本文以LAI作为结合点，讨论了利用复合型混合演化（SCE—UA）算法实现CERES—Wheat模型与遥感数据同化的可行性。CERES—Wheat模型同化后主要生育期和产量的模拟值分别与真实条件下模型相应模拟值以及实测值进行比较。结果表明，同化后CERES—Wheat模型的模拟精度对LAI外部同化数据的误差并不十分敏感。并且在LAI同化数据较少时，也可获得较好的同化结果。这一特点体现了SCE—UA算法应用于同化过程的优越性，为同化策略在区域冬小麦长势监测及估产中的应用提供了基础。

In this paper,the shuffled complex evolution(SCE_UA) method was used to assimilate remotely sensed data into CERES_Wheat model.In the process of model assimilation,leaf area index(LAI) was considered as the state variable.The simulated main growth stages and yields after assimilation were compared with simulated growth stages and yields with CERES_Wheat using actual input,and with measured data in the fields.The measured data was collected from four fields in different locations and planting conditions in Shunyi district and Beijing.The results show that the accuracy of simulation results of CERES_Wheat model after assimilation is not very sensitive to LAI errors and the number of LAI data.The advantage of the SCE_UA method will help to realize wheat growth monitoring and yield prediction.