Abstract: In this paper, the integrated prediction model of dynamic process for soil soluble salts is proposed. GM(1,1) model is used to simulate the trend, fourier transformation to abstract the periodic component and AR(p) model to s imulate stochastic process, and it is applied to study soil soluble salts qu antitatively in artificial grassland community. It is proved by theory and pract ice that this model is of higher precision on simulating and predicting dynamic process of soil soluble salts for a longer period of time. Moreover, it is simpl e and small for the task of calculation, and it is better than the traditional s ingle prediction model. Fig 2,Tab 3,Ref 8