Modeling and predicting the short term evolution of the Geomagnetic field

TL;DR

This paper models the Earth's core magnetic field evolution from 1900 to 2014 using an Ensemble Kalman filter, revealing two flow regimes and providing short-term predictions with quantified uncertainties.

Contribution

It introduces a coupled modeling approach with data assimilation to predict geomagnetic field evolution and distinguish flow regimes, improving understanding of core dynamics.

Findings

Identified two distinct flow regimes with different temporal behaviors.

Achieved accurate short-term predictions of the geomagnetic field and length of day variations.

Demonstrated limitations in long-term magnetic field predictions beyond 2000 years.

Abstract

The coupled evolution of the magnetic field and the flow at the Earth's core mantle boundary is modeled within the 1900.0-2014.0 time period. To constraint the dynamical behavior of the system with a core field model deriving from direct measurements of the Earth's magnetic field we used an Ensemble Kalman filter algorithm. By simulating an ensemble of possible states, access to the complete statistical properties of the considered fields is available. Furthermore, the method enables to provide predictions and to assess their reliability. In this study, we could highlight the cohabitation of two distinct flow regimes. One associated with the large scale part of the eccentric gyre, which evolves slowly in time and posses a very long memory of its past, and a faster one associated with the small scale velocity field. We show that the latter can exhibit rapid variations in localized areas. The combination of the two regimes can predict quite well the decadal variations in length of day, but it can also explain the discrepancies between the physically predicted and the observed trend in these variations. Hindcast tests demonstrate that the model is well balanced and that it can provide accurate short term predictions of a mean state and its associated uncertainties. However, magnetic field predictions are limited by the high randomization rate of the different velocity field scales, and after approximately 2000 years of forecast, no reliable information on the core field can be recovered.