Multi-temporal image co-registration improvement for a better representation and quantification of risky situations: the Belvedere Glacier case study

Scientific applications dealing with natural hazards make wide use of digital geographical data and change detection techniques. If the attention is focused on changes affecting surfaces' geometry, multi-temporal aerial photogrammetry can represent an effective tool. In this case, the degree of spatial coherence between measurements at different times is an important issue to deal with. Reliability and accuracy of measured differences strictly depend on the strategy used during image processing. In this paper, a simultaneous multi-temporal aerial image bundle adjustment approach (MTBA) is compared against two more traditional strategies for aerial stereo-pair adjustment to map surface changes of the Belvedere Glacier (Italian north-western Alps) in the period 2001-2003. Two aerial stereo pairs (of 2001 and 2003) were used to generate the correspondent digital surface models. These were then compared to map glacier shape differences and calculate ablation and accumulation volumes. Results demonstrate that the proposed MTBA approach improves and maximizes accuracy and reliability of measured differences also when available reference data are low quality ones. Final uncertainty for both direct (surface height differences) and derived (volume changes) measurements were quantified and mapped.