Tasks

TASK A1

Prediction of Weather Hazards under extreme Climate (Co-PI: A. Lau)

Objectives

To develop an innovative real-time prediction system to make site-specific forecasts of extreme weather events conducive to landslides, combining state-of-the-art numerical weather prediction (NWP) with big data analytics. Present and future climate scenarios will be considered.

Deliverables

A novel location- and intensity-specific rainfall forecast system to now-cast reliably for six hours
A new model of the fine-scale distribution of rainfall over complex terrain
Provide guidance for long-term planning for extreme weather events in HK and surrounding region

TASK A2

Innovative Field Monitoring (Co-PI: B. Huang)

Objectives

Develop a unique multi-source stereoscopic monitoring system applying deep learning analysis to data from remote sensing satellites, unmanned aerial vehicles (UAVs), and terrestrial sensors
Provide reliable field observation data such as high-resolution and high-frequency multi-level groundwater monitoring data to calibrate and verify findings from physical and laboratory tests
Use UAVs to characterize the types of hillside vegetation in Hong Kong and provide reliable data/images to determine if a particular type of vegetation is more vulnerable to natural terrain landslides

Deliverables

A unique multi-source stereoscopic monitoring system applying deep learning analysis to data from remote sensing satellites, UAVs and terrestrial sensors
Quantitative evaluation model of slope failure mechanisms using more reliable, accurate and relevant monitored data
Characterization of the types of hillside vegetation that are vulnerable to landslides

TASK B1

Artificial Intelligence and Machine Learning (Co-PI: L. Chen)

Objectives

Deep learning-powered heterogeneous data forecasting for reliable prediction of landslides
An in-depth understanding of the factors triggering landslides
Comprehensive and user-friendly support for real-time decision-making
Warning signal detection for landslides
Real-time visualization of heterogeneous data

Deliverables

To develop new techniques for high-quality forecasting for spatial-temporal sequences
Probabilistic reasoning for in-depth analysis of landslide-inducing factors
A user-friendly visualization platform to ease the interpretation of results from deep learning models and support risk-informed decisions

TASK B2

Dynamic Data-driven Landslide Stress-testing and Risk Assessment (Co-PI: L. Zhang)

Objectives

To develop a dynamic data-driven stress-testing method for managing landslide risks and empowering community resilience, and a data-driven dynamic landslide risk analysis using AI techniques
To propose grid-based risk criteria for multi-tiered landslide warning system

Deliverables

The first dynamic data-driven stress-testing method for managing landslide risks and enhancing community resilience
A data-driven dynamic landslide risk analysis method using AI techniques
Grid-based probabilistic risk criteria for multi-level landslide warning systems

TASK C1

Modelling and Understanding Slope Failure Mechanisms (PC: C.W.W. Ng)

Objectives

To understand landslide initiation and slope failure mechanisms under extreme weather conditions
To understand the interaction between entrained channel beds, debris flow and barrier system

The findings will provide the scientific basis for green risk mitigation measures and the early warning systems to be developed in Task C2

Deliverables

Unique large-scale physical test results to establish design guidelines for landslides and flow-bed-barrier interaction and a unique environmental chamber in centrifuge
State-of-the-art constitutive model for cyclic THM behaviour of unsaturated vegetated soils with desiccation cracks
New computational platform for large-deformation analysis of landslides triggered by extreme weather conditions

TASK C2

Innovative Green Mitigation and Early Warning System (Co-PI: W.K. Pun)

Objectives

To develop holistic landslide mitigation strategies by integrating expertise from geotechnical engineering, psychology and bioengineering to create new knowledge on landslide dynamics, debris-structure interaction, societal perception on landslide hazard and bioengineering techniques

Deliverables

A multi-tiered landslide early warning system with risk levels and distinct countermeasures to minimize casualties and economic loss
A smart flexible barrier system capable of detecting an impact event reliably
An innovative bio-engineering techniques including a landslide bio-shield system using native trees and novel methods for restoration of landslide trails using UAV

TASK C3

Immersive Education (Co-PI: T.C. Pong)

Objectives

To provide an immersive education environment in slope safety via MOOC platform using emerging AR/VR technologies and narrative visualization-based technologies
A blended learning model will be explored and developed and its impact on learning evaluated

The ultimate goal of this evidence-based learning model is to empower communities and improve their disaster preparedness. The immersive education will target the public, educators, authorities, decision-makers, engineers and technologists.

Deliverables

New AR/VR and visualization tools for engaging learners in a stimulating environment
Two new MOOCs on slope safety awareness and technologies
Educational campaigns for engaging and arousing the interest of the general public
A blended learning model using traditional and innovative modalities for enhancing learning motivation and impact for learners of different levels