Inferring vector magnetic fields from HMI observations of a solar plage region: one-component versus two-component Milne-Eddington inversions

The Very Fast Inversion of the Stokes Vector (VFISV) is a Milne-Eddington (ME) inversion code that was specifically designed to operate in the HMI data processing pipeline. To speed up the inversion process, the pipeline VFISV code (hereafter, VFISV_1C) employs one component of an ME model atmosphere with the magnetic filling factor of unity for all HMI pixels. Recently, VFISV_1C was modified to enable inversion of observed HMI Stokes profiles with two ME components, allowing the filling factor as a free parameter. Here we test whether the updated inversion code (hereafter, VFISV_2C) can improve the quality of HMI vector magnetic field data, in particular, for decaying active regions with intermediate field strength. This test is done with a long-lived plage of which the magnetic field is supposed to slowly evolve over a couple of solar rotations. The plage can be therefore considered as a quasi-steady target for comparison of HMI magnetic field observations at different times. For a large portion of the plage pixels, the inverted vector magnetic field with VFISV_1C shows a systematic overestimation in the transverse field, which causes a sign reversal of both the zonal (Bp, east-west) and meridional (Bt, north-south) components across the central meridian of the solar disk. We find that the east-west hemispheric sign reversal is substantially mitigated with VFISV_2C for most pixels in the plage with magnetic flux density greater than 300 Mx/cm². We attribute this improvement to the fact that the vector magnetic field inverted with VFISV_2C is determined in favor of having a smaller inclination angle from the line of sight and consequently a weaker transverse component, when compared to the one with VFISV_1C. The updated VFISV_2C can help us to more reliably infer magnetic fields from HMI observations of plages and use them for simulations.