成功大學博碩士論文系統

簡易檢索 / 詳目顯示

回結果列表

研究生：	鍾崴丞 Chung, Wei-Cheng
論文名稱：	獵風者高解析度全球導航衛星系統反射接收儀延遲-都卜勒映射圖於颱風情境下之多特徵型態分析 Mluti-feature Morphological Analysis of High-resolution GNSS-R Delay-Doppler Maps from TRITON in Typhoon Conditions
指導教授：	莊智清 Juang, Jyh-Ching
學位類別：	碩士 Master
系所名稱：	電機資訊學院 - 電機工程學系 Department of Electrical Engineering
論文出版年：	2026
畢業學年度：	114
語文別：	英文
論文頁數：	127
中文關鍵詞：	獵風者、TRITON 、全球導航衛星系統反射接收儀、延遲 - 都卜勒映射圖、颱風、DDM 形態分析、海面粗糙度
外文關鍵詞：	TRITON, GNSS-R, Delay-Doppler Map, typhoon, DDM morphology, sea-surface roughness
相關次數：	點閱：5 下載：0
分享至:	分享至facebook 分享至twitter

查詢本校圖書館目錄查詢臺灣博碩士論文知識加值系統勘誤回報

全球導航衛星系統反射接收儀（GNSS-R）可利用既有導航衛星訊號之海面反射特性觀測海洋狀態，在颱風等強風與降雨環境下具有持續觀測潛力。既有 GNSS-R 颱風相關研究多以 CYGNSS 等任務資料為基礎，將延遲 - 都卜勒映射圖（DDM）轉換為延遲－都卜勒圖平均、前緣斜率、正規化雙基地雷達散射截面等純量觀測量，用於熱帶氣旋海面風速反演、風場分析、颱風預報同化或颱風結構偵測。這些研究說明 GNSS-R 觀測對颱風海面環境具有應用價值，但其主要目標多集中於風速或位置等物理量估計，較少直接檢查高解析度 DDM 在颱風條件下的二維功率分布是否呈現穩定的形態重分布。

本研究以獵風者（TRITON）衛星 GNSS-R L1 DDM 為主要資料，結合 IBTrACS 颱風路徑與 ERA5 波高、風速資料，建立颱風樣本與遠場背景樣本。資料涵蓋 2024--2025 年 23 個西北太平洋颱風個案，包含 6,506 筆篩選後颱風樣本（經入射角與 SNR 配對後保留 6,385 筆）與約 437,000 筆颱風活躍期遠場背景樣本。分析上，本研究除使用 DDMA 與 LES 外，亦整理 Doppler 變異度、Doppler 有效展寬、核心－翼部對比與 3 dB 方向性寬度比例等 DDM 形態摘要量，並以入射角與 SNR 配對、重抽樣、隨機標籤對照、跨年驗證與海況配對檢查結果穩定性。

結果顯示，颱風樣本相對於背景樣本在 Delay-Doppler 平面上呈現可重現的形態偏移，而不只是單一純量觀測量的增減。經二維配對後，Doppler 變異度、Doppler 有效展寬、核心－翼部對比與 3 dB 方向性寬度比例大致保留一致方向；由上述四項指標組成的核心形態組合，在跨年測試中的保留年份 AUC 約為 0.64--0.66，高於僅使用 DDMA 與 LES 的純量反應組。此結果表示，高解析度 TRITON DDM 的二維形態資訊可補充傳統純量 GNSS-R 觀測量，提供描述颱風條件下海面反射形態改變的另一種分析角度。本文結果應解讀為弱至中等的探索性排序訊號，而非作業化颱風偵測系統。

Global Navigation Satellite System Reflectometry (GNSS-R) processes reflected navigation satellite signals to observe ocean surface conditions. As the navigation signals are in the L band, the signals are not subject to the rain affection and have the potential to provide information about strong-wind and rainfall in the presence of typhoons. A typhoon-related GNSS-R study is the NASA CYGNSS mission. In the mission, Delay-Doppler maps (DDMs) are observed as scalar observables including DDMA (Delay-Doppler Map Averaging), LES (Leading Edge Slope), and NBRCS (Normalized Bistatic Radar Cross Section) are obtained for tropical-cyclone wind-speed retrieval, wind-field analysis, forecast assimilation, or storm-structure detection. These studies demonstrate the value of GNSS-R observations for typhoon-related ocean environments, but they have focused mainly on estimating physical quantities such as wind speed or storm position. Less attention has been given to analyze whether the two-dimensional power distribution of high-resolution DDMs exhibits consistent morphological redistribution under typhoon conditions.

