EGU25-9680, updated on 14 Mar 2025
https://doi.org/10.5194/egusphere-egu25-9680
EGU General Assembly 2025
© Author(s) 2025. This work is distributed under
the Creative Commons Attribution 4.0 License.
Poster | Friday, 02 May, 14:00–15:45 (CEST), Display time Friday, 02 May, 08:30–18:00
 
vPoster spot 2, vP2.3
Temporal Fusion Transformers for Improved Coastal Dynamics Forecasting in the Western Black Sea 
Maria Emanuela Mihailov1, Miruna Georgiana Ichim1, Alecsandru Vladimir Chirosca2, Gianina Chirosca2,3, Lucian Dutu1, and Petrica Popov1
Maria Emanuela Mihailov et al.
  • 1Maritime Hydrographic Directorate, Constanta, Romania (emanuela.mihailov@dhmfn.ro)
  • 2Faculty of Physics, University of Bucharest, Magurele - Ilfov, Romania (alecsandru.chirosca@unibuc.ro)
  • 3National RD Institute for Optoelectronics “INOE 2000”, Magurele-Bucharest, Romania (gianina.chirosca@inoe.ro)

The paper investigates the potential of Artificial Intelligence (AI) and Machine Learning (ML) techniques, specifically Temporal Fusion Transformers (TFTs), to enhance the prediction of coastal dynamics along the Western Black Sea coast. We aim to bridge the gap between in-situ observations from five meteo-oceanographic stations and modelled geospatial marine data from the Copernicus Marine Service. TFTs are employed to refine predictions of shallow water dynamics by considering atmospheric influences, focusing on wave-wind correlations. Atmospheric pressure and temperature are treated as latitude-dependent constants, with specific investigations into extreme events like freezing and solar radiation-induced turbulence.  

The analysis utilizes a dataset of meteorological information collected by the Maritime Hydrographic Directorate (MHD) since 2015. The study relies on data gathered from seven automated weather stations at lighthouses along the Romanian coastline. The stations, part of the Romanian Navy - Marine Meteorological Surveillance Network, continuously gather meteorological parameters at specific ground-level heights, including wind speed and direction. The Copernicus Marine Service (CMEMS) wave reanalysis dataset for the Black Sea provides a comprehensive record of wave conditions with a spatial resolution of approximately 2.5 km and hourly temporal resolution.  

Explainable AI (XAI) is exploited to ensure transparent model interpretations and identify key influential input variables, including static, encoder, and decoder variables. Data attribution strategies address missing data concerns, while ensemble modelling enhances overall prediction robustness. The models demonstrate a significant improvement in prediction accuracy compared to traditional methods. This research provides a deeper understanding of atmosphere-marine interactions and demonstrates the efficacy of AI/ML in bridging observational and modelled data gaps for maritime safety and coastal management along the Western Black Sea coast.

 

Acknowledgements: The research of the M.E.M., P.P., M.G.I., and L.D. was conducted as part of the "Forecasting and observing the open-to-coastal ocean for Copernicus users" FOCCUS Project (https://foccus-project.eu/), funded by the European Union (Grant Agreement No. 101133911). Views and opinions expressed are, however, those of the author(s) only and do not necessarily reflect those of the European Union or the European Health and Digital Executive Agency (HaDEA). Neither the European Union nor the granting authority can be held responsible for them. 

The presented results of the M.E.M., P.P., M.G.I., and L.D. have been carried out with financial support from the Sectorial Research-Development Plan of the Romanian Ministry of National Defence, PSCD 2021–2024 Project (097/2021, 092/2022, 097/2023, 097/2024): „Development of an integrated monitoring system to increase the quality of hydro-oceanographic data in the area of responsibility of the Romanian Naval Forces".
Thanks are extended to the relevant departments of INOE-2000 for their help through the "Core Program with the National Research Development and Innovation Plan 2022-2027" with the support of MCID, project no. PN 23 05/2023, contract 11N/2023.

How to cite: Mihailov, M. E., Ichim, M. G., Chirosca, A. V., Chirosca, G., Dutu, L., and Popov, P.: Temporal Fusion Transformers for Improved Coastal Dynamics Forecasting in the Western Black Sea , EGU General Assembly 2025, Vienna, Austria, 27 Apr–2 May 2025, EGU25-9680, https://doi.org/10.5194/egusphere-egu25-9680, 2025.