中文摘要：

采用变分资料同化技术，结合最优控制思想，对一个海气耦合模型的模式参数和强迫项进行了反演，结果表明，采用该方法对模式进行优化，既可以补偿模式参数不准确性给预报带来的误差，又可以对模式参数本身进行修正和估计，为将来在实际应用中改善更复杂的预报模式、提高预报准确率提供了一个可借鉴的思路。

英文摘要：

For the prediction of ENSO, the accuracy of the model including the parameters, initial value and others of the model is important, which can be retrieved by the variational data assimilation methods developed in recent years. However, when the nonlinearity of the model is quite strong, the effect of the improvement made by the 4-D variational data assimilation may be poor due to the bad approximation of the tangent linear model to the original model. So in the paper the ideas in the optimal control is introduced to improve the effect of 4-DVAR in the inversion of the parameters of a nonlinear dynamic ENSO model. The results indicate that when the terminal controlling term is added to the cost functional of 4DVAR, which originated from the optimal control, the effect of the inversion may be largely improved comparing to the traditional 4DVAR, as can be especially obvious from the phase orbit of the model variables. The results in the paper also suggest that the method of 4DVAR in combination with optimal control cannot only reduce the error resulting from the inaccuracy of the model parameters but also can correct the parameters itself. This gives a good method in modifying the model and improving the quality of prediction of ENSO.