Modelling geomagnetic jerks with core surface flow derived from satellite gradient tensor elements of secular variation

The Swarm mission provides along- and across-track differences of magnetic field measurements, making it possible to generate spatial gradients of the geomagnetic field and its secular variation (SV). Similar data are obtainable from the CHAMP mission by taking along-track differences. These can be combined into a spatial gradient tensor of SV. We compare core-surface flow inversions from vector and tensor datasets, with a particular focus on the equatorial geomagnetic jerks observed by the CHAMP and Swarm missions. Our models are obtained directly from the SV data, without relying on numerical simulations for prior information or enforcing any flow geometry. We develop three different flavours of model, all damped to minimise spatial complexity and acceleration between epochs, and find all provide good fits to the data. With these, we scrutinise the extent of equatorial asymmetry required by core-surface flow to fit the data, and relate the flow to observations of changes in length-of-day. We find that using spatial gradients for flow-inversions improves the spatial resolution compared to using vector measurements, resolving ∼ 1.4 times as many flow coefficients for the Swarm models and ∼ 1.2 for the CHAMP models. During the 2017 and 2020 Pacific region geomagnetic jerks, our models show pulses in azimuthal flow acceleration, time-centred between the two jerks, and a new pulse occurring in mid-2022. This suggests that a new geomagnetic jerk in this region will occur at the end of 2024. We propose that the observed azimuthal acceleration pulses may occur when previously hypothesised Alfvén wave-packets interact with flow at the surface of the core. • Using the spatial gradients tensor of secular variation for core-surface flow inversion significantly improves the spatial resolution, compared to using vector data. • Approximately 10 % equatorial asymmetry in core-surface flow is required to fit the secular variation data. • 21st century geomagnetic jerks can be explained by pulses in azimuthal flow acceleration. This is in agreement with numerical simulations, and could allow short-term forecasting of geomagnetic jerks.