수중글라이더 관측 기반 하이브리드 CNN을 활용한 음속최소심도 추정

김, 근태¹ ; 임, 규창² ; 박, 종진³^{, *}

1경북대학교 해양학과 석사과정
2경북대학교 경북해양과학연구소 연구교수
3경북대학교 지구시스템과학부/경북해양과학연구소 교수

Estimation of Sound Minimum Depth Using a Hybrid CNN Based on Underwater Glider Observations

Geuntae Kim¹ ; Gyuchang Lim² ; Jong-Jin Park³^{, *}

1Master's program, Department of Oceanography, Kyungpook National University
2Research professor, Kyungpook Institute of Oceanography, Kyungpook National University
3Professor, School of Earth System Sciences/Kyungpook Institute of Oceanography, Kyungpook National University

Correspondence to: ^*Jong-Jin Park Daehak-ro 80, Buk-gu, Daegu 41566, Republic of Korea E-mail: jjpark@knu.ac.kr

초록

본 연구는 SOFAR 채널의 음속최소심도(SMD)를 추정하기 위한 하이브리드 CNN 모델을 제안한다. SMD는 원거리 탐지·음영 구역 활용 등 잠수함 전술에 핵심적이나, 실시간 직접 측정은 은밀성을 저해한다. 본 연구는 전술적 접근 가능 수심대(0~150 m) 내 수온·염분 프로파일만으로 심해 SMD를 추정하는 방법론을 제시한다. 동해 수중글라이더 자료로 검증한 결과, 제안 모델은 선형 회귀 대비 최대 30.1%의 오차 감소를 달성하였으며, 소용돌이·전선 등 비선형 해역에서도 승률 73.26%, 자연 변동성 대비 상대 오차율 50% 미만을 유지하였다. 복잡한 해양 환경일수록 예측 우위가 두드러지며, 잠수함이 위치 노출 없이 심해 음향 환경을 실시간으로 파악할 수 있는 전술 수단을 제공한다.

Abstract

This paper proposes a hybrid CNN model for estimating the sound minimum depth (SMD) of the SOFAR channel. Direct real-time measurement, however, compromises stealth critical to submarine operations. This paper presents a methodology that estimates deep-water SMD from minimal temperature and salinity profiles within the tactically accessible depth range (0–150 m). Validation using underwater glider observations in the East Sea demonstrated up to a 30.1% reduction in error over a linear regression baseline, with a win rate of 73.26% and relative error below 50% of the natural variability threshold (66.1 m) in nonlinear regimes such as eddies and fronts. The model's predictive advantage grows with oceanic complexity, offering submarines an effective method to assess the deep-water acoustic environment in real time without revealing their position.

Keywords:

Underwater Glider, Sound Minimum Depth, Deep Learning, Hybrid CNN, East Sea

키워드:

수중글라이더, 음속최소심도, 딥러닝, 하이브리드 합성곱 신경망, 동해

Acknowledgments

이 논문은 2026년도 해양수산부 재원으로 해양수산과학기술진흥원의 지원을 받아 수행된 연구임(RS-2023-00256005(2520000893), 고품질 준실시간 해양그리드 데이터서비스 체계 개발).

References

Robert J. Urick, Principles of Underwater Sound (3rd ed.), McGraw-Hill, 1983.
Paul C. Etter, Underwater Acoustic Modeling and Simulation (4th ed.), CRC Press, 2013.
Roy Burcher & Louis Rydill, Concepts in Submarine Design, Cambridge University Press, 1995. [https://doi.org/10.1017/CBO9781107050211]
D. C. Webb, P. J. Simonetti, & C. P. Jones, ‘SLOCUM: An Underwater Glider Propelled by Environmental Energy,’ IEEE Journal of Oceanic Engineering, VOL. 26, NO. 4, 2001, pp. 447-452. [https://doi.org/10.1109/48.972077]
JongJin Park, ‘First Observational Finding of Submesoscale Intrathermocline Eddy in the East Sea Using Underwater Glider,’ Journal of the Korean Society of Oceanography, VOL. 24, NO. 2, 2019, pp. 332-350.
Bartolomé Garau, Simón Ruiz, Weifeng G. Zhang, Ananda Pascual, Emma Heslop, John Kerfoot, & Joaquín Tintoré, ‘Thermal Lag Correction on Slocum CTD Glider Data,’ Journal of Atmospheric and Oceanic Technology, VOL. 28, NO. 9, 2011, pp. 1065-1071. [https://doi.org/10.1175/JTECH-D-10-05030.1]
Gyuchang Lim & JongJin Park, ‘Report About First Repeated Sectional Measurements of Water Property in the East Sea Using Underwater Glider,’ Journal of the Korean Society of Oceanography, VOL. 29, NO. 1, 2024, pp. 56-76.
Gyuchang Lim & JongJin Park, ‘Vertical Structural Variability of Diurnal Internal Tides Inside a Mesoscale Anticyclonic Eddy Based on Single Virtual-Moored Slocum Glider Observations,’ Frontiers in Marine Science, VOL. 9, 2022, 920049. [https://doi.org/10.3389/fmars.2022.920049]
Gyuchang Lim & JongJin Park, ‘Glider-Resolved Sonar Performance in Mesoscale Eddy Fields of the East Sea: Seasonal Comparison of HMS and VDS at 3 kHz,’ Journal of KNST, VOL. 8, NO. 4, 2025, pp. 872-881. [https://doi.org/10.31818/JKNST.2025.12.8.4.872]