<p>Recent changes in Antarctic surface air temperature (SAT) have received growing attention. However, due to sampling limitations in surface observations, detecting SAT trends in Antarctica carries significant uncertainty. This study estimates SAT data using the more comprehensive Moderate Resolution Imaging Spectroradiometer (MODIS) land surface temperature (LST) in Antarctica, based on the linear relationship between LST and SAT provided by the ERA5 reanalysis. Considering factors such as the ice-snow albedo model and cloud contamination, which cause biases in the raw satellite LST data, the bias is corrected using in-situ temperature measurements through a probability density function (PDF) matching method. The results show that the bias-corrected MODIS-derived SAT has shown notable improvements in terms of representativeness of Antarctic SAT changes compared to both ERA5 and in-situ measurements. Based on various datasets from 2001 to 2023, Antarctica exhibits notable spatial heterogeneity in average temperature, with the Antarctic Peninsula, West Antarctica, and coastal regions being relatively warmer, while the interior of East Antarctica remains comparatively colder. During this period, the Antarctic Peninsula has shown a warming trend, whereas West Antarctica has exhibited a cooling trend. However, given the fitting and data uncertainties, the SAT trends for Antarctica and its sub-regions are not statistically significant at the 95% confidence level, suggesting that anthropogenic warming has not yet exceeded the level of internal climate variability in Antarctica.</p>