On November 11, 2019, the ground shocks in the Rhone Valley south of Montelimar. The strength of the tremor felt in southeast France was 5.4 on the Richter scale. It left four injured and caused severe damage in several villages in Ardish and Drum, especially in the village of Til, which is closest to the epicenter. This event reminds us of the reality of the seismic hazard in the region and the importance of an accurate assessment of this hazard *, especially since the seismic rupture occurred not far from the nuclear facilities: ten kilometers from the power plant in Croissant and about 20 kilometers from the power plant in Croas. Tricastin Power Plant.
All of these facilities are subject to regulations that specify the method for operators to follow in determining the dimensions of their structures and equipment with regard to earthquake risks and ensuring nuclear safety. For its part, IRSN verifies that the data and assumptions that nuclear operators take into account are consistent with state of knowledge. To this end, within the institute, BERSSIN not only studies recent earthquakes and historic earthquakes, listed 1,000 years ago, in the region, but also any phenomenon likely to affect ground movements.
Nevertheless, the Till earthquake recordings conducted by IRSN clearly highlighted the phenomenon the BERSSIN team had been working on in the Rhone Valley since the 2000s: an effect capable of amplifying the Earth’s motions locally, thus greatly increasing the risk. What is it about? “When the Earth trembles and we measure the movement of the Earth at a certain point, this movement depends on the source of the earthquake – mainly on its energy – and on the spread of waves between that source and the point of the“ recording. ”The so-called“ location effect ”occurs in the last hundreds of meters and is related to the geological nature and stratigraphy. Superficial, ”explains Celine Gillis, a seismologist at Persen.
The site effect occurs in typical formations such as valleys filled with sediments. In these loose formations, seismic waves propagate at a much slower speed than in the stiffer country rocks. Because of this extreme contrast, waves that reach the sediments are amplified and trapped there: they are reflected off the edges and bottom of the valley without being able to escape. As a result, the amplitude and duration of the earth’s movement increases, and thus buildings in such locations are exposed to stronger and longer vibrations. This phenomenon was detected in 1985, during the magnitude 8 earthquake that shook Mexico City, which was erected on an ancient dry lake, and then this phenomenon was studied in different active regions of the planet (Mexico, the United States, Japan), but it remained Poorly known in the United States. The low seismic context of the Rhone Valley. How can the impact of the site be measured in this region? What are the most relevant methods for doing this?
To answer these questions, the BERSSIN team uses an experimental and numerical approach. The first is to record on rocks and sediments the movements of the Earth caused by the same earthquake and compare these seismic charts. It makes it possible to directly detect the amplification and frequencies of the waves occurring in it. The second aims to simulate the effect of this location by modeling seismic wave propagation.
In 2016, this team began implementing the experimental approach near the Tricastin nuclear site that spans over 600 hectares between Valence and Avignon that brings together several facilities including enrichment-related plants and the four reactors of the EDF power plant. According to the complete reconstructions of the geological history of the area by Mark Edward Cushing, a geologist at Persen, the site is located on an ancient valley. It was excavated by the Rhone River in extremely hard alabaster and limestones about 6 million years ago, when the Mediterranean Sea experienced a 1,500-meter drop in its level due to the closure of the Strait of Gibraltar. “After about 500,000 years,” the geologist explains, “the strait opened up, the waters rose to the Mediterranean, and soft sediments, like sand and mud, were deposited in the deep ancient valley that ranges from 400 to 500 meters per site.” On the surface, its width today can reach a few kilometers. Due to the strong contrast between limestone on the one hand and clay and sand on the other hand, all of its features identified it as a very good candidate for site effect.
First benchmarking campaign
During the first measurement campaign from August 2016 to June 2017, three seismic stations – two installed directly over the valley and one on limestone outside (see opposite map) recorded about ten interesting earthquakes, with the closest earthquake epicenter located about a hundred kilometers away. Comparison of the same earthquake signals shows an amplification of the seismic motion from 0.5 Hz of frequency. It could be 6 times stronger directly over the canyon than the limestone: this is especially the case for the seismic waves generated by the 6.5-magnitude earthquake that occurred on October 30th. 2016 in Norcia, Italy, about 700 kilometers from the Rhone Valley. Supplementary measurements also made it possible to make a preliminary estimate of shear wave velocities – the most energetic and damaging of buildings – at the stations: they are about 1000 m / s in sediments while they reach 2,500 at 3000 m / s in limestone. This factor of 2.5 to 3 clearly indicates strong contrast.
