HR: 17:15h
AN: T22E-06 INVITED     [PDF]
TI: What are the links between landslide distributions, topographic relief and erosion rates?
AU: * Stark, C P
EM: cstark@ldeo.columbia.edu
AF: Lamont-Doherty Earth Observatory, Columbia University, 61 Route 9W, Palisades, NY 10964 United States
AU: Guzzetti, F
EM: Fausto.Guzzetti@irpi.cnr.it
AF: IRPI-CNR, via della Madonna Alta 126, Perugia, 06128 Italy
AU: Vecer, J
EM: vecer@stat.columbia.edu
AF: Dept of Statistics, Columbia University, New York, NY 10027 United States
AB: If we wish to understand the pattern and rates of erosion in mountains, we need to understand the coupling between hillslope failure events and the evolution over time of hillslopes. There must be a link of some kind between the process that triggers the landslides and the morphology of the hillslopes sculpted by them. This link presumably determines the size, frequency and location of landslides, in combination with runoff and channel erosion processes, and together they determine the flux of sediment into the channel network. Unfortunately, it is very difficult to quantify the patterns and rates of landslide erosion, in part because individual landslides occur on scales (of area and volume) that are wildly variable, from failures a few meters across to several kilometers in size, and in part because the timing and location of landslides is complex. These problems make both data acquisition and mathematical analysis very challenging. Nevertheless, many landslide maps and inventories now exist, and with the advent of widely available digital elevation data, there is an excellent opportunity to study both in tandem. We present an analysis of several dozen landslide inventories and show that the size (planform area) distribution of landslides has more or less the same shape regardless of geographical location, bedrock lithology, or trigger mechanism. Each distribution has a heavy, power-law tail, with a similar exponent in most data sets. In some very good inventories the full shape of the distribution can be estimated with confidence, along with the modal average landslide size. We present a theoretical model that links these observations to the morphology of the landscape on which they were triggered. The model predicts a probability distribution for landslide sizes that matches very well our observations. The result is quite promising: it makes a simple link between hillslope failure and hillslope geometry, and it suggests that, in combination, high-resolution data for both may be able to tell us something about the physical parameters that drive landslide erosion and landscape evolution.
UR: http://geomorph.ldeo.columbia.edu
DE: 1815 Erosion and sedimentation
DE: 1824 Geomorphology (1625)
DE: 1869 Stochastic processes
DE: 3210 Modeling
DE: 8110 Continental tectonics--general (0905)
SC: Tectonophysics [T]
MN: 2003 Fall Meeting