In this paper, we analyse the calving activity of the Bowdoin Glacier, north-western Greenland, in 2015 by combining satellite images, UAV (unmanned aerial vehicle) photogrammetry and ice flow modelling. In particular, a high-resolution displacement field is inferred from UAV orthoimages taken immediately before and after the initiation of a large fracture, which induced a major calving event. A detailed analysis of the strain rate field allows us to accurately map the path taken by the opening crack. Modelling results reveal (i) that the crack was more than half-thickness deep, filled with water and getting irreversibly deeper when it was captured by the UAV and (ii) that the crack initiated in an area of high horizontal shear caused by a local basal bump immediately behind the current calving front. The asymmetry of the bed at the front explains the systematic calving pattern observed in May and July-August 2015. As a corollary, we infer that the calving front of the Bowdoin Glacier is currently stabilized by this bedrock bump and might enter into an unstable mode and retreat rapidly if the glacier keeps thinning in the coming years. Beyond this outcome, our study demonstrates that the combination of UAV photogrammetry and ice flow modelling is a promising tool to horizontally and vertically track the propagation of fractures responsible for large calving events.