The non-stationary characteristics of the climate system have been widely recognized. The occurrence of this non-stationary phenomenon is caused by the hierarchical structure of the climate system. As a high-level system, the external driving forcing changes with time, which leads to the non-stationary phenomenon of atmospheric movement. Slow feature analysis (SFA) method can extract the slow-changing features from fast-changing non-stationary signal. The SFA has been applied to attribution analysis of the climate system. In this paper, we use the SFA method to extract the driving force signal from the non-stationary time series obtained by the Henon mapping model to test its extraction capability. Then we extract the external driving force signal from Beijing monthly average temperature time series, and analyze the scale characteristics and physical mechanism of external driving forcing signals combined with wavelet transform. The results show that the long-period external driving forcing signal and the short-period external driving forcing signal jointly work on the climate system. At the same time, the long-period external driving forcing signal also works on short-period external driving forcing signal. This work contributes to understanding the hierarchical characteristics of the climate system.