These first observations demonstrate the existence of amplification of seismic waves at this site, but such a local measurement is insufficient to assess the diversity and complexity of the phenomenon on the Old Valley scale. Then Berenice Froment – who is also a seismologist at Persen -, Celine Gillis, and Mark Edward Cushing designed an even more ambitious project, the DARE project. The goal of course is to better describe the effect of the site in the Tricastin region, but also to make it a written case for testing and creating a new methodology. Indeed, if today, thanks to effective progress, it is possible to deploy networks of hundreds of sensors to multiply the measurement points, then it remains that we in low-earthquake regions are limited by the few events that must be recorded. Hence the idea of using another source of information: ambient seismic noise. By this we mean all vibrations that permanently cross the surface geological layers, whether they result from distant natural processes such as the inflation of the Atlantic Ocean or the Mediterranean, or more locally by human activities such as the passage of a TGV, or cars on a nearby highway, or Even factory noise.
So the first challenge was to show that this “new” seismic data could be used to measure the site effect on Tricastin. The second is to demonstrate that it can also be used to build a 3D model of the velocity of the shear waves needed to properly simulate wave propagation and entrainment in a valley. Today, the most efficient imaging method for creating such a model at this scale is those that qualify as “active seismic” applied in the field of oil exploration. Not only is it equipping the examined area with a large number of sensors, but also creating a well-controlled seismic source, generally by means of a vibrating truck. But the cost of these campaigns is still quite high.
“The so-called” negative “images, because it is sufficient to allow the sensors to record seismic background noise without the need to generate the signals, thus providing a new, much-accessible capability,” said Berenice Froment, the project’s pioneer. DARE. Already, this technique has been shown to produce images of the Earth’s crust (at a scale of ten kilometers) of seismic vibrations caused by the enlargement of the ocean (at frequencies below one hertz). Here the target scale is a few hundred meters, higher frequency signals (greater than one Hz), generated by more local operations are needed. These signals are more choppy and highly dependent on the location of the study, and they are also more difficult to use.
To meet all these challenges, the BERSSIN team, with the help of companies EGIS and SIPROBE, deployed 400 sensors in February 2020 over an area of about 10 kilometers by 10 kilometers covering the entire valley and the host rocks with the aim of the first measurement. Seismic noise in this industrial environment. In one month of the acquisition, they logged nearly 1.2 terabytes of data. According to the first analyzes, the signals related to local industrial activity are of high quality, reaching several Hz, and should allow an estimate of the amplification of seismic motion. This work is in progress.
Complex engineering valley
At the same time, the initial shear wave velocity model obtained from these data confirms strong discrepancies – with a factor of 2.5 – between filling sediments and host rocks. It reveals the complex three-dimensional geometry of the entire valley. It shows, for example, that the valley is deeper and deeper in the south than in the north and precisely delineates the meeting point between the ancient valley of Ardèche and the Rhône to the southwest. This model needs to be improved before it can be used in numerical simulations of a site effect, but it is indeed very promising.
Since the comprehensive methodology is new, the results will need to be validated with a more traditional approach. To this end, the project benefits from the support of the French National Research Agencies (ANR) and German (DFG) as part of a partnership with the University, the Helmholtz Center in Potsdam and the University of Grenoble to launch a complementary campaign of measures. This time, 50 seismic stations have been calibrated to record earthquakes, which have spread over the same area since September 2020, and will collect data for at least six months. The results of this Franco-German cooperation are expected by 2022. In addition to the specific study of the Tricastin site, they must prove the value of this abundant and rapid data to be obtained, namely, seismic noise to assess the effects of the site.
* Seismic hazard is the possibility that an area or site is exposed to an earthquake that has specific characteristics. Earthquake risk corresponds to the probability of loss of life and damage estimated according to risk.