Study Of Landslide Deformation Regular Based On The Impoundment Process In The Three Gorges Reservoir

The three gorges reservoir area has always been incidence of geological disasters, which due to its complicated natural conditions and geological conditions, and the strong human engineering activity. Since the water formally in June 2003, the three gorges has high filling after nearly three years. When the reservoir has entered into the phase of normal operations, water level only wave between 145m to 175m. The Raising and fluctuating of the water level contribute to the soaking and abrasion of water, ability to enhance bank rebuilding, reduce the stability of bank slope, even produce new landslide area, such as the Qian Jiang ping landslide happened in 2003 and the Liang Shui jing landslide happened in 2009.Meanwhile, the scale and intension of human engineering activities is increasing as the development of the society and economy. Artificial excavation, transport, stacking and other activities are contributed more than the natural geological processes to change the geological environment. When building the immigrant new town, rehabilitation of highway and other infrastructure engineering in the three gorges reservoir area, the activities, such as large scale excavation, slope cutting, handling and accumulation, etc., have a deep influence on the stability of the slope. For example, in 1993, in order to enlarge the immigrant building area, the government of Yun yang new town, violated to increase ramp loading, artificially induced the occur of landslide. In 1995, during the relocation process, the government of Ba dong, ignored the protection of slope, triggered the ancient landslide revival. The increasing scale and intension of human engineering activities, have an increasing serious impact on the natural environment, has become one of the important inducing factors of geological disasters in the area.Therefore, researching on the three gorges reservoir area landslide monitoring, early warning and assessment technology, mining the spatial distribution regularity between the control factors, such as strata rock group, geological structure, landform, hydrology condition, etc.; And on the basis of the research, study on the process of water storage, the landslide deformation evolution characteristics under the influence of inducing factors, such as rainfall and water level fluctuation and human engineering activities, implement the landslide deformation trend forecast analysis, which is required by the disaster prevention and mitigation in reservoir area and the guarantee of normal operation of the reservoir.This paper takes the theory of landslide prediction and landslide deformation mechanism as the foundation, takes the geographic information system, remote sensing, probability statistics and data mining methods as the main technical means, selects the three gorges library first area-Zigui to Badong as the study area, on the basis of the field investigation, related data collection and existing research results, start on the analysis of the regional landslide development features and typical professional monitoring, research from the three gorges reservoir water storage before 1993, in 2003 to 135m water storage,2006,2006 m water storage and completed in 2008,2008,175 m experimental landslide induced factors in the process of water after the water storage variation characteristics, including land use, the change of reservoir water and rainfall, explore the different types of landslide deformation law under the inducing factors, finally realizes the landslide deformation trend forecast analysis. The main results are as follows:(1) Analyzed the landslide spatial distribution and size characteristics in the study area.According to analysis the spatial distribution characteristics and scale of 275 catalog landslides in the study area, including the characteristic parameters, such as the type of landslides, the spatial distribution density, area, front and back edge elevation, etc., found the earthiness landslides in the study area size is given priority to with medium, and rocky landslides is mainly belong to large landslides; Flange height before and after the landslides are affected and controlled by human engineering activities; Landslide have regional space inhomogeneity and integration.(2) Mining summarizes the different types of landslide spatial development law in the study area.Using decision tree algorithm to mining of landslide development regularities in the study area, the results show that the soil/rock landslides in the study area was mainly influenced by basic geological conditions, including the formation rock group, fault distance and slope structure. In the study area, earthiness landslides are mainly distributed in clastic intermediated -acidic class, within 320 elevation, mainly for the northwest slope, and rock slope landslide are mainly distributed in v hill, elevation 280m-480m. In the study area, soil landslide disaster in the study area are mainly medium scale, and are mainly distributed in close of drainage, large range human engineering activities. The key elements of earthiness landslides in different scale development are basic geological and hydrological conditions. Rocky landslide mainly for large-scale landslide, compared with the soil landslide scale development, rocky landslide development scale in the study area affected more by the slope structure, elevation, and slope.(3)Analyzed the landslide deformation characteristics using professional monitoring data in the study area.Based on the statistically analysis to 61 professional monitoring landslides in the study area, found that professional monitoring landslides are mainly medium or deep landslide, the plane form is mainly like tongue, most development is step profile form. Through the analysis of four different types of landslide deformation characteristics, found that Baijiabao landslide and the left front of Tanjiawan landslide present a ladder-like development trend influenced by rainfall and reservoir. The cumulative displacement of Muyubao landslide and Daping landslide presents linear monotone increasing trend. The deformation of Muyubao landslide is mainly related to water level fluctuation, the displacement rate in the west of Daping landslide is big, is mainly affected by rainfall.