The East Antarctic Ice Sheet (EAIS), holding significant potential sea-level rise, has shown increased vulnerability in recent decades. While some marine-based catchments experience mass loss due to warm water intrusion, terrestrial catchments exhibit mass gain from increased accumulation. Long-term observations are crucial to distinguish recent trends from natural variability. Historical datasets, particularly aerial photographs from early expeditions, offer a unique opportunity to extend the observational record beyond the satellite era. Studies in Greenland and Svalbard have successfully used such historical data to assess long-term glacier response to climate change. However, in Antarctica, the scarcity of pre-1970s climate data and uncertainties in climate reanalysis estimates hinder a clear understanding of observed trends. This study re-examines the earliest large-scale aerial photography campaign in Antarctica (1936-37 Norwegian expedition), providing a unique record of glacier evolution extending back to the 1930s. This archive, combined with later Australian aerial photographs and modern satellite data, allows for quantification of decadal-scale glacier changes across the 20th and 21st centuries. The study focuses on 21 glaciers across three regions of the EAIS, totaling 2.6 million cubic kilometers of ice, regions which have maintained balance or gained mass in recent decades. The 1936-37 Norwegian expedition images, despite challenges in metadata and image quality, provide the earliest detailed view of a regional Antarctic coastline enabling historical glacier reconstruction.

Recent research highlights the increased vulnerability of the EAIS, with mass loss primarily concentrated in marine-based catchments due to Circumpolar Deep Water intrusion. Conversely, terrestrial catchments have shown mass gain due to increased accumulation. However, observational time series pre-dating the satellite era are limited, making it difficult to differentiate recent trends from natural fluctuations. While ice cores and geological data provide long-term context, their spatial and temporal resolutions are often limited. In contrast, historical aerial photographs offer broad spatial coverage and detailed temporal information, crucial for long-term analysis and model calibration. Prior studies using historical aerial images have been successful in Greenland and Svalbard, but their application to Antarctica has been limited due to data scarcity and challenges in processing oblique images and poor-quality scans.

This study utilized approximately 300 aerial images from three sources: the 1936-37 Norwegian Thorshavn IV expedition, subsequent Australian aerial campaigns (1954-1973), and modern satellite data. The 1936-37 images presented significant challenges due to limited metadata, oblique angles, and poor image quality. A manual image selection and geolocation process was employed, selecting images with sufficient overlap, contrast, and bedrock features for analysis. Structure-from-Motion (SfM) photogrammetric techniques were used to reconstruct past ice sheet configurations from the historical images. The SfM models were georeferenced using Ground Control Points (GCPs) derived from modern high-resolution satellite imagery. Elevation changes were quantified for 12 glaciers, with uncertainties determined by comparing historical surfaces to modern reference surfaces over stable bedrock. The generated Digital Elevation Models (DEMs) varied in pixel size (8-25 m) and elevation uncertainty (1.6-9.9 m). Orthophoto mosaics were also created (1-8 m pixel size). Historical flow speeds for four glaciers were determined by tracking crevasse movement between image acquisitions. Frontal changes for all 21 glaciers were measured from 1937 to 2023. In addition to the historical data, modern altimetry data (ICESat and ICESat-2) from 2003-2021 were included for comparative analysis. ERA5 reanalysis data were used to assess long-term trends in mean annual snowfall and mean austral summer air temperature for each region of interest. The sensitivity of results to circle placement and slight shifting of circle positions was tested through subsampling and repositioning. The overall elevation changes were calculated as the average of multiple circles to account for data gaps in DEMs.

The study revealed two distinct regional patterns: 1) Constant long-term ice elevations and historical frontal retreat in Lützow-Holm Bay. All six glaciers in this bay experienced net retreat between 1937 and the 1980s, followed by advancements, with the largest fluctuations observed at Shirase Glacier. Despite the frontal retreat, grounded sections of the glaciers showed insignificant surface elevation change. Modern altimetry confirms limited elevation changes, except for Shirase Glacier which showed thickening. 2) Regional frontal fluctuations and long-term ice thickening in Kemp & Mac Robertson Land and Ingrid Christensen Coast. While no regional long-term trend in frontal positions was observed, glaciers fluctuated between advances and retreats. Overall thickening was observed in these regions from historical DEMs, with the largest changes at Hoseason and Taylor Glaciers. Decadal variability was also observed, with periods of thinning superimposed on the long-term thickening trend. Historical flow velocities were found to be relatively constant since the 1950s for four glaciers, matching modern satellite-derived velocities. The long-term increase in snowfall corresponds with observed historical changes in glacier elevations, most significantly in Kemp and Mac Robertson Land, which show the greatest surface elevation increases.

The findings indicate that the long-term stability and growth of ice elevations observed in East Antarctica are not recent phenomena, but rather part of a trend spanning at least a century. The long-term thickening in Kemp and Mac Robertson Land and Ingrid Christensen Coast corresponds with an increase in snowfall in these areas, supporting the hypothesis that increased precipitation is the primary driver of observed ice thickening. The observed frontal changes in Lützow-Holm Bay are linked to changes in land-fast sea ice conditions, while the lack of corresponding elevation changes suggests limited buttressing from floating ice tongues on a decadal time scale. The minimal impact of surface melting on elevation changes in these areas is supported by the low summer temperatures in the region. The decadal variability observed within the long-term trends highlights the importance of long-term perspective when interpreting short-term changes in glacier dynamics. The study's results are consistent with basin-wide altimetry measurements, except in the Prince Olav Coast region (including Lützow-Holm Bay), where the basin-wide thickening trend does not apply locally. The mass gain in this study's regions has likely offset some mass loss from other sectors of the EAIS and WAIS. The observed trends are likely to persist given projections of increased snowfall across the EAIS.

This study provides a unique long-term perspective on glacier evolution in East Antarctica, demonstrating that recent ice growth is part of a longer-term trend spanning at least 85 years. The observed changes in ice thickness are strongly correlated with changes in snowfall, suggesting that atmospheric processes are dominant in shaping the mass balance of the terrestrial basins of the EAIS. Future studies could benefit from improving the accuracy of pre-satellite era climate data reconstructions to better understand the driving forces behind the observed changes. Further investigation into the potential for interactions between ocean and atmosphere in driving these long-term trends is warranted.

The study's limitations include challenges in processing the historical aerial photographs, primarily due to image quality, oblique angles, and limited metadata. The accuracy of elevation change estimations varies due to the resolution and quality of the available images and the presence of data gaps in the resulting DEMs. The study primarily focuses on glacier elevation changes and frontal positions, and further research is needed to thoroughly investigate other aspects of glacier dynamics such as ice velocity changes and mass balance. The ERA5 reanalysis data utilized may introduce uncertainties in pre-satellite era climate estimations.