[S-CG43] スラブ内地震とその発生メカニズム

[SCG43-03] Nearest-neighbour cluster analysis of intraslab seismicity in the Vrancea Seismic Zone, Romania

*Laura Petrescu¹、Bogdan Enescu^2,1 (1.National Institute for Earth Physics, Magurele, Romania、2.Kyoto University, Kyoto, Japan)

キーワード：Intraslab earthquakes, Seismicity clustering, Slab-crust coupling

The Vrancea Seismic Zone is an anomalous seismic nest of intermediate-depth earthquakes, located in the South-East Carpathians in Romania. Although far from active plate boundaries, Vrancea releases the largest strain in continental Europe, suggesting the existence of a relic slab that is extending and sinking into the mantle. We investigate possible correlations between earthquakes inside the seismogenic block (~60–180km) and the overlying crust (~0-50km) using the nearest-neighbour analysis in a combined multidimensional domain that considers location, time, and size of events, as well as b-values and fractal dimensions of the seismic volume. This method (Zaliapin and Ben-Zion, 2013) identifies deviations from a time-stationary, space-inhomogeneous marked Poisson process, allowing for systematic detection and analysis of earthquake clusters. Each cluster takes the form of a spanning tree with variably strong branches, or links to the mainshock, quantified by the exponent of the nearest-neighbour distance. To identify and classify such clusters in Vrancea, we use a period of 20 years (2001-2021) from the local ROMPLUS seismic catalog produced by the National Institute of Earth Physics (NIEP) in Romania. The magnitude of completeness varies with depth, time, and method used to estimate it. We enforce a minimum local magnitude of 2.5, the average between all obtained values. Previous analyses of classic plate boundary seismicity showed bimodal distributions of the nearest-neighbour earthquake distances. In contrast, our results show a skewed unimodal distribution, suggesting that Vrancea generates mostly background seismicity, consistent with the idea of a stochastic earthquake system. Commonly, the intersection of the two distributions is used to extract a threshold distance that separates background from meaningfully correlated seismicity. Since we do not find a clear bimodal distribution, we enforce a synthetic threshold value to investigate the patterns of strongly correlated seismicity. For the considered period we find several earthquake clusters with mainshocks M>4 in the subcrustal area and two swarms of similarly sized earthquakes in the upper crust. These clusters make up less than 1% of the observed seismicity. We do not find migrating clusters across a seismic hiatus at 50-60km depth and no strong correlations between subcrustal and crustal earthquakes. This suggests that deep earthquakes do not trigger shallow seismicity on short time-scales and the relic slab is at most weakly coupled with the overlying crust. Our investigation has fundamental implications for understanding earthquake mechanisms in anomalously subducting slabs and seismic nests worldwide and lays the ground for future earthquake forecasting design in intracontinental regions.