Unveiling coastal change across the Arctic with full Landsat collections and data fusion




Nylén, Tua; Calle, Mikel; Gonzales-Inca, Carlos

PublisherElsevier BV

NEW YORK

2025

Remote Sensing of Environment

Remote Sensing of Environment

REMOTE SENS ENVIRON

114696

322

27

0034-4257

1879-0704

DOIhttps://doi.org/10.1016/j.rse.2025.114696(external)

https://doi.org/10.1016/j.rse.2025.114696(external)

https://research.utu.fi/converis/portal/detail/Publication/491560595(external)



Arctic communities urgently need regional to local-scale information on the rapid coastal changes, caused by thawing permafrost, melting glaciers, and declining sea ice. We introduce a procedure for mapping coastal land cover change from satellite images in the challenging Arctic conditions (and beyond). Our approach utilizes data fusion and cloud computing in Google Earth Engine to process the full Landsat collections for the entire Arctic. It merges information from multiple Landsat sensors and utilizes complementary spatial data and two algorithms to enhance classification accuracy and processing efficiency. This mitigates issues with local illumination conditions and the low availability and quality of satellite data in the Arctic before 2010s. Calculating post-classification composites of coastal land cover over five-year time-steps effectively reduces the impacts of clouds, suspended sediment, and the tide. The procedure was iteratively developed in calibration sites with contrasting physical characteristics. Validation of the final product indicates an overall classification accuracy of more than 98 % (against manually labelled data) and a median shoreline error distance of c. 20 and 10 m in mesotidal and microtidal coasts, respectively. The resulting Arctic Coastal Change dataset presents coastal dynamics from 1984 to 2023 at a 30-m resolution, and highlights hotspots that experience coastal erosion or accretion at a rate of more than 10 m/a. The overall coherence of our results with 61 other studies across the Arctic shows the robustness of the procedure. However, exploring the dataset may uncover localized errors that call for procedure improvements through new collaborative Arctic coastal dynamics studies.


This work was funded by the Research Council of Finland - former Academy of Finland - [Tua Nylen, project ARIMPA, grant number 343338] ; and the Turku Collegium for Sciences, Medicine and Technology (TCSMT) [Mikel Calle] .


Last updated on 2025-23-04 at 10:32