Remote sensing applications for beach litter monitoring in the Arctic
Environmental protection in the Arctic – support of German activities in the Arctic Council in terms of a pilot study on monitoring plastic litter on arctic coastlines applying remote sensing techniques
Arctic-wide standardised surveys of beached marine litter are fundamental to draw up a regional action plan as proposed by the Protection of the Arctic Marine Environment working group (PAME) of the Arctic Council for 2021. Monitoring beach litter does not only provide information about the current degree of pollution but also serves as a basis for evaluation of the success of possible plans for action.
Remote sensing surveys of beach litter
In cooperation with AquaEcology GmbH & Co. KG, the Norwegian Polar Institute and WSP Arctic, the project investigated methods to monitor beach litter on Greenland and Svalbard on behalf of the German Environment Agency. Traditional methods of beach litter monitoring were supported and complemented by remote sensing methods as satellite imagery or drone surveys.
Satellite imagery served to identify areas with large quantities of plastic litter. Drone images were intended to identify and categorise plastic objects to save time in comparison to traditional survey methods. Due to limitations of the spatial resolution of all remote sensing techniques, this study concentrated on macro litter (> 2.5 cm) only.
Drone surveys
In drone surveys, manual inspection of the drone images and machine learning were tested and compared to the results of traditional monitoring following the OSPAR guidelines. Our WingtraOne drone with Sony QX1 RGB sensor and the MicaSense Altum multispectral sensor were used in these surveys, covering areas between 1.3 ha and 34 ha at a spatial resolution of 1.4 cm for the Sony QX1 and 3.4 cm for the MicaSense Altum sensor.
Evaluation of satellite imagery
To detect accumulations of plastic litter, WorldView-3 satellite imagery with a spatial resolution of 1.2 m in the visible (RGB) and near infrared (NIR) range and 3.7 m in the short-wavelength infrared (SWIR) range, taken a few days prior to drone surveys, was used to ensure comparability of both survey methods. Machine learning and spectral unmixing approaches were used to investigate a possible detection of plastic litter on a pixel or sub-pixel level in the Arctic.
On behalf of the German Environment Agency (UBA).
Project number: (FKZ) 3719 18 201 0
Project duration: 2019–2022
Contact
