2020ApJ...888..106T

We develop and implement a model to analyze the internal kinematics of galaxy clusters that may contain subpopulations of galaxies that do not independently trace the cluster potential. The model allows for substructures within the cluster environment and disentangles cluster members from contaminating foreground and background galaxies. We estimate the cluster velocity dispersion and/or mass while marginalizing over uncertainties in all of the above complexities. Using mock observations from the MultiDark simulation, we compare the true substructures from the simulation with the substructures identified by our model, showing that 50% of the identified substructures have at least 79% of its members are also members of the same true substructure, which is on par with other substructure identification algorithms. Furthermore, we show a ∼35% decrease in scatter in the inferred velocity dispersion versus true cluster mass relationship when comparing a model that allows three substructures to a model that assumes no substructure. In a first application to our published data for A267, we identify up to four distinct galaxy subpopulations. We use these results to explore the sensitivity of inferred cluster properties to the treatment of substructure. Compared to a model that assumes no substructure, our substructure model reduces the dynamical mass of A267 by ∼22% and shifts the cluster mean velocity by ∼100 km s^–1, approximately doubling the offset with respect to the velocity of A267's brightest cluster galaxy. Embedding the spherical Jeans equation within this framework, we infer for A267 a halo mass M₂₀₀ = (7.0 ± 1.3) x 10¹⁴ M_☉ h^–1 and concentration log₁₀c₂₀₀=0.71±0.38, consistent with the mass-concentration relation found in cosmological simulations.