3D Surface Reconstruction and Change Detection of Miage Glacier (Italy) from Multi-date Archive Aerial Photos

Historical aerial images provide valuable information for monitoring and change detection in different areas such as glaciers. Using aerial images, 3D models can be created and the geometry of the glaciers can be digitized manually to analyse their surface extent and evolution. Here, we use archive images to reconstruct 3D models of Miage glacier in the Val Veny region, Mont Blanc, Italy. First, the features of each dataset photos are inspected, and they are pre-processed to be used in the 3D reconstruction phase. This pre-processing is conducted to not only minimize the dataset’s problems, including shadows covering the glaciers and insufficient image overlapping over crucial places, but also to generate the best achievable dense 3D point clouds from each dataset. Point clouds of reconstructed 3D models may be used to detect variations in the Val Veny region and determine the rate of retreat and height differences for each glacier in the area over time. This could be done by computation of distances between points in cloud pairs in consecutive years. In this phase, no manual shifting and translating of point clouds are required as they were georeferenced in the 3D reconstruction phase and automatically co-registered with Iterative Closest Point algorithm. Miage glacier in the Val Veny area was chosen to be assessed considering the quality of generated point cloud during the 3D reconstruction phase. The results of the comparisons indicate how the Miage glacier volume has varied throughout time. While volume changes of the evaluated glacier were almost negligible until the mid-1990s, investigations demonstrate a significant rise in volume decrease at the beginning of 21st century.