Master of Science Henrik Fisser will Wednesday March 25th, 2026, at 12:15 hold his Thesis Defense for the PhD degree in Science. The title of the thesis is:
« Spatiotemporal and Statistical Distributions of Arctic Icebergs - Analyzing optical and synthetic aperture radar satellite data »
Icebergs are indicators of solid ice discharge from Arctic tidewater glaciers, and their melt in the ocean affects ocean circulations and the climate system. As obstacles in the ocean, icebergs are hazards to humans, ships, and industry. The background chapters of this thesis summarize established knowledge on tidewater glaciers, iceberg calving, iceberg drift and decay, and satellite remote sensing, laying the foundation for the three presented journal papers. Paper 1 quantifies uncertainties in the iceberg area retrieval from optical Sentinel-2 data with respect to varying solar illumination angles. We find that relative uncertainties in the threshold-based area retrieval depend on the iceberg size, and on the solar zenith angle. Inconsistent uncertainties above solar zenith angles of 65° imply an unreliable retrieval of iceberg areas from optical data in the spring and in the fall, depending on the latitude. Paper 2 targets uncertainties in the retrieval of iceberg areas from Sentinel-1 radar data validated with Sentinel-2 data, using an algorithm that detects and delineates icebergs as local outliers in radar images. Relative uncertainties in the iceberg area retrieval vary with the backscatter of the ocean in the surrounding of icebergs. Variations in iceberg sizes and in the iceberg backscatter additionally complicate the retrieval of iceberg areas. Gradient boosting regression models, developed in this paper, predict iceberg areas more accurately from Sentinel-1 images, by incorporating radar backscatter statistics of icebergs. Paper 3 studies Greenland icebergs in near-terminus mélange, glacial fjords, and along the drift into the ocean. Here, deep learning segmentation models are used to retrieve iceberg areas from Sentinel-1 radar data and from optical Sentinel-2 data during summer and fall months. Iceberg volumes are empirically derived from iceberg areas to quantify spatiotemporal and statistical variations in iceberg volumes. Results indicate lags between the near-terminus glacier velocity and the near-terminus iceberg volumes. Declines in iceberg volumes from glacial fjords into the ocean signify iceberg dispersion and melt. Changes in the statistical distribution with increasing distance from the terminus further depict iceberg melt and break-up during iceberg drift and decay. The developed and tested methods for the satellite-based iceberg size retrieval, and quantified uncertainties are relevant to applications in oceanography, glaciology, maritime safety, and offshore engineering. Thus, this thesis facilitates further research on ice-ocean interactions, iceberg meltwater fluxes, and iceberg drift and decay.
1st Opponent: Lecturer/Assistant Professor Dr. Anna Crawford, University of Stirling, United Kingdom
2nd Opponent: Researcher Dr. Mohamed Babiker, The Nansen Center (NERSC), Bergen, Norway
Internal member and leader of the committee: Professor Juha Vierinen, IFT, UiT