系统优化理论在CERES－大豆模式品种参数求解中的应用

APPLICATION OF OPTIMIZING THEORY TO DETERMINING GENETIC PARAMETERS INVOLVED IN CERES-SOYBEAN MODEL

根据田间试验资料，采用多维非线性单纯形法搜索求解模式中的品种参数，比较全面地评价了ＣＥＲＥＳ－大豆模式（ＳＯＹＧＲＯＶ５．４１）对我国大豆生长过程及其产量形成的模拟效果。结果表明：（１）若仅考虑某一单项的模拟效果，如发育期或收获产量等，而不涉及其余项目，则可以调整有关参数得到相当满意的模拟结果；若考虑所有模拟输出项的模拟效果，则其精度明显下降，相对误差在１０％～６０％之间。（２）对于某些生长过程，如叶面积指数动态和地上部分于物重的增重过程等，用ＣＥＲＥＳ大豆模式模拟得到的时间趋势与实例资料有较大差别；而对籽粒增重的模拟效果较好。

Based on field experimental measurements,CERES-Soybean Model (SOYGRO V5.41) is tested comprehensively on its simulation results of growth and yield production. Genetic parameters involved in this model are approached and computed out with multi-dimensional and non-linear simplex method.The simulation resuIts show that:(1) If only is some one of crop variables such as devlopment stage or harvest yield,concerned and the others are ignored,excellent simulation resuIts concerned variable can be obtained by readjusting the related parameters properly。 When all thesimulation outputs are taken into account for testing,the resuIts will not be so satisfied,the relative errors are between 10%～60%。(2) For some crop variables,such as thetime courses of leaf area index and canopy dry weight ,etc ,the simulated time trends bythis model are considerablv different from those of measurements。Whereas,thesirnulation resuIts of grain growth are better。