Assessing membership projection errors in star forming regions

Abstract

Context. Young stellar clusters harbour complex spatial structures emerging from the star formation process. Identifying stellar over-densities is a key step in better constraining how these structures are formed. The high accuracy of distances derived from Gaia DR2 parallaxes still do not allow us to locate individual stars within clusters of ≈1 pc in size with certainty. Aims. In this work, we explore how such uncertainty on distance estimates can lead to the misidentification of membership of sub-clusters selected by the minimum spanning tree (MST) algorithm. Our goal is to assess how this impacts their estimated properties. Methods. Using N-body simulations, we build gravity-driven fragmentation models that self-consistently reproduce the early stellar configurations of a star forming region. Stellar groups are then identified both in two and three dimensions by the MST algorithm, representing respectively an inaccurate and an ideal identification. We compare the properties derived for these resulting groups in order to assess the systematic bias introduced by projection and incompleteness. Results. We show that in such fragmented configurations, the dynamical mass of groups identified in projection is systematically underestimated compared to those of groups identified in 3D. This systematic error is statistically of 50% for more than half of the groups and reaches 100% in a quarter of them. Adding incompleteness further increases this bias. Conclusions. These results challenge our ability to accurately identify sub-clusters in most nearby star forming regions where distance estimate uncertainties are comparable to the size of the region. New clump-finding methods need to tackle this issue in order to better define the dynamical state of these substructures.