从非线性角度重新解读了vonNeumann提出的大气运动三个分类,指出大气过程预报问题中存在两类不确定性——初始条件的不确定性和外强迫条件的不确定性,它们分别成为第一类可预报性与第二类可预报性问题的研究对象。强调了大气过程可预报性的客观存在性,由于大气过程的复杂性及人类观测手段和认识水平的限制,人们只能给出这些可预报性的估计,逐日天气预报存在上限(2周左右)。分析了预报误差产生的来源是初始条件的不确定性和预报模式的不完善性,但根本原因是大气过程的混沌本质;还分析了混沌系统误差增长理论,但由于实际大气过程包含很多显著不同的时空尺度以及不同尺度间的相互作用,人们对大气过程复杂的误差增长规律的认识还不是很清楚;最后讨论了动力学数值模式中不同的空间分辨率会改变系统的性质,指出可预报性问题的研究必须考虑空间分辨率的影响。而在讨论初值不确定性对预报的影响时,不需要考虑控制参数微小变化产生的影响,当然条件是控制参数的微小变化不会引起系统性质的重大变化。

From the view of nonlinear theory,the three types of atmospheric motions issued by von Neumann(1955) are re-explained,and it is pointed out that in the prediction of atmospheric processes there exist two kinds of uncertainties-uncertainties in initial conditions and uncertainties in external forces,which become the studied objects of the first kind prediction and the second kind of prediction,respectively.In this paper,the objectivity in the predictability of atmospheric processes is emphasized.Owing to the complexity of atmospheric motions and the limitation of observation skill and knowledge of human being,the predictabilities of atmospheric processes can only be estimated.For example,the upper limit of daily weather is estimated to be about two weeks.The sources of prediction errors-the uncertainties in initial conditions and the imperfectability of prediction models are analyzed,and it is emphasized that the essential reason of prediction error growth is the chaotic character of atmospheric motions.