Strong ground motion data serve as the basis for establishing ground motion models. It is difficult to establish ground motion models in areas lacking sufficient strong motion data. This paper reviews several methods for establishing ground motion models in areas lacking strong motion data, including the numerical simulation method, the hybrid empirical method, and the referenced empirical approach. The numerical simulation method employs high-frequency and low-frequency ground motions simulated by stochastic and deterministic methods, respectively, to develop ground motion models. The hybrid empirical method can effectively solve the problem of lack of data by combining numerical simulation and actual observation data and applying the empirical ground motion model of the reference area to the target area by using the adjustment factor. The referenced empirical approach is based on the small earthquake records in the study area and adapts the existing empirical ground motion model to suit the specific regional situation with simplicity and effectiveness. Each of these three types of methods has its own characteristics, numerical simulation methods can take into account the characteristics of the seismic source, complex geological and site conditions, and the calculation results depend on the accuracy and precision of the source model and the subsurface velocity structure. The hybrid empirical method combines the flexibility of numerical simulation methods and the statistical characteristics of observed data, and can establish a relatively reliable model. The referenced empirical approach is quicker and simpler but is dependent on the data of the small earthquakes. Finally, this paper suggests that artificial intelligence and multi-source data fusion can be used to improve the accuracy and reliability of ground motion model in areas lacking strong motion data.