基于VMD和WinLSP相结合的GNSS-R海平面高度估测模型  

GNSS-R sea level height estimation model based on the combination of VMD and WinLSP

作　　者：胡媛[1] 袁鑫泰 刘卫[2] 胡庆松[1] 江志豪 钟李程 Hu Yuan;Yuan Xintai;Liu Wei;Hu Qingsong;Jiang Zhihao;Zhong Licheng(College of Engineering Science and Technology,Shanghai Ocean University,Shanghai 201306,China;Merchant Marine College,Shanghai Maritime University,Shanghai 201306,China)

机构地区：[1]上海海洋大学工程学院,上海201306 [2]上海海事大学商船学院,上海201306

出　　处：《海洋学报》2022年第11期170-178,共9页

基　　金：上海市自然科学基金(19ZR1422800);国家自然科学基金(52071199);国家重点研发计划(2019YFD0901303)。

摘　　要：全球导航卫星系统反射(Global Navigation Satellite System-Reflectometry,GNSS-R)技术是一种新兴的监测海平面高度变化的技术。本文依据GNSS-R技术中的信噪比分析法的原理,通过分析其分离趋势项和提取振荡频率的过程,建立了新的估测模型以提高反演精度。针对传统模型存在的信号分离不佳的问题,本文提出使用变分模态分解(Variational Mode Decomposition,VMD)算法替换传统的最小二乘拟合法(Least Squares Fitting, LSF)进行趋势项分量的分离。在此基础上,本文引入基于凯塞窗函数改进的LSP(Lomb-Scargle Periodogram)频谱分析法(记为WinLSP)来减弱因频谱泄露带来的反演误差。在瑞典翁萨拉的GTGU站和美国阿拉斯加州的SC02站开展的海平面高度反演实验结果表明,本文建立的估测模型相比于传统模型具有更高的反演精度。基于VMD+WinLSP估测模型得到的GTGU站反演结果的均方根误差(RMSE)、相关系数和反演点数分别为4.70 cm、0.98和5 647。与传统的LSF+LSP估测模型相比,反演精度和GNSS数据利用率分别提高了约29.7%和15.0%。SC02站的RMSE、相关系数和反演点数分别为14.34 cm、0.99和1 785,反演精度和GNSS数据利用率分别提高了约12.3%和9.4%。Global navigation satellite system-reflectometry(GNSS-R) technology is an emerging technology for monitoring sea level changes. Based on the principle of the signal to noise ratio(SNR) analysis method in GNSS-R technology, this paper established a new sea level height estimation model to improve the accuracy by analyzing the process of separating the trend term and extracting the oscillation frequency. Aiming at the problem of poor signal separation in the traditional model, this paper proposed to use the variational mode decomposition(VMD) algorithm to replace the traditional least squares fitting(LSF) to separate the trend term components. On this basis,this paper combined Lomb-Scargle Periodogram(LSP) spectral analysis method and Kaiser window function(referred to as WinLSP) to reduce the inversion error caused by spectral leakage. The results of sea level inversion experiments carried out at GTGU Station in Onsala, Sweden and SC02 Station in Alaska, USA show that the estimation model established in this paper has higher inversion accuracy than traditional model. The root mean square error(RMSE), correlation coefficient and number of inversion points of the inversion results of GTGU Station based on the VMD+WinLSP estimation model are 4.70 cm, 0.98 and 5 647, respectively. The inversion accuracy and GNSS data utilization are increased by about 29.7% and 15.0%, respectively;The RMSE, correlation coefficient,and inversion points of SC02 Station are14.34 cm, 0.99 and 1 785, respectively, and the inversion accuracy and GNSS data utilization are increased by about 12.3 % and 9.4%.

关 键 词：GNSS-R 信噪比 变分模态分解 凯塞窗函数 

分 类 号：P228[天文地球—大地测量学与测量工程]