Answers are grounded in this paper and cite their sources.
Over the past decades, coastal video-monitoring systems have proven to be highly useful tools in studying dynamic coastal processes, such as wave overtopping and coastal erosion, providing substantial contributions to the monitoring and management of these vulnerable areas. This work proposes a deep learning approach aimed at identifying wave overtopping events through the analysis of coastal imagery, relying on an architecture that integrates Convolutional Neural Networks (CNNs) with bidirectional Long Short-Term Memory (BiLSTM) units. The dataset consisted of sequences with and without wave overtopping events, derived from continuous video recordings by a coastal monitoring station at Praia dos Pescadores, an urban front facing the jetty of Ericeira harbor on the west coast of Portugal. The sequences were evenly distributed between overtopping events and normal conditions. The model was developed and optimized using Bayesian optimization techniques. After training and evaluation, the model showed high classification performance, reaching 0.97 in both precision and specificity, and 0.92 in recall. The F1-score and the Area Under the Receiver Operating Characteristic Curve (AUC-ROC) were each 0.94, reflecting the system’s capability to detect wave overtopping events accurately, while keeping the rate of false positives low. The proposed system offers a promising contribution toward the development of advanced operational coastal monitoring services, supporting improved coastal risk management and early warning strategies. Given its generalizable architecture, the approach may be adapted to other coastal settings, reinforcing its potential as a scalable solution for international applications in coastal management.
No takes yet. Share an insight, caveat, or question.
Santos et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: