Deep evaluation of numerical simulation of long-time tsunami propagation utilizing tsunami source constrained by offshore tsunami waveforms: Application to the 2016 Mw 6.9 off-Fukushima earthquake, Japan

• The ability of simulation to reproduce long-time tsunami propagation was evaluated. • Tsunami source constrained by offshore tsunami data was used in the simulation. • The largest tsunamis at nearshore and coastal areas were reproduced well. • Tsunami decay in nearshore areas was reproduced better than in coastal areas. In this work, the nonlinear long-wave simulation of the 2016 off-Fukushima tsunami was performed to evaluate the reproducibility of long-time tsunami data observed at nearshore (10–50 m in water depth) and coastal (<10 m in water depth) water-level stations located near the source. We used a previously published accurate tsunami source constrained by offshore tsunami waveforms recorded by a wide, dense seafloor observation network around Japan. The simulation accurately reproduced the tsunami waveforms up to the largest waves (90 min from the arrival time of the initial wave at each station) at most of the stations. Additionally, the long-time evolution of tsunami-amplitude envelopes (15 hours from the earthquake), including the decay process, were also reproduced at some stations. The reproducibility errors exceeded the uncertainties due to source estimation errors, indicating that the propagation modeling needs to be improved rather than the source modeling. Furthermore, the reproducibility was found to be higher for the nearshore station than for the coastal station in the same port. Our analyses suggest that numerical tsunami simulation is needed to be improved in modeling tsunami propagation in shallower waters than those where nearshore stations are located. Although the poor reproducibility of the entire envelope could not be completely reduced by the additional simulations with various setups in harbor structures and bottom friction, we found that each of the setups was effective for partial improvement. The findings in this study will contribute to future improvements of long-time tsunami propagation and real-time tsunami forecasting.