Wilks Daniel S
Department of Earth and Atmospheric Sciences, Cornell University, 1113 Bradfield Hall, Ithaca, NY 14853, USA.
Philos Trans A Math Phys Eng Sci. 2008 Jul 28;366(1875):2477-90. doi: 10.1098/rsta.2008.0005.
Conceptual climate models are very simple mathematical representations of climate processes, which are especially useful because their workings can be readily understood. The usual procedure of representing effects of unresolved processes in such models using functions of the prognostic variables (parametrizations) that include no randomness generally results in these models exhibiting substantially less variability than do the phenomena they are intended to simulate. A viable yet still simple alternative is to replace the conventional deterministic parametrizations with stochastic parametrizations, which can be justified theoretically through the central limit theorem. The result is that the model equations are stochastic differential equations. In addition to greatly increasing the magnitude of variability exhibited by these models, and their qualitative fidelity to the corresponding real climate system, representation of unresolved influences by random processes can allow these models to exhibit surprisingly rich new behaviours of which their deterministic counterparts are incapable.
概念性气候模型是气候过程非常简单的数学表示形式，特别有用，因为其运作易于理解。在这类模型中，使用不包含随机性的预报变量函数（参数化）来表示未解析过程的影响，通常会导致这些模型的变率远低于它们旨在模拟的现象。一种可行但仍很简单的替代方法是用随机参数化取代传统的确定性参数化，这可以通过中心极限定理从理论上得到证明。结果是模型方程成为随机微分方程。除了大大增加这些模型所表现出的变率大小以及它们与相应真实气候系统的定性逼真度之外，用随机过程表示未解析的影响还可以使这些模型展现出其确定性对应模型无法实现的惊人丰富的新行为。
