Impact of Systematic Modeling Uncertainties on Kilonova Property Estimation

The precise atomic structure and therefore the wavelength-dependent opacities of lanthanides are highly uncertain. This uncertainty introduces systematic errors in modeling transients like kilonovae and estimating key properties such as mass, characteristic velocity, and heavy metal content. Here, we quantify how atomic data from across the literature as well as choices of thermalization efficiency of r-process radioactive decay heating impact the light curve and spectra of kilonovae. Specifically, we analyze the spectra of a grid of models produced by the radiative transfer code \texttt{Sedona} that span the expected range of kilonova properties to identify regions with the highest systematic uncertainty. Our findings indicate that differences in atomic data have a substantial impact on estimates of lanthanide mass fraction, spanning approximately one order of magnitude for lanthanide-rich ejecta, and demonstrate the difficulty in precisely measuring the lanthanide fraction in lanthanide-poor ejecta. Mass estimates vary typically by 25-40\(\%\) for differing atomic data. Similarly, the choice of thermalization efficiency can affect mass estimates by 20\(\%\) to 50\(\%\). Observational properties such as color and decay rate are \textit{highly} model-dependent. Velocity estimation, when fitting solely based on the light curve, can have a typical error of \(\sim 100\%\). Atomic data of light r-process elements can strongly affect blue emission. Even for well-observed events like GW170817, the total lanthanide production estimated using different atomic datasets can vary by a factor of \(\sim6\).