Power-Law Characteristics of Spatiotemporal Evolution of Global Landslide Fatalities

Landslides are among the major natural hazards that disturb the balance between human societies and the environment. The severe fatalities they cause have far-reaching consequences for social safety and regional development. Because existing landslide databases do not record all landslide events, this study assumes that the distribution of landslide fatalities estimated from existing data can, to a considerable extent, approximates the underlying distribution for the fatalities if the complete landslide data are available. Based on a systematic examination of the NCDC database, which is much larger than the popularly used GLC and GFLD datasets, and employing a rigorous statistical framework based on goodness-of-fit evaluation using the Kolmogorov–Smirnov (KS) statistic and bootstrap procedure, we find that landslide-fatality magnitudes exhibit a power-law or shifted power-law behavior across global, national, and temporal scales. At the continental level, when fatalities from landslides of different magnitudes are collectively analyzed, power-law behaviors often do not emerge, due to spatial heterogeneity caused by different regions’ topographic conditions, drainage systems, levels of population exposure, and social vulnerability. When this happens, analysis of smaller spatial scales, such as the national scale, is more appropriate when the nation is not too large geographically.