Observations of root growth in stratified soils at the microscopic scale: Insights from micro-computed tomography

Enhancing the overall resilience of vegetated slopes against shallow mass movement can be achieved by better understanding soil-root interaction.  To predict the behaviour of vegetated slopes during design, parameters representing the root system structure, such as root distribution, length, orientation and diameter, should be considered in slope stability models. Microscale quantifications of how root growth influences soil characteristics, able to inform computational models, are scarce in the literature, especially for stratified soils. This study quantifies the relationship between soil physical characteristics and root growth, emphasising particularly on (1) how roots influence the physical architecture of the surrounding soil structure and (2) how soil structure influences root growth. A systematic experimental study is carried out using high-resolution X-ray micro-computed tomography (µCT) to observe the root behaviour in layered soil. In total, 2 samples are scanned over 15 days of growth, enabling the acquisition of 10 sets of images. A machine learning algorithm for image segmentation is trained to act at 3 different training percentages, resulting in the processing of 30 sets of images, with the outcomes prompting a discussion on the size of the training data set. An automated in-house image processing algorithm is developed to provide values of void ratio and root volume ratio for Regions of Interest at varying distance from the root. This work investigates the effect of stratigraphy on root growth, along with the effect of image-segmentation parameters on soil constitutive properties.