Multiscale edge-detection methods for the geometrical constraint of deformation sources in the volcanic environment

The development of the satellite remote sensing technologies is providing a great contribution to monitor volcanic phenomena. Specifically, the large amount of the ground deformation field data (i.e., DInSAR measurements) holds information about the changes of physical and geometrical parameters of deep and shallow volcanic reservoirs; therefore, the exploitation of these data becomes an important task since they actively contribute to the hazard evaluation.

Currently, DinSAR measurements are mostly used for modeling the volcanic deformation sources through the optimization and the inversion procedures; although the latter provide a physical and geometrical model for the considered volcanic site, their results strongly depend on the availability of a priori information and on the considered assumptions about the physical settings; therefore, they do not provide a single solution and they unlikely guarantees a correct analysis  for the multi-source cases.

In this scenario, we consider a new methodology based on the use of edge-detection methods for exploiting DInSAR measurements and characterizing the active volcanic sources. Specifically, it allows the estimation of the source geometrical parameters, such as its depth, horizontal position, morphological features and horizontal sizes, by using Multiridge, ScalFun and Total Horizontal Derivative (THD) methods. In particular, it has been proved the validity of Multiridge and ScalFun methods for modeling the point-spherical source independently from its physical features, such as the pressure variation, the physical-elastic parameters of the medium, such as the shear modulus, and low signal-to-noise ratio.

Now, we extend the proposed Multiridge and ScalFun methods from the hydrostatic-pressure point source to the tensile one, and then to the others (rectangular tensile-fault and the prolate spheroid analytical models) in order to investigate volcanic sources as sills, dikes and pipes.

Specifically, after the analysis of the physical and mathematical features of the considered models, we apply Multiridge and ScalFun methods to the synthetic vertical and E-W components of the ground deformation field. We carefully evaluate the advantages and the limitations which could characterize these cases, showing how to solve critical aspects. We especially focus on the sill-like source, for which the edge-detection methods provide very satisfying results. In addition, we perform a joint exploitation of the edge-detection methods to model the deformation source of Fernandina volcano (Galapagos archipelago) by analyzing COSMO-SkyMed acquisitions related to the 2012-2013 time interval.

In conclusion, this approach allows retrieving univocal information about the geometrical configuration of the analyzed deformation pattern. We remark that, although a subsequent analysis is required to fully interpret the ground deformation measurements, this methodology provides a reliable geometrical model, which can be used as a priori information to constrain the entire interpretation procedure during next analyzes.