The separation of contributions from different sources in the magnetic field signal measured at satellite altitude is an open challenge. An approach to this problem, using Principal Component Analysis, is here applied to geomagnetic external field series at Virtual Observatories (VOs). These series are computed from an enlarged data set of Swarm data covering all local times and all geomagnetic activity levels between January 2014 and December 2019. For each 30‐day time window, the Equivalent Source Dipole technique is used to reduce all measurements inside a cylinder to one single “observation” at its axis and 500 km altitude. Our results reveal a first principal mode with dipolar geometry and time variation following very closely the RC‐index of geomagnetic activity. They display a resolved second principal mode with annual periodicity and of approximately zonal quadrupolar radial pattern, reminiscent of results in a previous study using VO series from a filtered satellite data set and with lower time resolution. We resort to the recent comprehensive model CM6 to identify a possible source for this second mode. We propose that the dipolar mode is the expression of the magnetospheric ring current dynamics, at 30‐day time resolution, and the quadrupolar mode is the expression of the annual asymmetry between local summer and winter Sq current vortices. Two fainter modes could be related to the equinoctial amplification of Sq vortices and the ionospheric dynamo modulation by nonmigrating tides. We show that a more uniform local time sampling could contribute to better resolve ionospheric structures. Key Points Dense grid of external field series computed at Virtual Observatories from Swarm data, with 30‐day time resolution Principal Component Analysis tools recover a dominant zonal dipolar mode with sub‐annual variability strongly correlated with geomagnetic activity indices A less well‐resolved mode shows a nearly equatorial wavenumber‐4 structure that peaks at the equinoxes