# Opposite polarity magnetic field and convective downflows in a simulated sunspot penumbra

## TL;DR

This study uses 3D MHD simulations and synthetic spectral line modeling to analyze opposite polarity magnetic fields and convective downflows in sunspot penumbrae, highlighting observational limitations and the robustness of simulation results.

## Contribution

It provides a detailed analysis of opposite polarity magnetic fields and downflows in sunspot penumbrae through forward modeling, accounting for observational effects and validating simulation robustness.

## Key findings

- Good agreement between observations and simulations after considering instrumental effects.
- Opposite polarity magnetic fields and downflows are often hidden due to noise and spatial smearing.
- Simulation results are consistent across different grid resolutions.

## Abstract

Recent numerical simulations and observations of sunspots show a significant amount of opposite polarity magnetic field within the sunspot penumbra. Most of the opposite polarity field is associated with convective downflows. We present an analysis of 3D MHD simulations through forward modeling of synthetic Stokes profiles of the Fe\sci 6301.5 \AA~ and Fe\sci 6302.5 \AA~ lines). The synthetic Stokes profiles are spatially and spectrally degraded considering typical instrument properties. Line bisector shifts of the Fe\sci 6301.5 \AA~ line are used to determine line-of-sight velocities. Far wing magnetograms are constructed from the Stokes V profiles of the Fe\sci 6302.5 \AA~ line. While we find an overall good agreement between observations and simulations, the fraction of opposite polarity magnetic field, the downflow filling factor and the opposite polarity-downflow association are strongly affected by spatial smearing and presence of strong gradients in the line-of-sight magnetic field and velocity. A significant fraction of opposite polarity magnetic field and downflows are hidden in the observations due to typical instrumental noise. Comparing simulations that differ by more than a factor of two in grid spacing we find that these quantities are robust within the simulations.