测绘学报 ›› 2026, Vol. 55 ›› Issue (10): 1797-1813.doi: 10.11947/j.AGCS.2026.20260157

• 海洋测绘 • 上一篇    下一篇

星载全极化GNSS-R全球近海海面波周期估算

徐顺双1(), 布金伟1,2()   

  1. 1.昆明理工大学国土资源工程学院,云南 昆明 650093
    2.云南省自然资源智能监测与时空大数据治理重点实验室,云南 昆明 650051
  • 收稿日期:2026-04-27 修回日期:2026-09-17 出版日期:2026-11-16 发布日期:2026-11-16
  • 通讯作者: 布金伟 E-mail:202410109130@stu.kust.edu.cn;b_jinwei@kust.edu.cn
  • 作者简介:徐顺双(2006—),女,研究方向为GNSS反射遥感。E-mail:202410109130@stu.kust.edu.cn
  • 基金资助:
    国家自然科学基金(42404037; 42664002);云南省“兴滇英才支持计划”青年人才项目(KKXX202521022);云南省基础研究计划(202601AT070121; 202401CF070151);昆明理工大学高层次人才平台建设项目(20230041);国家级大学生创新训练计划(202510674023)

Global coastal wave period estimation of spaceborne full polarization GNSS-R

Shunshuang Xu1(), Jinwei Bu1,2()   

  1. 1.Faculty of Land Resources Engineering, Kunming University of Science and Technology, Kunming 650093, China
    2.Yunnan Key Laboratory of Intelligent Monitoring and Spatiotemporal Big Data Governance of Natural Resources, Kunming 650051, China
  • Received:2026-04-27 Revised:2026-09-17 Online:2026-11-16 Published:2026-11-16
  • Contact: Jinwei Bu E-mail:202410109130@stu.kust.edu.cn;b_jinwei@kust.edu.cn
  • About author:Xu Shunshuang (2006—), female, majors in GNSS reflectometry remote sensing. E-mail: 202410109130@stu.kust.edu.cn
  • Supported by:
    The National Natural Science Foundation of China(42404037; 42664002);The Xingdian Talents Support Project of Yunnan Province(KKXX202521022);The Yunnan Fundamental Research Projects(202601AT070121; 202401CF070151);Kunming University of Science and Technology High Level Talent Platform Construction Project(20230041);National College Student Innovation Training Program(202510674023)

摘要:

海浪波周期是衡量海浪能量与传播特性的核心指标,决定波浪破坏力与传播速度,在航海、海洋工程及波浪能开发等领域具有重要应用价值。当前,尽管星载GNSS-R技术在有效波高反演中已得到应用,但海浪波周期的研究还很少,特别是星载全极化GNSS-R观测数据(如土壤湿度主动被动反射计(SMAP-R))还未在波周期领域开展研究。因此,本文基于全极化SMAP-R数据,在剖析海浪参数与SMAP-R斯托克斯参数响应机制的基础上,利用机器学习模型探索极化信号增量价值并构建波周期反演模型。为验证全极化SMAP-R反演海浪谱峰周期的性能,分别采用欧洲中期天气预报中心(ECMWF)的第5代再分析(ERA5)和WaveWatch Ⅲ(WW3)数据作为参考,结果表明,以ERA5谱峰波周期为参考时,SMAP-R降轨数据训练的模型反演效果最佳,袋装树在4个机器学习模型中性能最优,相关系数(CC)、均方根误差(RMSE)和偏差(Bias)分别为0.84、1.694 s和0.020 s;以WW3谱峰波周期为参考时,4个机器学习模型反演得到的波周期结果与WW3谱峰波周期之间一致性良好,除决策树(DT)模型外,其他模型的CC均不低于0.74。本文证明了全极化SMAP-R数据在反演全球近海海浪波周期方面具有稳定且正向的信息增益贡献。

关键词: 全极化GNSS-R, SMAP-R, 斯托克斯参数, 海浪波周期, 机器学习

Abstract:

The wave period is a core indicator for measuring the energy and propagation characteristics of waves, determining the destructive force and propagation speed of waves. It has important application value in fields such as navigation, ocean engineering, and wave energy development. Currently, although spaceborne GNSS-R technology has been applied in effective wave height retrieval, there is still little research on wave period, especially in the field of wave period, where spaceborne fully polarized GNSS-R observation data (such as soil moisture active passive reflectometry (SMAP-R)) has not been studied. Therefore, based on fully polarized SMAP-R data, this article analyzes the response mechanism of wave parameters and SMAP-R Stokes parameters, explores the incremental value of polarization signals using machine learning models, and constructs a wave period retrieval model. To verify the performance of fully polarized SMAP-R retrieval of wave spectral peak periods, the fifth generation reanalysis (ERA5) from the European Centre for Medium Range Weather Forecasts (ECMWF) and WaveWatch Ⅲ (WW3) data were used as references. The results showed that when using the ERA5 spectral peak wave period as a reference, The model trained on SMAP-R orbit descent data has the best retrieval effect, and Bagging trees have the best performance among the four machine learning models. The correlation coefficient (CC), root mean square error (RMSE), and bias are 0.84, 1.694 s, and 0.020 s, respectively. When using the WW3 spectral peak period as a reference, the wave period results obtained by the four machine learning models have good consistency with the WW3 spectral peak period. Except for the decision tree (DT) model, the CC values of all other models are at least 0.74. This article demonstrates that fully polarized SMAP-R data has a stable and positive information gain contribution in the retrieval of global nearshore wave cycles.

Key words: full polarization GNSS-R, SMAP-R, Stokes parameters, ocean wave period, machine learning

中图分类号: