Do seismic hazard models kill? (by Massimiliano Stucchi)

Introduction. The appearance of an article, on the weekly magazine L’Espresso (http://espresso.repubblica.it/plus/articoli/2019/08/26/news/terremoto-calcoli-sbagliati-1.338128?ref=HEF_RULLO&preview=true), which took advantage of the 2016 Amatrice earthquake anniversary to discredit the Italian seismic hazard model and the national building code, based on it, using fake news and inconsistent arguments made me angry.
What follows is a comment written for the benefit of the international readers.
The original version in Italian which can be found here (https://terremotiegrandirischi.com/2019/08/27/la-colpa-e-dei-modelli-di-pericolosita-sismica-di-massimiliano-stucchi/), which can easily be translated by means of the improved https://translate.google.com/.

As in many countries, since 2008 the Italian building code (NTC08 and now NTC18) makes reference to design spectra; they are taken from the results of the 2004 PSHA model (1). A new PSHA assessment is been published soon.

On the other hand, since a few years a very small group of Italian researchers proposes a so called alternative method for the evaluation of seismic hazard, based on the neo-deterministic approach. No problem, it is a current scientific discussion. Things became more complex when this group claims that their method should be taken as a basis for the building code; it becomes boorish when, like in the above mentioned article, they claim that the seismic actions proposed by the PSHA model, adopted by the building code, have been overcome in recent earthquakes, and that casualties are due to that “wrong” seismic actions, by making use of “fake news”.
Let’s see.

The article repeats, once again, the fake news that recorded PGA overcame the PGA estimated by the PSHA model, in the occasion of 2009 (L’Aquila) and 2012 (Emilia-Romagna) earthquakes. It has been proved that the above statement is not true (2) (3); simply, comparison are wrongly made between soft ground recordings and hard ground estimates!
In the case of the 2016 Amatrice and Norcia earthquakes, yes, recorded PGA did overcome PSHA estimated PGA. Does it mean that the model did underestimate?

It must first be considered that estimated PGA comes with some % probability of being exceeded in xx years (the most common figures being 10% and 50 years). Moreover, the comparison should be made (if really needed) on the whole design spectrum, not on PGA which is not used for building design; but such comparisons are not recommendable (4). Finally,  the main point is that the PSHA model is a model; it offers various elaborations related to different probabilities of exceedance in different time-intervals. For example, some peak values recorded in 2016 are slightly higher than those related to the 2% probability of exceeding in 50 years, but are lower than those related to the 1% probability of exceeding in the same interval.

So the matter comes back to the main point: do we need to design against the maximum expected shaking (and how to assess it?), or to a shaking with a lower probability of exceedance with respect to the adopted one which, by the way, is adopted in many countries of the world?

This is not a seismological – nor a SHA – problem. SHA models offer a variety of possible solutions and then someone decides. It is a political decision which, usually, is in fact taken by engineers (cost-benefits analysis); unfortunately, this often happens without or with little explanation. We know for instance that source of the “475 return period” is close to casual, but it seems to represent a “satisfactory” compromise. Would be nice if it was explained better, however, so to allow that part of the public, which is not ready to follow scandal claims, to understand by itself.

Are the detractors of the PSHA model and the building code able to provide one example, only one, of a building, designed according the NTC08 without executions mistakes, which collapsed because the design spectra values were overcome by the recorded ones? It would be a good, practical case history, instead of a theoretical clash. This question was already asked for, without getting an answer.
And, even more important: why to give the wrong idea to the public that, as soon as design spectra are exceeded, buildings collapse?

The article, and the scientists behind, quotes the reconstruction of Norcia after the 1979 earthquake and the San Benedetto Basilica (which did stood it); both have little to do with NTC08 and related design spectra. The worst, however, comes with the reference to the collapse of the school in San Giuliano di Puglia (2002), “renovated according to inadequate criteria”. Shame on you, who mix this event with your crusade against PSHA and NTC08! That school was restored in the absence – at that time – of the building code (responsibility of the Ministries which delayed the expansion of the building code to all Italy with all their power), and according to a questionable design.

Seismic hazard models do not kill; buildings do, with the help of fake news!