隨著氣候變遷與極端氣候事件日益頻繁，瘋狗浪（Coastal Freak Waves）與越堤浪等突發性強波浪現象，對沿海地區的公共安全構成重大威脅。現行之波浪監測與預警系統多著重於開闊海域，對近岸突發海象事件之即時反應能力有限。為此，本研究提出一套結合即時監視影像與機器學習之沿岸危險波浪預警系統，應用機器學習（Long Short-Term Memory, LSTM）預測未來數秒內的異常浪襲事件，並於實地場域（宜蘭大里、臺東多良、臺南安平國際商港）進行驗證。
本系統透過灰階影像特徵擷取、波浪紋理摺積分析、小波轉換與時間序列模型整合，可有效辨識即將來襲之高強度波浪。研究顯示，在常態一般波浪情境下，5秒預警時間下之準確率達90%、召回率為80%；而於非常態颱風期間應用則達準確率74%、召回率76%。此外，本研究針對港口影像在可視角度受限且岸邊設有堤防狀態之下的越堤浪事件，亦提出初步判識模式，透過以小波理論再以摺積量化海浪紋理強度門檻與影像變異推估越堤潛勢，於2~15秒間可成功預測越堤浪發生之機率分別為，於強度門檻值為0.25狀態下，預測到越堤浪之準確率為92%；強度門檻值為0.3時，預測到越堤浪之準確率為86% 。
本方法所提出的預警模式受外海海相數據即時性，海相取樣點與監視區距離，監視器影像涵蓋面，影像解析度與清晰度等因素影響，早期預警的準確性將會有所差異，建議於光線條件不佳狀態下(夜間、光線不佳或夕陽折射等)，可輔以熱影像儀(FLIR)之夜間影像以納入分辨率與海浪特徵萃取分析，本研究成果可應用於沿海防災、港區管理與公共安全預警，並具備低成本、可擴充與實務應用潛力，為智慧海岸災防系統建置提供新思維。

This study proposes a real-time early warning system for hazardous nearshore waves using video surveillance and machine learning. A Long Short-Term Memory (LSTM) model predicts abnormal wave impacts within seconds, with field tests conducted in Yilan, Taitung, and Tainan. The system analyzes grayscale image textures and wavelet features to forecast intense waves, achieving up to 90% accuracy under normal conditions and 74% during typhoons. For overtopping waves, a separate model reached 92% accuracy. Thermal imaging is suggested for low-light conditions. The system is low-cost, scalable, and effective for coastal disaster prevention and port safety.

1. 華視新聞（2021年8月7日）。瘋狗浪強襲安平港 3人晨跑撲倒1死1傷。取自：https://news.cts.com.tw/cts/society/202108/202108072052160.html
2. Akila, D.S., Mori, I., Dusek, G., Davis, J., and Pang, A. (2021), Automated Rip Current Detection with Region based Convolutional Neural Networks. Coastal Engineering, 166, 103859.Andriolo, U.; Sánchez-García, E.; Taborda, R. Operational use of surfcam online streaming images for coastal morphodynamic studies. Remote Sens. 2019a, 11, 78.
3. Alizadeh, Mohamad., Javad Vahid Nourani (2024). Multivariate GRU and LSTM models for wave forecasting and hindcasting in the southern Caspian Sea. Ocean Engineering, 298.
4. Almar, R., Cienfuegos, R., Catalán, P. A., Michallet, H., Castelle, B., Bonneton, P., Marieu, V. (2012). A new breaking wave height direct estimator from video imagery. Coastal Engineering, 61, 42–48.
5. Tsai, C. H., Su, M. Y., Huang, and S. J. (2004), Observations and conditions for occurrence of dangerous coastal waves. Ocean engineering, 31(5-6), 745-760.
6. Amani, M., Moghimi, A., Mirmazloumi, S.M., et al. (2022). Ocean Remote Sensing Techniques and Applications: A Review (Part I). Water, 14, 3400. https://doi.org/10.3390/w14213400
7. Andriolo, U. (2019b). Near shore Wave Transformation Domains from Video Imagery. J. Mar. Sci. Eng., 7, 186.
8. Andriolo, U., Sánchez-García, E., Taborda, R. (2019a). Operational use of surfcam online streaming images for coastal morphodynamic studies. Remote Sens., 11, 78.
9. Callens, A., Morichon, D., Liria, P., Epelde, I., Liquet, B. (2021). Automatic Creation of Storm Impact Database Based on Video Monitoring and CNNs. Remote Sens., 13, 1933.
10. Chen, Y., & Li, B. (2020). Advanced wave prediction techniques for extreme ocean events. Ocean Engineering, 199, 106902.
