Synthesis and Source Characteristics of Tsunamis in the Sea of Japan Based on Normal‐Mode Method

The normal‐mode method, which is widely used in seismology, was applied to characterize the free oscillation and tsunami sources in the Sea of Japan. Improvements in matrix storage, and incorporation of a modern sparse eigensolver, enabled us to calculate 6,000 high‐resolution normal‐mode solutions for the Sea of Japan, down to a period of 8 min. We used kurtosis to group these modes into three types: 622 basin‐wide modes, 4,953 regional modes, and 425 local modes. By the superposition of these modes, tsunami waveforms were synthetized for the 1983 Sea of Japan earthquake (Mw 7.7). Comparisons of synthesized waveforms with those computed by the finite difference method as well as observations showed good agreement at periods longer than 8 min. The excitation weights for 60 fault models along the eastern margin of the Sea of Japan were computed, and they showed that the average excitation was larger for the sources located at shallower water depth or with larger magnitude. Among the three types of modes, the regional modes played the most important role and affected a wide coastal area. Key Points Six thousand high‐resolution normal‐mode solutions were obtained down to around 8‐min periods in the Sea of Japan Tsunami waveforms excited by the 1983 Sea of Japan earthquake were successfully synthesized using mode superposition Sixty potential submarine faults along the eastern margin of the Sea of Japan were characterized using excitation weights for the normal modes