Dr Leung is an Associate Professor in the Civil and Environmental Engineering and the Associate Director of the Geotechnical Centrifuge Facility at the Hong Kong University of Science and Technology (HKUST). His research expertise is unsaturated soil-vegetation interaction with emphasis on geotechnical engineering applications such as slope stabilisation. Dr Leung has published more than 80 SCI journal articles in the subject of soil-vegetation interaction and co-authors a textbook ‘Plant Soil Slope Interaction’ published by Taylor & Francis. Dr Leung is the awardee of the 2022 Geotechnical Research Medal of the ICE Publishing, 2022 Outstanding Young Geotechnical Engineer Award of the ISSMGE, 2019 Excellent Youth Scholar of the NSFC, 2019 Bright Spark Lecture Award of the ISSMGE and 2018 International Award for Innovation in Unsaturated Soil Mechanics from TC106 (Unsaturated Soils). He is currently the Editor-in-Chief of the ISSMGE Bulletin, the General Secretary of the Hong Kong Geotechnical Society (HKGES) and TC106 and TC107 (Tropical& Residual Soils).
When nature meets technology: AI-informed discovery of root–soil physical interactions
solution using vegetation has been considered as sustainable and
environmentally friendly approaches to improve slope performance. Vegetation can
affect the stability of shallow soils not only through mechanical root
reinforcement, but also through variations of soil matric suction associated
with root–water uptake process. Upon plant growth, roots explore the soil pore
space to search for nutrient and resources. Our recent work has demonstrated
that the process of root exploration in soil pore space could significantly
modify the water retention curve (WRC), which is a crucial soil hydraulic
property that underpins many key unsaturated soil properties that affects the
engineering design against slope stability. How fundamentally the soil pore
morphology might change and evolve with time upon root growth dynamics
including mortality and how this dynamic root–soil interaction at pore-level would
modify the WRC remain unclear. The lack of knowledge has made the existing
modelling of root-induced WRC mostly empirical or semi-empirical. Advanced
non-invasive imaging techniques such as mX-ray computer tomography (CT) have been
attempted to capture such interaction in-situ. However,
challenges exist to reliably segmentate the many phases involved in an
unsaturated rooted soil sample (soil, water, air, roots) from CT images because
some of these phases have rather similar X-ray attenuation coefficients.
The presentation will share our latest development of using advanced technology including artificial intelligence (AI) to aid the discovery of physical interaction between two natural materials, unsaturated soil and vegetation, and the more correct characterisation of the engineering properties of their composite. The presentation will start by introducing how a new laboratory apparatus is designed to enable unsaturated rooted soil samples to be subject to in-situ hydromechanical loading and X-ray imaging simultaneously. An AI-informed image processing technique will then be introduced to explain how the reliability of phase segmentation of the CT images of an unsaturated rooted soil sample can be significantly enhanced, making the detailed quantification of 3-D morphologies of soil pore and roots and the spatiotemporal distributions of pore air and water possible. Some new discoveries of how roots physically interact with the soil pore space including the dynamic changes in the distribution, orientation and connectivity of soil pore size and how this pore-level information can be used to explain the water retention properties will be discussed in the presentation.