Journal of Tropical Oceanography >
Research and application of global three-dimensional thermohaline reconstruction technology based on neural network
Received date: 2021-06-06
Revised date: 2021-07-28
Online published: 2021-08-16
Supported by
National Natural Science Foundation of China(41776004)
We apply the FOAGRNN (fruit fly optimization algorithm, FOA; generalized regression neural network, GRNN) method and SODA (simple ocean data assimilation) reanalysis data to construct a global ocean projection relationship model between sea-surface variables (sea surface height, SSH; sea surface temperature, SST; sea surface salinity, SSS) and subsurface thermohaline field. The remote sensing observations are utilized to evaluate the applicability of this global surface-subsurface reconstruction model. First, an ideal reconstruction test is executed using the independent SODA data in 2016. The idealized reconstruction results show that the global mean root mean square error (MRMSE) values of the reconstructed temperature and salinity are 0.36 ℃ and 0.08‰, which are reduced by about 50% and 60% compared to those of the WOA13 (World Ocean Atlas), respectively. Then, the satellite observations (Input field) and Argo profiles (verification field) are inputted to evaluate the practical application performance of the model. The results again indicate that our reconstruction model can reasonably reconstruct the thermohaline structures, and the MRMSE values of the reconstructed temperature and salinity are 0.79 ℃ and 0.16‰, which are 27% and 11% lower than those in the WOA13, respectively. Specifically, the RMSE of temperature is small at the sea surface and in the deep ocean, and the largest value exists in the thermocline layer with a maximum value of 1.35 ℃ at 100 m, and then quickly decreases to 0.81 ℃ at 250 m. The RMSE of salinity mostly decreases as depth increases, and has the largest peak of about 0.25‰ around 25 m. Finally, the analysis of Argo floats’ tracks and the statistics of regional water mass confirm that the reconstructed model can better describe the interior characteristics of the three-dimensional thermohaline field.
NIE Wangchen , WANG Xidong , CHEN Zhiqiang , HE Zikang , FAN Kaigui . Research and application of global three-dimensional thermohaline reconstruction technology based on neural network[J]. Journal of Tropical Oceanography, 2022 , 41(2) : 1 -15 . DOI: 10.11978/2021070
图1 果蝇算法参数优化流程(a)和FOAGRNN重构温盐场流程(b)图a流程为图b流程中的FOA优化小循环, S值对应spread参数, Smell对应重构结果与训练样本的均方根误差(RMSE), 满足条件结束循环输出最优spread参数。SSH: 海面高度; SST: 海面温度; GRNN: 广义回归神经网络模型 Fig. 1 The process of FOA and the process of FOAGRNN reconstruction of temperature and salinity field. The flow chart on the left corresponds to the FOA optimization mini-cycle on the right, S value corresponds to spread parameter, Smell corresponds to RMSE of reconstruction results and training samples; When condition is met, the end of the cycle, and outputs the optimal parameters of spread |
图2 SODA原始场与SODA温盐重构结果以及气候态资料之间的误差比较a. 2016年原始场与其他温度场的平均RMSE; b. 2016年原始场与其他盐度场的平均RMSE; c. 在南/北半球1/7月, SODA原始场与重构结果之间的温度场RMSE; d. 在南/北半球1/7月, SODA原始场与重构结果之间的盐度场RMSE Fig. 2 The changes of RMSE error between original field and SODA reconstruction results/climate state data with water depth. (a) Temperature field, in the worldwide scale and 2016 average; (b) salinity field, in the worldwide scale and 2016 average; (c) RMSE of temperature field between SODA original field and reconstructed results in the Northern/Southern Hemisphere and January/July; (d) RMSE of salinity field between SODA original field and reconstructed results in the Northern/Southern Hemisphere and January/July |