This thesis uses TRITON GNSS-R L1 DDMs, together with IBTrACS typhoon tracks and ERA5 wave-height and wind-speed data, to construct typhoon samples and far-field background samples. The dataset covers 23 western North Pacific typhoon cases during 2024--2025, including 6,506 screened typhoon samples (of which 6,385 remain after incidence-angle/SNR matching) and approximately 437,000 far-field background samples during active typhoon periods. In addition to DDMA and LES, this study examines DDM morphological summaries, including Doppler variance, Doppler effective spread, core-wing contrast, and the 3 dB directional width ratio, and checks their stability using incidence-angle/SNR matching, resampling, random-label controls, cross-year validation, and sea-state matching.

The results show that typhoon samples exhibit reproducible morphological shifts on the Delay-Doppler plane relative to background samples, rather than a simple increase or decrease in a single scalar observable. The four morphology indicators retain more consistent effect directions after matching, and their core morphology set achieves held-out cross-year AUC values of approximately 0.64--0.66, higher than the scalar-response set using only DDMA and LES. These findings suggest that high-resolution TRITON DDM morphology can complement traditional scalar GNSS-R observables by describing two-dimensional power-distribution changes that scalar quantities do not fully capture. The current results should be interpreted as weak-to-moderate exploratory ranking signals, not as an operational typhoon detection system.

摘要 I
Abstract III
致謝 V
Contents VI
List of Tables XI
List of Figures XII
List of Acronyms XIV
Chapter 1 Introduction 1
1 Research Motivation 1
2 Literature Review 2
3 Research Contributions 4
4 Thesis Organization 5
Chapter 2 Theoretical Background 7
1 GNSS-R Bistatic Scattering over the Sea Surface 7
2 Formation of the Delay-Doppler Map 8
3 Common GNSS-R Observables and Their Limitations 10
4 DDM Morphological Information and the Positioning of Observables 11
5 TRITON High-Resolution DDM and the Positioning of This Study 12
6 Summary 14
Chapter 3 Research Data and Methods 15
1 Overview of the Research Design 15
2 Data Sources 16
2.1 TRITON DDM Observation Data 16
2.2 IBTrACS Typhoon Data 17
2.3 ERA5 Sea-State Auxiliary Data 17
3 TY/BG Sample Definition and Quality Filtering 18
3.1 TY Samples and the Background Exclusion Radius 18
3.2 Typhoon Intensity Filtering 19
3.3 Definition of the BG Sample Pool 20
3.4 Quality and Geometric Filtering 20
4 DDM Preprocessing and Morphology Indicators 22
4.1 Purpose of Preprocessing and the Input Field 22
4.2 Non-Negativization and Peak Normalization 23
4.3 Conventional Scalar Response Indicators 24
4.4 Second-Moment Spread Indicators 25
4.5 Doppler Effective Spread 27
4.6 Contour Area and Core-Wing Contrast 30
4.7 3 dB Directional Width Ratio 33
4.8 Summary of Candidate Indicators 36
5 Incidence-Angle and SNR Matched Sampling and Bootstrap 37
5.1 Control Objectives of Two-Dimensional Matching 37
5.2 Incidence-Angle and SNR Binning and Matched-Cell Sampling 38
5.3 Nested Matched Resampling 41
5.4 Relative-Velocity-Magnitude Sensitivity Design 43
6 DDM Image Difference Analysis 44
7 Negative-Control Checks 46
8 Correlation Analysis and PCA 47
8.1 Purpose of the Analysis 47
8.2 Construction of the TY-Only PCA 47
8.3 Projection of the Matched TY/BG Samples 48
9 Feature-Group Ablation and Core Morphology Combination Selection 49
9.1 Candidate Subsets and the Bidirectional Cross-Year Framework 49
9.2 The L2-Regularized Logistic Regression under Matching 50
9.3 Within-Cell-Stratified AUC 53
9.4 Pareto Candidate Set and Feature-Combination Definition 54
10 Cross-Model Generalization of the Core Combination 54
10.1 Objective of the Check 54
10.2 Linear and Nonlinear Models 55
10.3 Resampling, Uncertainty, and Negative Controls 55
11 Robustness Check Using Local Sea-State Conditions 56
11.1 Sea-State Data and Control Purpose 56
11.2 Rationale for the Four-Dimensional Matching Design 57
11.3 Sea-State Binning 57
11.4 Sample Composition Diagnostics 59
12 Summary and Overview of the Analysis Framework 60
12.1 Review of the Methodological Workflow 60
12.2 Summary of the Analysis Framework 60
Chapter 4 Results and Analysis 62
1 Distribution of Typhoon Samples 62
1.1 Annual and Temporal Distribution 63
1.2 Typhoon Intensity Distribution 64
1.3 Distribution of Sample Sizes 65
2 Differences in Average DDM Morphology 65
2.1 Balance of Observation Conditions After Two-Dimensional Matching 65
2.2 DDM Image Differences Under Repeated Matching 67
2.3 Spatial Distribution of the Differences 69
2.4 Background-Sample Negative Control 69
3 Matched Indicator Effects 71
3.1 Matched Indicator Effects 71
3.2 Morphological Interpretation of the Indicator Differences 73
3.3 Relative-Velocity Geometry Sensitivity Check 74
3.4 Random-Label Permutation Check 77
4 Feature Structure: Correlation and PCA 78
4.1 Correlation Structure of the Candidate Indicators 79
4.2 Principal Axes of Variation Within TY Samples 81
4.3 Low-Dimensional Structure and the TY/BG Difference 82
5 Selection of the Core Morphology Subset 84
5.1 Exploratory Screening of All Candidate Subsets 85
5.2 Outer-Loop Validation of the Pareto Candidate 86
6 Cross-Model Generalization of the Core Subset 87
6.1 Cross-Year Score Distributions 87
6.2 Low-Dimensional Morphological Structure of the Core Combination 89
6.3 Nonlinear Model Sensitivity 91
6.4 Random-Label Negative Control and Regularization Sensitivity 93
7 Robustness Checks Using Local Sea-State Conditions 95
7.1 Four-Dimensional Sea-State Matching Design 95
7.2 Sea-State Stratification and Matching Balance 96
7.3 Conditional Heterogeneity Across Sea-State Intervals 99
7.4 Random-Label Check Within Stratified Cells 101
8 Summary 102
Chapter 5 Conclusions and Future Work 104
1 Conclusions 104
2 Limitations 105
3 Future Work 106
References 107
                                    

