物理耗散技术及其在MM4中的应用

PHYSICAL DISSIPATIVE TECHNIQUE AND ITS APPLICATION TO THE MM4

在传统的流体动力学框架下，数值天气预报问题往往被提为经典牛顿力学的确定论初值问题 。然而大气是多体系统，利用仅对三体以下的系统才能精确描述的理论来处理多体系统的运 动，会引起较大的误差和问题。为此，文中提出了一套在数值天气预报模式中引入“不可逆 热力学算子”的技术，按热力学第二定律的约束方式，来控制多体系统的演变方向，以提高 数值天气预报的精度。例如在ＭＭ４模式中引入不可逆热力学算子后，温度场、湿度场、高度 场及全风速场的平均相对均方根误差减小了约１３％，其中４８ ｈ预报的误差减小两成以上。鉴于文中提出的不可逆热力学算子引入技术是基于描述耗散性的物理定律，而非纯粹出于计算 上的考虑，故名为物理耗散技术。由于它所依据的物理原理对所有流体力学和大气数值模式 均适用，因此该项新技术在地球物理流体力学界将有广阔的应用前景。

Under the traditional framework of fluid dynamics, the problem of the numeric al weather prediction is often expressed as the deterministic initial value prob lem of the classical Newtonian mechanics. The atmosphere is, however, a many bo dy system, the methodology by which the system with two bodies could be precise ly solved would cause bigger errors and problems when handling the many body sy stem by it. A kind of technique to incorporate “the irreversible thermodynamic operators”, therefore, into the numerical weather prediction models is suggest ed in this paper, to control the evolutionary direction of the many body system according to the constraining way of the second law of thermodynamics, and thus the forecasting accuracy of the numeric al weather prediction has been noticeably improved: in(e.g.) the MM4 the average d relative root mean square error of the fields of the temperature , humidity , height and whole wind velocity has decreased by about 13%, among which the a vera ged error of the 48 h forecasts has decreased by more than 20%. Since the techn ique to introduce the irreversable thermodynamic operator suggested in this pape r is based on the physical lew that describes the dissipativity and comes out no t only from the computational consideration, the technique is named as the physi cal dissipative technique. In view of the universality of the principle incorporating the irreversible thermodynamics operators suggested in this paper for the fluid dynamics and atm ospheric numerical models, the applications and generalization of this incorpora ting technique would produce a great impact on the field of geophysical fluid dy namics.