After Amatrice: Unlocking seismic hazard. A race against time.

Italy’s earthquake in August raised questions about whether the seismic risk in the region has increased. Amatrice is an area of historical seismic activity where seismic hazard and risk studies carried out a few years ago anticipated that a future earthquake of this size was possible, given a relative period of inactivity. We discuss below whether other recent events had increased the probability of an event in the area, and what was the impact of the Amatrice earthquake on the seismic hazard of the region.

Good understanding of the level of hazard, as well as of portfolio characteristics, such as the exposure distribution and additional information regarding building vulnerability and/or retrofitting, is paramount to adequately capture the seismic risk of a portfolio.

Engaging risk and vulnerability

Italy is a country highly vulnerable to natural disasters, with significant seismic and volcanic hazard. According to the OECD, the cost of natural disasters in Italy amounted on average to 0.2% of GDP per annum over the past few years. Italy has very low residential insurance take-up (1%), so preparedness and mitigation plans to increase resilience and reduce economic impact are crucial.

Epicentre location compared to the Italian seismic hazard map. (Source: INGV)

Figure 1. Epicentre location compared to the Italian seismic hazard map. (Source: INGV)

The different building performance exhibited in adjacent regions Norcia and Amatrice was instructive. Despite being subjected to similar earthquake intensity, Norcia was more resilient with only minor damage recorded, compared to the extensive damage recorded in Amatrice. The significantly different response is the direct result of a seismic retrofitting program carried out in Norcia after the M5.8 earthquake that occurred in the city in 1979, whereas an earthquake had not occurred in Amatrice since 1883. Clearly Norcia had a different perception of risk than Amatrice, and mitigation actions pre-event had significantly reduced vulnerability to ground motion, even though seismic hazard levels at the two cities had been fairly similar.

Insured losses for the August earthquake are estimated to be around €60-65million, largely due to the very low insurance take-up in a residential area like Amatrice. Should an event such as the 1976 Friuli or the 1980 Irpinia take place again today, insured losses could be in excess of €2bn, with economic losses rising to multiples of that.

How has seismic hazard changed?

The epicentre of the August event was located along one of the Apennine faults and in a zone characterised by a high seismic hazard, as highlighted by the dark colors in the national seismic zonation map (figure 1), i.e. a region where significant earthquakes are likely to happen.

According to Bird et al. (2015), based on historical earthquake catalogues and GPS data, the event is estimated to have a 1% chance of occurrence per year, i.e. a recurrence interval of 100 years. The ground motions recorded were largely within the seismic code design ground motions, suggesting that the Amatrice earthquake, although with a low probability of occurrence, was to be expected.

2016 earthquake location compared to 1997 and 2009 earthquakes (source:

Figure 2. 2016 earthquake location compared to 1997 and 2009 earthquakes (Source:

Scientific studies after Umbria (1997) and L’Aquila (2009) events, along the Apennines, suggested that post-event stress transfer could be triggering fault reactivation in the area, increasing the likelihood of future earthquakes in the region (e.g. Serpelloni et al., 2012). Some experts even support the view that the 2016 earthquake has “filled the gap” between the two previous events, as highlighted in figure 2.

Significant research has focused on the interaction between faults and the transfer of seismic stress following an earthquake that could potentially impact on the occurrence of another event. This static stress transfer acts over short distances (generally less than 200 km) and time periods from minutes to decades after an earthquake.

Seismic stress transferred by L’Aquila earthquake (Source: Serpelloni et. al. (2012), as edited by

Figure 3. Seismic stress transferred by L’Aquila earthquake (Source: Serpelloni et. al. (2012), as edited by

Following the 2009 L’Aquila earthquake, Serpelloni et al. (2012) assessed the variation in the seismic stress in adjacent faults. They distinguished the relieved areas (illustrated in blue at right), from the ones which were subjected to an increase in stress, hence an increase in seismic hazard and consequently the probability of occurrence of future earthquakes. According to their calculation, the L’Aquila earthquake caused a marginal stress increase in the fault that recently ruptured, as denoted with the light red colours and the star in figure 3. It is therefore likely that the L’Aquila event facilitated the activation of the fault, which led to the earthquake on 24 August 2016.

