一阶模式误差订正方案的长期积分性质对初值的依赖性

Dependence of the First-Order Model Error Correcting Plan's Long-Term Integral Properties to Initial Value

采用最大简化气候模型,使用理想试验的方法,对模式用无误差的参数积分10 010 d,每隔10 d取一个值,获得1 000个初值,用任意一个初值积分500 000步,用所得的数据通过grapher软件可得出散点图和概率密度分布图,通过对模式长期积分在相空间中的概率密度分布分析得出参数无误差模式和有误差模式的长期积分性质对初值的依赖性不大而一阶模式误差订正方案长期积分性质对初值的依赖性很大,且与回溯阶有很大的关系。

This paper adopts the most simplified climate model,together with the ideal experimental model to compute the integral（10 010 days） on condition of non-error parameters. 1 000 initial values can be obtained from this trial（every 10 days for a value） and then any one of the listed initial value is selected to be computed 500 000 times,from which scatter diagram and profile of probability density can be acquired by the grapher software. From the study of the profile of probability density distribution conducted on long-term integration in the phase space,it can be found that the integral property has less dependence on initial values on condition of error free and existed parameters,while the first-order model error correction scheme＇s long-term integral properties depend considerably on the initial value and there are some correlations with the retrospective order.