Study On Risk Early Warning Of Rain-induced Landslide

Landslide is the most common geological disaster in China,which often causes heavy losses.Research shows that most of the landslides were induced by rainfall,especially rainstorm.Rainfall-induced landslides are characterized by wide distribution,high frequency and huge damage.Conducting the early warning research based on a scientific method is the key to reduce such kind of disasters.Considering that rainfallinduced landslide is the result of internal and external factors,the early warning research should combine the landslides assessment and rainfall analysis.However,in addition to the geological attributes of deformation and failure,landslides also have social attributes that affect the safety of life and property.Therefore,considering the element risk of landslide during the early warning will help refine the results of spatiotemporal warning and make disaster prevention work more targeted.In view of this,in order to organize a reasonable early warning method for rainfallinduced landslide,this thesis takes Cili County,Hunan Province,as an example.Through the technics of machine learning,physical deterministic model,statistical analysis and numerical simulation,the risk early warning research of rainfall-induced landslide was carried out.The method flow of rainfall-induced landslide early warning for both regional scale and specific landslide was established respectively,based on the collected data,evaluation accuracy level and actual early warning demand.Which aims to provide method support for the prevention of rainfall-induced landslide.The main research contents and achievements are as follows:(1)Summarize the requirements of rainfall-induced landslide early warning.Considering the differences in basic data details,model methods,and evaluation unit suitability among different research regions.Based on landslide susceptibility assessment,landslide rainfall threshold analysis,construction of rainfall early warning model and landslide risk analysis,the principle and method of rainfall-induced landslide early warning for regional scale and specific landslide are discussed respectively.(2)The internal relationship between geological environmental factors and historical landslides in the Cili County can be quickly established by the method based on machine learning.This method can quickly realize susceptibility assessment in a large area with sufficient samples and further guide landslides prevention.Random Forest is the best evaluation model in the study area among the commonly used machine learning methods.The Fast Shallow Landslide Assessment Model(FSLAM)which considering the physical characteristics of landslide was adopted to carry out the regional landslide susceptibility in Lingyang Town area.This method needs better data acquisition and higher evaluation accuracy.The slope unit automatically divided by the r.slopeunits program as the basic evaluation unit for the Town area in this thesis.Landslide-prone regions can be shown more directly by converting the 90 th percentile value of the grid calculation results inside the units to the slope units susceptibility results.The results based on slope units facilitate further risk warning studies using susceptibility results.(3)Statistical analysis shows that landslides in Cili county area were most significantly affected by the cumulative rainfall of the previous three days,with an optimal reduction factor of 0.6 for effective rainfall penetration.The rainfall threshold could be established quickly by the I-D curve.The boundaries of four warning levels of low/medium/high/very high were assigned to the three days of accumulated rainfall of38.07 mm / 108.48 mm / 244.93 mm,respectively.According to the rainfall values of the antecedent rainfall(Ra)and the event rainfall(Re)in 10,20,50,100 and 200 years return period,the rainfall scenario of the Lingyang Town was constructed.The FSLAM model was used to calculate the proportion of slope units with a probability of failure greater than 0.5 in the study area under different rainfall scenarios(i.e.,the risk degree index Ri).A quantitative relation between Ra,Re,and Ri was constructed and a 3D surface classification of rainfall threshold boundaries was obtained by fitting.Based on the results of the FSLAM calculations,a quantitative rain threshold function is established,and the effects of physical and mechanical effects are explicitly taken into account in the analysis.(4)The results of meteorological early warning in Cili county are better able to correspond to historical landslide disasters,with the accuracy of the early warning reaching 68.96%.By properly classifying the landslide risk index Ri,we have developed a meteorological early warning model for rain-induced landslides in Town areas.Risk warnings for rain-induced landslides have been implemented by combining research on weather warnings and regional vulnerability assessments.We develop an integrated vulnerability index for the units in the Lingyang Town area by considering the difference in proportion and importance of different disaster-bearing objects in the slope units.Regional integrated risk warning results based on vulnerability correction can effectively reduce the impact of high-level unit landslide warnings due to landslide susceptibility assessment and achieve a more reasonable early warning analysis.(5)The rainfall was the most significant factor that induced the development of the deformation of the Chenxiyu landslide.And the landslide is sensitive to rainfall infiltration,and the slope’s metamorphic suction is obviously affected by rainfall infiltration.We determined the soil-water characteristic curve of the top soil of the slope by unsaturated soil testing.Tests also showed that matric suction can increase the shear strength of soil by increasing its cohesive force.Single fracture seepage tests and simulations have shown that fracture width and coverage are key factors in fracture seepage,that fractures on the slope surface are the dominant channel for significant rise in percolation,and that the presence of fractures can considerably affect slope stability.The effect of "progressive rainfall" on Chenxiyu landslide is the most significant,and it is the most unfavorable condition in rainfall threshold analysis.(6)Geo-Studio software was used to analyze the shift of landslide risk under the action of rainfall.The quantitative relationship between rainfall,risk and failure probability was established,and the warning classification was realized according to the landslide control code.We use the DAN3 D software to simulate the unsteady motion characteristics of Chenxiyu landslide.The vulnerability of the Chenxiyu landslide after failure is quantitatively calculated based on the distribution of population and economic disaster bearing bodies in the landslide area.Under the extreme instability scenario,the total population casualty risk caused by the Chenxiyu landslide is 9 people,and the total economic disaster bearing body loss risk is 989,100 yuan.The risk warning of the Chenxiyu landslide was realized under different rainfall recurrence periods of 10 years,20 years,50 years,100 years,200 years.Which combined the simulation results of rainfall and landslide damage probability.