| Potential landslide volume is an important parameter expressing potential landslide magnitude.With potential landslide volume,the landslide hazard and risk assessment can be carried out,and decisions on mitigation plans and budget measures can be made.However,due to the lack of underground information related to potential failure surface,it is hard to make a prediction for landslide volume.Some physically-based modelling methods can be applied to predict landslide volumes by analyzing slope stabilities of slopes and determining potential failure surfaces.However,different models may take different assumptions to define potential failure surface and analyze slope stability,consequently,the predicted landslides and their volumes may be also different.On the other hand,different models have different input data requirements,thus the difficulties for a successful application could also be different.The main objective of this study is to compare two Geographic Information System(GIS)based models: OpenLISEM and Scoops3D,in terms of landslide volume prediction.Due to the difficulties related to the acquisition of input data,a hypothetical dataset has been developed to perform the model comparison.The hypothetical dataset represented a volcanic environment by assigning the properties of volcanic ash soils to the virtual soils.Three scenarios were developed based on the hypothetical volcanic environment.In scenario 1,the initial soil thickness was initially determined to be homogeneous 10 meters mimicking from an explosive eruption where volcanic ash homogeneously rains down and covers the whole terrain.Based on scenario 1,the optimal internal model parameters of OpenLISEM and Scoops3 D were determined,and a critical soil thickness was calculated using OpenLISEM by removing landslides from the terrain,which will later be used in Scenario 2 and 3.Wetting front and pore water pressure caused by groundwater level were considered as triggering factors for scenario 2 and 3,respectively.The landslides simulated by OpenLISEM and Scoops3 D in three scenarios were compared using four comparing strategies: landslide number,location,volume,and area-volume statistical relationship.This study has concluded that input data,as well as the calibration and validation data are essential in modelling,while high-quality data is not a guarantee for performing a good simulation.The main reason is that physical models cannot fully represent the exact geological environment.Assumptions have been made to simplify the simulation,thus the selection of physical models for application should be regarded with caution.The comparison results from sensitivity analysis on internal model parameters reveal that the internal model parameters can significantly influence the landslide volume prediction.These parameters should be very carefully selected when applying these two models.When determining critical soil thickness,it was found that there is no successive failures in OpenLISEM once the potential slope failures are cleared out of the terrain.Whereas there still are slope failures in Scoops3 D when potential slope failures are cleared out of the terrain.This behavior indicates that OpenLISEM is more logical than Scoops3 D.From comparisons under three scenarios,it was found that OpenLISEM and Scoops3 D produce very different landslides in terms of number,location,and volume.Then volumes produced by Scoops3 D are much larger than OpenLISEM,and Scoops3 D tends to produce relatively round landslides.Sometimes,Scoops3 D may cut into the bedrock and produce extremely large volume.This behavior also indicates that Scoops3 D may lead to unexpected results. |
PDF Full Text Request