How to extract external forcing information from continuous nonstationary systems is important for studying the variability of the climate system. Based on the slow feature analysis (SFA), in this paper, two independent driving forces with different periods are given to affect the modified Lorenz model to discuss whether the SFA approach can reconstruct the forcing signals. The results show that, to some extent, SFA can dig them out, and the result is related to the embedding dimension m, the analyzed variables, and the position of the forcing signal acting on the system. At the same time, combined with wavelet analysis, the two original forcing signals can be separated from the SFA extracted signal. In addition, we apply the above method to the climatic data through the 500-hPa geopotential height field and focus on the region where SFA-extracted signal have a significant scale of about 4 years over East Asia and speculate on the possible physical factors based on the distribution characteristics of the region. The results show this scale may be derived from the East Asian trough. This work may open up the possibility to improve the monthly prediction of 500-hPa geopotential height field over East Asia from the perspective of causality.

如何从连续的非平稳系统中提取外部强迫信息对于研究气候系统的变异性很重要。基于慢特征分析（SFA），本文给出了两个不同周期的独立驱动力来影响改进的Lorenz模型，讨论了SFA方法是否可以重构强迫信号。结果表明，在一定程度上，SFA可以将其挖掘出来，其结果与嵌入维数m有关。，分析的变量以及作用在系统上的强制信号的位置。同时，结合小波分析，可以将两个原始强制信号与SFA提取信号分离。此外，我们通过500-hPa的地理高度场将上述方法应用于气候数据，并重点研究了SFA提取的信号在东亚地区具有大约4年显着尺度的区域，并推测了可能的物理因素。该地区的分布特征。结果表明，该规模可能源自东亚海槽。从因果关系的角度来看，这项工作可能为改进东亚地区500hPa地势高度场的月度预报开辟了可能性。