11. Chen, Y.-W., Yu, T.-T., Peng, W.-F. (2025). Early Surge Warning Using ML with Real-Time Surveillance Camera. J. Mar. Sci. Eng., 13, 193. https://doi.org/10.3390/jmse13020193
12. Chien, H., Hsu, T.-W., & Lin, T.-L. (2015). Field observations of long-period wave groups during typhoons in southern Taiwan. Ocean Engineering, 109, 624–633.
13. Choi, Heejeong, Park, M., Son, G., et al. (2020). Real-time significant wave height estimation from raw ocean images. Ocean Engineering, 201, 107129.
14. Dankert, H., Horstmann, J., Lehner, S., Rosenthal, W. (2003). Detection of wave groups in SAR images. IEEE Trans. Geosci. Remote Sens., 41(6), 1437–1446.
15. Greenwood, C., Vogler, A., Morrison, J., Murray, A. (2018). Approximation of sea surface using X-band radar. Remote Sens. Environ., 204, 439–447.
16. Hochreiter, S., & Schmidhuber, J. (1997). Long short-term memory. Neural Computation, 9(8):1735–1780.
17. Huang, C., & Liu, X. (2021). Real-Time Wave Height Prediction with Computer Vision. Int. J. Remote Sens., 42(14), 5076–5086.
18. Huang, J., & Zhang, X. (2017). Prediction of extreme wave events using numerical ocean models. J. Coastal Res., 33(3), 611–619.
19. Huang, W., Zhao, X., Huang, W., Hao, W., Liu, Y. (2023). Deep learning-based wave height prediction. Ocean Engineering, 283.
20. Huntley, D., Saulter, A., Kingston, K., Holman, R. (2009). Use of video imagery to test surf height models. WIT Trans. Ecol. Environ., 126, 39–50.
21. Jorges, C., Berkenbrink, C., Stumpe, B. (2021). Ocean wave prediction from bathymetry using LSTM. Ocean Engineering, 232.
22. Krieger, K., & Roberts, S. (2018). Real-time rogue wave prediction. Marine Technology Society Journal, 52(4), 23–34.
23. Kharif, C., & Pelinovsky, E. (2003). Physical mechanisms of the rogue wave phenomenon. European Journal of Mechanics - B/Fluids, 22(6), 603–634.
24. Lin, Y. P., Huang, C. J., Chen, S. H., Doong, D. J., Kao, C. C. (2017). GNSS buoy for coastal monitoring. Sensors, 17(1), 172.
25. Liu, P.-C. (2018). Extreme wave conditions in the western Pacific: An overview. Journal of Marine Science and Technology, 26(5), 559–567.
26. Liu, P. C., Chen, H. S., Doong, D.-J., Kao, C. C., Hsu, Y.-J. G. (2009). Freak waves during Typhoon Krosa. Ann. Geophys., 27, 2633–2642.
27. Meng, Z., Chen, Z., Khoo, B. C., Zhang, A. (2022). Sea wave prediction using LSTM. Ocean Engineering, 262.
28. Morales-Márquez, V., Orfila, A., Simarro, G., et al. (2018). Operational beach management under storm forcing. Nat. Hazards Earth Syst. Sci., 18, 3211–3223.
29. Navarro, W., Velez, J. C., Orfila, A., Lonin, S. (2019). Sea state estimation with X-Band radar. IEEE Trans. Geosci. Remote Sens., 57(9), 6292–6310.
30. Pugliano, G., Robustelli, U., Di Luccio, D., et al. (2019). Statistical shoreline detection deviations. J. Marine Sci. Eng., 7(5), 137.
31. Smith, J., & Williams, D. (2019). Early warning systems for extreme wave events. J. Coastal and Marine Systems, 217, 102244.
32. Tian, X., & Sun, C. (2018). Early Warning System for Tsunami and Storm Surges. IEEE Access, 6, 74160–74169.
33. Tsai, C. H., Su, M. Y., Huang, S. J. (2004). Dangerous coastal wave conditions. Ocean Engineering, 31(5-6), 745–760.
34. Tsai, C.-P., et al. (2004). Study on freak waves along the northeast coast of Taiwan. Coastal Engineering Journal. 46(2), 109–121.
35. Wu, F., & Wang, L. (2019). Wave detection and forecasting using buoy and satellite data. J. Oceanography, 75(6), 753–764.
36. Wu, W., & Li, Z. (2019). Deep learning wave forecasting for coastal monitoring. J. Coastal Research, 35(5), 1069–1076.
37. Yamamoto, Y., & Nakamura, T. (2020). AI-based tsunami early warning. Tsunami Science, 16(2), 112–118.
38. Zhang, M., Tavakoli, S., & Hirdaris, S. (2023). A deep learning method for the prediction of focused waves in a wave flume. IOP Conference Series: Materials Science and Engineering, 1288(1), 012007.