Transient core surface dynamics from ground and satellite geomagnetic data

M. Istas, N. Gillet*, C. C. Finlay, M. D. Hammer, L. Huder

*Corresponding author for this work

Research output: Contribution to journalJournal articleResearchpeer-review


We present an update of the geomagnetic data assimilation tool pygeodyn, use it to analyse ground and satellite-based geomagnetic data sets, and report new findings on the dynamics of the Earth's outer core on interannual to decadal timescales. Our results support the idea that quasi-geostrophic Magneto-Coriolis waves, recently discovered at a period of 7 yr, also operate on both shorter and longer timescales, specifically in period bands centred around 3.5 and 15 yr. We revisit the source of interannual variations in the length of day and argue that both geostrophic torsional Alfven waves and quasi-geostrophic Magneto-Coriolis waves can possibly contribute to spectral lines that have been isolated around 8.5 and 6 yr. A significant improvement to our ensemble Kalman filter algorithm comes from accounting for cross-correlations between variables of the state vector forecast, using the 'Graphical lasso' method to help stabilize the correlation matrices. This allows us to avoid spurious shrinkage of the model uncertainties while (i) conserving important information contained in off-diagonal elements of the forecast covariance matrix, and (ii) considering a limited number of realizations, thus reducing the computational cost. Our updated scheme also permits us to use observations either in the form of Gauss coefficient data or more directly as ground-based and satellite-based virtual observatory series. It is thanks to these advances that we are able to place global constraints on core dynamics even at short periods.
Original languageEnglish
JournalGeophysical Journal International
Issue number3
Pages (from-to)1890-1915
Publication statusPublished - 2023


  • Earth rotation variations
  • Satellite magnetics
  • Inverse theory
  • Core


Dive into the research topics of 'Transient core surface dynamics from ground and satellite geomagnetic data'. Together they form a unique fingerprint.

Cite this