图3 选取代表区域和Argo浮标轨迹NP (北太平洋): 150°—170°E, 30°—40°N; NA (北大西洋): 40°—20°W, 40°—50°N; EQWP (赤道西太平洋): 170°E—170°W, 10°S—0; EQEP (赤道东太平洋): 110°—90°W, 10°S—0; SI (南印度洋): 50°—70°E, 40°—30°S; ANT (南大洋): 120°—60°W, 65°—55°S。Argo浮标编号: (1) 5902460, (2) 5903130, (3) 5905167, (4) 5904090, (5) 2902653, (6) 6901584; 红色曲线为Argo浮标轨迹, 黑点为轨迹起点, 每个浮标剖面数量在72~146之间。该图基于国家测绘地理信息局标准地图服务网站下载的审图号为GS(2016)1611的标准地图制作 Fig. 3 The selected representative area and Argo floats tracks. NP (North Pacific): 150°—170°E, 30°—40°N; NA (North Atlantic): 40°—20°W, 40°—50°N; EQWP (Equatorial Western Pacific): 170°—170°W, 10°S—0; EQEP (Equatorial Eastern Pacific): 110°—90°W, 10°S—0; SI (Southern indian Ocean): 50°—70°E, 40°—30°S; ANT (Antarctic Ocean): 120°—60°W, 65°—55°S. Argo float ID numbers: (1) 5902460, (2) 5903130, (3) 5905167, (4) 5904090, (5) 2902653, and (6) 6901584. These red curves are the Argo buoy track, and these black dots are the starting point of the tracks; the number of float profiles is between 72 and 146 |
表1 各区域SODA原始场与重构场以及气候态的垂向平均均方根误差(MRMSE)Tab. 1 Vertical MRMSE between original and reconstructed SODA field/ climate state in each region |
重构温度 MRMSE/℃ | WOA气候态MRMSE/℃ | SODA气候态MRMSE/℃ | 重构盐度MRMSE/‰ | WOA气候态MRMSE/‰ | SODA气候态MRMSE/‰ | |
---|---|---|---|---|---|---|
北太平洋 | 0.37 | 1.07 | 0.97 | 0.05 | 0.11 | 0.11 |
北大西洋 | 0.41 | 0.88 | 0.86 | 0.07 | 0.17 | 0.16 |
赤道西太 | 0.30 | 0.89 | 0.72 | 0.07 | 0.24 | 0.20 |
赤道东太 | 0.37 | 0.85 | 0.79 | 0.04 | 0.09 | 0.08 |
南印度洋 | 0.29 | 0.61 | 0.56 | 0.04 | 0.09 | 0.11 |
南大洋 | 0.23 | 0.65 | 0.61 | 0.04 | 0.12 | 0.12 |
平均 | 0.33±0.06 | 0.83±0.13 | 0.75±0.12 | 0.05±0.01 | 0.14±0.05 | 0.13±0.03 |
图4 温、盐场纬向截面分布a和d. SODA重构温盐截面; b和e. SODA原始数据温盐截面; c. 温度差值; f. 盐度差值。截面为经过00°15′E纬线, 时间为2016年1月29日, 图中空白表示陆地 Fig. 4 Distribution of temperature and salinity field in a zonal section. (a) and (d). SODA reconstructed temperature/salt profiles; (b) and (e). SODA raw data temperature/salt profiles; (c) and (f). their difference. The cross profile is along a latitude passing 00°15′E, and the time is January 29, 2016. The blank space represents land |
图5 温、盐场纬向截面分布a和d. SODA重构温盐截面; b和e. SODA原始数据温盐截面; c. 温度差值; f. 盐度差值。截面为经过60°S纬线, 时间为2016年1月29日 Fig. 5 Distribution of temperature and salinity field in a zonal section. (a) and (d). SODA reconstructed temperature/salt profiles; (b) and (e). SODA raw data temperature/salt profiles; (c) and (f). their difference. The cross profile is along a latitude passing 60°S, and the time is January 29, 2016 |
图6 Argo剖面观测资料分别与卫星数据温盐重构剖面以及WOA气候态资料之间的RMSE随水深的变化a. 2016年Argo数据与重构场/WOA气候态温度场的平均RMSE; b. 2016年Argo数据与重构场/WOA气候态盐度场的平均RMSE; c. 南/北半球1/7月温度场重构误差; d. 南/北半球1/7月盐度场重构误差 Fig. 6 Changes of RMSE between Argo field observation data, satellite data reconstruction results/WOA climate state data with water depth. (a) Temperature field, in the worldwide scale and 2016 average; (b) salinity field, in the worldwide scale and 2016 average; (c) temperature field reconstruction error, in the Northern/Southern Hemisphere and January/July; (d) salinity field reconstruction error, in the Northern/Southern Hemisphere and January/July |
图7 Argo浮标轨迹及背景场分图号代表Argo浮标编号: a. 5902460; b. 5904090; c. 5903130; d. 6901584; e. 5905167; f. 2902653。红色曲线为浮标轨迹, 蓝叉为轨迹起点, 黑叉为表层流场和海平面异常(SLA)所对应时刻 Fig. 7 Argo float track and background field. Argo float number. (a) 5902460; (b) 5904090; (c) 5903130; (d) 6901584; (e) 5905167; (f) 2902653. The red curve is float track, and the blue cross is the starting point of the track, and black cross is the corresponding moment of the surface flow field and SLA |
