利用深度学习和GNSS研究水负荷位移时空变化  

Research on Spatio-temporal Change of Water Loading Displacement Using Deep Learning and GNSS

作　　者：汤伟尧 TANG Weiyao(China Railway First Survey and Design Institute Group Co.,Ltd.,Xi’an 710043,China)

机构地区：[1]中铁第一勘察设计院集团有限公司,陕西西安710043

出　　处：《地理空间信息》2025年第3期5-8,共4页Geospatial Information

基　　金：基于北斗地基增强的铁路运维动态基准关键技术(SKLK18-01)。

摘　　要：基于长江中上游地区14个全球导航卫星系统(GNSS)测站的观测数据和重力恢复与气候实验(GRACE)Mascon数据,提出了一种基于卷积长短期记忆人工神经网络(LSTM)的水负荷位移估计方法。研究表明,在数据预处理时改正大气和非潮汐海洋负载,可明显减弱GNSS测站的非线性运动。与格林函数法相比,该方法的估计结果与GNSS真实观测值的差值标准差平均降低了20.2%。夏、秋季水负荷位移多为负值,而冬、春季则多为正值,且呈现出显著的周年和半周年周期性以及近1/3年周期变化。湖北省西南部水负荷位移振幅较大,而西部地区尤其是西北部地区振幅较小;中部地区水负荷位移普遍偏大,东部地区振幅逐渐减小。研究结果可为长江中上游地区水负荷位移的监测和分析提供新的视角。Based on observational data from 14 global navigation satellite system(GNSS)stations in the middle and upper reaches of the Yangtze River and gravity recovery and climate experiment(GRACE)Mascon data,we put forward a water loading displacement estimation method based on a convolutional long short-term memory(LSTM)neural network.The results show that the nonlinear motion of GNSS station is obviously weakened by correcting atmospheric and non-tidal ocean loading during data preprocessing.Compared to the Green’s function method,this method shows better consistency in the discrepancy between estimated water loading displacement and actual GNSS observations,with an average standard deviation reduction of 20.2%.The water loading displacement tends to be negative in summer and autumn,and positive in winter and spring.Additionally,significant annual and semi-annual periodic changes in water loading displacement are observed,along with a near one-third of a year signal.Spatially,the southwest of Hubei Province exhibits larger water loading displacement amplitudes,while the western region,especially the northwest,shows smaller amplitudes.Central regions generally have larger displacements,and the amplitude gradually decreases in the eastern areas.This study offers a new perspective for monitoring and analyzing water loading displacement in the middle and upper reaches of the Yangtze River.

关 键 词：水负荷位移 GNSS GRACE 卷积LSTM 格林函数 

分 类 号：P228.4[天文地球—大地测量学与测量工程]