利用一个大气大洋耦合环流模式做了ＥＮＳＯ预测研究。通过用实测的热带太平洋区域海表温度异常叠加到耦合模式模拟的气候海温上强迫预报模式得到了预测所需的初始场，在此基础上进行了系统性的时效为 ２年的后报试验，时间从 １９８１年 １１月至 １９９７年 １２月。对后报结果的检验表明，在预测的前１５个月里，Ｎｉｎｏ３区海表温度距平总体预测相关技巧达到０．５４以上，均方根误差不超过０．９℃。预测系统对８０年代的ＥＮＳＯ现象具有较高的预测能力，而对９０年代的ＥＮＳＯ事件，其预测水平相对较低。进一步分析表明，预测技巧具有季节依赖性，从秋末和冬季开始的预测相对较差而从春季到秋初开始的预测则很好。值得一提的是从每年３月份开始的预测，其０．６以上的相关技巧可达８个月，这对中国夏季降水预测是非常重要的。该预测系统对１９９７至２０００年的ＥＮＳＯ现象的实际预测较为成功。 ENSO prediction was made using a coupled ocean-atmosphere atmospheric circulation model. The initial field needed for prediction was obtained by using the observed superposition of the sea surface temperature anomalies over the tropical Pacific region to the coupled model for simulating the climate over the sea surface temperature. Based on this, a systematic 2-year post-test , From November 1981 to December 1997. Examination of the results of the hindcasts shows that in the first 15 months of the forecast, the overall prediction of sea surface temperature anomalies in the Nino3 area has reached above 0.54 and the root mean square error does not exceed 0.9 ° C. The forecasting system has a high predictive ability for the ENSO phenomenon in the 1980s and a relatively low level of prediction for the ENSO events in the 1990s. Further analysis shows that the forecasting techniques are seasonally dependent, with relatively poor predictions from late autumn and early winter, and good from spring to early autumn. It is worth mentioning that the forecast from March of each year, the related skills of more than 0.6 up to 8 months, which is very important for China’s summer precipitation forecast. The forecast system has been more successful in predicting the ENSO phenomenon from 1997 to 2000.