There has been NO warming in Antarctica since 1979 (www.nature.com)

A recent study has developed a new, spatio-temporally complete Antarctic surface air temperature (SAT) dataset using deep learning techniques, marking a significant advancement in climate data for the region. Utilizing a resolution of 1° x 1° from multiple in situ observation sources since 1979, the dataset addresses longstanding discrepancies among existing temperature datasets. Key to this development, researchers trained their deep learning model on daily SAT data from three global reanalysis datasets, enhancing its ability to capture complex spatial temperature patterns more accurately than traditional interpolation methods. The model was validated against observations from both staffed and automated meteorological stations, demonstrating improved alignment with recorded patterns, particularly in capturing long-term temperature trends. This new dataset is essential for the AI and climate science communities, as it not only enhances our understanding of the Antarctic climate system—but also serves as a vital resource for a variety of scientific disciplines. Its implications extend beyond regional studies, providing insights into climate change effects that could impact global sea levels and local ecosystems. The continuous updating capability of this deep-learning-derived dataset positions it as a critical tool for climate impact assessments and model validation in Antarctic research, facilitating more robust discussions surrounding the challenges posed by climate change.