From subsidence to uplift at Campi Flegrei, with an eye to Vesuvio

Geodetic data measure deformation of various origins (e.g., tectonic, volcanic, hydrological, human-induced). To attempt separating the different signals underlying the observations, data can be analysed as in a blind source separation (BSS) problem. A common technique to tackle BSS problems is the Independent Component Analysis (ICA), which decomposes the dataset into a set of independent components (ICs) under the assumption of a linear mix of the non-moving sources.

We use a variant of ICA (vbICA, variational Bayesian ICA; Choudrey and Roberts, 2003) to analyze ground displacement time series from ERS-ENVISAT SAR images in a large area including Campi Flegrei and Vesuvio volcanoes (Italy) between 1993 – 2010. The time series, which have gaps, were obtained through the SBAS technique and provided to us by IREA-CNR.

Our analyses evidence two significant ICs. Their spatial and temporal features indicate that one of the ICs (IC1 in what follows) is related to Campi Flegrei only and describes the subsidence occurred before 2000, the mini-uplifts started in 2000 and 2005, and a post-2005 uplift. The other significant IC (IC2 in what follows) is related to the subsidence occurred at Vesuvio after 2001 (Amoruso & Crescentini, 2023) and an additional (with respect to IC1) post-2001 Campi Flegrei dynamics.

The IC1 spatial pattern confirms the results in Amoruso et al. (2014): it is satisfied by the joint effects of two pressurized sources embedded in an elastic layered half-space, i. e. a sill at a depth of about 3.5 km and a small spheroid at a depth of about 2 km; the two sources satisfy large-scale and local (Solfatara fumarolic field) deformation, respectively.

The IC2 spatial pattern is consistent with Amoruso & Crescentini (2023) as regards Vesuvio. As for the Campi Flegrei area, it is quite complex and difficult to ascribe to a single source, even as general as a moment tensor. If the whole IC2 spatial pattern is inverted for small (with respect to depth) pressurized spheroids (point sources), two deep sources (depths larger than about 8 km) acting in the Campi Flegrei area are needed in addition to the deflating source beneath Vesuvio. The three sources appear inactive before 2000; afterwards the two deep sources beneath Campi Flegrei show opposite behaviour.

We show that the scenario outlined by our results is consistent with geophysical (e.g., gravity) and petrological data. It is also consistent with subsequent pressurization of a sill-like source at about 8 km depth, inflating at least since 2015 (Amoruso & Crescentini, 2022).

References

Amoruso et al. (2014), J. Geophys. Res: SE, 119, 858–879.

Amoruso, A., and Crescentini, L. (2022). Remote Sens., 14, 5698.

Amoruso, A., and Crescentini, L. (2023). Remote Sens., 15, 3038.

Choudrey, R., and Roberts, S. (2003). Neural Computation, 15(1), 213-252.