[1] V. U. Zavorotny, S. Gleason, E. Cardellach, and A. Camps, "Tutorial on remote sensing using GNSS bistatic radar of opportunity," IEEE Geosci. Remote Sens. Mag., vol. 2, no. 4, pp. 8–45, Dec. 2014, doi: 10.1109/MGRS.2014.2374220.
[2] C. Ruf, D. McKague, M. Morris, D. Posselt, and M. Moghaddam, "The GNSS-R CYGNSS mission: An update," in Proc. IEEE Int. Geosci. Remote Sens. Symp. (IGARSS), Yokohama, Japan, 2019, pp. 5171–5172, doi: 10.1109/IGARSS.2019.8900604.
[3] M.-Y. Chen, H. Chien, W.-H. Yeh, L.-C. Lin, and Y.-F. Liou, "Evaluation of the effect of satellite motion on GNSS-R wind speed retrieval: Insights from TRITON," Front. Remote Sens., vol. 6, 2025, Art. no. 1657576, doi: 10.3389/frsen.2025.1657576.
[4] V. U. Zavorotny and A. G. Voronovich, "Scattering of GPS signals from the ocean with wind remote sensing application," IEEE Trans. Geosci. Remote Sens., vol. 38, no. 2, pp. 951–964, Mar. 2000, doi: 10.1109/36.841977.
[5] M. P. Clarizia and C. S. Ruf, "Wind speed retrieval algorithm for the Cyclone Global Navigation Satellite System (CYGNSS) mission," IEEE Trans. Geosci. Remote Sens., vol. 54, no. 8, pp. 4419–4432, Aug. 2016, doi: 10.1109/TGRS.2016.2541343.
[6] C. S. Ruf and R. Balasubramaniam, "Development of the CYGNSS geophysical model function for wind speed," IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 12, no. 1, pp. 66–77, Jan. 2019, doi: 10.1109/JSTARS.2018.2833075.
[7] L. Ricciardulli, C. Mears, A. Manaster, and T. Meissner, "Assessment of CYGNSS wind speed retrievals in tropical cyclones," Remote Sens., vol. 13, no. 24, Dec. 2021, Art. no. 5110, doi: 10.3390/rs13245110.
[8] M. P. Clarizia, V. U. Zavorotny, D. S. McKague, and C. S. Ruf, "CYGNSS Level 2 wind speed retrieval algorithm theoretical basis document," Univ. Michigan, Ann Arbor, MI, USA, UM Project Doc. 148-0138, Rev. 6, Sep. 14, 2020. [Online]. Available: https://podaac.jpl.nasa.gov/dataset/CYGNSS_L2_V3.0. Accessed: Jul. 14, 2026.
[9] M. M. Al-Khaldi, S. J. Katzberg, J. T. Johnson, Y. Kang, E. J. Kubatko, and S. Gleason, "Matched filter cyclone maximum wind retrievals using CYGNSS: Progress update and error analysis," IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 14, pp. 3591–3601, 2021, doi: 10.1109/JSTARS.2021.3066406.
[10] K.-J. Lin, S.-C. Yang, and S. S. Chen, "Improving analysis and prediction of tropical cyclones by assimilating radar and GNSS-R wind observations: Ensemble data assimilation and observing system simulation experiments using a coupled atmosphere–ocean model," Weather Forecasting, vol. 37, no. 9, pp. 1689–1711, Sep. 2022, doi: 10.1175/WAF-D-21-0202.1.
[11] M. Asgarimehr, C. Arnold, T. Weigel, C. Ruf, and J. Wickert, "GNSS reflectometry global ocean wind speed using deep learning: Development and assessment of CyGNSSnet," Remote Sens. Environ., vol. 269, Feb. 2022, Art. no. 112801, doi: 10.1016/j.rse.2021.112801.
