GNGTS 2016 - Atti del 35° Convegno Nazionale

130 GNGTS 2016 S essione 1.1 paleoseismological record (e.g. Delaby, 2001; Becker et al ., 2005; Kagan et al. , 2005). However, the difficulty in quantitative modelling of the observed deformation, and its direct attribution to a geometrically-constrained seismogenic source is a major issue (Lacave et al. , 2004; Becker et al. , 2006). Lacave et al. (2000) investigated the range of fundamental natural frequencies and the damping of speleothems, and established that most of the broken speleothems are a direct indicator of the peak ground acceleration (PGA) during past earthquakes. Based on the analysis of the mechanical behavior of speleothems through static tests, Lacave et al. (2004) stipulated the PDF (probability density functions) for the bending stress leading to rupture, and established different vulnerability curves (probability of breaking as a function of PGA) for speleothems according to their shapes. On the other hand, unbroken speleothems may be used to define an upper limit of the “strength” for earthquakes that could have ever occurred during the speleothems lifetime. Based on these results, we started developing speleothem-based vulnerability curves for selected areas within the axial seismogenic belt of the Apennines. The study area in southern Italy is in northern Calabria (Pollino Range), and represents a gap in the belt of active faults and current seismicity. Previous works on broken speleothems from caves distributed around suspect active faults established past episodes of speleothem deformation, tentatively related to earthquakes (Ferranti et al. , 1997; Ferranti and Maschio, 2007), but no investigation on vulnerability was carried. The study area in central Apennines is located ~10 km west of the Fucino Plain, one of the largest tectonic basin of the Apennines, affected by the large 1915 earthquake (M=7). In detail, the cave is located close to the Liri fault, a ~42 km long and southwest dipping normal fault, considered by some authors (e.g. Roberts and Michetti, 2004) active and seismogenic. Research strategy. The starting step of the research was to select within the target areas the caves with required features (concretions within near-surface rooms, wide range of speleothem shapes, easy access, and so on). We visited four caves scattered on a ~20 km stretch on the southern flank of the Pollino Range, and one cave in Abruzzi. In both settings, caves are in the immediate footwall of suspect active normal faults (Pollino- Castrovillari, and Liri faults, respectively). The shape of speleothems was recorded by measuring length and diameter of intact and deformed speleothems. Vulnerability analysis of the speleothem population based on the length and diameter ratio was initially carried according to published curves (Lacave et al. , 2012). In a following step, we performed static tests on a representative speleothem population from the different settings in order to correctly define the speleothem vulnerability curves (in terms of probability to be broken), and thus past PGA thresholds, for each investigated area. In addition, we performed theoretical and numerical modeling in order to estimate the values of the horizontal ground acceleration required to failure the speleothems. In particular we used a finite element method (FEM), with the SAP200 software, starting from the detailed geometry of the speleothems and their mechanical properties. Laboratory analysis included U/Th and radiometric dating of the speleothem deformation events in order to define, for each area, the paleoseismological frame. For unbroken speleothems, dating of the most recent speleothem layer was used to define the time interval of the limited PGA threshold experienced by speleothems. The speleothem-based estimate was then compared with the acceleration predicted by the national seismic hazard model (MPS04, 2004). Discussion. In the Pollino area, we found a good chronological correlation between deformation events recorded by single speleothems of comparable size from an individual cave. At a larger scale, type and age correlation of events among different caves located along a ~20 km stretch of the southern flank of the range led to identification of a well-constrained, regional collapse event at ~5-7 ka, and a previous similar event at ~23 ka. Because these events are recorded by speleothem with moderate vulnerability, we regard them as related to strong, but infrequent earthquakes. Conversely, we did not found evidence of significant seismic shacking between ~40-23 ka, ~23-7 ka, and after~5-7 ka (for this latter time span there are no evidence even for more vulnerable speleothems).