Based on the daily precipitation data observed at 696 meteorological stations across China during the period from 1960 to 2017, a frequency-domain model(PRS DAR(1)) was established by integrating the first order discrete autoregressive model (DAR(1)) with phase randomization simulation(PRS) method. The performance of proposed model was compared with that of the classical WeaGETS model in terms of distribution characteristics, temporal coherence, and multi site spatial coherence. The results indicate that PRS DAR(1) model can more accurately reproduce the statistical characteristics of daily precipitation sequences, and it outperforms WeaGETS model especially in terms of temporal and spatial coherence. Moreover, PRS DAR(1) model exhibits strong applicability across different daily average precipitation intervals and spatial distributions. The results in this paper provide a new approach for improving traditional precipitation generation models and enhancing the accuracy of daily precipitation simulation.