Authors: James Crowley (National Solar Observatory, CU Boulder), Kevin Reardon (National Solar Observatory, CU Boulder)

New-generation solar telescopes like DKIST are providing high-resolution multi-height observations of the solar atmosphere. With the unprecedented quantity and quality of data being produced, it’s especially important now to understand the mapping between spectropolarimetric observations and the solar atmosphere. By applying machine learning clustering techniques to both DKIST observations and synthesized observations from MHD simulations, we explore how spectropolarimetric observations can be leveraged to learn more about the solar atmosphere at multiple heights. These techniques could be powerful tools to accelerate inversion of solar spectra, assist in the synthesis of observations from MHD simulations, and constrain thermodynamic and magnetic properties of the Sun across different heights.