An LSTM-Based Framework for Classification of Marine Oil Spills and Related Pollution Events in the Bulgarian Black Sea and Danube Coastal Zone
Year: 2026
Type: conference-paper
Abstract: Marine oil spills pose a serious threat to marine ecosystems, coastal environments, and maritime economic activities. Rapid detection and accurate classification of pollution events are essential for effective environmental monitoring and risk management. This paper proposes an LSTM-based framework for the classification of marine oil spills and related pollution events using multivariate environmental time series data. The approach integrates hydrometeorological parameters, water quality indicators, and remote sensing features to capture temporal patterns associated with different types of marine pollution. A balanced dataset containing environmental observations is used to train and evaluate the proposed model. The results demonstrate that Long Short-Term Memory networks are capable of effectively learning temporal dependencies and accurately classifying pollution events. The proposed framework contributes to the development of intelligent monitoring systems supporting predictive environmental risk management in the Bulgarian Black Sea and Danube coastal zone.
Language: English
Subfield: Pollution
Field: Environmental Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7213498554