The slow feature analysis (SFA) can extract slowly varying external forcing information from non-stationary time series. In recent years, the SFA method has been applied to the climate change field to explore the potential driving forces of climate change and related dynamic mechanisms. This study extracts the slowly varying external forcing information of the global land surface air temperature (LSAT) based on the SFA method. It investigates the spatial structure characteristics of the global LSAT slow varying driving force and the main driving factors of low-frequency variability. The LSAT slowly varying driving force extracted by the SFA method has a significant correlation with global radiative forcing (GRF) and the main modes of the global sea surface temperature (SST) (i.e., Atlantic Multidecadal Oscillation, tropical Pacific El Niño–Southern Oscillation variability, and Interdecadal Pacific Oscillation), indicating that the LSAT variability in most parts of the world is significantly affected by GRF and the three SST modes. The influence of GRF on the LSAT variability has global consistency, while that of the three SST modes on the LSAT variability has obvious regional characteristics. In addition, the interpretation variance of the LSAT variability of the GRF and SST modes significantly improved because the SFA method can effectively reduce the explanatory random noise in the original LSAT sequence, further showing that the GRF and SST modes are the main driving factors of the global LSAT low-frequency variability. Finally, the results of the historical sea surface temperature-driven Atmospheric General Circulation Model test, which is also known as the Atmospheric Model Intercomparison Project (AMIP)test, verify the significant influence of the three SST modes on the regional LSAT variability.