Evaluation of laboratory methods to quantify particle size segregation using image analysis in landslide flume tests

Landslides comprised of a wide range of particle sizes (e.g. debris flows) exhibit flow structures arising from particle size segregation. Segregation influences the mobility of the flow, the development of debris fans, and the resulting impact forces to be expected when designing barriers and containment structures. In order to capture the flow dynamics of segregable materials in numerical simulations, experimental datasets quantifying segregation in the final deposit are required. However, the measurement of segregation is not a straightforward task as segregation observed at an external transparent boundary may not be indicative of segregation within the bulk of the landslide mass due to sidewall friction. In this paper, we explore the use of four different strategies to optically measure particle size segregation in large landslide flume tests, comparing measurements taken (i) at the external transparent flume boundary; (ii) using a thin transparent plane as a splitter plate along the centre of the flow; and using a (iii) vertically or (iv) horizontally inserted transparent plate into the static deposit after flow arrest. Relationships between concentrations measured by projected area (i.e. sidewall image) to concentrations by mass are derived and validated for a tridisperse mixture to assess which sampling method most closely represented the original source volume. Of the four strategies tested, the transparent splitter plane method was identified to cause the least amount of out-of-plane segregation of particles, provides a rich database of highly detailed observations of segregation of tridisperse granular flows that can be used to evaluate future numerical model outcomes, and is recommended for future laboratory flume investigations.