基于北斗三号PPP-B2b同震位移与深度学习的震级快速估算——以2025年定日Mw 7.1和曼德勒Mw 7.7地震为例

doi:10.11947/j.AGCS.2026.20260112

测绘学报 ›› 2026, Vol. 55 ›› Issue (7): 1171-1182.doi: 10.11947/j.AGCS.2026.20260112

• 人工智能的大地测量与卫星重力反演前沿 • 上一篇

基于北斗三号PPP-B2b同震位移与深度学习的震级快速估算——以2025年定日Mw 7.1和曼德勒Mw 7.7地震为例

李贺¹(), 陈克杰¹(), 崔文峰¹, 柴海山¹, 方荣新², 彭朝勇³, 孙丽⁴

^1.南方科技大学地球与空间科学系，广东　深圳　518055
^2.武汉大学卫星导航定位技术研究中心，湖北　武汉　430079
^3.中国地震局地球物理研究所地震动力学与强震预测全国重点实验室，北京　100081
^4.中国地震局台网中心，北京　100045

收稿日期:2026-03-27 修回日期:2026-05-07 发布日期:2026-08-18
通讯作者: 陈克杰 E-mail:lihe12100@gmail.com;chenkj@sustech.edu.cn
作者简介:李贺（2003—），男，硕士生，主要从事基于深度学习的大震震级实时估算研究。 E-mail：lihe12100@gmail.com
基金资助:
国家自然科学基金(42474046; 42274025);应急管理部重点科技计划(2024EMST040402);国家重点研发计划(2024YFC3012800)

Rapid magnitude estimation based on BDS-3 PPP-B2b coseismic displacement and deep learning: a case study of the 2025 Tingri Mw 7.1 and Mandalay Mw 7.7 earthquakes

He Li¹(), Kejie Chen¹(), Wenfeng Cui¹, Haishan Chai¹, Rongxin Fang², Chaoyong Peng³, Li Sun⁴

^1.Department of Earth and Space Sciences, Southern University of Science and Technology, Shenzhen 518055, China
^2.Satellite Navigation and Positioning Technology Research Center, Wuhan University, Wuhan 430079, China
^3.State Key Laboratory of Earthquake Dynamics and Forecasting, Institute of Geophysics, China Earthquake Administration, Beijing 100081, China
^4.China Earthquake Networks Center, China Earthquake Administration, Beijing 100045, China

Received:2026-03-27 Revised:2026-05-07 Published:2026-08-18
Contact: Kejie Chen E-mail:lihe12100@gmail.com;chenkj@sustech.edu.cn
About author:Li He (2003—), male, postgraduate, majors in real-time magnitude estimation of large earthquakes based on deep learning. E-mail: lihe12100@gmail.com
Supported by:
The National Natural Science Foundation of China(42474046; 42274025);Key Science and Technology Program of Ministry of Emergency Management(2024EMST040402);The National Key Research and Development Program of China(2024YFC3012800)

摘要/Abstract

摘要：

快速准确的震源表征是提升地震预警时效性与准确性的核心。基于全球历史大震高频GNSS波形构建的通用训练数据集，本文开发了一种名为CTSequence的时空序列深度学习模型，旨在利用基于PPP-B2b技术的实时GNSS观测流，同步捕捉地震破裂的时空演化特征并确定震级。将该模型应用于2025年我国西藏定日Mw 7.1地震和缅甸曼德勒Mw 7.7地震，以验证其泛化性能。结果显示，模型在P波到达首个台站后约30 s内即可提供有价值的震级估算，并在震后75 s左右迅速收敛至真实震级附近。在测试集中，200 s时最终准确率为99.6%。此外，模型展现出优异的稳健性，即便在台站稀疏（N=8）的观测条件下，仍能保持稳定的估算性能。本文模型可为地震预警系统提供稳定、快速且具备不确定性表征能力的震级估算支撑，尤其适用于台站稀疏或网络通信中断的极端环境。

关键词: 高频GNSS, 地震预警, 深度学习, 震级估算, PPP-B2b

Abstract:

Rapid and accurate source characterization is central to improving the timeliness and accuracy of earthquake early warning. Based on a universal training dataset constructed from high-rate GNSS waveforms of historical global major earthquakes, this study develops a spatiotemporal sequence deep learning model named CTSequence, which is designed to simultaneously capture the spatiotemporal evolution of earthquake rupture and determine magnitude using real-time GNSS observation streams based on PPP-B2b technology. The model is applied to the 2025 Tingri, Xizang Mw 7.1 earthquake and the Mandalay, Myanmar Mw 7.7 earthquake to validate its generalization performance. Results show that the model can provide valuable magnitude estimates within approximately 30 s after the P-wave arrival at the first station and converges rapidly to the true magnitude around 75 s after the earthquake origin time. In the test dataset, the final accuracy at 200 s reaches 99.6%. Furthermore, the model demonstrates excellent robustness, maintaining stable estimation performance even under sparse station conditions (N=8). This approach can provide stable, rapid, and uncertainty-characterized magnitude estimation support for earthquake early warning systems, and is particularly suitable for extreme environments with sparse station coverage or network communication disruptions.

