Identifying plausible historical scenarios for coupled lake level and seismicity rate changes: The case for the Dead Sea during the last 2 millennia

Studies of seismicity induced by water level changes in reservoirs and lakes focus typically on well-documented contemporary records. Can such interactions be explored on a historical timescale when the two data types suffer from severe uncertainties stemming from the different nature of the data, methods and resolution? In this study, we show a way to considerably improve the correlation between interpolated records of historical Dead Sea level reconstructions and discrete seismicity patterns in the area, over the period of the past 2 millennia. Inspired by the results of our previous study, we carefully revise the historical earthquake catalog in the Dead Sea to exclude remote earthquakes and include small local events. For addressing the uncertainties in lake levels, we generate an ensemble of random interpolations of water level curves and rank them by correlation with the historical records of seismic stress release. We compute a synthetic catalog of earthquakes, applying a Mohr-Coulomb failure criterion. The critical state of stress at hypocentral depths is achieved by static poroelastic deformations incorporating the change in effective normal stress (due to the best-fit water level curve) superimposed on the regional strike-slip tectonic deformations. The earthquakes of this synthetic catalog show an impressive agreement with historical earthquakes documented to have damaged Jerusalem. We refine the seismic catalog by searching for small local events that toppled houses in Jerusalem; including all local events improves the correlation with lake levels. We demonstrate for the first time a high correlation between water level changes and the recorded recurrence intervals of historical earthquakes.