Volcanic mass flows constitute an important trigger of tsunamis, resulting in more local, although sometimes larger, impacts comparable to earthquake-induced tsunamis. Bearing in mind the destructive potential of these phenomena, the tsunamigenic landslide that occurred on 30 December 2002 on the unstable NW slope of Stromboli volcano in Italy has been re-examined here, by means of unpublished helicopter-borne (visible and thermal) images, and slope stability analysis. The main result of this study is that the sequence of landslides triggering the 2002 Stromboli tsunami can be defined as the final stage of a lateral magma intrusion that exerted a high thrust at high altitude, destabilizing the entire slope. This study, allowing a fuller understanding of the event that took place in Stromboli on 30 December 2002, also allows us to generalize the results, making a contribution to comprehending the volcanic edifice failures that can trigger tsunamis.

The slope’s stability at Stromboli is evaluated with Slope Stability Analysis Program (SSAP; Borselli et al., 2020), a free software that investigates the Factor of Safety (FS) of sliding surfaces through bi-dimensional LEM analysis by assigning the geo-mechanical parameters (according to Mohr-Coulomb, generalized Hoek and Brown or Barton-Bandis non-linear failure criterion), setting the water table position, the loading conditions and the presence of retaining structures, already used to estimate the stability of volcanic edifices in different geological contexts (Borselli et al., 2011; Dondin et al., 2017). In this study, the Factor of Safety is estimated through the Morgenstern–Price method (Morgenstern and Price, 1965), a general method of slices developed on the basis of limit equilibrium. It requires satisfying the/an equilibrium of forces and moments acting on individual blocks. The research of critical surfaces is carried out through the “Random Search” engine, that generates possible sliding surfaces through the Monte Carlo methods. SSAP generally adopts a feature called “Dynamic Surface Search Attractor” that progressively reduces the initial research area set by the user or provided automatically by the software, according to the surfaces with lower FS gradually identified. This function is extremely useful because it allows the user to find ever more critical surfaces within a specific area. At the same time, the software tends to discard potentially unstable areas if not identified in the early stage of the research. SSAP gives as output a graphical representation of the surfaces with lowest FS, internal distribution of forces and pressure, distribution of local FS, local distribution of the numerical reliability of general FS numerical solution (namely RHO index) and a 2D color map with distribution of average local FS obtained by local stress distribution, defined as quasi-Finite Element Analysis (qFEM; Schofield and Wroth, 1968; Griffiths and Lane,1999) and over stress ratio (OSR) maps (Farias and Naylor, 1998). These maps, based on a combination of LEM and FEM procedures, can highlight the presence of local areas within the slope that may display a critical state of stress. Although the actual state of deformation is not calculated by the software (SSAP does not allow the user to include the strain’s parameters in the slope characterization), this feature provides a graphical representation of those areas in which progressive failure can originate. Based on Schofield and Wroth (1968) and Griffiths and Lane (1999), the qFEM map describes the local distribution of the FS, including a representation of the potential direction of plasticization for those areas where FS<1. A local OSR map, based on Farias and Naylor (1998), exhibits in terms of local mean stress (principal stress and path stress) the areas where the maximum local shear stress is greater than the local shear strength. The areas with OSR >1.0 are the most likely to develop progressive failure.

The slope’s topography is obtained by digitalizing the contour lines included in Tommasi et al. (2008) and Baldi et al. (2008). Then a slope profile line that passes through the area mostly affected by failures and deformation during the 30 December 2002 landslide sequence was selected for the LEM analysis (maximum number of 100 nodes imposed by the SSAP). In each simulation performed in static and seismic conditions, the water table is assumed to be at the sea level, and above it the rock properties refer to dry condition. This assumption is justified by the absence of springs a few meters above the sea level, the general dry state of investigated outcrops, and the low average annual precipitation on the island (Revil et al., 2011).

The SdF behavior in response to both static and loading conditions has been reproduced considering a mechanical stratigraphy as follows:

- a shallow layer in the subaerial part of the slope made up of rockfill material (Tommasi et al., 2004). The failure’s envelope of the rockfill material is described by the Barton-Kjaernsli criterion (1981) implemented in SSAP as described by Barton & Bandis (1991), Barton (2013) and Prassetyo et al. (2017);

- Lava-Breccia unit (Apuani et al., 2005a), considered as a lower limit for the slope’s geotechnical characterization. Non-linear failure envelopes as the Generalized Hoek and Brown criterion GHB (Hoek et al. 2002), implemented in SSAP following Carranza-Torres (2004) and Lei et al. (2016), accurately/efficiently describe the behavior of the volcanic edifice both at shallow and higher depths;

- a stronger substratum mainly composed of the Lava Unit (Apuani et al., 2005a). The Mohr- Coulomb failure criterion is considered for this zone.

File Content

All files can be read using the SSAP software, which can be downloaded for free from the website https://www.ssap.eu/downloads.html

The software developer is Prof. Lorenzo Borselli, co-author of the article, who can be contacted at lborselli@gmail.com.

Below is the list of files and their meaning.

falda.fld -> water table file

geomeccanici.geo -> gemechanical parameters

Impostazioni.par -> setting file

modello_6_mod.mod -> input file for SSAP (making the software read this file, the software automatically reads all the others as well)

profilo_2016_SdF_D_Factor.dat

strato_sup_modificato.jrc

Magma Thrust\10kn.dxf -> failure surface

Magma Thrust\s_new.dxf -> failure surface

Magma Thrust\s3_20kn.dxf -> failure surface

Magma Thrust\s3_100kn.dxf -> failure surface

Magma Thrust\singola_3.sin -> input file for SSAP for failure surface stability analysis (making the software read this file, the software automatically reads all the others as well)

For questions

Federico Di Traglia (fditraglia@inogs.it; federico.ditraglia@unifi.it)

Lorenzo Borselli (lborselli@gmail.com)