A Hazard Assessment Model For Regional Shallow Landslide And Its Application

China is a geohazards-prone country.Among various geohazards,landslides account for the largest proportion,and many of them are characterized by large-scale,complex mechanisms and high risks,which are typical worldwide.Influenced by heavy precipitation in rainy seasons and periodic fluctuation of reservoir water level,the Three Gorges Reservoir is an area where landslides are concentrated developed in China.These landslides post huge risk to residents and their properties.Hence,regional landslide hazard assessment is critical for landslide risk reduction of this region.Additionally,in the present times that global environments change quickly,landslide activities are affected by various dynamic factors,among which the land use change and climate change are the most evident factors.Therefore,assessment of landslide hazard under different environmental modes are significant for knowing landslide risk and deciding the alternatives for risk reduction and mitigation.Benefit from the fast development of geographic information system(GIS)techniques,various landslide spatial modelling methods have been proposed and applied.Compared with the other types of models,deterministic models can combine the landslide mechanism and physical properties of rock-soil masses,which is helpful for knowing the failure processes of slopes.However,there are still several issues existing on this topic.One is that the computational time is long because many present models use Richards equation for the rainfall infiltration in unsaturated soil.The other one is that many models cannot address the uncertainty of soil spatial distribution,so they only use a specific value for input parameters.These issues limit the application and extension of deterministic models in large-scale landslide hazard assessment.Considering these issues,this study proposed a deterministic model called FSLAM(Fast Shallow Landslide Assessment Model)to conduct fast landslide hazard analysis at a regional scale.The Chinese Three Gorges Reservoir area and the Val d'Aran region in the Pyrenees(Spain)were taken as the study areas,respectively.The modelling process of FSLAM was described,and its application and potential prospect in regional landslide hazard assessment were discussed.The main findings and conclusions are as follows:(1)The FSLAM model principles were introduced,as well as the code programming based on Fortran language.There are three parts in the FSLAM model: the first one is the geotechnical model by using the infinite slope model based on the Mohr Coulomb theory.The second part is the simplified hydrology modelling considering lateral flow and vertical flow of underground water.We combine the antecedent effective rainfall and event rainfall to calculate the water table.The third one is the runoff model,which applies the “curve number” method proposed by United States Geological Survey(USGS)to compute the peak discharge.Meanwhile,the FSLAM model includes the stochastic method,which allows to input the range of parameters,not only a specific value.This addresses the uncertainty of soil properties.(2)The importance of parameters in the model were calculated by the sensitivity analysis.There are 12 parameters in the FSLAM model,including 2 geometry parameters(slope angle,drainage area),6 soil properties(density,porosity,hydraulic conductivity,soil depth,soil cohesion,friction angle),2 vegetation-related parameters(root cohesion and curve number),and 2 rainfall-related parameters(antecedent effective rainfall and event rainfall).A homogenous slope with 25°angle and 10-m resolution was created to compute the sensitivity of the factor of safety to each parameter.The results show that soil cohesion,root cohesion and friction angle are the most importance input parameters,compared with the other parameters.The Pareto method was also used to analyze the importance of input parameters with the significance level of 0.99,and the results were consistent with that of the sensitivity analysis.(3)The Wanzhou County in the Three Gorges Reservoir area was taken as the study area,and the shallow landslide hazard under rainfall conditions with different return periods was assessed.The remote sensing dataset was used to extract the 30-day antecedent rainfall and daily extreme rainfall.A 7-day temporal uncertainty of initial landslide inventory was determined by analyzing the relationship between shallow landslides and rainfall data.After incorporating the temporal uncertainty of landslide inventory,we utilized the water balance theory and statistical distribution model to conduct regional rainfall mapping.Shallow landslides during 1995 and 2005 were used to calibrate the parameters,and the FSLAM modelling was performed.The results showed that the computational time for Wanzhou with about 7.5 million cells was less than 2 minutes.The landslide hazard in the area increased evidently when the return period level of the input rainfall increased.However,the change of landslide hazard in different villages varies.The overall hazard level in middle and west parts of the region increased more greatly than the other areas.Overall,the FSLAM model can be applied to effectively assess and predict the landslide hazard under rainfall conditions with different return periods.(4)The Val d'Aran region in the Pyrenees(Spain)was taken as a study area,and the FSLAM was used to analyze the impacts of future environmental changes on regional landslide hazard.The landslide episode in 2013 due to heavy rainfall and snowmelt was used as a case,and the peak discharge values at two gauge stations were used to calibrate parameters.The analysis of trend of future rainfall showed that the extreme rainfall will increase evidently in the future.The daily rainfall will increase between 14% and 26% by the end of 21 th century assuming a return period of 100 years.The simulation of land cover change in the region showed that the area of forest will increase in the future whereas the areas of shrubs and grassland will decrease.The FSLAM model was used to quantify and compare the impacts of the two changes on landslide hazard.The results indicated that the stabilizing effect due to the increase pf forest(increase of root cohesion)was larger than the destabilizing effect of increase of rainfall.Hence,the overall stability condition in the region will increase when the two changes are considered at the same time.