Fire accident seriously threatens the stable operation and personnel security in the subway station, and clarifying the smoke temperature distribution is conducive to arranging reasonably fire alarm. In this study, model experiments and numerical simulations are conducted to investigate the ceiling temperature profiles, mainly altering the heat release rate (HRR) and fire locations in a large subway station. The repeatability of the experimental data is confirmed, and the numerical model is validated by the measurement data. During a fire happening in a large-wide space, the smoke movement beneath the ceiling can be divided into radial and one-dimensional flow, whose prediction models are proposed under fire source located at the center. On this basis, it can be found that the radial temperature profile is independent of fire source location. While, the one-dimensional temperature distribution model is modified by introducing the effects of transvers and longitudinal fire locations respectively. This work could provide data support and theoretical reference for clarifying the smoke movement for large subway station fire.