Towards the new seismic hazard model of Italy (interview with Carlo Meletti)


In 2004 a small research group, coordinated by INGV, released the Map of Seismic hazard of the Italian territory (MPS04), compiled as required by the Ordinance n. 3274 of the President of the Council of Ministers (2003). The map was to serve as a reference for the Regions, whose task is to update the seismic classification of the respective territories. The map was then made “official” by the Ordinance n. 3519 of the President of the Council of Ministers (28 April 2006) and subsequently published on  the Official Gazette (No. 108 of 11 May 2006).
In the following, other elaborations were added to the map using the same conceptual structure. It  represents the first modern seismic hazard model for Italy. For the first time estimates for different return periods and for various spectral accelerations were released. This model has been then used as the basis for the building code contained in the 2008 Technical Regulations (NTC08), which became operational in 2008 and was also adopted by the 2018 Technical Regulations.
Features and events related to the success of MPS04 are described, among other things, in two posts of this blog:

https://terremotiegrandirischi.com/2016/09/26/che-cose-la-mappa-di-pericolosita-sismica-prima-parte-di-massimiliano-stucchi/
https://terremotiegrandirischi.com/2016/10/05/la-mappa-di-pericolosita-sismica-parte-seconda-usi-abusi-fraintendimenti-di-massimiliano-stucchi/

As usual in many seismic countries, since a few years a research group is compiling a new hazard model, which uses updated data and techniques.
Massimiliano Stucchi discusses about it with Carlo Meletti who, after its important contribution to MPS04, coordinates the new initiative through the INGV Seismic Hazard Center.

MPS04,  even if compiled  “in a hurry” in order to meet the State requirements, had a considerable success, both in the technical-administrative field and – after a few years – at the public level. What drives a new model to be built?

There is the awareness that after more than 10 years we are able to better describe the seismic hazard in Italy. A hazard model is the synthesis of knowledge and data  available at the time of its compilation. In the meantime, we have accumulated a lot of new or updated data (starting from an important revision of not only the historical catalog of earthquakes, but also of the fault and seismogenic sources database, as well as all the accelerometric records of the strong Italian earthquakes of the last 10 years).
Therefore, we decided to check how much the hazard assessment does change. It is a regular practice in the most developed countries (every 6 years in the United States, every 5 in Canada, every 10 in New Zealand). We were driven by a scientific need, but also the Civil Protection Department supported this initiative to verify the potential impact on the building code (seismic classification of municipalities and building regulations).

Can you briefly summarize the phases of this new initiative and anticipate, if possible, the release date of the new model?

Our project started in mid-2015 and we wanted to finish in 2 years. However, our intent was very ambitious: for instance, we involved a very large scientific community (about 150 researchers who, for various reasons, participated). Also the number of activities that we have foreseen was very wide, in how much we wanted to be sure not to neglect any fundamental aspect for the estimate of the hazard. Obviously the intention was to confirm that Italy stands at the same level as the most advanced countries in the world. Because of this reason, deadlines have been postponed respect to the initial project planning. The Department of Civil Protection itself gave priority to the quality of the model rather than to quantity of time necessary to obtain it, given that in the meanwhile the MPS04 model is still considered reliable.
In part,  the process was lengthened because of the continuous interaction with the Great Risks Commission, which assessed the first results during the course of work, suggested changes, and asked for further checks. To this day, we think we are really close to the project conclusion and we can finish by the end of 2018.

Can you anticipate, even without going into detail, if there are any important changes in the distribution of seismic hazard?

In general there is a tendency to have higher acceleration values today than the estimates of the past. Stations located near the faults have recorded accelerations never observed before, for all large earthquakes (also in Italy), due to the increase of number and density of seismic stations worldwide. As a consequence, the ground motion attenuation models have been updated: this now produces higher accelerations compared to the models used in past. This is one of the most important innovations produced by the improvement of available data, in this case the accelerometric recordings.
As for the geographical distribution, compared to the 2004 model, the high seismic hazard areas (Friuli, the entire Apennine chain up to Calabria) will be more restricted and concentrated on the seismogenic structures responsible for the strongest earthquakes. On the other hand, some areas, particularly in Southern Italy, may be a little less dangerous than previously known, thanks to the revision of many data on the seismicity of the past. It is an aspect that we are checking carefully in these weeks.

The new model follows the probabilistic approach like the previous one and like the vast majority of models worldwide. Can you summarize the reasons for this choice?

There is no “unique” seismic hazard but, depending on the use that you want to make, it may be appropriate to use different indicators. The probabilistic approach offers the estimation of expected shaking values from very high to very low probabilities (in other words, it is like saying from frequent to rare). The choice of the parameter to be used for buildings does not concern the hazard danger but the level of safety to be assured to citizens and buildings. The choice is up to the legislator since the size of the economic investment varies. In contrast, the deterministic approach, which defines shaking scenarios for maximum events, provides a single extreme parameter related to a certain event that may not even occur. therefore, the design of buildings could be too burdensome, as engineers have also well explained in this blog.

After each of the most recent earthquakes, rumors claiming that the “map” MPS04 “would underestimate” the seismic hazard for a variety of reasons (approach, models and data used, etc.) raised. Some of these rumors were clearly based on flawed arguments, for example, the type of soil was not taken into account; others just considered the classic “map”, while the MPS04 model offers a variety of seismic hazard assessments that lend themselves to more pondered analyzes. Is that true?

