Skip to content

Assessing cloud points co-registration methods efficiency based on UAS images acquired in an active volcanic environment (Stromboli volcano, Italy)

Pierre-Yves Tournigand1, 2, Benoît Smets2, 3, Tullio Ricci4, Riccardo Civico4, Riky Centeno5, 6, Matthieu Kervyn2

  • Affiliations:  1Laboratoire Magmas et Volcans, CNRS, IRD, OPGC, Université Clermont Auvergne, France; 2Department of Geography, Vrije Universiteit Brussel, Belgium; 3Department of Earth Sciences, Royal Museum for Central Africa, Belgium; 4Istituto Nazionale di Geofisica e Vulcanologia, Rome, Italy; 5Instituto Geofísico del Perú, Arequipa, Perú; 6Universidad Tecnológica del Perú, Arequipa, Perú 

  • Presentation type: Poster

  • Presentation time: Monday 16:30 - 18:30, Room Poster Hall

  • Poster Board Number: 141

  • Programme No: 3.1.33

  • Theme 3 > Session 1

Abstract

Active volcanoes are challenging environments to study morphological processes, due to their continuous evolution, unpredictability and limited accessibility. Recent technological developments including rapid photogrammetric surveys conducted with Unoccupied Aircraft Systems (UAS) have provided new opportunities for such studies. However, most of the conventional ways to perform surveys (e.g., automated flights, ground control points) and to post-process data (e.g., Real Time Kinematic, Post Processed Kinematic, multi-epoch co-alignment) cannot be applied or do not provide the same data quality as in more stable environments. More specifically, the co-registration of UAS surveys time-series, crucial to accurately assess the morphological evolution of volcanic edifices, proves challenging and necessitates adjustments in the processing techniques used. In this study, we evaluate multiple post-processing techniques (e.g., multi-epoch co-alignment, manual markers, Iterative Closest Point) to accurately co-register the survey time-series of the Stromboli volcano crater terrace. Data have been acquired during two field campaigns in May 2022 and October 2023, with a volcanic activity generating explosive events every 20 to 30 minutes. Our results show that techniques prevailing in the scientific community, such as the multi-epoch co-alignment, may not be the most effective in this context with an average vertical difference over stable areas of 1.95 +/- 0.29 m. On the other hand, less popular approaches, such as manual markers method, may generate higher accuracy results with an average vertical difference over stable areas of 0.28 +/- 0.1 m. We finally provide a best practices protocol for operators willing to work with photogrammetric time-series in active volcanic environments.