图8 Argo浮标5902460 (位于北大西洋)数据图像a. Argo观测温度场; b. 卫星数据重构温度场; c. Argo观测盐度场; d. 卫星重构盐度场; e. 图a中虚线对应时刻的Argo/卫星重构/ WOA13温度剖面; f. 盐度场; g. 观测温度场与卫星重构/WOA13气候态的RMSE; h. 盐度场; i. Argo剖面和重构剖面计算的比容海面高度。图a中的红色竖线为 Fig. 8 Argo float 5902460, located in the North Atlantic Ocean. (a) Argo observed temperature field; (b) satellite data reconstruction temperature field; (c) Argo observed salinity field; (d) satellite data reconstruction salinity field; (e) The Argo/satellite reconstruction/WOA13 temperature profile at the time corresponding to the dotted line in fig. a; (f) salinity field; (g) RMSE of observed temperature field and reconstructed temperature field/WOA13 climate state; (h) salinity field; (i) steric height calculated by Argo profile and reconstructed profile. The red vertical line in fig. a is the corresponding time of the black cross in fig. 7 |
图9 Argo浮标5904090 (位于黑潮延伸体)数据图像a. Argo观测温度场; b. 卫星数据重构温度场; c. Argo观测盐度场; d. 卫星重构盐度场; e. 图a中虚线对应时刻的Argo/卫星重构/ WOA13温度剖面; f. 盐度场; g. 观测温度场与卫星重构/WOA13气候态的RMSE; h. 盐度场; i. Argo剖面和重构剖面计算的比容海面高度。图a中的红色竖线为 Fig. 9 Argo float 5904090, located in the Kuroshio extension. (a) Argo observed temperature field; (b) satellite data reconstruction temperature field; (c) Argo observed salinity field; (d) satellite data reconstruction salinity field; (e) The Argo/satellite reconstruction/WOA13 temperature profile at the time corresponding to the dotted line in fig. a; (f) salinity field; (g) RMSE of observed temperature field and reconstructed temperature field/WOA13 climate state; (h) salinity field; (i) steric height calculated by Argo profile and reconstructed profile. The red vertical line in fig. a is the corresponding time of the black cross in fig. 7 |
图10 Argo浮标5903130 (位于南大西洋)数据图像a. Argo观测温度场; b. 卫星数据重构温度场; c. Argo观测盐度场; d. 卫星重构盐度场; e. 图a中虚线对应时刻的Argo/卫星重构/ WOA13温度剖面; f. 盐度场; g. 观测温度场与卫星重构/WOA13气候态的RMSE; h. 盐度场; i. Argo剖面和重构剖面计算的比容海面高度。图a中的红色竖线为 Fig. 10 Argo float 5903130, located in the South Atlantic Ocean. (a) Argo observed temperature field; (b) satellite data reconstruction temperature field; (c) Argo observed salinity field; (d) satellite data reconstruction salinity field; (e) The Argo/satellite reconstruction/WOA13 temperature profile at the time corresponding to the dotted line in fig. a; (f) salinity field; (g) RMSE of observed temperature field and reconstructed temperature field/WOA13 climate state; (h) salinity field; (i) steric height calculated by Argo profile and reconstructed profile. The red vertical line in fig. a is the corresponding time of the black cross in fig. 7 |
图11 Argo浮标6901584 (位于南印度洋)数据图像a. Argo观测温度场; b. 卫星数据重构温度场; c. Argo观测盐度场; d. 卫星重构盐度场; e. 图a中虚线对应时刻的Argo/卫星重构/ WOA13温度剖面; f. 盐度场; g. 观测温度场与卫星重构/WOA13气候态的RMSE; h. 盐度场; i. Argo剖面和重构剖面计算的比容海面高度。图a中的红色竖线为 Fig. 11 Argo float 6901584, located in the southern Indian Ocean. (a) Argo observed temperature field; (b) satellite data reconstruction temperature field; (c) Argo observed salinity field; (d) satellite data reconstruction salinity field; (e) The Argo/satellite reconstruction/WOA13 temperature profile at the time corresponding to the dotted line in fig. a; (f) salinity field; (g) RMSE of observed temperature field and reconstructed temperature field/WOA13 climate state; (h) salinity field; (i) steric height calculated by Argo profile and reconstructed profile. The red vertical line in fig. a is the corresponding time of the black cross in fig. 7 |
