This analysis is an extension of the studies based on the space–time–magnitude domain nearest-neighborhood distances (ƞ) obtained from two successive earthquakes. We included epicentral depth as a parameter, and established a new analytical solution for seismicity clustering in the space–time–depth–magnitude domain. The proposed technique was applied to SW Japan seismic observation data for 2011–2018, and revealed that the temporal, spatial, and epicentral (T, R, E) components of the joint distribution for ƞ were prominently bimodal, indicating the presence of two, statistically distinct, earthquake populations. We used the Lomb–Scargle periodogram (LSP) to generate a power spectrum preceding the Kumamoto earthquake, and detected the periodic component of unequally spaced points for log₁₀ƞ (logarithmic scale of ƞ). The LSP showed a dominant peak corresponding to the time period between foreshocks and aftershocks, due to the highest number of recorded ƞ observations. The technique presented in this paper can be applied globally for any time period and any seismotectonic zone preceding large earthquakes of interest.

这种分析是一种基于空间-时间-幅度域中的最近邻距离（研究的延伸ƞ从两个连续的地震获得）。我们将震中深度作为参数，并为时空-深度-幅度域的地震活动性聚类建立了新的分析解决方案。拟议的技术应用于日本西南部2011-2018年地震观测数据，揭示了ƞ联合分布的时间，空间和震中（T，R，E）分量明显是双峰的，表明存在两个统计学上不同的地震种群。我们所使用的博士伦-Scargle周期图（LSP），以产生地震熊本之前的功率谱，并检测出的不等间距点的周期性分量日志₁₀ ƞ（的对数标度ƞ）。由于记录的ƞ观测数量最多，因此LSP显示了一个与前震和余震之间的时间段相对应的显性峰。本文介绍的技术可以在感兴趣的大地震之前的任何时间段和任何地震构造带中全局应用。