Carla Braitenberg ^*

• Department of Mathematics, Informatics and Geosciences, University of Trieste, Trieste, Italy

This study explores the dynamic evolution of the Hunga volcano, Kingdom of Tonga, covering two volcanic unrests, focusing on the creation and subsequent disappearance of a new island between Hunga Tonga (HT) and Hunga Ha’Apai (HH) from 2013 to 2023. The island expanded in 2015 and vanished in January 2022 due to a massive eruption (VEI = 6), featuring a 57 km high volcanic cloud and generating multiple Tsunami that caused damage across the Pacific Ocean. Utilizing remote sensing techniques, including multispectral imaging from Sentinel 2, Landsat 8-9 and SAR imaging from Sentinel 1, the research employs a supervised Random Forest classification algorithm to individuate the changing subaerial surface area of the volcano. This approach documents size variations in the islands, particularly during weeks surrounding two volcanic unrests. The classifier, trained on nearly cloud-free multispectral images, automatically delineates surface area changes over the years. The temporal resolution of area change, limited to images with less than 5% cloudiness, encompasses about 50% of Landsat and 20% of Sentinel 2 images between 2013 and 2023, selected from 739 available images. The multispectral observations are complemented by 215 Sentinel 1 SAR images, penetrating clouds, though with limited bands. Despite higher noise, the classifier on Sentinel 1 successfully distinguishes land from ocean. Sentinel 1 observations, starting in 2014, cover the volcano unrest of 2014/2015. Earth Engine, a cloud computing data facility, is used for processing. Analysis indicates a slight decrease in area change post-2015 island formation and identifies the disappearance of the island bridge connecting HT and HH, along with two smaller islands south of HT and HH in 2022. The 2022 explosion is preceded by an increase in island area weeks before the eruption. The global satellite coverage suggests potential applicability to automatically detect changes in oceanic areas, distinguishing water from new volcanic islands, offering a means to identify volcanic unrests and document their evolution.