Seepage-slope Stability Analysis Under The Influence Of Soil Hydraulic Parameter Uncertaint

Triggering of landslides under extreme rainfall are related to lower soil suction and weaker shear strength in response to more infiltration and higher pore water pressure propagation to the potential slip surface.Hydromechanical models can be formulated based on two mechanisms: soil moisture dynamics affect self-weight and total stress,and variations of pore water pressure affect effective stress.Particularly,for unsaturated soil hydrology,soil moisture content and pore water pressure are commonly simulated with either analytical or numerical solutions,and the simulation accuracy is governed by the parameterization strategy of soil hydraulic properties.In hydrological modeling,uncertainties often originate from theoretical flaws in model conceptualizations,errors in initial/boundary conditions and observations,and selection of parameter sets.In order to explore the uncertainties of soil hydraulic parameters affect the simulations of unsaturated soil hydrology,two parameter optimization approaches were proposed based on the Generalized Likelihood Uncertainty Estimation(GLUE)method.In Strategy I,soil hydraulic parameters were randomly drawn from the entire parameter space(refer to GLUE-random);while in Strategy II,soil hydraulic parameters were constrained by Rosetta pedotransfer function(refer to GLUE-Rosetta).Firstly,we conducted a synthetic numerical experiment for nine typical soil types,the soil moisture simulated using typical parameter sets were regarded as true,which assisted obtaining the posterior distribution of soil hydraulic parameters based on either GLUE-random and GLUE-Rosetta to explore the uncertainties of soil hydraulic parameters.Then the proposed approaches were validated in a landslide deposit in Yindongzi valley,Dujiangyan Country,Sichuan Province of China.The field monitoring of soil moisture content and rainfall was used as an input for a hydromechanical analysis integrated with an infinite slope stability approach to investigate the uncertainties of soil hydraulic parameter in the landslide material.Particle Batch Smoother integrated with GLUE-random and GLUE-Rosetta to explore the uncertainty of joint estimation of soil hydraulic parameters and soil moisture dynamic.The main conclusions are as follows:(1)Both of soil hydraulic parameters selected by GLUE-random and GLUE-Rosetta shows significant uncertainties and equifinality phenomenon in parameter sets.Compare with the prior soil hydraulic parameter range,the posterior PDF of soil hydraulic parameters selected through GLUE-random have wider range indicated significant uncertainties.GLUE-random and GLUE-Rosetta selected a number of soil hydraulic parameter sets leading to satisfying performance in simulated soil moisture dynamics.However,the results suggested that GLUERosetta outperformed GLUE-random,which may provide evidence that GLUE-Rosetta can obtain better simulation of soil moisture dynamics as well as reduce the model uncertainties.We also provided numerical experiments to exam whether the selected soil hydraulic parameters by GLUE-random and GLUE-Rosetta also performed well in other rainfall conditions.The results suggest GLUE-Rosetta also outperformed than GLUE-random.Therefore,GLUE-Rosetta can also reduce the uncertainties of soil hydraulic dynamics by reducing the uncertainties of soil hydraulic parameter.(2)Combined the soil hydraulic parameters generated by GLUE-random and GLUERosetta with infinite slope stability analysis approach to analysis the slope stability.The comparison of GLUE-random and GLUE-Rosetta to estimate pore water pressure and slope stabilities show that GLUE-Rosetta constantly yields more reliable estimates in pore water pressure to support soil slope stability analysis.GLUE-Rosetta can reduce the uncertainties of soil hydraulic parameters to provide reference in current early-warning system for landslide disasters.(3)Using the data assimilation method of Particle Batch Smoother to estimate soil hydraulic parameters through updating by time windows.The results show that Particle Batch Smoother can reduce the uncertainties of soil hydraulic parameters,soil moisture dynamics and safety factory.