Long-term glacier elevation change at Gran Campo Nevado since 1945

Scanned historical aerial photographs acquired with film cameras from the early twentieth century to the early 2000s are the longest and richest archive of Earth observation data for reconstructing past topography. Those with stereoscopic acquisition enable the generation of Digital Elevation Models (DEMs) and orthoimages when processed with photogrammetric techniques, extending the assessment of environmental change beyond the time scale of modern satellite observations.  

In this study, we present a long-term (1945-2020) dataset of glacier surface elevation for the Gran Campo Nevado ice field in southern Chile. The dataset is based on aerial photographs acquired in 1945 using a Trimetrogon camera and in the 1980s and 1990s using nadir-looking film cameras from the Chile60 and Geotec flight campaigns, complemented by a 2020 Pléiades satellite–derived DEM made available through the Pléiades Glacier Observatory program (Berthier et al., 2023). To process the historical photographs, we developed an open-source pipeline that builds on structure-from-motion (SfM) principles and incorporates learning-based feature-detection and matching algorithms, such as SuperPoint and LightGlue. Absolute image orientation is achieved through automated detection of ground control points derived from the Pléiades DEM and orthoimage. DEMs accuracy was evaluated over stable terrain by comparing them with the Pléiades reference DEM. As well, the reconstructed DEMs are compared with those obtained using an established SfM processing workflow (HSfM; Knuth et al., 2023). The resulting DEMs provide a reconstruction of glacier surface elevation spanning more than seven decades, and glacier elevation changes are quantified from the DEM time series. By using reproducible, open-source methodologies, this presentation demonstrates opportunities for the research community to leverage other historical datasets and extend analyses beyond what is possible with modern satellite observations alone.