APPLICATION OF THE FOUR-DIMENSIONAL VARIATIONAL DATA ASSIMILATION TECHNIQUE ON OPTIMIZING THE INITIAL CONDITIONS OF Z-C MODEL

The four-dimensional variational data assimilation (4DVAR) technique is applied in the Zebiak-Cane ENSO forecasting model for optimizing its initial conditions by using the adjoint method. The performance of adjoint model is checked via an ideal experiment. It suggests that ENSO forecasting model based on the adjoint model of 4DVAR technique has an ability of assimilating the observed sea surface temperature data. Comparing to the observational Nino3 Index, the simulated and forecasted one produced using the present model generally agree with that of observational during the period of 1997-1999. This suggests that Zebiak-Cane ENSO forecasting model has been improved by using the 4DVAR technique.