Data Assimilation

Data from latest HMI Magnetogram, to be assimilated(With angular distance from disk center)
Magnetic boundary condition for the MHD code MAS, derived from HipFT(SDO assimilation region plus later SO view and assimilated regions are indicated)
f 0/– · 0.0s

Maps of the photospheric magnetic field are the key boundary condition for the MHD model of the solar corona. Solar observatories often create “synoptic” maps of the surface field, built up over the solar rotation (these are really diachronic maps). As we did in 2024, we are using the surface flux transport model (HipFT) (Caplan et al. 2025) to assimilate and evolve the solar magnetic flux to create the boundary maps used in the model. HipFT is part of the Open-source Flux Transport (OFT) model suite. The data assimilation is performed with MagMAP.

The primary data source for the model are magnetograms from SDO/HMI (beginning August 1st, magnetograms will also be assimilated from SO/PHI). The radial magnetic field, BrB_{r}, is the key input to HipFT. Previous surface flux transport models have typically inferred BrB_{r} from the line-of-sight (LOS) field, assuming the field is primarily radial where it is measured. This assumption known to be a poor approximation in active regions. For the2024 eclipse, we assimilated HMI LOS magnetograms, but this year, MagMAP has been extended to ingest HMI vector magnetograms (special thanks to the HMI team for making these available in real time). However, the noise in the transverse field is much higher than in the LOS field, making it less reliable in weak field regions. We combine BrB_{r} from the vector field in strong field regions with BrB_{r} inferred from BLOSB_{LOS} in weak field regions to assimilate BrB_{r}. We correct the inferred BrB_{r} in weak field regions based on the center-to-limb angle of the measurement, similar to the correction described by Yang et al. (2026).

Flux-Balanced Correction

A difficulty with real-time assimilation is that artifacts can easily be introduced into the SFT. A common example is when only a portion of an active region is observed, such that one polarity is unbalanced. We mitigate this issue with a flux-balanced weighting that reduces the input of any unbalanced polarity. We ensure that the weighted, assimilated data does not change the flux balance of the map, ensuring that a net flux of zero is maintained for the entire map.

This year, these issues present a more significant challenge because the Solar Dynamics Observatory is in the “eclipse season” portion of its geosynchronous orbit, which means HMI data is not available for certain hours in the morning UTC. We assimilate all data HMI as soon as it is available, leading to data gaps or “jumps” in the assimilation.

Data assimilated into HipFT, after weighted, flux balanced correction applied(With angular distance from disk center)
HipFT model result(SDO assimilation region plus later SO view and assimilated regions are indicated)
f 0/– · 0.0s
IAC LogoNASA LogoNSF Logo
ACKNOWLEDGEMENTS

This research is funded by NASA and NSF. We are grateful to Instituto de Astrofísica de Canarias (IAC) for providing a dedicated allocation at the LaPalma supercomputer and proud to support the North African Telescope Eclipse (NATE) project, led by IAC. We also acknowledge dedicated resources from the ACCESS program and Expanse supercomputer at the San Diego Supercomputer Center (SDSC). We thank NASA's Advanced Supercomputing Division (NAS) for allocations on the Athena supercomputer that enabled many of the visualizations on this page to be produced rapidly in real-time. We also thank the SDO/HMI team of the Solar Physics Group at Stanford University for their support in providing timely access to HMI vector magnetograph data. Lastly, we are particularly excited to work with the Solar Orbiter PHI team, to conduct the farside assimilation experiment in near-real-time.