Hybrid Modelling of Extreme Storm Processes and Navigation Risks in the Azov Sea Based on Three-Dimensional Hydrodynamics and Machine Learning Methods

Introduction. Extreme storms with wind speeds exceeding 30–35 m/s pose a significant threat to navigation and coastal infrastructure in the Azov Sea. The complex bathymetry, shallow water, and coastal geometry amplify wave and surge effects, causing severe destruction. The increasing frequency of extreme weather events requires next-generation forecasting systems capable of capturing nonlinear multiscale interactions between wind, waves, and currents.
Materials and Methods. A hybrid approach was developed, combining three-dimensional numerical hydrodynamic modelling based on the Navier-Stokes equations with Large-Eddy Simulation (LES) turbulence closure, ensemble probabilistic forecasting, and machine learning methods — including Physics-Informed Neural Networks (PINNs) and Fourier Neural Operators (FNOs). Atmospheric and oceanographic data from ERA5 and CMEMS reanalyses were used to reconstruct storm scenarios for 2010–2024. Ship-wave interactions were modeled in six degrees of freedom, while coastal infrastructure fragility was evaluated using probabilistic vulnerability curves. Validation was performed using Sentinel-1/3 satellite data processed by the “LBP-neural_network” software package and Copernicus Marine Service products.
Results. Three representative storm scenarios were simulated. The significant wave height in the central Azov Sea reached up to 5.2 m, with surge amplitudes up to 1.5 m. The most hazardous conditions occurred in the Kerch Strait, where current velocities reached 1.1 m/s. Under wind speeds of 30–35 m/s, the probability of exceeding the critical 4 m wave height was 42%. Resonant ship motions with roll amplitudes up to 25° were detected, indicating a high capsizing risk. Risk maps identified the most vulnerable zones near Taganrog, Yeysk, and Port Kavkaz. The integration of PINNs and FNOs accelerated ensemble simulations by a factor of 10–12 while maintaining prediction errors below 8%.
Discussion. The proposed hybrid methodology proved highly effective for modelling extreme hydrodynamic processes and navigation risks. The LES framework accurately reproduced wave breaking and vortex generation processes, while coupling with neural network surrogates combined physical consistency with computational efficiency.
Conclusion. The approach improved forecast accuracy by 25–30% compared with conventional spectral models (SWAN, WAVEWATCH III). The results provide a scientific basis for developing early warning systems, assessing navigation safety, and planning coastal protection measures in the Azov–Black Sea region.

1. Amarouche K., Akpinar A., Rybalko A., Myslenkov S.A. Assessment of SWAN and WAVEWATCH-III models regarding the directional wave spectra estimates based on Eastern Black Sea measurements. Ocean Engineering. 2023;272:113944. https://doi.org/10.1016/j.oceaneng.2023.113944

2. Reduan Atan R., Nash S., Goggins J. Development of a nested local scale wave model for a 1/4 scale wave energy test site using SWAN. Journal of Operational Oceanography. 2017;10:59–78. https://doi.org/10.1080/1755876X.2016.1275495

3. Masselink G., Russell P., Rennie A., Brooks S., Spencer T. Impacts of climate change on coastal geomorphology and coastal erosion relevant to the coastal and marine environment around the UK. MCCIP Science Review. 2020;158–189. https://doi.org/10.14465/2020.arc08.cgm

4. Yaitskaya N. The Wave Climate of the Sea of Azov. Water. 2022;14(4):555. https://doi.org/10.3390/w14040555

5. Veldman A., Luppes R., Bunnik T., Huijsmans R., Duz B., Iwanowski B., Wemmenhove R., Borsboom M., Wellens P., van der Heiden H., Plas P. Extreme Wave Impact on Offshore Platforms and Coastal Constructions. In: Proceedings of the International Conference on Offshore Mechanics and Arctic Engineering (OMAE). Rotterdam, Netherlands, 2011;7. https://doi.org/10.1115/OMAE2011-49488

6. Elmisaoui S., Kissami I., Ghidaglia J.-M. High-Performance Computing to Accelerate Large-Scale Computational Fluid Dynamics Simulations: A Comprehensive Study. In: Lecture Notes in Computer Science. 2024. https://doi.org/10.1007/978-3-031-54318-0_31

7. Dietterich H., Lev E., Chen J., Richardson J., Cashman K. Benchmarking computational fluid dynamics models of lava flow simulation for hazard assessment, forecasting, and risk management. Journal of Applied Volcanology. 2017;6:9.

8. International Maritime Organization (IMO). Second Generation Intact Stability Criteria. IMO Guidelines. 2023. London, UK. URL: https://www.imo.org (accessed: 03.09.2025).

9. Chu Van T., Ramirez J., Rainey T., Ristovski Z., Brown R. Global impacts of recent IMO regulations on marine fuel oil refining processes and ship emissions. Transportation Research Part D: Transport and Environment. 2019;70:123–134.

