In this paper, the movement and deformation of surface is regarded as the random and dynamic time-varying system. Forecast of the state of surface movement is divided into two parts: one is the forecast of time-varying parameter influencing the factor of surface movement, the other is the forecast of state of surface movement. Timevarying parameters can be forecasted by means of AR(n) model or the approximate mean. The outcome of the above example shows that the forecasting accuracy of the surface movement state is improving by forecasting time-varying parameters firstly.