融合星载GNSS-R数据和多变量参数全球海洋有效波高深度学习反演法

doi:10.11947/j.AGCS.2024.20230050

测绘学报 ›› 2024, Vol. 53 ›› Issue (7): 1321-1335.doi: 10.11947/j.AGCS.2024.20230050

融合星载GNSS-R数据和多变量参数全球海洋有效波高深度学习反演法

布金伟¹(), 余科根²(), 汪秋兰¹, 李玲惠¹, 刘馨雨¹, 左小清¹, 常军³

^1.昆明理工大学国土资源工程学院，云南　昆明　650093
^2.中国矿业大学环境与测绘学院，江苏　徐州　221116
^3.自然资源部第一大地测量队，陕西　西安　710054

收稿日期:2023-02-22 发布日期:2024-08-12
通讯作者: 余科根 E-mail:b_jinwei@kust.edu.cn;kegen.yu@cumt.edu.cn
作者简介:布金伟（1992—），男，博士，研究方向为GNSS反射遥感。E-mail：b_jinwei@kust.edu.cn
基金资助:
云南省基础研究计划项目(202401CF070151);昆明理工大学高层次人才平台建设项目(20230041);国家自然科学基金(42174022);云南省大学生创新训练计划项目(S202310674221)

Deep learning retrieval method for global ocean significant wave height by integrating spaceborne GNSS-R data and multivariable parameters

Jinwei BU¹(), Kegen YU²(), Qiulan WANG¹, Linghui LI¹, Xinyu LIU¹, Xiaoqing ZUO¹, Jun CHANG³

^1.Faculty of Land Resources Engineering, Kunming University of Science and Technology, Kunming 650093, China
^2.School of Environmental Science and Spatial Informatics, China University of Mining and Technology, Xuzhou 221116, China
^3.The First Geodetic Surveying Brigade of MNR, Xi'an 710054, China

Received:2023-02-22 Published:2024-08-12
Contact: Kegen YU E-mail:b_jinwei@kust.edu.cn;kegen.yu@cumt.edu.cn
About author:BU Jinwei (1992—), male, PhD, majors in GNSS-R. E-mail: b_jinwei@kust.edu.cn
Supported by:
The Yunnan Fundamental Research Projects(202401CF070151);The Platform Construction Project of High-level Talent in Kunming University of Science and Technology(20230041);The National Natural Science Foundation of China(42174022);The Innovative Training Plan Program for College Students of Yunnan Province(S202310674221)

摘要/Abstract

摘要：

星载GNSS-R作为一种新兴的观测方法，最近被应用于有效波高反演。现有研究通常使用从延迟多普勒图中提取的特征值以构建经验地球物理模型函数反演SWH。然而，使用多个变量参数作为模型输入具有很大挑战。为此，本文提出了一个融合星载GNSS-R数据和多变量参数反演全球海面SWH的深度学习网络模型（GloWH-Net）。为了验证本文模型的性能，ERA5、WaveWatch Ⅲ和AVISO SWH数据被用作广泛测试的参考数据，以评估GloWH-Net模型和先前模型（即经验模型和机器学习模型）的SWH反演性能。结果表明，当分别使用ERA5、WaveWatchⅢ和AVISO SWH作为参考值时，所提的GloWH-Net模型反演SWH的均方根误差分别为0.330、0.393和0.433 m，相关系数分别为0.91、0.89和0.84。相比基于最小方差估计器的经验组合模型反演SWH的均方根误差分别降低了53.45%、48.06%和40.63%；相比袋装树机器学习模型反演SWH的均方根误差分别降低了21.92%、18.72%和4.47%。表明了本文方法在反演全球海面SWH方面具有显著优势。

关键词: GNSS-R, 延迟多普勒图, 海洋有效波高, 经验模型, 深度学习模型

Abstract:

Global navigation satellite system-reflectometry (GNSS-R), as an emerging observation method, has recently been applied to the retrieval of significant wave height (SWH). Existing studies typically use extracting features from delay Doppler maps (DDMs) to construct empirical geophysical model functions (GMFs) for SWH retrieval. However, using multiple variable parameters as model inputs poses significant challenges. Therefore, this article proposes a deep learning network model (named GloWH-Net) that integrates spaceborne GNSS-R data and multivariate parameters to invert global sea surface SWH. To verify the performance of the proposed model, ERA5, Wavewatch Ⅲ (WW3), and AVISO SWH data were used as reference data for extensive testing to evaluate the SWH retrieval performance of the GloWH-Net model and previous models (i.e. empirical and machine learning models). The results showed that when ERA5, WW3, and AVISO SWH were used as reference data respectively, the root mean square error (RMSE) of the proposed GloWH-Net model for retrieving SWH were 0.330 m, 0.393 m, and 0.433 m, respectively, the correlation coefficients (CC) were 0.91, 0.89, and 0.84, respectively. Compared with the empirical combination model based on the minimum variance estimator (MVE), the RMSE of SWH retrieval is reduced by 53.45%, 48.06%, and 40.63%, respectively; Compared to bagging tree (BT) machine learning model, the RMSE of SWH retrieval decreased by 21.92%, 18.72%, and 4.47%, respectively. This indicates that the deep learning method proposed in this article has significant advantages in retrieving global sea surface SWH.

