Characterising multi-stage landslide activity rates with synthetic aperture radar satellite data

Landslides are a significant hazard in mountainous environments.

The advent of earth observation from space has hugely increased the scope of landslide studies and improved our understanding in terms of hazard mitigation, early warning, triggering mechanisms and mass-wasting effects.

Occurrences of new landslides can be observed in optical satellite images, while slow-moving landslides can be monitored using satellite radar interferometry (InSAR).

However, while the spatial coverage of landslide studies has been expanded by the availability of remote sensing datasets, a complete picture of landslide activity remains difficult to obtain from satellite imagery: optical satellite images are best-suited to detection of new landslides in vegetated environments, while inSAR is limited to slow-moving landslides.

Current methods therefore struggle to detect multi-stage failure or reactivation of pre-existing landslide scars for fast-moving or incoherent deformation.

Here I will develop new SAR-based techniques using amplitude and coherence time series to detect multi-stage failure and reactivations.

I will test and apply these techniques at a range of spatial scales (individual large landslides up to regional inventories).I will apply to techniques to two case study areas (Nepal and Papua New Guinea) that have experienced landslides triggered by sequences of both earthquakes and rainfall.

The case where landslides are triggered by a sequence of events is one where detection of multi-stage failure is particularly important: whether a landslide fails once or several times has implications for both hazard and erosion.

By applying the new methods here alongside traditional remote sensing techniques, we hope to obtain a more comprehensive view of landslides than is currently possible.