Stochastic modeling for trajectories drift in the ocean: Application of Density Clustering Algorithm

TL;DR

This paper presents a stochastic Leeway model combined with density clustering to predict and analyze oceanic drift trajectories of sea objects, with applications to long-term simulations and accident reconstruction in the Mediterranean Sea.

Contribution

It introduces a novel spatial clustering algorithm to identify probable search areas and applies stochastic modeling to improve drift prediction accuracy.

Findings

Density clustering effectively identifies high-probability drift zones.

Draft-limited drifters better match observed trajectories during accidents.

Model scenarios show different forcing fields impact drift predictions.

Abstract

The aim of this study is to address the effects of wind-induced drift on a floating sea objects using high--resolution ocean forecast data and atmospheric data. Two applications of stochastic Leeway model for prediction of trajectories drift in the Mediterranean sea are presented: long-term simulation of sea drifters in the western Adriatic sea (21.06.2009-23.06.2009) and numerical reconstruction of the Elba accident (21.06.2009-23.06.2009). Long-term simulations in the western Adriatic sea are performed using wind data from the European Center for Medium-Range Weather Forecast (ECMWF) and currents from the Adriatic Forecasting System (AFS). An algorithm of spatial clustering is proposed to identify the most probable search areas with a high density of drifters. The results are compared for different simulation scenarios using different categories of drifters and forcing fields. The reconstruction of sea object drift near to the Elba Island is performed using surface currents from the Mediterranean Forecasting System (MFS) and atmospheric forcing fields from the ECMWF. The results showed that draft-limited to an upper surface drifters more closely reproduced target trajectory during the accident.