The 1967 Steinholtshlaup: The use of Structure from Motion (SfM) photogrammetry to understand its geomorphological legacy

• Main Author: Fisher, Jamie
• Other Authors: Dugmore, Prof. Andrew
• Format: Master Thesis
• Language: English
• Published: The University of Edinburgh 2021
• Subjects: Steinholtshlaup; Jökulhlaup; Iceland; Landslide; Structure from Motion; DEM
• Online Access: https://hdl.handle.net/1842/38944; https://doi.org/10.7488/era/2196

Landslide-induced jökulhlaup are an increasingly pertinent glacial hazard with climate change enacting paraglacial landscape responses which are instigating more frequent landslide and proglacial lake interactions, a matter of utmost relevance in areas of ongoing deglaciation like the Himalayas. This study used the 1967 Steinholtshlaup in Iceland as a case study to explore the feasibility of using an Earth observation approach through Structure from Motion (SfM) photogrammetry, establishing geomorphological landscape changes, and providing a better understanding of the mechanisms that control jökulhlaup interactions with the landscape and the enduring geomorphological legacy. Historical aerial photography and a modern Unmanned Aerial Vehicle (UAV) survey of the valley were used to derive Digital Elevation Models (DEM) and orthomosaics, used to analyse geomorphic change and boulder feature tracking over a 60 year time period between 1960 to 2019. The results highlighted the challenges inherent with combining datasets of varying spatial and temporal resolution but demonstrated the ability of Earth observation to better understand the impacts of jökulhlaup phenomena by quantifying time-series terrain modification. A clear pattern of decaying volume of deposited debris was identified along the valley floor, and a relationship was found between the pre-existing valley characteristics and the evolving behaviour and impact of the hlaup on the landscape. This led to the discovery that the distal boulder deposits of up to 80m³ were picked up enroute by the flood from the end of the valley, rather than being carried 5km from the landslide’s origin, notably disagreeing with Kjartansson’s (1967) original first-hand observations. This study serves to emphasise the current importance of remote sensing in measuring the impacts of these increasingly frequent hazards across often inaccessible glacial landscapes.