Utilization of crowdsourced macroseismic observations to distinguish damaging from harmless earthquakes globally within minutes of an event

Rapid assessment of an earthquake’s impact on the affected society is a crucial step in the early phase of disaster management, determining the further organization of civil protection measures. In this study, we demonstrate that felt-reports containing macroseismic observations, collected via the LastQuake service of the European Mediterranean Seismological Center, can be utilized to rapidly estimate the probability of a felt earthquake to be “damaging” rather than “harmless” on a global scale. In our fully data-driven, transparent, and reproducible approach, we first map the reported observations to macroseismic intensities according to the EMS-98 macroseismic scale. Subsequently, we compare the distribution of felt-reports to documented earthquake impact in terms of economic losses, number of fatalities, and number of damaged or destroyed buildings. Using the distribution of felt-reports as predictive parameters and an impact measure as the target parameter, we infer a probabilistic model utilizing Bayes’ theorem and Kernel Density Estimation, that provides the probability of an earthquake to be “damaging”. For 22% of felt events in 2021, a sufficient number of felt-reports to run the model is collected within 10 minutes after the earthquake. While a clean separation of “damaging” and “harmless” events remains a challenging task, correct and unambiguous assessment of a large portion of “harmless” events in our dataset is identified as a key strength of our approach. We consider our method an inexpensive addition to the pool of earthquake impact assessment tools, that can be utilized instantly in all populated areas on the planet. Being fully independent of seismic data, the suggested framework poses an affordable option to potentially improve disaster management in regions that lack expensive seismic instrumentation today and in the near future.