Landslide Kinematical Analysis through Inverse Numerical Modelling and Differential SAR Interferometry

• Published in: Pure and applied geophysics 2015-11, Vol.172 (11), p.3067-3080
• Main Authors: Castaldo, R., Tizzani, P., Lollino, P., Calò, F., Ardizzone, F., Lanari, R., Guzzetti, F., Manunta, M.
• Format: Article
• Language: English
• Subjects: Creep (materials); Earth and Environmental Science; Earth Sciences; Geophysics/Geodesy; Interferometry; Inverse; Kinematics; Landslides; Landslides & mudslides; Mathematical models; Methodology; Modelling; Numerical analysis; Remote sensing; Synthetic aperture radar; Viscous flow
• ISSN: 0033-4553; 1420-9136
• DOI: 10.1007/s00024-014-1008-3

The aim of this paper is to propose a methodology to perform inverse numerical modelling of slow landslides that combines the potentialities of both numerical approaches and well-known remote-sensing satellite techniques. In particular, through an optimization procedure based on a genetic algorithm, we minimize, with respect to a proper penalty function, the difference between the modelled displacement field and differential synthetic aperture radar interferometry (DInSAR) deformation time series. The proposed methodology allows us to automatically search for the physical parameters that characterize the landslide behaviour. To validate the presented approach, we focus our analysis on the slow Ivancich landslide (Assisi, central Italy). The kinematical evolution of the unstable slope is investigated via long-term DInSAR analysis, by exploiting about 20 years of ERS-1/2 and ENVISAT satellite acquisitions. The landslide is driven by the presence of a shear band, whose behaviour is simulated through a two-dimensional time-dependent finite element model, in two different physical scenarios, i.e. Newtonian viscous flow and a deviatoric creep model. Comparison between the model results and DInSAR measurements reveals that the deviatoric creep model is more suitable to describe the kinematical evolution of the landslide. This finding is also confirmed by comparing the model results with the available independent inclinometer measurements. Our analysis emphasizes that integration of different data, within inverse numerical models, allows deep investigation of the kinematical behaviour of slow active landslides and discrimination of the driving forces that govern their deformation processes.