An Objective Method for the Intercomparison of Terrain Stability Models
Location
Space Dynamics Laboratory
Event Website
http://water.usu.edu/
Start Date
3-26-2004 2:00 PM
End Date
3-26-2004 2:15 PM
Description
A terrain stability map quantifies the propensity for landslide initiation at a point in space. There are a number of process based models used for terrain stability mapping. SHALSTAB and SINMAP are two such models that combine steady-state hydrology assumptions with the infinite slope stability model to quantify slope stability. Inputs are digital elevation data from which slope and drainage area are evaluated as well as hydrologic and soil parameters. SHALSTAB quantifies terrain instability in terms of the critical effective rainfall required to trigger pore pressure induced instability. SINMAP quantifies terrain stability in terms of the probability that the infinite slope stability model factor of safety is greater than one over uniform probability distributions quantifying the uncertainty in model parameters. These are two models with similar physical basis but they use different indices to quantify instability. Because of this the relative performance of these models, when compared to observed landslide locations, is difficult to assess. A new statistic based on the cumulative fraction of the stability index at locations of observed landslides, relative to the cumulative fraction of the stability index over the terrain has been developed. This statistic quantifies the discriminatory capability of a stability map in terms of the degree to which a terrain stability map is successful in "capturing" observed landslides in areas mapped as unstable while minimizing the extent of these areas. This statistic provides a way to compare models that use different stability indices and is a quantity that can be used for the calibration or optimization of model parameters using observed landslide initiation locations. This is useful when physical data necessary to estimate model parameters is limited. We show results where the performance of these models that use different indices are compared using data from the Chetwynd area in east central British Columbia where 696 landslide initiation locations were mapped. We found that in this particular region slope was the dominant variable that discriminated terrain instability. The optimally calibrated SHALSTAB and SINMAP models that also use drainage area as a predictor offered only marginal improvement over slope alone. The additional flexibility of the probability framework used by SINMAP gave it a slight advantage over SHALSTAB in discriminating unstable terrain.
An Objective Method for the Intercomparison of Terrain Stability Models
Space Dynamics Laboratory
A terrain stability map quantifies the propensity for landslide initiation at a point in space. There are a number of process based models used for terrain stability mapping. SHALSTAB and SINMAP are two such models that combine steady-state hydrology assumptions with the infinite slope stability model to quantify slope stability. Inputs are digital elevation data from which slope and drainage area are evaluated as well as hydrologic and soil parameters. SHALSTAB quantifies terrain instability in terms of the critical effective rainfall required to trigger pore pressure induced instability. SINMAP quantifies terrain stability in terms of the probability that the infinite slope stability model factor of safety is greater than one over uniform probability distributions quantifying the uncertainty in model parameters. These are two models with similar physical basis but they use different indices to quantify instability. Because of this the relative performance of these models, when compared to observed landslide locations, is difficult to assess. A new statistic based on the cumulative fraction of the stability index at locations of observed landslides, relative to the cumulative fraction of the stability index over the terrain has been developed. This statistic quantifies the discriminatory capability of a stability map in terms of the degree to which a terrain stability map is successful in "capturing" observed landslides in areas mapped as unstable while minimizing the extent of these areas. This statistic provides a way to compare models that use different stability indices and is a quantity that can be used for the calibration or optimization of model parameters using observed landslide initiation locations. This is useful when physical data necessary to estimate model parameters is limited. We show results where the performance of these models that use different indices are compared using data from the Chetwynd area in east central British Columbia where 696 landslide initiation locations were mapped. We found that in this particular region slope was the dominant variable that discriminated terrain instability. The optimally calibrated SHALSTAB and SINMAP models that also use drainage area as a predictor offered only marginal improvement over slope alone. The additional flexibility of the probability framework used by SINMAP gave it a slight advantage over SHALSTAB in discriminating unstable terrain.
https://digitalcommons.usu.edu/runoff/2004/AllAbstracts/16