The expansion of metro system can bring varying degrees of impact to the surrounding environment. To study this complex system problem, this paper discusses the temporal and spatial impact by metro system from the perspective of land use change simulation and scenario analysis. The traditional cellular automata (CA) model can realize the simulation of land use change under various scenarios through system dynamics or Markov chain to control the long-term demand forecasting. However, this type of model ignores the filtering of noise data from imageries and increases uncertainty of the system. Therefore, based on the Future Land Use Simulation (FLUS) model, this paper integrates Kalman filter to control the stochastic process of the state-space system, and predicts the spatio-temporal evolution of land use change impacted by metro system in Nanjing from 2019 to 2035. The results show that: (1) The proposed CA-Kalman filter model can realize the optimized simulation of land use change with good accuracy; (2) Urban patches impacted by metro system will emerge from the existing urban boundaries at the cost of occupation of cultivated land, although there is still significant expansion of urban land and construction land, it will reach the upper limit in 2050.