A new software and a new database chart the course towards an increasingly accurate reconstruction of the seismic history of the territories
A team of researchers from theIstituto Nazionale di Geofisica e Vulcanologia (INGV)In partnership with Southern University of Science and Technology Shenzhen (SUSTech, China), developed a algorithm that integrates for the first time three fundamental geological measurements for the study of past earthquakes: rejection, length of crustal rupture and age of earthquake.
The algorithm was presented in a study recently published in the scientific journal Bulletin of the Seismological Society of America (BSSA)The paper provides magnitude estimates with all uncertainties explicitly quantified and shows that, in the central Apennines, paleoseismological time series capture almost exclusively extreme magnitudes, i.e., magnitudes greater than 6.5, with important implications for seismic hazard assessment.
Reconstructing the seismic history of an area prior to the advent of seismographs is an operation as essential as it is delicate.. Sufficiently energetic earthquakes “they break” the Earth's surface: by locating these fractures, measuring how much the surface has moved in a single event (i.e. the "rejection") and dating the sediments that record its age, Paleoseismologists can identify prehistoric and historical earthquakesHowever, translating these ground measurements into a reliable magnitude is far from trivial, fraught with measurement uncertainties, conservation issues, and modeling challenges.
At the basis of the work there is PaleoEcA, an A new database collects 44 paleoearthquakes in the central Apennines that occurred over the last 27.000 years..
“We have harmonized decades of paleoseismological investigations, associating each measurement with its uncertainty”, explains Deborah Di Naccio, researcher at INGV and first author of the article. “Without this transparency, any subsequent processing risks producing magnitudes that are only apparently precise.”.
The results presented in the study are based on a new software, PaleoBAYES, an algorithm open-source which uses the statistical method of Bayesian inference to simultaneously fuse three geological data: the rejection, the length of the surface rupture, and the age of the event.
“PaleoBAYES takes into account not only observational errors, but also two often overlooked factors.”, he adds Davide Zaccagnino, a researcher at SUSTech in Shenzhen and co-author of the article. On the one hand, the probability of preservation decreases with the passage of time: an older trace is more likely to have been erased. On the other hand, it is well known that the distribution of earthquakes over time is not uniform: according to the Gutenberg-Richter law, the frequency of events decreases exponentially with increasing magnitude. Taking this property into account when interpreting the events detected in paleoseismological trenches is therefore more realistic than assuming a uniform distribution of different magnitudes. In this way, the final estimate is not a single number, but a credibility interval: for our 44 events, the 95% uncertainty typically lies within ±0.3-0.7 magnitude units..
The test bed was historical earthquakes of known magnitude. For the 1915 Fucino earthquake, for example, the method returned a magnitude of M 6.83 ± 0.13, fully consistent with instrumental and macroseismic estimates (M 6.7-7.0). But the study's real revelation lies elsewhere. “In the central Apennines, earthquakes with magnitudes below about 6.5 have a very low probability of leaving traces that survive long enough to be observed today.”, he claims Michele Carafa, researcher at INGV and co-author of the study. This implies that the paleoseismological catalog samples almost exclusively the extreme tail of the magnitude distribution, that is, the very large events. Any hazard assessment that ignores this threshold of effective completeness—and that doesn't rigorously propagate uncertainties—can introduce systematic biases..
The study also shows that the results depend heavily on the chosen scaling laws, reinforcing the need for a unique and reproducible probabilistic approach. With PaleoBAYES and the PaleoEcA example, the scientific community now has a toolbox for a transition from the deterministic estimation of pre-instrumental earthquake magnitudes to the systematic quantification of uncertainties in paleoseismology.
Link to the PaleoBAYES software and the PaleoEcA database
Useful links:
Istituto Nazionale di Geofisica e Vulcanologia (INGV)
Southern University of Science and Technology Shenzhen (Guangdong, China)
Photos: The Monte Morrone fault at Colle delle Vacche, Central Apennines, Italy (Photo credit: Vanja Kastelic).