Multi-Temporal Block Adjustment For Aerial Image Time Series: The Belvedere Glacier Case Study

While considering monitoring actions aimed at natural hazards management, an historical research about the investigated phenomenon is mandatory, in order to acquire those data necessary to understand its dynamics and to forecast possible evolutions. In this paper, some results referred to an experience of multitemporal aerial triangulation are presented, with the purpose of demonstrating accuracy improvements in stereo model orientation. The case study is the Belvedere Glacier that belongs to the Monte Rosa east face (Italian Western Alps) and since 1868 has been being interested by superglacial lakes outbursts, ice avalanches and surge-type movements, awakening the interest of the scientific community in studying these phenomena and generally to consider a glacier monitoring. Approaching the multitemporal analysis of the geometric features of moving surfaces, the most critical factor is always the planimetric coherence between the oriented stereo models of the different periods. This work tries to give a preliminary answer to this problem. Two digitized aerial stereo-pairs (years 2001 and 2003) acquired at different scales were used for the tests . Three Aerial Triangulation approaches were considered and compared: a) a traditional Single Block Bundle Adjustment (SBBA) ; b) a Single Block Sequential Bundle Adjustment (SBSBA); c) a Multi-temporal Block Bundle Adjustment (MTBBA). Statistical parameters referring to some Check Points sited in no-moving areas demonstrate that MTBBA produces a better coregistration among blocks, guaranteeing a better coherence between measurements and consequently a higher precision in determining displacements measured at the ground level. The declared intention of testing the results in an operative context, suggested to carry out image orientation exploiting existing data for the identification of the Ground Control Points. Even if the image mean scale suggests the use of high precision Ground Control Points, a faster approach was chosen, measuring: a) their planimetric coordinates on the available orthoimages having a nominal scale of 1:10.000; b) extracting the elevation values from the Piemonte Region Digital Elevation Model. According to the selected Check Points, resulting coordinate differences were successively compared and statistical parameters were calculated. The stereo models, oriented according to the MTBBA approach, were then employed in the computation of a map describing glacier surface variations, aimed to the description of its dynamics in the referring period (2001-2003).