GRSG Conference 2022: Orbit to Outcrop
Title: Space-time landslide hazard modelling via Ensemble Neural Networks
Author: Ashok Dahal
Abstract:
For decades, a full numerical description of the spatio-temporal dynamics of a landslide could be achieved only via physics-based models. The part of the geomorphology community focusing on data-driven model has instead focused on predicting where landslides may occur via susceptibility models. Moreover, they have estimated when landslides may occur via models that belong to the early-warning-system or to the rainfall-threshold themes. In this context, few published research have explored a joint spatio-temporal model structure. Furthermore, the third element completing the hazard definition, i.e., the landslide size, has hardly ever been modeled over space and time. However, the technological advancements of data-driven models have reached a level of maturity that allows to model all three components (Where, When and Size) mentioned above.
This work, takes this direction and proposes for the first time a solution to the assessment of landslide hazard in a given area by jointly modeling landslide occurrences and their associated areal density per mapping unit, in space and time. To achieve this ambitious task, we have used a spatio-temporal landslide database generated for the Nepalese region affected by the Gorkha earthquake on the 25th of April 2015. The model relies on a deep-learning architecture trained using an Ensemble Neural Network, where the landslide occurrences and densities are aggregated over a squared mapping unit of 1×1 km and classified/regressed against a nested 30~m lattice.
At the nested level, we have expressed predisposing and triggering factors. As for the temporal units, we have used an approximately 6-month resolution depending on the mapped inventory dates. The results are promising as our model performs satisfactorily both in the classification (susceptibility) and regression (density prediction) tasks. We believe that the model we propose brings a level of novelty that has the potential to create a rift with respect to the common susceptibility literature, finally proposing an integrated framework for hazard modeling in a data-driven context.
depending on the mapped inventory dates. The results are promising as our model performs satisfactorily both in the classification (susceptibility) and regression (density prediction) tasks. We believe that the model we propose brings a level of novelty that has the potential to create a rift with respect to the common susceptibility literature, finally proposing an integrated framework for hazard modelling in a data-driven context.
Strong earthquakes not only induce co-seismic mass wasting but also exacerbates the shear strength of hillslope materials and cause higher landslide susceptibility in the subsequent years following the earthquake. Previous studies have investigated post-seismic landslide activity mainly by using landslide inventories. However, landslide inventories do not provide information on deformation given by ground shaking and limit our observations with only failed hillslopes. As a consequence, we lack comprehensive, quantitative analysis revealing how hillslopes behave in post- seismic periods.
Satellite-based synthetic aperture radar interferometry (InSAR) could fill this gap and provide millimeter-scale measurements of ground surface displacements that can be used to monitor hillslope deformation. Here we use the Persistent Scatterer Interferometry technique to monitor pre- and post- seismic hillslope deformation for the area affected by the Mw 6.9 Nyingchi, China earthquake that occurred on the 18th of November 2017 in addition to several other moderate to strong ones (Mw>5) occurred after the Nyingchi earthquake.
We use Copernicus Sentinel-1 satellite data acquired between 2016 and 2022 to examine hillslope deformations and generate deformation time series for five time windows representing three intra-seismic windows as well as pre- and post- seismic periods associated with those earthquakes that hit the same area under consideration. Based on these five different deformation time series, we identify actively deforming and stable hillslopes over the study area and monitor their evolution through the 6-year time window.
Results show that hillslopes gain elevated deformation rates following the Mw 6.9 Nyingchi earthquake compared to its pre-seismic counterpart and then 1.7 years after the earthquake, surface deformations decrease again. Although several other earthquakes of magnitude between 5 and 5.1 occurred in 2019, 2020 and 2021, none of them influences hillslope deformations as significant as the main earthquake does. Overall, our results show that after a series of earthquakes occurred in the last five years in the study area, the deformation rates have not been returned to the pre-seismic level yet.