Large landslides and debris flows, or lahars, pose some of the greatest threats to people and property downstream from stratovolcanoes such as those of the Cascade Range. In order to accurately forecast areas that are likely to be impacted by landslides and lahars, USGS scientists develop and test computer models based upon statistical analyses of flow-path geometries and mathematical analyses of flow physics.
Lahar and debris flow models produce accurate hazard maps
Laharz_py is a statistically based computer program capable of generating debris-flow and lahar hazard maps very rapidly, even in the absence of site-specific information other than topography. The software calculates a hazard zone on and around a volcano and estimates hazard zones downstream far from the volcano as areas of potential inundation. It is a python-language computer program built on a GIS platform (compatible with ArcGIS version 10 or later), and is based on statistical trends of worldwide data. The model has also been applied to and calibrated using real-world scenarios of non-volcanic debris flows and rock avalanches.
D-Claw is based on the physics of landslide and debris-flow motion. The generality of D-Claw enables it to be applied to nearly any landslide or debris flow–in either volcanic or non-volcanic settings–but the model requires geometric input data of a prospective landslide or debris-flow source area as well as the physical composition of debris. The model is a suite of fortran-language computational routines that solve equations describing coupled evolution of two components of flow velocity as well as flow depth, debris porosity, and basal pore-fluid pressure. D-Claw's computational architecture is derived from the Clawpack family of programs, devised to solve diverse systems of equations that express physical conservation laws.
Both LaharZ and D-Claw require three-dimensional digital topographic data as inputs. The required resolution of topographic data depends on the size of the simulated landslide or debris flow. For very large volcanic events, a digital elevation model (DEM) with horizontal resolution of roughly 10 m (about 30 ft) is commonly adequate. Simulation of smaller events requires higher-resolution DEMs, which are commonly generated using airborne Lidar scanning of topography.
Mathematical models proofed with real-world experiments
Model development and testing also requires reliable, reproducible data that document the nature of landslide and debris-flow events. Such data are generated through utilization of the USGS debris-flow flume, a unique, large-scale experimental facility constructed near Blue River, Oregon in 1991. To date, more than 150 experiments have been conducted at the USGS flume, and nearly all are viewable in an on-line video archive. Results of these experiments, as well as explanations of model development and testing, are available in numerous published reports, many of which are accessible via Dr. Richard Iverson's USGS Professional Page.