图12 Argo浮标5905167 (位于东澳大利亚洋流)数据图像Argo观测温度场; b. 卫星数据重构温度场; c. Argo观测盐度场; d. 卫星重构盐度场; e. 图a中虚线对应时刻的Argo/卫星重构/ WOA13温度剖面; f. 盐度场; g. 观测温度场与卫星重构/WOA13气候态的RMSE; h. 盐度场; i. Argo剖面和重构剖面计算的比容海面高度。图a和图c中空白为Argo数据缺测导致; 图b和图d中空白以及图i中曲线不连续情况为卫星数据缺测导致。图a中的红色竖线为 Fig. 12 Argo float 5905167, located in the East Australian Current. (a) Argo observed temperature field; (b) satellite data reconstruction temperature field; (c) Argo observed salinity field; (d) satellite data reconstruction salinity field; (e) The Argo/satellite reconstruction/WOA13 temperature profile at the time corresponding to the dotted line in fig. a; (f) salinity field; (g) RMSE of observed temperature field and reconstructed temperature field/WOA13 climate state; (h) salinity field; (i) steric height calculated by Argo profile and reconstructed profile. The blank in (a) and (c) is caused by the lack of Argo data; The blank in (b), (d) and the curve discontinuity in (i) are caused by the lack of satellite data. The red vertical line in fig. a is the corresponding time of the black cross in fig. 7 |
图13 Argo浮标2902653 (位于西太平洋附近)数据a. Argo观测温度场; b. 卫星数据重构温度场; c. Argo观测盐度场; d. 卫星重构盐度场; e. 图a中虚线对应时刻的Argo/卫星重构/ WOA13温度剖面; f. 盐度场; g. 观测温度场与卫星重构/WOA13气候态的RMSE; h. 盐度场; i. Argo剖面和重构剖面计算的比容海面高度。图d中空白和图i中曲线不连续情况为卫星数据缺测导致。图a中的红色竖线为 Fig. 13 Argo float 2902653, located near the western Pacific. (a) Argo observed temperature field; (b) satellite data reconstruction temperature field; (c) Argo observed salinity field; (d) satellite data reconstruction salinity field; (e) The Argo/satellite reconstruction/WOA13 temperature profile at the time corresponding to the dotted line in fig. a; (f) salinity field; (g) RMSE of observed temperature field and reconstructed temperature field/WOA13 climate state; (h) salinity field; (i) steric height calculated by Argo profile and reconstructed profile. The blank in (d) and the curve discontinuity in (i) are caused by the lack of satellite data. The red vertical line in fig. a is the corresponding time of the black cross in fig. 7 |
图14 各代表区域所有标准深度0~1000m的温度和盐度图a. 北太平洋; b. 北大西洋; c. 赤道西太平洋; d. 赤道东太平洋; e. 南印度洋; f. 南大洋。灰色表示Argo观测资料, 绿色表示卫星数据重构, 蓝色表示WOA气候态 Fig. 14 Temperature and salinity maps of all standard depths from 0 to 1000 m in each representative area. (a) North Pacific Ocean; (b) North Atlantic; (c) western equatorial Pacific; (d) eastern equatorial Pacific; (e) southern Indian Ocean; (f) Antarctic Ocean. Argo observation data (gray), satellite data reconstruction (green), and WOA climate state (blue) |
表2 各区域Argo观测数据与卫星观测数据重构场以及WOA气候态的垂向平均均方根误差(MRMSE)Tab. 2 Vertical MRMSE of Argo observation data, satellite observation data reconstruction field and WOA climate state in each region |
卫星重构温度MRMSE/℃ | WOA13温度MRMSE/℃ | 卫星重构盐度MRMSE/‰ | WOA13盐度MRMSE/‰ | |
---|---|---|---|---|
北太平洋 | 0.98 | 1.56 | 0.17 | 0.15 |
北大西洋 | 0.93 | 1.35 | 0.25 | 0.24 |
赤道西太 | 0.77 | 1.01 | 0.14 | 0.26 |
赤道东太 | 0.92 | 1.04 | 0.09 | 0.10 |
南印度洋 | 0.53 | 0.75 | 0.13 | 0.10 |
南大洋 | 0.48 | 0.70 | 0.17 | 0.10 |
平均 | 0.77±0.16 | 1.07±0.26 | 0.16±0.04 | 0.16±0.06 |
[1] |
鲍森亮, 张韧, 王辉赞, 等, 2018. 基于海表信息的海洋盐度剖面重构[C]// 第九届海洋强国战略论坛论文集. 北海: 海洋出版社: 206-215. (in Chinese)
|
[2] |
范良, 赵国忱, 苏运强, 2013. 果蝇算法优化的广义回归神经网络在变形监测预报的应用[J]. 测绘通报, (11): 87-89.
|
[3] |
潘文超, 2011. 果蝇最佳化演算法: 最新演化式计算技术[M]. 台中: 沧海书局. (in Chinese)
|
[4] |
王喜冬, 韩桂军, 李威, 等, 2011. 利用卫星观测海面信息反演三维温度场[J]. 热带海洋学报, 30(6): 10-17.
|
[5] |
|
[6] |
|
[7] |
|
[8] |
|
[9] |
|
[10] |
|
[11] |
|
[12] |
|
[13] |
|
[14] |
|
[15] |
|
[16] |
|
[17] |
|
[18] |
|
[19] |
|
[20] |
|
[21] |
|
[22] |
|
[23] |
|
/
〈 | 〉 |