Key words: high-rate GNSS, earthquake early warning, deep learning, magnitude estimation, PPP-B2b

中图分类号:

P228

李贺, 陈克杰, 崔文峰, 柴海山, 方荣新, 彭朝勇, 孙丽. 基于北斗三号PPP-B2b同震位移与深度学习的震级快速估算——以2025年定日Mw 7.1和曼德勒Mw 7.7地震为例[J]. 测绘学报, 2026, 55(7): 1171-1182.

He Li, Kejie Chen, Wenfeng Cui, Haishan Chai, Rongxin Fang, Chaoyong Peng, Li Sun. Rapid magnitude estimation based on BDS-3 PPP-B2b coseismic displacement and deep learning: a case study of the 2025 Tingri Mw 7.1 and Mandalay Mw 7.7 earthquakes[J]. Acta Geodaetica et Cartographica Sinica, 2026, 55(7): 1171-1182.

导出引用管理器 EndNote|Reference Manager|ProCite|BibTeX|RefWorks

链接本文: http://xb.chinasmp.com/CN/10.11947/j.AGCS.2026.20260112

http://xb.chinasmp.com/CN/Y2026/V55/I7/1171

图/表 10

图1

图2

表1

图3

图4

图5

图6

图7

图8

图9

参考文献 33

[1]	Allen R M, Brown H, Hellweg M, et al. Real-time earthquake detection and hazard assessment by ElarmS across California[J]. Geophysical Research Letters, 2009, 36(5): L00B08.
[2]	孙丽. 地震预警的新发展：新型传感器网络的应用[J]. 地球与行星物理论评(中英文), 2023, 54(1): 105-107.
	Sun Li. The new development of EEW: the application of new sensor networks[J]. Reviews of Geophysics and Planetary Physics, 2023, 54(1): 105-107.
[3]	Melgar D, Bock Y, Sanchez D, et al. On robust and reliable automated baseline corrections for strong motion seismology[J]. Journal of Geophysical Research: Solid Earth, 2013, 118(3): 1177-1187.
[4]	Meier M A, Heaton T, Clinton J. Evidence for universal earthquake rupture initiation behavior[J]. Geophysical Research Letters, 2016, 43(15): 7991-7996.
[5]	Hoshiba M, Iwakiri K, Hayashimoto N, et al. Outline of the 2011 off the Pacific coast of Tohoku Earthquake (Mw 9.0): earthquake early warning and observed seismic intensity[J]. Earth, Planets and Space, 2011, 63(7): 547-551.
[6]	Marcou S, Lux A I, Akimov A, et al. Exploring end-to-end earthquake early warning performance in large earthquakes using the February 2023 Kahramanmara ş, Türkiye sequence[J]. Scientific Reports, 2026, 16: 327.
[7]	陈克杰, 柴海山, 李佳峰, 等. GNSS海啸预警：地表形变监测与电离层扰动探测[J]. 中国科学(地球科学), 2026, 56(3): 928-942.
	Chen Kejie, Chai Haishan, Li Jiafeng, et al. An overview of GNSS for tsunami early warning: From lithosphere to ionosphere[J]. Scientia Sinica (Terrae), 2026, 56(3): 928-942.
[8]	Bock Y, Melgar D, Crowell B W. Real-time strong-motion broadband displacements from collocated GPS and accelerometers[J]. Bulletin of the Seismological Society of America, 2011, 101(6): 2904-2925.
[9]	Chen Kejie, Ge Maorong, Babeyko Andrey, et al. Retrieving real-time co-seismic displacements using GPS/GLONASS: a preliminary report from the September 2015Mw 8.3 Illapel earthquake in Chile[J]. Geophysical Journal International, 2016, 206(2): 941-953.
[10]	Zumberge J F, Heflin M B, Jefferson D C, et al. Precise point positioning for the efficient and robust analysis of GPS data from large networks[J]. Journal of Geophysical Research: Solid Earth, 1997, 102(B3): 5005-5017.
[11]	张小红, 李星星, 李盼. GNSS精密单点定位技术及应用进展[J]. 测绘学报, 2017, 46(10): 1399-1407. DOI: . doi: 10.11947/j.AGCS.2017.20170327
	Zhang Xiaohong, Li Xingxing, Li Pan. Review of GNSS PPP and its application[J]. Acta Geodaetica et Cartographica Sinica, 2017, 46(10): 1399-1407. DOI: . doi: 10.11947/j.AGCS.2017.20170327
[12]	柴海山, 陈克杰, 魏国光, 等. 北斗三号与超高频GNSS同震形变监测：以2021年青海玛多Mw 7.4地震为例[J]. 武汉大学学报(信息科学版), 2022, 47(6): 946-954.
	Chai Haishan, Chen Kejie, Wei Guoguang, et al. Coseismic deformation monitoring using BDS-3 and ultra-high rate GNSS: a case study of the 2021 Maduo Mw 7.4 earthquake[J]. Geomatics and Information Science of Wuhan University, 2022, 47(6): 946-954.
[13]	Chen Kejie, Chai Haishan, Li Jiafeng, et al. An overview of GNSS for tsunami early warning: from lithosphere to ionosphere[J]. Science China Earth Sciences, 2026, 69(3): 898-912.
[14]	中国卫星导航系统管理办公室. 北斗卫星导航系统空间信号接口控制文件：精密单点定位服务信号(PPP-B2b)(1.0版)[S]. 北京: 中国卫星导航系统管理办公室, 2020.
