GRSG 35th Conference 2024 Abstract

Title: Understanding coastal landslide processes using low-cost virtual outcrop models

Author: Ryan Kromer

Organisation: University of Leeds

***This work will be presented by Samuel Hinton, MSc candidate at the University of Leeds, or myself. ***

The coastline of SW England facing the English Channel possesses some of the most spectacular cliff faces exposing thick successions of Jurassic-aged geology. The coastline is a UNESCO world heritage site and attracts a wide variety of visitors. Landslide events are a common occurrence which may pose a hazard to visitors. The Environmental Agency has adopted a minimalistic coastal management plan, which allows for natural coastal retreat. At some sites, such as the 42 m-high vertical sandstone cliffs at West Bay, Dorset, a combination of factors including construction of a breakwater, unfavourable geological structure, and a rising sea level, could be contributing to an observed increase in landslide activity. This study aims to better understand the structural controls, the mechanisms of failure and the frequency of landslide activity at the West Bay Cliffs using low-cost close-range remote sensing. The results will be used to better communicate the hazards at the site and to inform the placement of in-place monitoring sensors.

A low-cost UAV and a hand-held DSLR camera were used to collect data for Structure-from-Motion (SfM) virtual outcrop model (VoM) generation in March 2023 and July 2024. The VoMs were used to interpret the geology, geohazards, geomorphic processes and landslide change at the site. For SfM change detection, a co-alignment approach was used, which builds multi-temporal models in a single block during alignment. This ensures VoM consistency over time and results in a high comparative accuracy. The limits of detection at 95% confidence range from 5 cm for the handheld VoMs to 10 cm for the UAV-based models.

Analysis of the VoMs and change results revealed basal failure resulting from wave action, wedge, and toppling failure up to 19 000 m3 resulting from two sets of cross cutting joints, and pervasive smaller scale rock fall resulting from overhanging more resistant sandstone beds and intersecting joint structures. There is also evidence of extensional faulting and cambering induced gulls. A total of 40 gulls up to 2 m wide at the head of the cliff were identified, some containing sand infill and others creating sea caves at beach level. These gulls seem to be associated with a higher degree of weathering and are conduits for water infiltration. A higher rate of activity and larger rock slope failures were observed closer to the breakwater, where the beach has shrunk since its construction.

Conceptual models of the processes driving the landslide activity were created alongside a mapping of areas susceptible to the larger wedge and topple failures. The study was limited in data collection frequency which didn’t allow the detection of slope displacements and didn’t allow for precise timing of the rock slope failures. The study highlights the possibility of obtaining VoMs and change results using low-cost remotely sensing equipment, which can be invaluable to better understanding coastal hazards and change.