Overview of my research interests and current and past projects



Bjarke Nilsson
PhD Student

Contact


Research


Overview of my research interests and current and past projects.


Satellite Altimetry

Satellite altimeters measure the height of the sea surface from space. Conventional altimeters observe a narrow line directly below the satellite (nadir), while the SWOT mission, launched in 2022, maps a two-dimensional swath of the ocean surface in a single pass.

My work centres on processing and evaluating SWOT data for geodesy, for example assessing how well SWOT observes the static ocean topography (GRL 2025), alongside data from missions such as CryoSat-2, ICESat-2, SARAL/AltiKa and Sentinel-3/6.

Illustration of nadir and wide-swath (SWOT) altimetry, and how the sea surface is converted to gravity and inverted to bathymetry
From sea surface height to gravity and bathymetry: nadir altimetry versus wide-swath altimetry from SWOT.

Marine Geodesy

The mean sea surface and the marine gravity field are fundamental reference fields for both geodesy and oceanography. I developed the DTU25 mean sea surface, which incorporates SWOT observations (ESSD 2026; dataset). I am now working on DTU26GRA, a new global marine gravity field from SWOT with spatially adaptive filtering.

Global map of the DTU25MSS mean sea surface
The DTU25MSS global mean sea surface.
Global map of the DTU26GRA free-air marine gravity anomaly
Free-air gravity anomaly from the DTU26GRA marine gravity field (in preparation).

Bathymetry

Only around a quarter of the seafloor has been mapped directly by ships. Elsewhere, ocean depth is predicted from marine gravity derived from satellite altimetry. The higher resolution of SWOT reveals smaller seamounts and seafloor features than before. Together with David Sandwell's group at Scripps Institution of Oceanography, I work on predicting global bathymetry from SWOT-based gravity.

Global map of bathymetry predicted with machine learning
My global bathymetry prediction, made with machine learning.

Machine Learning

Converting gravity into ocean depth has traditionally relied on linear inversion. I develop machine-learning methods that learn this relationship from ship soundings and quantify how reliable the predictions are away from mapped areas (Paper 1 and Paper 2, JGR Solid Earth 2026). In 2026 I extended this work towards joint gravity inversion during a research stay at NYCU, Taiwan.

Maps of distance to nearest ship sounding, prediction confidence level and predicted elevation for three regions
Distance to the nearest training sounding, prediction confidence and predicted seafloor elevation for three example regions.