 The study proposes a new cellular automata, CA, approach based on Bayesian spatially varying coefficient, BSVC, to simulate multi-temporal urban sprawl in Hefei and Qingdao, China. The results show that the proposed CAHPS VC model outperforms the geographically weighted regression, GWR, model in terms of accuracy and reduced bias in reproducing urban sprawl patterns. The study extends urban simulation methods by accounting for spatial non-stationarity. This article was authored by Shurei Chen, Yongzhou Feng, Chen Yi and others.