碎波角(Peel angle)(亦稱浪崩角)是兩不同時間的碎波邊界連線與碎波前進方向之夾角，是評估衝浪活動適宜性的一個參數，本研究旨在建立一個自動分析碎波角的模式。
本文採用深度學習方法裡的卷積神經網路(Convolutional neural network, CNN)來分析碎波角，本研究透過模式超參數的率定來提高辨識準確率，結果顯示CNN模式中之激活函數選擇Sigmoid函數，並使用2000次迭代以及64的批次大小可獲得最佳模式。本研究分析三個現場觀測海域共30部空拍影片(影像時序列)，對於碎波帶的判識準確性達80%以上，而對於碎波角的分析，獲得其平均誤差為5.9o，在可以接受之範圍。除此之外，考量便利性，本研究亦提出以單張影像取代影像時序列來計算碎波角，結果發現，單張影像與前述影像時序列分析結果相近，兩者誤差約3.9o，此結果確認未來可以單張影像來進行分析。
最後，本文分析台灣10處衝浪熱點海域的碎波角，結果發現各地的碎波角約介於30o~70o之間，其季節變化並不顯著，分析也發現部分海域受海岸地形影響，碎波角有較大空間變異性，民眾從事相關活動時宜留意。

Peel angle is the angle between the boundary line of the breaking wave and the advanced direction of the breaking wave at two separate times. It is a parameter used to assess the suitability of surfing activities. This research intends to build a model for the automatic analysis of peel angles.
In this study, the Convolutional Neural Network (CNN) is used to calculate the peel angel. The calibration of the model hyperparameters has been implemented to increase the identification precision. The results indicate that the optimal CNN mode can be obtained by using the Sigmoid activation function,2000 iterations, and a batch size of 64. In three on-site observation sea areas, a total of 30 aerial videos (time-series images) were analyzed, and the identification accuracy of the surf zone was greater than 80%. The average error for the analysis of the peel angle was 5.9 degrees. In addition, replacing time-series images with a single image is proposed in this study in order to determine the angle of breaking wave. The results indicate that the picture and video analysis results are comparable, with a difference of approximately 3.9 degrees. The results of the study confirm that the analysis method of a single image is feasible.
Eventually, the peel angle of ten surfing spots in Taiwan is analyzed in this study. The results indicate that peel angles range between 30 and 70 degrees, and seasonal variations are negligible. The investigation also revealed that certain coastal regions are affected by topography and the spatial distribution of the peel angle is different. Therefore, the public should take precautions when engaging in associated activities.

[1] 國家海洋研究院(2021)，海洋運動能力分級與海洋環境條件關係調查分析，國立臺灣海洋大學。
[2] Andriolo, U. (2019). Nearshore Wave Transformation Domains from Video Imagery. J. Mar. Sci. Eng. 7, 186.
[3] ASBPA, (2011). Science and Technology Committee, SURFERS AS COASTAL PROTECTION STAKEHOLDERS. American Shore & Beach Preservation Association White Paper.
[4] Battjes, J. A. (1974). Surf Similarity. Proc. 14th Coastal Engineering Conference, ASCE, 466-480
[5] Beuzen, T., Goldstein, E. B. and Splinter, K. D. (2019). Ensemble models from machine learning: An example of wave runup and coastal dune erosion. Nat. Hazards Earth Syst. Sci., 19, 2295–2309.
[6] Black, K. and Mead, S. (2001). Design of the Gold Coast Reef for Surfing, Public Amenity and Coastal Protection: Surfing Aspects. J. Coast. Res. 29, 115–130.
[7] Bryan, K., Davies-Campbell, J., Hume, T. and Gallop, S. (2019). The Influence of Sand Bar Morphology on Surfing Amenity at New Zealand Beach Breaks. J. Coast. Res. 87 (Suppl. 1), 44–54.
[8] Buscombe, D. and Carini, R. (2019). A Data-Driven Approach to Classifying Wave Breaking in Infrared Imagery. Remote Sens., 11, 859.
[9] Buscombe, D., Carini, R., Harrison, S., Chickadel, C. and Warrick, J. (2020). Optical wave gauging using deep neural networks. Coast. Eng. 155, 103593.
