ABSTRACT
Landslides are complex phenomena influenced by many factors, including soil and rock types, bedding planes, topography, and moisture content. Landslide events are single to many thousands of landslides, generally associated with a trigger such as an earthquake, a large storm, a rapid snowmelt, or a volcanic eruption. A landslide event may be quantified by the frequencyarea distribution of the triggered landslides. We have recently shown (Malamud et al. 2004a) that the frequency-area statistics of three substantially complete landslide inventories are well approximated by the same probability density function, a three-parameter inverse-gamma distribution. We also introduced a landslide-event magnitude scale mL log(NLT), with NLT the total number of landslides associated with the landslide event, in analogy to the Richter earthquake magnitude scale. We argue that the statistics of triggered-landslide events under a wide variety of conditions follow a general statistical behaviour to a good approximation. Such a ‘general’ statistical behaviour
is widely accepted for the frequency-magnitude statistics of earthquakes, which are also complex, but generally follow a power-law relationship between the number of earthquakes and the earthquake rupture area, the Gutenberg-Richter relation.
