Retrieval and multi-temporal characterization of oil spills from multi-sensor earth observation imagery

Authors

Pelich R., La T.V., Chini M., Matgen P.

Reference

2022 IEEE International Workshop on Metrology for the Sea; Learning to Measure Sea Health Parameters, MetroSea 2022 - Proceedings, pp. 388-392, 2022

Description

In this study we propose an image classification method that allows to delineate oil spills from multi-sensor earth observation (EO) data, i.e. Synthetic Aperture Radar (SAR) and multi-spectral imagery. By making use of the SAR intensity and an index derived from multi-spectral data, we perform a multiscale-based bimodal distribution classification, represented in our case by the oil spill and sea clutter, respectively. The proposed method is applied to a sequence of images acquired with a daily frequency allowing to characterise the temporal and spatial of evolution of the oil spill. In addition, we address the surface wind and currents corresponding to each satellite image in order to investigate their impact on the oil spill evolution. The experimental results are focused on two different oil spill events: one in the waters around Mauritius after a Japanese bulk carrier, MV Wakashio, ran aground on a coral reef, and one in the Persian golf which is the largest offshore oil development area.

Link

doi:10.1109/MetroSea55331.2022.9950927