Slope stability and climate change in high-mountain regions: Understanding and predicting geo-hazards.

HIGHLIGHTS
------------------
• Timely project: a game changing approach is urgently needed to improve understanding, model and forecast landslides in mountainous areas.
• Teamwork effort: This PhD project will benefit from a wider group collaboration, with a team of researchers with interest in Engineering Geology, Risk Management, Cryo-Hazards, Hydrology and Computer science. A wealth of experimental observations on landslide kinematics, external forcing and geology/geotechnics will be available from our project partners in Switzerland (Research Center in Alpine Environment) and Norway (Landslide Unit - Norwegian Water and Energy directorate).
• Significance: Key aspects of landslide science are not fully understood yet, including the mechanisms behind a series of cascade events as a result of global warming, permafrost degradation, progressive degradation of the slope and the transition from slow-moving landslides to catastrophic failures.


PROPOSED RESEARCH
------------------------------------
We propose an exciting and truly multi-disciplinary PhD opportunity to create a paradigm shift in methodologies for modelling and forecasting the temporal occurrence of a catastrophic landslides in High Mountain areas, where the instabilities are controlled both by a progressive strength reduction -associated with freeze/thaw cycles and permafrost degradation- and a seasonally intermittent water flow through deep fractures.

You will start investigating a recent collapse in Norway using a series of high quality datasets on slope kinematics (both in-situ and remote sensing techniques) and environmental forcing [1]. Exceptional data of an active slope failure captured in real-time by NVE, and a series of time-dependent models of brittle failure, which importantly apply to all brittle slope failures worldwide ([2], [3]) will be used to back analyse the slope during the first stage of this PhD project. Accessing to these fundamental observations, you will unpick and model the highly non-linear landslide response to both the environmental forcing and the progressive movement of the slope, using a new physically-based time-invariant model for forecasting slope kinematics (fig. 2) [4]. Back and forward analysis of tipping point behaviours will be carried out in order to model better extreme and often unexpected events ([5]).

According to background education and to your particular research interests, this research will involve one or several of the following elements:
(1) Advance data analytics + artificial Intelligence: You will be analysing big-data in the form of time-series (see above) either using data-driven approaches (e.g. support vector machine, neural networks systems, multivariate regression analysis, Bayesian theories, etc.) or physically-based models –or both- in order to improve landslide forecasting. The complex links between external forcing (rainfall, temperature), infiltration and slope kinematics will be investigated using state of the art computational techniques in order to propose a new generation of Early Warning Systems, that will be developed and implemented in close collaboration with project partners in Switzerland and Norway (CREALP, NVE).
(2) Remote Sensing: using state of-the-art Earth Observation techniques (GB-Radar, drones, time-lapse cameras, satellite imagery, etc.) and in-situ sensors you will investigate both slope kinematics and gradual slope damage of the slope by looking at freeze-thaw cycles + permafrost degradation, including the links with changing climate. A series of recent landslide events at regional scale will be investigated in high mountain areas (Swiss Alps / Norway) in close collaboration with our project partners (CREALP, NVE) that will be providing logistical support.
(3) Laboratory-based experiments: you will be using the facilities in our rock mechanics and engineering geology laboratories –and partner facilities- to better parametrize and model the gradual slope damage by studying progressive strength degradation of geologic materials through time (crack growth, freeze-thaw cycles, rock bridges, etc.) using a series of experimental methods, which has key implications for slope stability.


SCIENTIFIC BACKGROUND AND METHODS
--------------------------------------------------------------
The complexities that obstruct a straightforward forecasting of slope failures in alpine areas include complex model parameterization and a series of epistemic and parametric uncertainties. Well-established strain-rate failure models, such as the power-law acceleration rate [7] and inverse failure models that were built under the assumption of constant stress conditions may explain the lack of agreement between these simple models and field observations, questioning the ability to predict landsliding. In addition, current models for progressive slope degradation in permafrost do not yet take into account the time-varying behaviour of the slope [8]. At present, critical strains, detectable limits and the transition from secondary to tertiary creep, are maturing conceptually in this field, but we will test in this PhD whether the availability of Big Data together with the use of modern data-driven techniques will shed light in predicting slope accelerations as response to rainfall.