Criticisms might be the proof of the “success” of MPS04. It has become a popular document, very common on the web, and it is shown every time there is talk about earthquakes in Italy. Part of the criticism derives also from our lack of ability to communicate what the values exactly represent and therefore the colors on the map (horizontal acceleration on rigid and horizontal ground, with a certain probability of being exceeded in 10% of cases in 50 years is not the maximum value possible!). On the other side, there are commentators and colleagues who attack the MPS04 model just because they did not take part in its compilation. Unfortunately, most of the times critics are based on wrong or bad faith arguments, or arguments that show how people are not able to grasp the concept of probability. Moreover, when MPS04 came out, it was blasted because its values were considered as too high, whereas after the strong earthquakes of the last 10 years, it is criticized of underestimating data?. They should make up their mind!
It is also worth remembering that a probabilistic model is not evaluated after a single event, but after a significant period of time in which all the events occurred with the model estimates are compared. Finally, I also remember that the current model relies on over 2000 different parameters for each site, precisely to better describe the danger, but people look at only one data (10% in 50 years).

In short, it seems to me that sometimes someone tends to attribute the cause of some collapses to (wrong) seismic hazard assessments, when there should be a discussion on how to build and, perhaps adopt –  the classical reference parameters from an engineering point of view (10% probability of overcoming in 50 years, often assumed as a dogma and never sufficiently explained). Do you agree?

Actually, I am still waiting for someone to show me, a collapsed building that had been designed (correctly) with the NTC08 in Emilia or in central Italy. Then, as a completion to the previous answer, I would like to mention that the PGA parameter used to represent seismic hazard is not used directly to design buildings. It is not the eventual “low” estimated PGA that causes buildings collapse. The PGA is the “Peak Ground Acceleration” that is recorded during an earthquake and it is only one of the many parameters that can be used. For example, a much more significant parameter is the acceleration related to a period of oscillation close to the one of the building.
With reference to the parameters, it is common knowledge that one should always use only 10% of probability of exceeding in 50 years for ordinary buildings. The choice of the value of probability of overcoming is not up to the seismologists, but to the authors of the legislation. In the United States, for example, only the danger with a 2% probability of exceeding in 50 years is taken into consideration. In order to be more precautionary than in the past the code in Italy can decide to use a longer return period and therefore more stringent parameters. Furthermore, it should be remembered that the provisions of the legislation represent the minimum value according to which buildings must be designed: no one forbids the adoption of higher reference values.

The new model does not refer to a formal “normative” request. Can you make any predictions about its possible use?

It is essentially an updated scientific document and it represents what the scientific community can tell about the seismic hazard today in Italy. I insisted on starting this project to show that our knowledge is not blocked at 2004. We would have started much earlier if the model of danger had not been implemented by too little time in the construction regulations.
Its use will essentially depend on the results we will obtain; so it was agreed by the Great Risks Commission in 2015. The differences between the old and the new model will be weigh and if significant will justify changes in the input for the code. The requirements of the new model were however discussed and defined with DPC and the two centers of competence for seismic engineering, namely ReLuis and Eucentre, and with them the future use will be discussed.

The “classical” map MPS04 has been used by many as a “popular” description of the Italian seismic hazard: this has led – through the media but not only – to various exaggerated simplifications (eg: “in areas with low seismic hazard you cannot have very strong earthquakes “), which were the causes of many misunderstandings.
Personally, I have always maintained that the description of the seismic hazard for information purposes requires multiple maps, each dedicated to a different aspect (eg: maximum magnitude, maximum observed intensities etc.), accompanied by well-made explanations. It is also necessary that those who want to get information should devote time and patience to these aspects, avoiding the search for “spots”. Are we in the right direction, in your opinion?

I have always thought that representing the seismic hazard through a map is  an understatement of the complex articulation of the model. But it is also true that that the map is published in the Official Gazette.
When looking at a map, it is easy to misjudge. People tend to compare different places (L’Aquila is worse off than Milan, to give an example), or to think  that the areas we represented with green and light blue were exempt from the possibility of a strong earthquake. It is difficult for us too, but I think we should strive to describe the hazard with several representations at the same time. For example, by using more the seismic hazard curves for a single site, showing the different shaking values with the respective different frequency of exceedance, or with maps that represent the different probabilities of a certain level of shaking, in which it is possible to see that the zero probability of having an earthquake actually in Italy does not exist. In addition to that, the availability of other types of maps would help, such as the one of the maximum magnitude, which is one of the input elements for the seismic hazard.
But first of all we should commit ourselves to better describe this type of work even better and with simpler language. to change the degree of acceptance of the new model.

Rispondi

Inserisci i tuoi dati qui sotto o clicca su un'icona per effettuare l'accesso:

Logo WordPress.com

Stai commentando usando il tuo account WordPress.com. Chiudi sessione /  Modifica )

Google+ photo

Stai commentando usando il tuo account Google+. Chiudi sessione /  Modifica )

Foto Twitter

Stai commentando usando il tuo account Twitter. Chiudi sessione /  Modifica )

Foto di Facebook

Stai commentando usando il tuo account Facebook. Chiudi sessione /  Modifica )

Connessione a %s...