loading page

Pumice Raft Detection Using Machine-Learning on Multispectral Satellite Imagery
  • +2
  • Maggie Zheng,
  • Tushar Mittal,
  • Kristen Fauria,
  • Ajit Subramaniam,
  • Martin Jutzeler
Maggie Zheng
Massachusetts Institute of Technology

Corresponding Author:[email protected]

Author Profile
Tushar Mittal
Massachusetts Institute of Technology
Author Profile
Kristen Fauria
Vanderbilt University
Author Profile
Ajit Subramaniam
Columbia University
Author Profile
Martin Jutzeler
University of Tasmania
Author Profile


Most of Earth’s volcanic eruptions occur underwater, and these submarine eruptions can significantly impact large-scale earth systems. In this study, we develop a new semi-automated analysis framework to detect submarine eruptions through the supervised classification of satellite images on Google Earth Engine (GEE). We present a case study from the Rabaul caldera region in Papua New Guinea and find a large number of new unreported pumice rafts (in ~16% of images from 2017–present). After analysis of the spatial pattern of raft sightings and ancillary observations, we interpret that these rafts are not the result of a new eruption. Instead, we posit that the observed rafts represent remobilization of pumice clasts from previous historical eruptions. This novel process of raft remobilization may be common at near-shore/partially submarine caldera systems (e.g., Rabaul, Krakatau) and has significant implications for new submarine eruption detection, volcanic stratigraphy, and biological dispersal by rafts.