GNGTS 2023 - Atti del 41° Convegno Nazionale

Session 1.2 GNGTS 2023 Volcanic and Seismic source Modelling (VSM) is an open-source Python tool to model ground deformation detected by satellite and terrestrial geodetic techniques. It allows to choose one or more geometrical sources as forward model among sphere, spheroid, ellipsoid, fault, and sill. It supports geodetic from several techniques: interferometric SAR, GNSS, levelling, Electro-optical Distance Measuring, tiltmeters and strainmeters. Two sampling algorithms are available, one is a global optimization algorithm based on the Voronoi cells and the second follows a probabilistic approach to parameters estimation based on the Bayes theorem. VSM can be executed as Python script, in Jupyter Notebook environments or by its Graphical User Interface (GUI). Its broad applications range from high level research to teaching, from single studies to near real-time hazard estimates. It is freely available on GitHub (https://github.com/EliTras/VSM) . In this contribution the functionalities of VSM will be shown. The VSM python tool VSM is designed to estimate source parameters of the most common analytical sources of deformation by using geodetic data (Fig. 1). It is structured as a typical overdetermined non-linear inverse problem (Tarantola, 1987), in which data are known while the parameters of the source(s) that generated the observed data are unknown. The scheme of VSM is reported in Fig. 2. The first block is the VSM input, consisting of the collection of geodetic observations, the selection of one or more deformation sources as forward model (with related search range for each parameter) and the selection of the sampling algorithm (with few setting parameters). The data inversion block is the VSM execution, based on the chosen algorithm. The last block is the VSM output. It consists of several products such as the sampled parameter space, the parameters distributions, the optimal parameters, and synthetic data. Fig. 2. Scheme of the VSM. The input of VSM consists of at least one dataset. The forward model is one or more analytical sources among those reported in Figure 1. One of the two available sampling algorithms can be selected. After data inversion is completed, several products are generated as output.