# Testing the Accuracy of Data Driven MHD Simulations of Active Region Evolution

## TL;DR

This study evaluates the accuracy of data-driven magnetohydrodynamic (MHD) simulations of active region evolution in the solar corona, emphasizing the importance of appropriate temporal sampling of photospheric magnetic field data.

## Contribution

It introduces a validation framework using simulated active region formation to assess how data-driven MHD models depend on data cadence and provides a sampling condition for accurate modeling.

## Key findings

- ARs emerging over 25 hours can be modeled with ~1% error at 12-minute cadence.
- Under-sampling rapid features causes large errors and incorrect coronal morphology.
- Higher-cadence data improves model accuracy for dynamic features.

## Abstract

Models for the evolution of the solar coronal magnetic field are vital for understanding solar activity, yet the best measurements of magnetic field lie at the photosphere, necessitating the development of coronal models which are \textit{"data-driven"} at the photosphere. We present an investigation to determine the feasibility and accuracy of such methods. Our validation framework uses a simulation of active region (AR) formation, modeling the emergence of magnetic flux from the convection zone to the corona, as a ground-truth dataset, to supply both the photospheric information, and to perform the validation of the data-driven method. We focus our investigation on how the data-driven model accuracy depends on the temporal frequency of the driving data. The Helioseismic and Magnetic Imager on NASA's Solar {\jl Dynamics} Observatory produces full-disk vector magnetic field measurements at a 12 minute cadence. Using our framework we show that ARs that emerge over 25 hours can be modeled by the data-driving method with only $\sim$1\% error in the free magnetic energy, assuming the photospheric information is specified every 12 minutes. However, for rapidly evolving features, under-sampling of the dynamics at this cadence leads to a strobe effect, generating large electric currents and incorrect coronal morphology and energies. We derive a sampling condition for the driving cadence based on the evolution of these small-scale features, and show that higher-cadence driving can lead to acceptable errors. Future work will investigate the source of errors associated with deriving plasma variables from the photospheric magnetograms as well as other sources of errors such as reduced resolution and instrument bias and noise.