MS003681 -
计量
- 文章访问数: 24
- HTML全文浏览量: 7
- PDF下载量: 3
- 被引次数: 0
