基于深度学习联合WOA18温盐模型构建声速场  

Construction of Sound Velocity Field Based on Deep Learning Combined With WOA18 Temperature and Salinity Model

作　　者：唐安 贺凯飞[1] 吴宇 王松[1] TANG An;HE Kaifei;WU Yu;WANG Song(College of Oceanography and Space Informatics,China University of Petroleum(East China),Qingdao 266580,China)

机构地区：[1]中国石油大学(华东)海洋与空间信息学院,山东青岛266580

出　　处：《海洋科学进展》2024年第4期816-829,共14页Advances in Marine Science

基　　金：国家自然科学基金项目(42174021);山东省自然科学基金项目(ZR2021MD060)。

摘　　要：声速变化是影响水下精密定位的重要因素,受制于现有的声速剖面获取手段,目前的声速代表性误差严重影响着水下定位精度。针对现实中难以实现一定海域内时间空间的连续观测,本文以地转海洋学实时观测阵(Array for Real-time Geostrophic Oceanography, Argo)温盐度数据作为真值,利用美国国家海洋和大气管理局(National Oceanic and Atmospheric Administration, NOAA)提供的2018世界海洋地图集(The World Ocean Atlas 2018, WOA18)中的历史温盐数据,基于添加注意力机制的长短期记忆神经网络模型(Long Short Term Memory, LSTM),来构建局部海域声速场。结果表明,该方法可用于反演局部海域500~1 500 m深度范围内较为精准的声速剖面,且添加注意力机制的LSTM神经网络模型反演的声速在太平洋局部海域均方根误差为0.34 m/s,在大西洋局部海域声速均方根误差为0.78 m/s,相比传统反向传播神经网络(Back Propagation Neural Network, BPNN)和添加了遗传因子的反向传播神经网络(Back Propagation Neural Network+Genetic Algorithm, BPNN+GA)在精度上得到了改善。The change in sound velocity is an important factor affecting the precise positioning of underwater.Due to the existing sound velocity profile acquisition methods,the current sound velocity representation error seriously affects the underwater positioning accuracy.In view of the difficulty in realizing continuous temporal and spacial observation in a certain sea area in reality,this paper uses the temperature and salinity data of the Array for Real-time Geostrophic Oceanography(Argo)as the true value and a neural network model based on Long Short Term Memory(LSTM)with added attention mechanisms as well as the World Ocean Atlas 2018(WOA18)historical thermohaline data to construct local ocean sound velocity field.The results show that this method can be used to retrieve relatively accurate sound velocity profiles in the depth range of 500-1500 m in local waters of the Pacific Ocean,and the root-mean-square error of sound velocity inversion by LSTM neural network model with added attention mechanism is 0.34 m/s.The root-mean-square error of sound velocity in the local waters of the Atlantic Ocean is 0.78 m/s.Compared with the traditional Back Propagation Neural Network,the accuracy of BPNN and Back Propagation Neural Network Genetic Algorithm(BPNN+GA)neural networks with added genetic factors has been improved.

关 键 词：声速剖面反演 WOA18 ARGO 深度学习 

分 类 号：P229.2[天文地球—大地测量学与测量工程]