10. You J., Faltinsen O.M. A numerical investigation of second-order difference-frequency forces and motions of a moored ship in shallow water. Journal of Ocean Engineering and Marine Energy. 2015;1:157–179. https://doi.org/10.1007/s40722-015-0014-6

11. Protsenko E.A., Panasenko N.D., Protsenko S.V. Mathematical Modelling of Catastrophic Surge and Seiche Events in the Azov Sea Using Remote Sensing Data. Computational Mathematics and Information Technologies. 2024;8(2):33–44. (In Russ.) https://doi.org/10.23947/2587-8999-2024-8-2-33-44

12. Sukhinov A., Protsenko E., Protsenko S., Panasenko N. Wind Wave Dynamic’s Analysis Based on 3D Wave Hydrodynamics and SWAN Models Using Remote Sensing Data. In: Fundamental and Applied Scientific Research in the Development of Agriculture in the Far East (AFE-2022). Zokirjon ugli K.S., Muratov A., Ignateva S. (eds). Springer, Cham, Switzerland. Lecture Notes in Networks and Systems. 2023;733:1–12. https://doi.org/10.1007/978-3-031-37978-9_39

13. Kantamaneni K. Coastal infrastructure vulnerability: an integrated assessment model. Natural Hazards. 2016;84:139–154. https://doi.org/10.1007/s11069-016-2413-y

14. Protsenko E.A., Protsenko S.V., Sidoryakina V.V. Predictive Mathematical Modeling of Sedimentation and Coastal Abrasion Relief Transformation Processes. Journal of Mathematical Sciences. 2024;284:126–139. https://doi.org/10.1007/s10958-024-07331-6

15. Meyers J., Sagaut P. On the model coefficients for the standard and the variational multi-scale Smagorinsky model. Journal of Fluid Mechanics. 2006;569:287–319. https://doi.org/10.1017/S0022112006002850

16. Janssen P.A.E.M. Quasi-linear Theory of Wind-Wave Generation Applied to Wave Forecasting. Journal of Physical Oceanography. 1991;21(11):1631–1642. https://doi.org/10.1175/1520-0485(1991)021<1631:QLTOWW>2.0.CO;2

17. Faltinsen O.M. Sea Loads on Ships and Offshore Structures. 1st ed. Cambridge University Press: Cambridge, UK; 1990. Pp. 1–5.

18. Raissi M., Perdikaris P., Karniadakis G.E. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations. Journal of Computational Physics. 2019;378:686–707. https://doi.org/10.1016/j.jcp.2018.10.045

19. Luo Y., Li Y., Sharma P., Shou W., Wu K., Foshey M. et al. Learning human–environment interactions using conformal tactile textiles. Nat Electron. 2021;4:193–201. https://doi.org/10.1038/s41928-021-00558-0

20. Sukhinov A.I., Protsenko S.V., Panasenko N.D. Mathematical modeling and ecological design of marine systems taking into account multi-scale turbulence using remote sensing data. Computational Mathematics and Information Technologies. 2022;6(3):104–113. (In Russ.) https://doi.org/10.23947/2587-8999-2022-6-3-104-113

21. Myslenkov S.A., Arkhipkin V.S. Recurrence of Storm Waves in the Sea of Azov according to Modeling. Russian Meteorology and Hydrology. 2024;49:1061–1066. https://doi.org/10.3103/S1068373924120045

22. The official website of The Copernicus Climate Change Service (C3S). URL: https://climate.copernicus.eu/ (accessed: 12.09.2025).

23. The official website of International Hydrographic Organization. URL: https://iho.int/ (accessed: 12.09.2025).

24. Protsenko E.A., Sukhinov A.I., Protsenko S.V. Numerical Modelling of Hydrodynamic Wave Processes in the Azov Sea Based on the WAVEWATCH III Wind-Wave Model. Computational Mechanics of Continuous Media. 2025;17(4):422–431. (In Russ.) https://doi.org/10.23947/2587-8999-2025-17-4-422-431

25. Sukhinov A.A., Ostrobrod G.B. Efficient face detection on Epiphany multicore processor. Computational Mathematics and Information Technologies. 2017;1(1):1–15. (In Russ.) https://doi.org/10.23947/2587-8999-2017-1-1-1-15

26. Sukhinov A., Panasenko N., Simorin A. Algorithms and programs based on neural networks and local binary patterns approaches for monitoring plankton populations in sea systems. E3S Web of Conferences. 2022;363:02027.

27. Panasenko N.D. Forecasting the coastal systems state using mathematical modeling based on satellite images. Computational Mathematics and Information Technologies. 2023;7(4):54–65. (In Russ.) https://doi.org/10.23947/2587-8999-2023-7-4-54-6

28. The official website of Earth Observing System. URL: https://eos.com/landviewer/account/pricing (accessed: 12.09.2025).