论文部分内容阅读
基于NCEP/NCAR再分析资料和COADS海洋资料中的全球月平均海平面气压场、850hPa纬向风场及海表温度场 ,利用Matlab中的NeuralNetworkToolbox仿真环境和BP模型改进算法比较准确地仿真和反演出了南方涛动指数、赤道纬向风指数和滞后的赤道东太平洋海温之间的动力结构和预报模型。该模型具有很好的拟合精度和可行的预报效果 ,可在一定时效内预测赤道东太平洋月平均海温的变化趋势。由于所建系统是具有直接因果关系的预报模型 ,因此不仅可直接用于预测 ,而且可有效避免类似非线性微分方程组在积分过程中由于对初值敏感性而可能产生的对预报结果的不确定性
Based on the NCEP / NCAR reanalysis data and the global monthly average sea level pressure field, the 850hPa zonal wind field and the sea surface temperature field in the COADS oceanographic data, the NeuralNetworkToolbox simulation environment in Matlab and the BP model improvement algorithm are used to accurately simulate and reverse The dynamic structure and prediction model for the Southern Oscillation Index, the Equatorial Zonal Wind Index, and the Lagging Equatorial Pacific SSTA have been performed. The model has good fitting accuracy and feasible forecasting effect, and can predict the monthly mean SST in the eastern equatorial Pacific over a certain period of time. Because the system is a forecasting model with a direct causal relationship, it can not only be directly used in forecasting, but also can effectively avoid similar nonlinear differential equations in the integration process due to the sensitivity of the initial value may be generated for the forecast results do not Certainty