Modelling InSAR-derived hillslope velocities with multivariate statistics: A first attempt to generate interpretable predictions

Spatiotemporal patterns of earth surface deformation are influenced by a combination of the geologic, topo-graphic, seismic, anthropogenic, meteorological and climatic conditions specific to any landscape of interest. These have been mostly modelled through machine learning tools. However, these influences are yet to be explored and exploited to train interpretable data-driven models and then make predictions on the deformation one may expect in space or time. This work explored this aspect by proposing the first multivariate model dedicated to InSAR-derived deformation data. The results we obtain are promising for we suitably retrieved the signal of environmental predictors, from which we then estimated the mean line of sight velocities for a number of hillslopes affected by seismic shaking. The importance of such models resides in its potential for opening an entirely new research line for slope instability modelling.

Automated summary

Spatiotemporal patterns of earth surface deformation are influenced by a combination of various factors, including geology, topography, seismic activity, human activities, meteorology, and climate conditions. This study proposes a multivariate model dedicated to InSAR-derived deformation data, aiming to explore these influences and make predictions on deformation. The obtained results are promising and have the potential to open up new research opportunities for slope instability modeling.