Authors: Alan Hsu (Harvard University), Jenna Samra (Smithsonian Astrophysical Observatory), Steven Tomczyk (Solar Scientific LLC), Katharine Reeves (Smithsonian Astrophysical Observatory)

Remote sensing instruments estimate the coronal magnetic field from measurements of coronal emission line polarization, where the polarization state is measured by inverting modulated intensity measurements with the polarimeter demodulation matrix. Calibration procedures commonly estimate the demodulation matrix by parameterizing it and computing point estimates through optimization processes. We present a novel Bayesian model for spectropolarimeters and a corresponding probabilistic calibration procedure. Our method computes the joint posterior of the free parameters and propagates the calibration uncertainty downstream to the magnetic field estimates. We validate our method with Stokes observations generated from a global MHD model of the corona. We show that the spread of the estimated distributions agrees with the ground truth parameters, and that our reconstructed magnetic field is photon-noise limited rather than calibration-noise limited. We also test our calibration procedure on a polarimeter prototype in the lab and demonstrate that the method works on real hardware.