Many of the present earthquake early warning (EEW) systems quickly determine an event’s hypocenter and magnitude and then predict strengths of ground motions. The Mw 9.0 Tohoku earthquake, however, revealed some technical issues with such methods: (1) underprediction at large distances due to the large extent of the fault rupture and (2) overprediction because the system was confused by multiple aftershocks that occurred simultaneously. To address these issues, we propose a new concept for EEW, in which the distribution of the present wavefield is estimated precisely in real time (real‐time shake mapping) by applying a data assimilation technique, and then the future wavefield is predicted time evolutionally by simulation of seismic‐wave propagation. Information on the hypocenter location and magnitude is not necessarily required. We call this method, in which physical processes are simulated from the precisely estimated present condition, numerical shake prediction because of its analogy to numerical weather prediction in meteorology. By applying the proposed method to the 2011 Tohoku earthquake and the 2004 Mid‐Niigata Prefecture earthquake (Mw 6.7), we show that numerical shake prediction can precisely and rapidly predict ground motion in real time.
Online Material: Animations as examples of numerical shake prediction.