Key words: global navigation satellite system-reflectometry, delay Doppler maps, ocean significant wave height, empirical model, deep learning model

中图分类号:

P237

布金伟, 余科根, 汪秋兰, 李玲惠, 刘馨雨, 左小清, 常军. 融合星载GNSS-R数据和多变量参数全球海洋有效波高深度学习反演法[J]. 测绘学报, 2024, 53(7): 1321-1335.

Jinwei BU, Kegen YU, Qiulan WANG, Linghui LI, Xinyu LIU, Xiaoqing ZUO, Jun CHANG. Deep learning retrieval method for global ocean significant wave height by integrating spaceborne GNSS-R data and multivariable parameters[J]. Acta Geodaetica et Cartographica Sinica, 2024, 53(7): 1321-1335.

图/表 14

图1

表1

图2

表2

图3

表3

图4

图5

图6

表4

表5

图7

表6

图8

参考文献 29

[1]	SOULAT F, CAPARRINI M, GERMAIN O, et al. Sea state monitoring using coastal GNSS-R[J]. Geophysical Research Letters, 2004, 31(21):L21303.
[2]	SHAH R, GARRISON J L, EGIDO A, et al. Bistatic radar measurements of significant wave height using signals of opportunity in L-, S-, and Ku-bands [J]. IEEE Transactions on Geoscience and Remote Sensing, 2016, 54(2):826-841.
[3]	ALONSO-ARROYO A, CAMPS A, PARK H, et al. Retrieval of significant wave height and mean sea surface level using the GNSS-R interference pattern technique: results from a three-month field campaign [J]. IEEE Transactions on Geoscience and Remote Sensing, 2015, 53(6):3198-3209.
[4]	王鑫, 孙强, 张训械, 等. 中国首次岸基GNSS-R海洋遥感实验[J]. 科学通报, 2008, 53(5):589-592.
	WANG Xin, SUN Qiang, ZHANG Xunxie, et al. The first shore-based GNSS-R ocean remote sensing experiment in China[J]. Chinese Science Bulletin, 2008, 53(5):589-592.
[5]	邵连军, 张训械, 王鑫, 等. 利用GNSS-R信号反演海浪波高[J]. 武汉大学学报(信息科学版), 2008, 33(5):475-478.
	SHAO Lianjun, ZHANG Xunxie, WANG Xin, et al. Sea surface wave height retrieve using GNSS-R signals[J]. Geomatics and Information Science of Wuhan University, 2008, 33(5):475-478.
[6]	金玲. GNSS-R接收机及有效波高反演方法研究[D]. 北京: 北京化工大学, 2016.
	JIN Ling. Research on receiver and inversion method of significant wave height based on GNSS-R[D]. Beijing: Beijing University of Chemical Technology, 2016.
[7]	李颖, 朱雪瑗, 崔璨, 等. 船载GNSS-R有效波高测量的初步研究[J]. 海洋环境科学, 2016, 35(2):180-183.
	LI Ying, ZHU Xueyuan, CUI Can, et al. Preliminary study on ship-borne significant wave height measurement using GNSS-R signals[J]. Marine Environmental Science, 2016, 35(2):180-183.
[8]	徐飞, 孙协昌, 刘馨宁, 等. 利用机载GNSS-R的有效波高反演技术[J]. 飞行器测控学报, 2017, 36(3):212-218.
	XU Fei, SUN Xiechang, LIU Xinning, et al. A method of retrieval of significant wave height using airborne GNSS-R[J]. Journal of Spacecraft TT & C Technology, 2017, 36(3):212-218.
[9]	QIN Lingyu, LI Ying. Significant wave height estimation using multi-satellite observations from GNSS-R[J]. Remote Sensing, 2021, 13(23):4806.
[10]	俞永庆. 岸基GNSS反射信号有效波高反演研究[J]. 无线电工程, 2021, 51(10):1075-1079.
	YU Yongqing. Retrieval of significant wave height using coastal GNSS reflectometry[J]. Radio Engineering, 2021, 51(10):1075-1079.
[11]	张一, 周立. 基于NARX回归神经网络的岸基GNSS-IR有效波高反演模型分析[J]. 测绘通报, 2022(2):90-94.
	ZHANG Yi, ZHOU Li. Study on inversion model of significant wave height from shore-based GNSS-IR by using NARX recurrent neural network[J]. Bulletin of Surveying and Mapping, 2022(2):90-94.
[12]	WANG Xiaolei, HE Xiufeng, SHI Jian, et al. Estimating sea level, wind direction, significant wave height, and wave peak period using a geodetic GNSS receiver[J]. Remote Sensing of Environment, 2022, 279:113135.
