Volume unbalance on the 2016 Amatrice - Norcia (Central Italy) seismic sequence and insights on normal fault earthquake mechanism

Abstract

We analyse the M_w 6.5, 2016 Amatrice-Norcia (Central Italy) seismic sequence by means of InSAR, GPS, seismological and geologic data. The >1000 km² area affected by deformation is involving a volume of about 6000 km³ and the relocated seismicity is widely distributed in the hangingwall of the master fault system and the conjugate antithetic faults. Noteworthy, the coseismically subsided hangingwall volume is about 0.12 km³, whereas the uplifted adjacent volumes uplifted only 0.016 km³. Therefore, the subsided volume was about 7.5 times larger than the uplifted one. The coseismic motion requires equivalent volume at depth absorbing the hangingwall downward movement. This unbalance regularly occurs in normal fault-related earthquakes and can be inferred as a significant contribution to coseismic strain accomodated by a stress-drop driven collapse of precursory dilatancy. The vertical coseismic displacement is in fact larger than the horizontal component, consistent with the vertical orientation of the maximum lithostatic stress tensor.

Figure 1

Map of the 2016 M_w 6.5 Amatrice-Norcia seismic sequence. 1 to 6 are the cross-sections of the seismicity and coseismic vertical motion shown in Fig. 2. (A,B) Is the trace of the cross-section shown in Fig. 3.

Figure 2

Cross-sections of the seismicity occurred during the 2016 Amatrice-Norcia seismic sequence with the associated vertical displacement recorded by SAR data. The dashed red lines represent the main inferred fault planes. The zero of the vertical displacement shown below each section represents the datum plane before the earthquake. Each section shows structural differences, illustrating the irregular shape of ruptures delimiting the prismatic volume of the graben or half-graben. In some sections, the SW-dipping master normal fault is associated with an antithetic NE-dipping conjugate faults. In all sections it occurs a low-angle NE-dipping decollement in which the overlying seismic volume is lying. The maximum coseismic subsidence developed in the central part of the sequence where the largest M_w 6.5 event occurred. Earthquakes data after.

Figure 3

Geological cross-section of the area affected by the M_w 6.5 Amatrice-Norcia earthquake. Geological data after.

Figure 4

3D view of the seismicity related to the 2016 Amatrice-Norcia seismic sequence. The figure shows the spatial distribution of the more than 100,000 aftershocks and the focal mechanisms of the two M_w 6.0 and M_w 6.5 mainshocks occurred on August 24^th 2016 and October 30^th 2016. The mechanisms show N150°–160° trending normal faulting. The area affected by the sequence is elongated NW-SE, about 70–90 km long and 10–15 km wide. The distribution of the seismicity demonstrates the shape of the involved upper crustal volume rather than a simple planar fault.

Figure 5

(A) Map showing the cumulated displacements occurred from September 2015 and November 9, 2016. It is recorded by the ALOS2 DInSAR data, showing the areas collapsed and uplifted during the seismic sequence between M_w 6.0 August 24^th and M_w 6.5 October 30^th 2016 assuming that no pre-seismic deformation occurred. Coseismic uplift is marginal with respect to subsidence. The largest deformation is concentrated in the hangingwall of the master WSW-dipping normal fault system. Maximum coseismic subsidence was around 100 cm, whereas the highest uplift in the hangingwall (i.e., to the west) was about 10–12 cm. The estimated collapsed volume is 0.12 km³. The uplifted volume in the hangingwall is about 7.5 times smaller, posing the question of the unbalance of the volumes. According to error estimates (see method section), the values ranging between −3 and 3 cm are masked out. The dashed black and magenta polygons refer to the areas selected for subsided and uplifted volume calculation, respectively, reported in Supplementary Information Tables S1 and S2. (B,C) are 3D views of the vertical deformation map (A). The vertical exaggeration is 5000 times. The grey areas refer to the masked-out deformation values ranging between −3 cm and 3 cm (see main text), and colour code is equal to 2D map in (A). The subsided volume has a depocenter in the center of the asymmetric graben. The subsided volume is much larger than the uplifted one.

Figure 6

(A) Horizontal and vertical components associated with the October 30^th 2016 M_w 6.5 mainshock (data after). Notice the larger coseismic vertical displacement, in agreement with the vertical maximum stress tensor (σ₁). The fault dip correctly represents the vector sum of the horizontal and vertical components of the displacement. (B) Surface rupture associated with the two seismic events. (C) Numerical modelling of interseismic and coseismic deformation in a simplified brittle upper crust and visco-plastic lower crust. During the interseismic the fault is locked in the upper crust, whereas is shearing in steady state in the lower crust. A dilated volume forms above the brittle-ductile transition to accommodate the strain partitioning. At the coseismic stage, the fault hangingwall collapses and recover the previously formed dilation. See shear stress associated with the two stages (modified after).

Figure 7

During the 2016–2017 Amatrice-Norcia sequence, the subsided volume (A) was about 7.5 times larger than the uplifted volume (B). This volume unbalance can be explained either by an elastically stretched crust or alternatively permeated by a large number of fractures formed during the interseismic period in the brittle upper crust. This would require the existence at depth of a dilated volume able to accommodate the fall of the normal fault hangingwall. The collapse of the overlying prismatic volume may be triggered by the final loss of strength in the dilated wedge and along the fault.

Figure 8

Comparison between two models to explain the phenomenology associated with normal fault-related earthquakes. Normal fault activates either by elastic rebound due to the crustal stretching (lower left) or to the gravitational collapse of the hangingwall (lower right). See text for further explanation. Notice that the elastic rebound is at odds with a number of observed data, such as (i) the larger vertical coseismic displacement with respect to the horizontal component, (ii) in spite of the extensional setting, the crust is in compressional state of stress, while the elastic rebound model requires horizontal interseismic stretching; (iii) at the bottom of the normal fault there is both positive and negative volume unbalance where the ductile crust cannot absorb instantaneously such deformation, particularly where the largest differential stress is required at the brittle-ductile transition (BDT); (iv) the coseismic and post-seismic symmetric footwall uplift and hangingwall subsidence (A = B) do not occur. These inconsistencies could rather be satisfied by the existence of a pre-existing dilated wedge formed during the interseismic period, which will eventually loose strength, allowing the hangingwall to collapse gravitationally and explaining the larger collapsed volume with respect to the uplifted one. Near fault black arrows indicate the state of stress required by the two compared models.