Antarctica remains one of the most data-poor regions on Earth for surface air temperature (SAT) monitoring.

Even after the 1957–58 International Geophysical Year expanded stations, effective observational coverage over the continent is often ~10% or less in gridded products.

Vast interior areas (e.g., the East Antarctic Plateau) and pre-1961 records have huge gaps. Conventional interpolation methods (kriging, etc.) produce smooth but uncertain results, limiting confidence in long-term trends, regional variability, and Antarctica’s role in global temperature reconstructions.

Led by Chenxi Ouyang, Qingxiang Li, Zichen Li, and Sihao Wei, the team developed the China global Artificial Intelligence Reconstructed Surface Temperature 20CR/CMIP6 (C-AIRST R/M) datasets using partial convolutional neural networks (PConv), a form of “image inpainting” that treats incomplete temperature fields like damaged photos and learns to fill gaps intelligently.

Key details:

• Training: Models learned from 20th Century Reanalysis (20CR) data and CMIP6 climate model outputs.
• Inputs: Merged China global Land Surface Air Temperature version 2.1 (C-LSAT2.1) with two sea-surface temperature datasets (ERSSTv6 and HadSST4). The HadSST4 merge was preferred for better physical consistency.
• Output: Spatially complete global monthly surface temperature anomaly fields from 1850–2024 at 5° × 2.5° resolution.
• Focus: Dramatically improved Antarctic coverage, where traditional methods have struggled.

Performance highlights (validated post-1961, where more independent station data exist for testing):

• Spatial correlations with reanalysis data >0.99 in tests.
• Against 14 independent Antarctic stations (not used in training): average correlation ~0.72, RMSE ~0.59–0.61°C.
• The model captures large-scale patterns, variability, and trends far better than simple interpolation in data-sparse zones.

The reconstructions indicate a gradual Antarctic warming trend since 1961 (statistically significant at the 0.05 level in the study), with stronger signals in parts of West Antarctica.

This aligns with some reanalysis products but contrasts with debates over limited continent- wide warming in certain satellite- era records.

The AI approach reduces uncertainty from missing data but does not resolve underlying questions about physical drivers or model biases (since it learns patterns from reanalysis and climate models).

This work adds to a growing list of AI applications in climate science: from analyzing glacier flow and iceberg debris in sediments to speeding up image analysis of seafloor life. It demonstrates AI’s value in handling incomplete geophysical datasets, potentially reducing uncertainty in Antarctic contributions to global temperature series.

_____________________________________________________________________________________

AI Proves Its Mettle in Reconstruction of Antarctic Temperatures

From Science Under Attack

By Ralph B. Alexander

In a 2025 post, I described the largely unsuccessful attempt of an AI to spearhead research in climate science. Now, however, another AI appears to have succeeded in the more technical task of accurately reconstructing surface air temperatures across Antarctica – something that standard temperature datasets have been unable to achieve. The work is reported in a recent paper by a team of Chinese researchers.

The figure below illustrates the geographical distribution of available observational data in Antarctica from 1979 to 2023. The data comes from a number of manned and automatic weather stations, together with meteorological observations over the ocean collected from ships and buoys. As can be seen, the majority of observations are in coastal or near-coastal regions, precluding full spatial coverage of the continent.

To overcome this shortfall, the limited station observations have traditionally been interpolated using various reanalyses. But it’s difficult for reanalysis datasets to capture complex spatial patterns, say the researchers, and such datasets often contain significant uncertainties. Moreover, deriving surface air temperatures from reanalysis datasets depends in part on model simulations rather than actual instrumental measurements.

In the light of these limitations to an accurate reconstruction of Antarctic temperatures, the Chinese research team has applied deep learning methods. This approach has already been utilized successfully to reconstruct Arctic temperatures.

Antarctic temperatures were reconstructed using daily surface air temperature data from the various sources depicted in the figure above. Daily average temperatures were calculated from observations made at 3-hour intervals for some sources, 1-hour intervals for others. Training data for the deep learning model was provided by surface air temperatures from the three reanalysis datasets that showed the best agreement with observed temperatures.

The training data, which covered the period from 1979 to 2005, totaled 29,211 daily temperature samples. Using reanalysis data from different time periods, such as 1995 to 2021, as the training set had little impact on the temperature reconstruction. Validation of the training data and testing of the reconstructed temperature set employed reanalysis data from 2006 to 2012 and 2013 to 2018, respectively.

Testing of the reconstructed Antarctic temperatures was conducted for three specific days in 2015: January 1, July 1 and November 1. For these days, the reconstructed temperatures were found to be highly correlated with their reanalysis counterparts, with spatial correlation coefficients >0.99. The researchers say this correlation shows that the trained deep learning model is capable of accurately reproducing Antarctic surface air temperatures, even with the limited observational data available.

Just how different the reconstructed surface temperatures are from global observational temperature datasets for Antarctica is depicted in the next figure. The figure shows linear trends in annual Antarctic surface air temperatures from 1979 to 2023, measured in degrees Celsius per decade. The datasets are: (a) this reconstruction; (b) Berkeley Earth; (c) ERA reanalysis; (d) NOAAGlobalTemp5.1; (e) GISTEMPv4; and (f) HadCRUT5.

You can see that none of the standard datasets exhibit the pronounced cooling trend in East Antarctica in (a), something that was inferred earlier from the ERA reanalysis dataset by a different group of Chinese researchers. Nevertheless, all datasets show warming in the Antarctic Peninsula (on the left of the continent in the maps above).

Differing from the annual trends is the pattern for the summer months (November to April) only, presented in the figure below for the period from 1989 to 2022. Although the cooling trend still dominates in East Antarctica, warming is no longer prominent in the Peninsula but is found in West Antarctica and the southern portion of East Antarctica.

East Antarctica actually experienced a summer heat wave in 2022, when the temperature soared to -10.1 degrees Celsius (13.8 degrees Fahrenheit) at the Concordia weather station, located at the 4 o’clock position from the South Pole, on March 18. This balmy reading was the highest recorded hourly temperature at that station since its establishment in 1996, and 20 degrees Celsius (36 degrees Fahrenheit) above the previous March record high there. Remarkably, the temperature remained above that record for three consecutive days, including nighttime.

But Antarctica is nothing if not unpredictable. Despite the 2022 heat wave, the mercury dropped to -51.2 degrees Celsius (-60.2 degrees Fahrenheit) on January 31, 2023. This marked the lowest January temperature recorded anywhere in Antarctica since the first meteorological observations there in 1956. Two days earlier on January 29, the nearby Vostok station, about 400 km (250) miles closer to the South Pole, registered a low temperature of -48.7 degrees Celsius (-55.7 degrees Fahrenheit), that location’s lowest January temperature since 1957.

Such swings from record highs to record lows remain a puzzle, but the present reconstruction at least helps to characterize long-term trends.