[10] Ellenson, A., Simmons, J., Wilson, G., Hesser, T. and Splinter, K. (2020). Beach State Recognition Using Argus Imagery and Convolutional Neural Networks. Remote Sens. 12, 3953.
[11] Galvin, C. J. (1968). Breaker-Type Classification on Three Laboratory Beaches. Journal of Geophysical Research, 73, 12.
[12] Henriquez, M. (2005). Artificial Surf Reefs, Delft, Holland: Delft University of Technology, Master’s thesis, 53.
[13] Holman, R. and Haller, M. C. (2013). Remote Sensing of the Nearshore. Annu. Rev. Mar. Sci. 5, 95–113.
[14] Hutt, J. Black, K. and Mazeiraud, V. (2001). Improving the Surfing Climate of Narrowneck Beach. In Coasts & Ports 2001: Proceedings of the 15th Australasian Coastal Ocean Engineering Conference, the 8th Australasian Port Harbour Conference., Institution of Engineers, Australia: Barton, Australia, 114.
[15] LeCun, Y., Bottou, L., Bengio, Y. and Haffner, P. (1998). Gradient-based learning applied to document recognition. Proceedings of the IEEE. 86(11), 2278-2324.
[16] Lewis, J., Hunt, S. and Evans, T. (2015). Quantification of Surfing Amenity For Beach Value And Management. In Proceedings of the Coastal Conference, Forster, Australia, 11–13.
[17] Mckenzie, S. and Green, R. (2018). Gnarometer Surfcam Live Inspector. In Proceedings of the International Conference on Image and Vision Computing, Auckland, New Zealand. 1–6.
[18] Mead, S. T., Black, K. P. and McComb. (2001). Westshore Coastal Process Investigation. Report for Napier City Council, September.
[19] Molines, J., Herrera, M. P., Gómez-Martín, M. E. and Medina, J. R. (2019). Distribution of individual wave overtopping volumes on mound breakwaters. Coast. Eng., 149, 15–27.
[20] Peres, D., Iuppa, C., Cavallaro, L., Cancelliere, A. and Foti, E. (2015). Significant wave height record extension by neural networks and reanalysis wind data. Ocean Model., 94, 128–140.
[21] Scarfe, B. E., Elwany, M. H. S. and Mead, S. T. (2003). The Science of Surfing Waves Surfing Breaks - A Review. Scripps Institution of Oceanography Technical Report, 1-13.
[22] Scarfe, B. (2002). Categorising Surfing Manoeuvres Using Wave and Reef Characteristics. Master’s Thesis, The University of Waikato, Hamilton, New Zealand, 181.
[23] Shand, T., Weppe, S., Quilter, P., Short, A., Blumberg, B. and Reinen-Hamill, R. (2020). Assessing the effect of Earthquake-Induced Uplift and Engineering Works on a Surf Break of National Significance. In Proceedings of the ICCE, Online. 36, 23.
[24] Shand, T., Bailey, D. and Shand, R. (2012). Automated Detection of Breaking Wave Height Using an Optical Technique. J. Coast. Res. 28, 671–682.
[25] Stringari, C., Guimarães, P., Filipot, J., Leckler, F. and Duarte, R. (2021). Deep neural networks for active wave breaking classification. Sci. Rep. 11, 3604.
[26] Thompson, M., Zelich, I., Watterson, E. and Baldock, T. E. (2021). Wave Peel Tracking: A New Approach for Assessing Surf Amenity and Analysis of Breaking Waves. Remote Sensing. 13(17), 3372.
[27] Walker, J., Palmer, R. and Kukea, J. (1972). Recreational surfing on Hawaiian reefs. In Proceedings of the 13th Coastal Engineering Conference, Vancouver, BC, Canada, 10–14, 2609–2628.
[28] Yoo, J., Fritz, H., Haas, K., Work, P. and Barnes, C. (2011). Depth Inversion in the Surf Zone with Inclusion of Wave Nonlinearity Using Video-Derived Celerity. J. Waterw. Port Coast. Ocean Eng. 137, 95–106.