Description

Intense precipitation can trigger large populations of landslides in mountainous regions. As a result, significant losses can be suffered during a storm. The hydrological literature is rich with scientific applications where potential areas under the threat of floods are estimated in near-real-time as the incoming storm approaches a given landscape. However, when it comes to landslides, a much more limited research has been developed to an operation stage, up to the point where decision makers can issue spatially distributed alert levels on the basis of probabilistic estimates.

Objectives and Methodology

This MSc topic aims at laying the foundations for a near-real-time forecasting system. To achieve this, a multidisciplinary approach which involves geomorphology, statistics and meteorology will be pursued. Several study areas can be selected across the globe. A potential candidate location could be in Indonesia or Fiji or more.

A reference statistical model will be built to estimate the functional relation between a rainfall-induced landslide inventory and a set of predisposing and triggering factors. From such model, a simulation step will follow, where subsequent rainfall patterns will be plugged-in to measure the variation in the landslide susceptibility patterns over time. These will be tested against subsequent landslide inventories to check the actual predictive power of such procedure. Whether the simulations will prove to be a valid tool to predict landslide occurrences over space, this will support the implementation for actual near-real-time applications.

Further reading