	China Satellite Navigation Office. BeiDou navigation satellite system signal in space interface control document: precise point positioning service signal (PPP-B2b)(V1.0)[S]. Beijing: China Satellite Navigation Office, 2020.
[15]	Fang Rongxin, Lü Huanghui, Hu Zhigang, et al. GPS/BDS precise point positioning with B2b products for high-rate seismogeodesy: application to the 2021Mw 7.4 Maduo earthquake[J]. Geophysical Journal International, 2022, 231(3): 2079-2090.
[16]	Mousavi S M, Beroza G C. Deep-learning seismology[J]. Science, 2022, 377(6607): eabm4470.
[17]	LeCun Y, Bengio Y, Hinton G. Deep learning[J]. Nature, 2015, 521(7553): 436-444.
[18]	Meier M A, Ross Z E, Ramachandran A, et al. Reliable real-time seismic signal/noise discrimination with machine learning[J]. Journal of Geophysical Research: Solid Earth, 2019, 124(1): 788-800.
[19]	Zhu Weiqiang, Beroza G C. PhaseNet: a deep-neural-network-based seismic arrival time picking method[J]. Geophysical Journal International, 2019, 216(1): 261-273.
[20]	Perol T, Gharbi M, Denolle M. Convolutional neural network for earthquake detection and location[J]. Science Advances, 2018, 4(2): e1700578.
[21]	Kuang Wenhuan, Yuan Congcong, Zhang Jie. Real-time determination of earthquake focal mechanism via deep learning[J]. Nature Communications, 2021, 12: 1432.
[22]	Mousavi S M, Beroza G C. A machine-learning approach for earthquake magnitude estimation[J]. Geophysical Research Letters, 2020, 47(1): e2019GL085976.
[23]	Taroni M, Spassiani I, Laskin N, et al. How many strong earthquakes will there be tomorrow?[J]. Frontiers in Applied Mathematics and Statistics, 2023, 9: 1152476.
[24]	Frankel A, Wirth E, Marafi N, et al. Broadband synthetic seismograms for Magnitude 9 earthquakes on the cascadia megathrust based on 3D simulations and stochastic synthetics, part 1: methodology and overall results[J]. Bulletin of the Seismological Society of America, 2018, 108(5A): 2347-2369.
[25]	Lin J T, Melgar D, Thomas A M, et al. Early warning for great earthquakes from characterization of crustal deformation patterns with deep learning[J]. Journal of Geophysical Research: Solid Earth, 2021, 126(10): e2021JB022703.
[26]	Cui Wenfeng, Chen Kejie, Wei Guoguang, et al. Simultaneous magnitude and slip distribution characterization from high-rate GNSS using deep learning: case studies of the 2021Mw 7.4 Maduo and 2023 Turkey doublet events[J]. Geophysical Journal International, 2024, 238(1): 91-108.
[27]	Quinteros-Cartaya C, Quintero-Arenas J, Padilla-Lafarga A, et al. A deep learning pipeline for large earthquake analysis using high-rate global navigation satellite system data[J]. Earth Science Informatics, 2025, 18(4): 516.
[28]	Ruhl C J, Melgar D, Geng Jianghui, et al. A global database of strong-motion displacement GNSS recordings and an example application to PGD scaling[J]. Seismological Research Letters, 2019, 90(1): 271-279.
[29]	肖懿思, 方荣新, 李瑜, 等. 利用高频GNSS观测研究2025年西藏定日Ms 6.8地震地表形变及预警震级[J]. 武汉大学学报(信息科学版), 2025, 50(11): 2176-2186.
	Xiao Yisi, Fang Rongxin, Li Yu, et al. Surface displacement and early warning magnitude of the 2025 Dingri, Xizang Ms 6.8 earthquake using high-rate GNSS observations[J]. Geomatics and Information Science of Wuhan University, 2025, 50(11): 2176-2186.
[30]	Melgar D, Crowell B W, Geng Jianghui, et al. Earthquake magnitude calculation without saturation from the scaling of peak ground displacement[J]. Geophysical Research Letters, 2015, 42(13): 5197-5205.
[31]	Laske G, Masters G, Ma Z, et al. Update on CRUST1.0: a 1-degree global model of Earth's crust[C]//Proceedings of 2013 Geophysical Research Abstracts. Munich: European Geosciences Union, 2013, 15: EGU2013-2658.
[32]	Crowell B W, Melgar D, Bock Y, et al. Earthquake magnitude scaling using seismogeodetic data[J]. Geophysical Research Letters, 2013, 40(23): 6089-6094.
[33]	Jiang Weiping, Zhao Qile, Li Min, et al. The progress of IGS analysis center at Wuhan University[J]. Journal of Geodesy and Geoinformation Science, 2023, 6(3): 46-57.

