The forecasting method is important because it can predict phenomena in the future. The accuracy of the forecast depends on the model and the initial conditions. In addition the predictability measurement method is important can be check the accurate of forecasts. In this research is winter monsoon forecasts in Thailand by the shallow water model. The data from The Bjerknes Centre for Climate Research (BCCR) University of Bergen Norway. The global climate model is Bergen Climate Model (BCM) Version 2.0 (BCCR-BCM2.0) of the Intergovernmental Panel on Climate Change (IPCC) is used. The Lyapunov exponent (LE) is the predictability measurement method for verify the efficiency of model and establish the new predictability measurement method by the time series analysis. The result to show that the new predictability measurement method by the time series analysis can be measure the efficiency of the winter monsoon forecasts in Thailand by the shallow water model for December 2015 to December 2056 are suitable.

The shallow water model is a mathematical model was developed from the shallow water equations. It consists of the continuous flow and the momentum equation of non-linear in the x-axis and y-axis. The boundary conditions of the model is the cyclic continuity from east to west and the open boundary conditions from north to south. The northeast monsoon forecast by the shallow water model can’t be predicted in a long time period because the boundary conditions of the model has a problem. Thus improvement the boundary conditions of shallow water model by using wall boundary condition from north to south. The results of this research to found that the shallow water model can correctly forecast the northeast monsoon period in the long term.