Automated probabilistic interpretation of rotary soundings and Bayesian spatial prediction of quick clay presence

Quick clay landslides pose a significant hazard to human life, infrastructure, and the environment in several countries. Detecting and mapping quick clay is essential for effective hazard management. The rotary sounding method is widely used for quick clay detection in Norway due to its effectiveness and robustness. There are, however, inherent uncertainties associated with this approach arising from the data interpretation and the limited number of soundings normally available for an area of interest. This study introduces a two-step approach for detection and spatial interpolation of quick clay pockets using rotary soundings. First, an automated procedure is developed to interpret the rotary sounding data with depth. The procedure is based on a heuristic approach that mimics the visual interpretation process and requires the choice of two threshold parameters. These parameters are determined through data analysis by comparing rotary sounding predictions with direct ground-truth samples from three quick clay sites. The interpreted data is then used to estimate the probability of quick clay presence beyond the sounding locations using a hierarchical Bayesian model. The effectiveness of the proposed method is demonstrated using rotary sounding data from a quick clay test site in Norway.