by After an earthquake, it is crucial early in disaster management to obtain a rapid assessment of the severity of the impact on the affected population in order to be able to initiate adequate emergency measures. A quick and good first assessment of whether an earthquake causes major or minor damage is often not possible after only 10 minutes through information from affected people about the “perceived intensity” of the earthquake. This is shown in a recent study by researchers led by Henning Lilienkamp and Fabrice Cotton from the GFZ German Research Center for Geosciences, now published in the journal The Seismic Record. In their new approach, they evaluate the information transmitted by people after an earthquake through a website or app of the European Mediterranean Seismological Center’s Quake Last service. Since no seismic measurement data is required, this low-cost approach can help future disaster management, especially in regions with few measurement devices.
Background: Assessing the impact of an earthquake
It is vital for decision makers and disaster managers to be able to assess the impact of an earthquake as quickly as possible, as it directly affects the measures taken to protect lives and limit further damage.
In some cases, such as the series of catastrophic earthquakes in Turkey and Syria in February 2023, it is immediately clear that a large-scale emergency response is urgently needed. But this is not always true.
“For example, in the 5.9 magnitude earthquake that struck remote areas of Afghanistan and Iran on June 12, 2022, which left more than 1,000 dead, it was unclear for hours whether significant impacts were expected or which was not, according to the European Seismological Mediterranean. Center (EMSC),” explains Henning Lilienkamp, PhD student in the department “Earthquake Hazards and Dynamic Risks” at the German GFZ Research Center for Geosciences and lead author of the current study.
There are rapid assessment systems such as PAGER, developed by the US Geological Survey. However, it currently takes about 30 minutes on average to give already quite extensive estimates of the effects of an earthquake. And it is based on the ShakeMap method, so it requires ground acceleration data and other seismic observations as well as reports from the population.
A new approach based exclusively on felt reports
“Many types of data can be included in the assessment of an earthquake and its immediate consequences, and a differentiated analysis is crucial for successful long-term disaster management,” emphasized Lilienkamp. He and his colleagues are now able to show that an initially rough but often more rapid assessment is already possible based on information from the population in question alone. In addition to Henning Lilienkamp and Fabrice Cotton, head of the “Earthquake Hazards and Dynamic Risks” section at the GFZ and professor at the University of Potsdam, other researchers from the University of Potsdam, the EMSC, and the University of Bergamo were also involved. in the study, published in the journal “The Seismic Record.”
In their new approach, they use data on the “perceived intensity” of an earthquake from people affected. They transmit their personal post-quake assessment based on a graph and a comment via the EMSC’s LastQuake service website or app. It was developed to warn people as quickly as possible immediately after an earthquake. Already in the first 10 minutes after an event, a large set of data may be available for evaluation, although this number of course depends on the strength of the earthquake and the level of participation of the communities.
For example, for the February 6 earthquake sequence in Turkey, the LastQuake service collected about 6500 reports for the first 7.8 magnitude earthquake and about 4800 reports for the second 7.5 magnitude earthquake, according to Lilienkamp. For the first tremor, it took about 4.5 minutes to collect 50 reports – the minimum number needed to run the model developed here – and after 10 minutes, 1232 reports were already available.
“We were convinced that this database is too valuable to ignore in the long run, because it is collected efficiently and on a global scale, including in regions where seismic instrumentation is not expensive,” says Lilienkamp.
Development of a model for rapid assessment of earthquake effects
The researchers had access to more than 1.5 million reports collected worldwide on the perceived intensity of more than 10,000 earthquakes of any size from 2014 to 2021. Based on the “crowdsourced data”, they developed a probabilistic model that can be used to estimate whether an earthquake is high or low impact.
For this purpose, in step one, the detected intensity data of past earthquakes was converted into representative parameters, such as a “dual intensity value” that quantifies the magnitude of the shaking. Another value describes the spatial extent of the area where the earthquake was felt.
In a second step, the affected earthquakes were classified using more detailed information from global earthquake impact databases. The study defines earthquakes with strong impacts as those associated with at least one of the following impacts: one building destroyed, at least 50 buildings damaged, at least two deaths or financial losses documented.
Ultimately, this results in a statement about how likely it is that an earthquake will have strong impacts with the “feel” information transmitted.
Step one can also be used to classify new earthquakes.
Validation and limitation of the new approach
The researchers then tested their model on eleven earthquakes from 2022. “A key strength of our approach is that it is able to accurately and unambiguously assess a large number of low-impact events,” summarizes Henning Lilienkamp.
The public may be able to recognize an earthquake as having a low impact, because earthquakes of this type – although not very strong – can still be felt and may cause concern. with the researchers in their paper.
“We see with high-impact events that it is still a challenge to clearly distinguish them from the lower-impact ones as well,” says Lilienkamp. This could also be because there are far fewer severe earthquakes in the data base that the model is “learning” on. With data collected over time, this could improve further, according to the researchers.
A natural limitation of the approach – especially during strong earthquakes – is the lack of very early reports from the area where the shaking was the most intense. “This effect is well known and reflects the fact that people in such dire circumstances naturally prioritize finding shelter and rescuing people in danger, over sending in reports on their smartphones,” a Lilienkamp explains. In addition, the global analysis shows that the LastQuake service is still mainly used in Europe at the moment (75 percent of the data comes from there).
“We see our method as a cost-effective addition to the pool of earthquake impact assessment tools that are completely independent of seismic data and can be used in many population areas around the world. Although it is still a task open to further develop our method into a tool for practical use, we demonstrate the potential to support disaster management in regions that currently do not have expensive seismic instruments,” says Fabrice Cotton from GFZ.
Outlook: Application capability
Lilienkamp and colleagues suggest that their method could be used to develop a “traffic light” system based on impact scores, where green-level scores would not require further action by decision-makers, with yellow prompting further investigation and red potentially raise an alert.
“As seismologists, we need to better understand exactly how decision makers and emergency services such as fire departments act in an emergency, what kind of information is useful, and at what high impact probabilities they would prefer to raise. alarm,” said Lilienkamp. “Careful communication of the capabilities of our model and the individual needs of potential end users will be critical to the practical implementation of traffic light systems.”