[12] W. Guo, H. Du, C. Guo, B. J. Southwell, J. W. Cheong, and A. G. Dempster, "Information fusion for GNSS-R wind speed retrieval using statistically modified convolutional neural network," Remote Sens. Environ., vol. 272, Apr. 2022, Art. no. 112934, doi: 10.1016/j.rse.2022.112934.
[13] S. Arabi, M. Asgarimehr, M. Kada, and J. Wickert, "Hybrid CNN-LSTM deep learning for track-wise GNSS-R ocean wind speed retrieval," Remote Sens., vol. 15, no. 17, Aug. 2023, Art. no. 4169, doi: 10.3390/rs15174169.
[14] S. Yang, S. Jin, Y. Jia, and M. Ye, "Significant wave height estimation from joint CYGNSS DDMA and LES observations," Sensors, vol. 21, no. 18, Sep. 2021, Art. no. 6123, doi: 10.3390/s21186123.
[15] J. Zhen, Y. Zhu, D. Yang, G. Zhang, and F. Wang, "Detection of typhoons and estimation of eye position using satellite-based GNSS-reflectometry," J. Beijing Univ. Aeronaut. Astronaut., vol. 51, no. 6, pp. 2106–2118, 2025, doi: 10.13700/j.bh.1001-5965.2023.0395 (in Chinese).
[16] M. P. Clarizia, C. S. Ruf, P. Jales, and C. Gommenginger, "Spaceborne GNSS-R minimum variance wind speed estimator," IEEE Trans. Geosci. Remote Sens., vol. 52, no. 11, pp. 6829–6843, Nov. 2014, doi: 10.1109/TGRS.2014.2303831.
[17] R. Balasubramaniam and C. S. Ruf, "Azimuthal dependence of GNSS-R scattering cross-section in hurricanes," J. Geophys. Res. Oceans, vol. 125, no. 7, Jul. 2020, Art. no. e2020JC016167, doi: 10.1029/2020JC016167.
[18] S. Soisuvarn, Z. Jelenak, F. Said, P. S. Chang, and A. Egido, "The GNSS reflectometry response to the ocean surface winds and waves," IEEE J. Sel. Topics Appl. Earth Observ. Remote Sens., vol. 9, no. 10, pp. 4678–4699, Oct. 2016, doi: 10.1109/JSTARS.2016.2602703.
[19] Taiwan Space Agency (TASA), "TRITON algorithm theoretical basis document," ver. 2.0, 2025. [Online]. Available: https://tacc.cwa.gov.tw/v2/en/triton_download.html. Accessed: Jul. 14, 2026.
[20] Taiwan Analysis Center for COSMIC (TACC), "TRITON GNSS-R data products," Central Weather Administration, Taipei, Taiwan, 2024. [Online]. Available: https://tacc.cwa.gov.tw/v2/en/triton_download.html. Accessed: Jul. 14, 2026.
[21] K. R. Knapp, M. C. Kruk, D. H. Levinson, H. J. Diamond, and C. J. Neumann, "The International Best Track Archive for Climate Stewardship (IBTrACS): Unifying tropical cyclone best track data," Bull. Amer. Meteorol. Soc., vol. 91, no. 3, pp. 363–376, Mar. 2010, doi: 10.1175/2009BAMS2755.1.
[22] J. Gahtan, K. R. Knapp, C. J. Schreck, III, H. J. Diamond, J. P. Kossin, and M. C. Kruk, "International Best Track Archive for Climate Stewardship (IBTrACS) Project, Version 4.01," Western Pacific basin subset, NOAA National Centers for Environmental Information, 2024, doi: 10.25921/82ty-9e16. Accessed: Jul. 14, 2026.
[23] H. Hersbach et al., "The ERA5 global reanalysis," Q. J. Roy. Meteorol. Soc., vol. 146, no. 730, pp. 1999–2049, Jul. 2020, doi: 10.1002/qj.3803.
[24] Copernicus Climate Change Service (C3S), "ERA5 hourly data on single levels from 1940 to present," Climate Data Store (CDS), 2018, doi: 10.24381/cds.adbb2d47. Accessed: Jul. 14, 2026.

簡易檢索 / 詳目顯示

相關論文