Even though a lot of research has been done in past decades in the field of probabilistic and deterministic seismic hazard assessment, estimating the time frame for the next large earthquake is challenging with an acceptable degree of confidence. However, recent studies on the dynamics of faults and the transmission of stress have helped to better understand the physics of earthquakes. Studies such as those mentioned above indicate that an event of such a magnitude in the Amatrice area was probably in line with the expected hazard and stress state following the L’Aquila earthquake.

Going forward, seismologists from Istituto Nazionale di Geofisica e Vulcanologia (INGV) have analysed data recorded after the Amatrice event to assess the slip distribution along the fault plane (green/red rectangles in figure 4), a process that allows them to better understand the segments of the fault that ruptured.

More importantly, INGV recently calculated the seismic stress induced by the ruptured fault to adjacent faults (figure 4). They found that the Norcia fault is characterised by a reduction of the stress level (denoted by the blue colours in figure 4), whereas the edges of Vettore-Bove and Gorzano faults have been subjected to an increase of their stress level (highlighted in yellow/red in figure 4). The study therefore suggests that the seismicity in the area is not expected to be reduced in the future, and that most likely has shifted away from the resilient Norcia to other areas that may be less resilient, as mentioned above.

Distribution of the seismic stress transferred by the rupture of the fault(s) along adjacent faults. (Source INGV and Gruppo Emergeo)

Figure 4. Distribution of the seismic stress transferred by the rupture of the fault(s) along adjacent faults. (Source INGV and Gruppo Emergeo)

In alignment with the above, Stein and Sevilgen (2016) also highlighted the potential for another event (not necessarily imminent) in the area. According to their study, shown in figure 2, the gap in the ruptured faults between the 1997 and 2009 earthquakes is likely not completely filled by the recent earthquake sequence, and some adjacent faults may still be activated in the future. The segments of the fault that did not rupture are approximately 20km long each and would be capable of causing a M6.0 earthquake. These parts of the fault have a history of moderate or large earthquakes, so they certainly are seismic.

It is important to highlight that the last large magnitude earthquake in the fault segment northwest of the 2016 rupture (toward Vettore-Bove), occurred in 1859. Therefore another event around M6.0 in that part of the fault could take place in the coming years. It is worth mentioning that, as far as we know, there have not been any other retrofitting programs in the area and therefore vulnerability could be high.

Given the expected increase in seismic hazard and the likely change in risk perception following the Amatrice earthquake, a window of opportunity may exist now for risk mitigation actions and measures to address the lack of insurance protection. (See Tim Edwards’ blog for further discussion.)



This post was co-written by Myrto Papaspiliou and Crescenzo Petrone.

Myrto Papaspiliou joined Willis Re International in 2012 and is the earthquake expert of the Model Research & Evaluation team. She has extensive experience in evaluating and validating earthquake catastrophe models throughout the world and has been heavily involved in developing the Willis View of Risk. Myrto holds a PhD in Earthquake Geotechnical Engineering from Imperial College London and prior to her time in Willis she worked as an earthquake engineer in a leading engineering consulting company in London.

Crescenzo Petrone is a Research Associate at University College London and is currently seconded at Willis Re in the framework of a knowledge exchange project supported by the Willis Research Network. He has a broad experience in assessing the impact of different natural hazards on the built environment. Crescenzo holds a PhD in Earthquake Engineering from University of Naples Federico II, Italy, and has a strong track record of collaborations with several international academic institutions (University of Ljubljana, University of Bristol and University at Buffalo)



About Rosa Sobradelo

Rosa Sobradelo is part of the Willis Research Network where she coordinates research activities in the areas of geo…
Categories: Natural Catastrophe, Reinsurance | Tags: ,

Leave a Reply

Your email address will not be published. Required fields are marked *