回归系数	文献[30]方法/cm	文献[28]方法/m	文献[32]方法/cm
A	－6.687	－5.919	－4.434
B	1.5	1.009	1.047
C	－0.214	－0.145	－0.138

基于北斗三号PPP-B2b同震位移与深度学习的震级快速估算——以2025年定日Mw 7.1和曼德勒Mw 7.7地震为例

Rapid magnitude estimation based on BDS-3 PPP-B2b coseismic displacement and deep learning: a case study of the 2025 Tingri Mw 7.1 and Mandalay Mw 7.7 earthquakes

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 10

参考文献 33

相关文章 15

编辑推荐

Metrics

本文评价

[1]	时天东, 赵玲, 赵文豪, 齐霁, 崔浩, 彭程里, 张新长. 时空信息显式引导的高分光学遥感影像可控生成方法[J]. 测绘学报, 2026, 55(5): 894-908.
[2]	王家耀, 陈琳, 程士源, 王利军, 熊思奇. 人工智能赋能地图科学数智化[J]. 测绘学报, 2026, 55(3): 381-389.
[3]	禄小敏, 张志义, 闫浩文, 何毅, 苏小宁. 融合深度图信息最大化和多层感知机的建筑物群组模式识别方法[J]. 测绘学报, 2026, 55(3): 425-438.
[4]	季顺平, 刘瑾, 高建, 龚健雅. 多视影像深度学习密集匹配三维重建智能框架[J]. 测绘学报, 2025, 54(9): 1633-1646.
[5]	张继贤, 顾海燕, 倪欢, 李海涛, 杨懿, 丁少鹏, 隋淞蔓. 遥感智能变化检测的深度学习方法：演变与发展趋势[J]. 测绘学报, 2025, 54(8): 1347-1370.
[6]	方帅, 刘加恩, 张晶. 自适应参考特征引入与多尺度特征聚合的时空融合算法[J]. 测绘学报, 2025, 54(8): 1476-1488.
[7]	孟妮娜, 李凤梅, 周校东. 数据与认知双驱动的建筑物群制图综合结果与尺度一致性识别[J]. 测绘学报, 2025, 54(7): 1318-1331.
[8]	王亚青, 王中辉. 异构图卷积网络支持下的河系自动选取方法[J]. 测绘学报, 2025, 54(7): 1332-1345.
[9]	安晓亚, 郭伟茹, 张鹏鑫, 李欣欣, 石磊. 顾及几何位置和移动特征相似性的船舶轨迹聚类方法[J]. 测绘学报, 2025, 54(6): 1107-1121.
[10]	王超, 陈天宇, 张同, AhmedTanvir, 纪立强, 谢涛, 杨佳俊, 王帅. 基于全局差分增强模块和平衡惩罚损失的多源光学遥感影像变化检测[J]. 测绘学报, 2025, 54(5): 873-887.
[11]	罗卿莉, 李雪岩, 黄国满, 陈红辉, 薛铭龙, 李健. AOSN：α-最优网络模型的山区单通道SAR高程重建方法[J]. 测绘学报, 2025, 54(5): 888-898.
[12]	涂伟, 池向沅, 赵天鸿, 杨剑, 朱世平, 陈德莉. 城市排水管网流量预测多视图时空图神经网络模型[J]. 测绘学报, 2025, 54(2): 334-344.
[13]	张志力, 姜慧伟, 胡翔云. 面向极简交互的遥感地物精确批量提取框架[J]. 测绘学报, 2025, 54(10): 1863-1876.
[14]	张正华, 陈国良. 一种轻量且旋转不变的激光雷达位置识别网络[J]. 测绘学报, 2025, 54(1): 90-103.
[15]	石岩, 王达, 邓敏, 杨学习. 时空异常探测：从数据驱动到知识驱动的内涵转变与实现路径[J]. 测绘学报, 2024, 53(8): 1493-1504.