The massive Sargassum algae beachings observed over the past decade are a new natural hazard currently impacting the island states of the Caribbean region (human health, environmental damages, and economic losses). This study aims to improve the prediction of the surface current dynamic leading to beachings in the Lesser Antilles using clustering analysis methods. The input surface currents were derived from the Mercator model and the Hybrid Coordinate Ocean Model (HYCOM) outputs in which we integrated the windage effect. Past daily observations of Sargassum beaching on Guadeloupe coasts and satellite-based Sargassum offshore abundance were also integrated. Four representative current regimes were identified for both Mercator and HYCOM data. The analysis of the current sequences leading to beachings showed that the recurrence of two current regimes is related to the beaching peaks respectively observed in March and August. The performance score of the predictive model showed that the HYCOM data seem more suitable to assess coastal Sargassum hazard in the Lesser Antilles. For 1 year of tests (i.e., 2021), the decision tree accuracy respectively reached 70.1 % and 58.2 % for HYCOM and Mercator with a temporal uncertainty range ±3 d around the forecast date. The present clustering analysis predictive system, requiring lower computational resources compared to conventional forecast models, would help improve this risk management in the islands of the region.

In view of the high vulnerability of the small islands of the Lesser Antilles to cyclonic hazards, realistic very fine scale numerical simulation of hurricane-induced winds is essential to prevent and manage risks. The present innovative modeling aims at combining the most realistically simulated strongest gusts driven by tornado-scale vortices within the eyewall and the most realistic complex terrain effects. The Weather Research and Forecasting (WRF) model with the nonlinear backscatter and anisotropy (NBA) large eddy simulation (LES) configuration was used to reconstruct the devastating landfall of category 5 Hurricane Irma (2017) on Saint Barthélemy and Saint Martin. The results pointed out that the 30 m scale seems necessary to simulate structures of multiple subtornadic-scale vortices leading to extreme peak gusts of 132 m s⁻¹ over the sea. Based on the literature, such extreme gust values have already been observed and are expected for category 5 hurricanes like Irma. Risk areas associated with terrain gust speed-up factors greater than 1 have been identified for the two islands. The comparison between the simulated gusts and the remote sensing building damage highlighted the major role of structure strength linked with the socio-economic development of the territory. The present modeling method could be easily extended to other small mountainous islands to improve the understanding of observed past damage and to develop safer urban management and appropriate building standards.

This study proposes an implementation of an incremental neural network (INN) that was initially designed for affective computing tasks. INNs are a family of machine learning algorithms that combine prototype-based classifiers with neural networks. They achieve state-of-the-art performance with less data than traditional approaches. In this research, we conduct an in-depth review of INN mechanisms and present a research-grade framework that enables the use of INNs on arbitrary data. We evaluated our implementation on two different datasets, including the AVEC2014 Challenge, which involved predicting depressive state from auditive and visual modalities. Our results are encouraging, demonstrating the potential of INNs in situations where approaches have to be explainable or when data are scarce.

In view of the high vulnerability of the Lesser Antilles small islands to cyclonic hazards, realistic very fine scale numerical simulation of hurricane-induced winds is essential to prevent and manage risks. The present innovative modeling aims at combining the most realistic simulated strongest gusts driven by tornado-scale vortices within the eyewall and the most realistic complex terrain effects. The Weather Research and Forecasting (WRF) model with the Nonlinear Backscatter and Anisotropy (NBA) Large Eddy Simulation (LES) configuration was used to reconstruct the devastating landfall of category 5 Hurricane Irma (2017) on Saint Barthélemy and Saint Martin islands. The results pointed out that the 30-m scale seems necessary to simulate intense 400-m scale vortices leading to extreme peak gusts like 132 m s⁻¹ over sea. Risk areas associated with terrain gust speed-up factors greater than one have been identified for the two islands. The comparison between the simulated gusts and the remote sensing building damages highlighted the major role of structure strength linked with the socio-economic development of the territory. The present modeling method could be easily extended to other small mountainous islands to improve the understanding of observed past damages and to develop safer urban management and appropriate building standards.