GRSG 34th Conference 2023
Title: Applications of UAV-Born Hyperspectral Data for The Detection and Characterization of Oil Spills
Author: Rebecca Scafutto
Oil spills endangers marine and coastal ecosystems. Effective response to contamination relies on quick actions to constrain and treat the leaked oil before it reaches the shore. In this context, previous knowledge of the characteristics of the spilled oil is crucial to decide the more appropriate strategy to deal with the oil spill. In this context, remote sensing tools have been used to provide a wide range of data to assist in the monitoring and characterization of offshore oil spills. Most of the data are acquired with sensors onboard orbital or airborne platforms. UAV-born sensors, however, are showing potential to assist in monitoring and remediation tasks, providing high-resolution data in much finer scale.
Here, we aim to evaluate UAV-born hyperspectral imagery collected by the HySpex Neo VS-620 camera in the VNIR-SWIR range for the characterization of oil spills. A controlled experiment was designed to simulate oil spills in large scale. The experiment comprised three tanks of 50.000 m³ filled with a saline solution simulating sea water, where 5 and 30 liters of crude oil was leaked in the first two tanks and the third remained clean as a control tank. Images were acquired regularly for 11 weeks, between August – October 2022. The tanks remained exposed to weathering during the entire experiment. The results revealed that the crude oils could be mapped and separated from the water. Besides, variations were identified with the hyperspectral data related to oil thickness and exposition time. During the first days of the experiment, the oil split in different fractions, following Bonn agreement patterns.
The sheen (0.04 – 0.3 µm), metallic (5 – 50 µm), and continuous true color (>200 µm) categories were visually observed and mapped individually in the hyperspectral data. Furthermore, variations were also observed in the geometry and depth of the main hydrocarbon feature, centered around 1700 nm, related to the exposition time and weathering of the oil. In this case, the feature was shorter and. These results demonstrate that UAV-born data can be used for the qualitative characterization of oil spills, with the potential to be used to estimate exposition time and oil thickness and volume based on Boon agreement parameters.