[13]	ALPERS W, HASSELMANN K. Spectral signal to clutter and thermal noise properties of ocean wave imaging synthetic aperture radars[J]. International Journal of Remote Sensing, 1982, 3(4):423-446.
[14]	PENG Qin, JIN Shuanggen. Significant wave height estimation from space-borne cyclone-GNSS reflectometry[J]. Remote Sensing, 2019, 11(5):584.
[15]	YANG Shuai, JIN Shuanggen, JIA Yan, et al. Significant wave height estimation from joint CYGNSS DDMA and LES observations[J]. Sensors, 2021, 21(18):6123.
[16]	BU Jinwei, YU Kegen. Significant wave height retrieval method based on spaceborne GNSS reflectometry[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19:1503705.
[17]	BU Jinwei, YU Kegen. A new integrated method of CYGNSS DDMA and LES measurements for significant wave height estimation[J]. IEEE Geoscience and Remote Sensing Letters, 2022, 19:1505605.
[18]	布金伟, 余科根, 韩帅. 星载GNSS-R海浪有效波高反演模型构建[J]. 测绘学报, 2022, 51(9):1920-1930. DOI: 10.11947/j.AGCS.2022.20210284.
	BU Jinwei, YU Kegen, HAN Shuai. Construction of spaceborne GNSS-R ocean waves significant wave height retrieval model[J]. Acta Geodaetica et Cartographica Sinica, 2022, 51(9):1920-1930. DOI: 10.11947/j.AGCS.2022.20210284.
[19]	ASGARIMEHR M, ARNOLD C, WEIGEL T, et al. GNSS reflectometry global ocean wind speed using deep learning: development and assessment of CyGNSSnet[J]. Remote Sensing of Environment, 2022, 269:112801.
[20]	WANG Feng, YANG Dongkai, YANG Lei. Retrieval and assessment of significant wave height from CYGNSS mission using neural network[J]. Remote Sensing, 2022, 14(15):3666.
[21]	WANG Changyang, YU Kegen, ZHANG Kefei, et al. Significant wave height retrieval based on multivariable regression models developed with CYGNSS data[J]. IEEE Transactions on Geoscience and Remote Sensing, 2023, 61:4200415.
[22]	BU Jinwei, YU Kegen, ZHU Feiyang, et al. Joint retrieval of sea surface rainfall intensity, wind speed, and wave height based on spaceborne GNSS-R: a case study of the oceans near China[J]. Remote Sensing, 2023, 15(11):2757.
[23]	BU Jinwei, YU Kegen, PARK H, et al. Estimation of swell height using spaceborne GNSS-R data from eight CYGNSS satellites[J]. Remote Sensing, 2022, 14(18):4634.
[24]	CLARIZIA M P, RUF C S, JALES P, et al. Spaceborne GNSS-R minimum variance wind speed estimator[J]. IEEE Transactions on Geoscience and Remote Sensing, 2014, 52(11):6829-6843.
[25]	REYNOLDS J, CLARIZIA M P, SANTI E. Wind speed estimation from CYGNSS using artificial neural networks[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2020, 13:708-716.
[26]	WANG Changyang, YU Kegen, QU Fangyu, et al. Spaceborne GNSS-R wind speed retrieval using machine learning methods[J]. Remote Sensing, 2022, 14(14):3507.
[27]	ERTUGRUL Ö F. A novel type of activation function in artificial neural networks: trained activation function[J]. Neural Networks, 2018, 99:148-157.
[28]	BU Jinwei, YU Kegen, NI Jun, et al. Combining ERA5 data and CYGNSS observations for the joint retrieval of global significant wave height of ocean swell and wind wave: a deep convolutional neural network approach[J]. Journal of Geodesy, 2023, 97(8):81.
[29]	SOISUVARN S, JELENAK Z, SAID F, et al. The GNSS reflectometry response to the ocean surface winds and waves [J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016, 9(10):4678-4699.

