近几年，我国地震灾害形势严峻，地震烈度速报需求迫切。然而，我国地震监测台网间距大，很难实现大震发生时地震烈度分布图的快速编制。本文基于随机方法，利用历史小震数据反演地震动估计模型的参数，参考地震发生时获取的少量强震动数据，选取合适的震源参数，正演地震动场的分布。根据地震烈度与地震动参数的经验关系，实现地震烈度速报。本方法可以模拟大震的近断层地震动特征，对高烈度区的判断较目前常用的烈度速报方法更为合理。研究成果既可以用于首张烈度分布图的发布，还能够用于地震烈度的动态修正。

Several destructive earthquakes occurred in China in recent years, which resulted in urgent demands of seismic intensity repaid report (SIRR). However, the distance interval of seismic monitoring network is too large to issue the distribution of seismic intensity immediately. Based on stochastic model, this study inverts the model parameters, such as source parameter, ground motion attenuation parameters and site amplification parameter, by using the ground motion records of small earthquakes. When moderate or large earthquakes come up in the future, the source parameter can be modified by a few strong ground motion records. Combined with the inverted ground motion attenuation parameters and site amplification parameters, new model can simulate the ground motion of moderate or large earthquakes. According to the relationship between ground motion parameters and seismic intensity, the SIRR can come true. The result of this study can be applied to issue the first-time seismic intensity report and the dynamic update of the report. In comparison with the ground motion attenuation relationship used in SIRR, stochastic model can not only predict the location and shape of Meizoseismic area more rationally, but also simulate the characteristics of near-fault ground motion, which can estimate the area of high seismic intensity more accurately.