Reconstruction of the collapse of the 'Azure Window' natural arch via photogrammetry

TL;DR

This study uses underwater photogrammetry with a diver-mounted camera to reconstruct and analyze the collapse of the Azure Window arch, providing insights into the collapse dynamics and post-event changes.

Contribution

It demonstrates a novel application of high-resolution underwater photogrammetry for documenting and understanding complex geomorphological collapse events.

Findings

Collapse initiated by erosion at the pillar base

Collapse involved separation along pre-existing joints

Post-collapse site changes documented

Abstract

The Azure Window was a natural arch situated in the west coast of Gozo (Maltese Archipelago) that collapsed in March 2017. We employ a Diver Propulsion Vehicle-mounted camera system to capture data for the 3D-modelling of this collapsed arch via photogrammetry. We demonstrate use of this method to document complex underwater geomorphology spread across a large area, and draw up a geomorphic assessment of the site and collapse event on the basis of this 3D-model. The methodology enables a reconstruction and understanding of the collapse event. On account of the high-resolution attained, we are able to cross-match the principal submerged components with their pre-collapse location; this enables an understanding of the dynamics of the collapse, confirmation of rock break-up along existing joints, and mapping the distribution of the rock and debris from the collapse event. We conclude that the key stages in the collapse of the Azure Window entailed erosion at the base of the pillar, leading to the latter's collapse in the southwest direction, breaking into two main sections that separated along the lithological boundary. We also find clear evidence that separation of some sections of the pillar followed pre-existing joints. The bridge collapsed vertically upon loss of support from the pillar, breaking into two main components and many other fragments. We also document further changes at the site post-collapse. We show that this approach can be utilised to understand and characterise such events even when significant time has elapsed since collapse, and rocks have already undergone erosion and significant marine growth.