|We extract tsunamigenic earthquakes related to subduction zones from the NOAA database (coloured according to magnitude; Dunbar et al., 2008; 2009) from 1962 - 2016 and assign the events to subduction zone segments (black; Heuret et al., 2011).|
The role of tectonics in tsunami generation is at present poorly understood. However, the fact that some regions produce more tsunamis than others indicates that tectonics could influence tsunamigenesis. We complement a global earthquake database that contains geometrical, mechanical, and seismicity parameters of subduction zones with tsunami data. We statistically analyse the database to identify the tectonic parameters that affect tsunamigenesis. The Pearson’s product-moment correlation coefficients reveal high positive correlations of ~0.65 between, amongst others, the maximum water height of tsunamis and the seismic coupling in a subduction zone. However, these correlations are mainly caused by outliers. The Spearman’s rank correlation coefficient results in more robust correlations of ~0.60 between the number of tsunamis in a subduction zone and subduction velocity (positive correlation) and the sediment thickness at the trench (negative correlation). Interestingly, there is a positive correlation between the latter and tsunami magnitude. In an effort towards multivariate statistics, a binary decision tree analysis is conducted with one variable. However, this shows that the amount of data is too scarce. To complement this limited amount of data and to assess physical causality of the tectonic parameters with regard to tsunamigenesis, we conduct a numerical study of the most promising parameters using a geodynamic seismic cycle model. An increase in sediment thickness on the subducting plate results in a shift in seismic activity from outerrise normal faults to splay faults. Besides that, the splay fault is the preferred rupture path for a strongly velocity strengthening friction regime in the shallow part of the subduction zone, which increases the tsunamigenic potential. A larger updip limit of the seismogenic zone results in larger vertical surface displacement.
|The Pearson’s product-moment correlation coefficients. The correlations are low or mainly caused by outliers. + and - indicate positive and negative correlations, respectively.|
This project is in collaboration with Silvia Brizzi (Roma Tre University), Francesca Funiciello (Roma Tre University), and Arnauld Heuret (Univerity of Guyane).