NERC ONE Planet DTP
The UK is facing more intense and frequent extreme weather conditions due to climate change, leading to dramatically increased landslide hazards and in turn immense threat to the economy and life. For instance, the prolonged precipitation in the year 2012 has caused a five-fold increase in the number of landslide events. It is of pressing need to implement early warning of regional-scale landslide hazards for the proactive mitigation of these destructive events, but slope stability analyses require geological, geotechnical, meteorological, and geomorphological data which are typically sparse and limited in engineering practice. The dominant driver of the landslide hazard in the UK is rainfall, but network infrastructure operators primarily rely on static rainfall thresholds based on localised rainfall intensity and duration (or antecedent rainfall amounts), giving no strategic information on where the landslide hazard is the largest or evolves the quickest.
This project will propose a novel physically-based early warning procedure for the landslide hazard that provides timely near real-time hazard analyses over wide areas. The new procedure will provide dynamic hazard maps that enable a better understanding of the slope response (failure mode, failure time, failure location, and failure scale) under extreme weather conditions, facilitate a targeted deployment of field monitoring instruments, and ultimately economical loss reductions through proactive preventative or mitigative efforts. To address the knowledge gap mentioned above, open-access km-scale high-resolution rainfall radar data developed by the Met Office will be used to offer slope-scale detail of rainfall delivery at 5-minute intervals. Machine learning of the physically based numerical analyses of rainfall-induced landslides will establish a computationally efficient slope response model that can rapidly evaluate responses of thousands of slopes within minutes. For the first time, the wide-scale slope failure model would be optimized and calibrated through backanalyses of the historical landslides and field monitoring data collected at key sites. Laboratory tests of soil samples extracted from the site will provide refined details of soil properties and contribute to a further updating of the landslide hazards. New rainfall thresholds will be developed using future climate scenarios from the UK climate projection produced by the Met Office and new early warning protocols will be established.
The candidate will gain skills in machine learning methods, probabilistic and back analysis of slope stability, and engineering experience through collaboration with industry partners.
Key Research Gaps and Questions:
- How sensitive are different landslide failure mechanisms (both shallow and deep-seated failures) to the specific parametric assumptions required to enable efficient, widescale, dynamic slope analyses?
- How effective are optimised, dynamic, widescale slope failure models driven by actual spatio-temporal rainfall patterns at predicting the occurrence of actual slope behaviour (both failures and non-failures)?
- How do hazards of slope failure evolve spatially over an infrastructure network-scale area under projected future rainfall scenarios?
Applicants will normally hold a first or 2:1 undergraduate degree or a master’s degree in a subject relevant to the research project (including but not limited to civil engineering, geotechnical engineering, mathematics, and engineering) and desirably have knowledge and experience in slope stability analysis and laboratory tests of soil properties.