Laboratoire de Meteorologie Dynamique (CNRS and IPSL), Ecole Normale Superieure, 75231 Paris Cedex 05, France.
Chaos. 2009 Dec;19(4):043109. doi: 10.1063/1.3230497.
In a simple, one-layer atmospheric model, we study the links between low-frequency variability and the model's fixed points in phase space. The model dynamics is characterized by the coexistence of multiple "weather regimes." To investigate the transitions from one regime to another, we focus on the identification of stable manifolds associated with fixed points. We show that these manifolds act as separatrices between regimes. We track each manifold by making use of two local predictability measures arising from the meteorological applications of nonlinear dynamics, namely, "bred vectors" and singular vectors. These results are then verified in the framework of ensemble forecasts issued from "clouds" (ensembles) of initial states. The divergence of the trajectories allows us to establish the connections between zones of low predictability, the geometry of the stable manifolds, and transitions between regimes.
在一个简单的单层大气模型中，我们研究低频可变性与模型相空间中固定点之间的联系。该模型的动力学特征是存在多个“天气型”。为了研究从一种天气型向另一种天气型的转变，我们重点关注与固定点相关的稳定流形的识别。我们表明，这些流形充当了天气型之间的分隔线。我们通过利用非线性动力学在气象学中的两个局部可预测性度量来跟踪每个流形，即“繁殖向量”和奇异向量。然后，我们在从初始状态的“云”（集合）中发布的集合预报的框架内验证这些结果。轨迹的发散使我们能够建立低可预测性区域、稳定流形的几何形状以及天气型之间转变之间的联系。