模型	DDMs	与DDM相关	与接收机相关	与几何相关	其他辅助参数
BT/SVM	BRCS，effectivescattering area	DDMA，LES，TES，LEWS，TEWS，SNR	gps_eirp，sp_rx_gain	sp_inc_angle，sp_lat，sp_lon，RCG	风速、风向、水深、降雨量
ANN	—
GloWH-Net	BRCS，effectivescattering area

模型	拟合参数			拟合性能指标
模型	a	b	c	RMSE/m	R²
经验模型（DDMA方法）	0.953 3	-0.204 9	-0.094 5	0.712	0.25
经验模型（LEWS方法）	0.857 0	-0.228 7	-0.013 9	0.709	0.26

模型	RMSE/m	Bias/m	CC	MAPE/（%）
经验模型（DDMA方法）	0.715	0.004	0.49	31.18
经验模型（LEWS方法）	0.709	0.001	0.51	30.78
经验模型（DDMA+LEWS方法）	0.709	0.002	0.51	30.85
BT	0.423	-0.013	0.85	15.14
SVM	0.564	0.047	0.72	22.00
ANN	0.455	0.001	0.83	16.75
GloWH-Net	0.330	0.032	0.91	12.19

模型	RMSE/m	Bias/m	CC	MAPE/（%）
经验模型（DDMA方法）	0.757	0.050	0.48	32.02
经验模型（LEWS方法）	0.757	-0.031	0.48	33.44
经验模型（DDMA+LEWS方法）	0.757	0.046	0.48	32.09
BT	0.483	0.033	0.83	16.65
SVM	0.625	0.097	0.69	23.17
ANN	0.502	0.039	0.81	17.98
GloWH-Net	0.393	0.073	0.89	14.29

模型	RMSE/m	Bias/m	CC	MAPE/（%）
经验模型（DDMA方法）	0.707	-0.015	0.46	30.19
经验模型（LEWS方法）	0.801	-0.371	0.47	39.25
经验模型（DDMA+LEWS方法）	0.730	-0.193	0.47	33.93
BT	0.453	-0.021	0.82	16.51
SVM	0.551	0.042	0.72	21.75
ANN	0.469	-0.012	0.81	17.36
GloWH-Net	0.433	0.019	0.84	15.16

融合星载GNSS-R数据和多变量参数全球海洋有效波高深度学习反演法

Deep learning retrieval method for global ocean significant wave height by integrating spaceborne GNSS-R data and multivariable parameters

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 14

参考文献 29

相关文章 11

编辑推荐

Metrics

本文评价

辅助变量参数	RMSE/m	Bias/m	CC	MAPE/m
未考虑	0.413	-0.027	0.86	15.28
降雨（RI）	0.422	0.004	0.86	15.26
风速（WS）	0.336	0.035	0.91	12.19
风向（WD）	0.345	-0.034	0.91	13.15
水深（wd）	0.413	-0.007	0.86	14.95
RI+WS+WD+wd	0.257	-0.024	0.95	9.88

[1]	黄令勇, 李世忠, 夏俊明, 王海岩, 孙越强, 杨日新, 杜起飞, 黄志勇. 岸基条件下的星载GNSS-R干涉测高精度验证评估[J]. 测绘学报, 2024, 53(2): 239-251.
[2]	宋敏峰, 何秀凤, 王笑蕾, 肖儒雅, 贾东振, 李伟强. 顾及地球曲率及椭球高的GNSS-R几何计算方法[J]. 测绘学报, 2023, 52(6): 884-894.
[3]	布金伟, 余科根, 韩帅. 星载GNSS-R海浪有效波高反演模型构建[J]. 测绘学报, 2022, 51(9): 1920-1930.
[4]	陶庭叶, 李江洋, 朱勇超, 汪俊涛, 陈皓, 时梦杰. 阶段模型修正的星载GNSS-R土壤湿度反演方法[J]. 测绘学报, 2022, 51(9): 1942-1950.
[5]	宋越, 李厚朴, 翟国君. 机载激光测深波形去噪算法对比分析[J]. 测绘学报, 2021, 50(2): 270-278.
[6]	吴学睿, 夏俊明, 白伟华, 张兴刚. GNSS-R/IR监测地表冻融状态对延迟多普勒波形和多路径数据影响分析[J]. 测绘学报, 2019, 48(8): 1059-1066.
[7]	吴学睿, 金双根. 全极化GNSS-R陆面参数延迟多普勒图模型[J]. 测绘学报, 2018, 47(S0): 101-108.
[8]	杨磊, 杨东凯, 朱云龙, 高超群. 利用GNSS-R信号进行交通车流检测[J]. 测绘学报, 2018, 47(3): 370-375.
[9]	王泽民, 刘智康, 安家春, 林国标. 基于GPS和北斗信噪比观测值的雪深反演及其误差分析[J]. 测绘学报, 2018, 47(1): 8-16.
[10]	王娜子, 鲍李峰, 高凡. 逐历元GNSS-R测高单差和双差算法[J]. 测绘学报, 2016, 45(7): 795-802.
[11]	鲍艳松王军战,张友静,屈建军,张伟民,陈晨. 合成孔径雷达影像地形校正半经验模型[J]. 测绘学报, 2011, 40(4): 0-441.