2002 Denver Annual Meeting (October 27-30, 2002)

Paper No. 11
Presentation Time: 10:30 AM

A UNIVERSAL LANDSLIDE DISTRIBUTION AND AN ASSOCIATED LANDSLIDE-EVENT MAGNITUDE SCALE


MALAMUD, Bruce D.1, TURCOTTE, Donald L.2, GUZZETTI, Fausto3 and REICHENBACH, Paola3, (1)Department of Geography, King's College London, Strand, London, WC2R 2LS, United Kingdom, (2)Earth and Atmospheric Sciences, Cornell Univ, Snee Hall, Ithaca, NY 14853, (3)CNR - IRPI Perugia, via della Madonna Alta, 126, Perugia, 06128, Italy, bruce@malamud.com

An earthquake, a rapid snowmelt, or a large storm can trigger a landslide event, ranging from a few individual landslides to hundreds of thousands. The frequency-area (or frequency-volume) distribution of a triggered-landslide event describes the frequency of landslides that occur at different sizes. We examine three landslide events, from different parts of the world and with different triggering mechanisms, and find that all three have the same “universal” landslide distribution, a three-parameter inverse gamma distribution which for small events (less than about 0.0004 km2) has a “roll-over” and for medium and large events is a power-law (a “fat” tail) distribution. The three landslide inventories were each compiled shortly after the triggering event, and are complete down to 0.0001-0.0002 km2: (1) 11,111 landslides triggered by the 1994 Northridge earthquake, California, compiled by Harp and Jibson (1994), (2) 4,369 landslides triggered by a 1997 snowmelt event in the Umbria region, Italy, compiled by Cardinali et al (2000), and (3) 9,871 landslides triggered by 1998 Hurricane Mitch (heavy rainfall) in Guatemala, compiled by Bucknam et al (2001). All three landslide data sets fall directly on our “universal” landslide distribution. We further introduce a landslide-event magnitude scale, with measure log(N), N the total number of landslides triggered by the event. If an inventory of triggered landslides is not complete (i.e., only the largest landslides have been compiled), the density of landslides of a given size can be compared with the “universal” landslide distribution, and the corresponding landslide event magnitude inferred. One can extend this technique to inventories of historic/geologic landslides—which we have done for Italy, Japan, and Bolivia—inferring the number of landslides that occurred over geologic time, and how many of these were erased from the landscape by erosion and mass-wasting. Landslides are clearly extremely complex, and include many different physical processes over different scales. However, earthquakes are also extremely complex. The Richter earthquake magnitude scale is certainly approximate, but has found great use since its introduction. Even with its obvious limitations, we suggest that our proposed landslide magnitude scale will also be very useful.