(3) Carried out the change rule of land use under different water phases in the study area.Using high, medium resolution satellite images and aerial data of the study area in 1993 (before water storage),2003,2006,2008,2010 and 2013, considering spectral characteristics, texture characteristics and NDVI, DEM and slope to find land use information and analysis the land use change rule in the process of filling in the study area, to explore the influence of land use change on the landslide stability. According to the analysis, between the Three Gorges Reservoir impoundment 10 years ago, the land use changes are mainly focused on the conversion of arable land, grassland and woodland, where most shrubs, grassland had been converted to cropland. And ten years later, the change is mainly farmland to grassland and bush water, followed by the conversion of woodland shrub to grassland and farmland.(4) Using association rule to mine the influence of land use change on landslide stability.By studying, during 1993-2003, in the study area, the land use of soil and rock landslides show a strong relevance with arable-shrub grassland and woodland-arable, in addition, soil landslide are associated to arable-water body. Studies show that landslides are mainly within the scope of arable land changes, namely strong influenced by the human activities. Finally, during 2003-2013, landslide stability of the study area is affected by human activity water level changes together.(5) Using association rules analysis the deformation trend of soil landslides (the Baijiabao landslide and the Tanjiawan landslide) and rock landslides (the Muyubao landslide and the Daping landslide) in the study areaBy studying, found that the four landslides in the study area are affected by landslide rainfall and reservoir water level together, especially when the water level drops quickly, landslides may occur rapid deformation. The main reason is in the study area, soil landslide body substance are mostly poor permeability clay, mudstone, when the water level drops quickly, groundwater and reservoir water level to form a positive gap, pointing to the outside of the hydrodynamic pressure the landslide, which is not conducive to landslides stability. The rock landslides are affected more by rainfall than water level, mainly due to the rock landslides material are mostly good permeability sandstone, when there is a rapid rainfall, landslide weight increases, cause widespread groundwater level uplift, make pores of osmotic pressure, while softening the sliding zone soil to reduce the shear strength and impact landslide stability.(6) Analyzed the power-law distribution characteristics of deformation rate of monitoring landslide.To analysis the speed deformation of these Four typical landslides, found that the while speed of displacement is out of the rate, there is a linear between the speed of deformation and its frequency, four landslides fitting linear correlation coefficients of above 0.94, shows that between landslide rate and distribution of frequency power law rules, the landslide deformation rate has a god self-similarity and scale-free property, which provide an evidence for the critical phenomenon in the evolution process of the landslide.(7) Carried out the Baishuihe landslide deformation prediction based on rough set and particle swarm optimization support vector machine (SVM)By analyzing the characteristic of Baishuihe landslide deformation, this paper has found that continuous heavy rainfall and water level drop is the main influence factors of landslide deformation. Based on the influence factors such as rainfall and reservoir water level and the feature of landslide deformation response relationship, using the theory of rough set attribute to reduction of 10 factors on the deformation of landslide to find the nuclear subset, and then establish subset based on the particle swarm optimization support vector regression model to make the prediction of landslide displacement rate is. Validated, mean absolute error and relative error of the test sample is 0.234 and 0.891% respectively, and trend prediction results and the measured values are basically identical. Through the analysis of the landslide deformation rules of rough sets, this paper has found the Baishuihe landslide is at the uniform deformation stage, only when the rainfall and water level is heavy, the landslide is in the stage of accelerating deformation. When the monthly average rainfall is less than 10.598 mm, continuous rainfall is no more than 36.9 mm, the rate of decline of reservoir water level is below 0.235 m and the monthly average reservoir water level is below 146.59 m, the Baishuihe landslide is in uniform deformation stage.While the Baishuihe landslide deform is in accelerated phase, the average rainfall in the month is no less than 10.598mm, maximum continuous rainfall is greater than 36.9mm and reservoir water level dropped by an average rate of more than 0.235m. When the reservoir water changes consistent, large rainfall may also affects the deformation of landslide. Even less rainfall, the large change of the water level will prompt landslide deformation.(8) Carried out Shuping landslide groundwater level change rules based on the classification and regression tree.By analyzing Shuping landslide groundwater change and the factors such as rainfall, water level response relationship, this paper has found that the change of underground water level and water level was consistent with each other. The fluctuation of underground water level is the same as the change of water level. When the water level rise, the underground water level rises, when the water level decreases, underground water level decreases too. Compared with reservoir level, rainfall has less effect on the landslide at the front of the underground water level. When the water level change is slow, and rainfall is large, the underground water level will show uplift. All above show that the dynamic changes of the groundwater in front of the Shuping landslide is affected by factors such as water level and rainfall comprehensive. Based on the influence factors such as rainfall, water level and the change of underground water level dynamic response relationship, this paper has selected appropriate subset of the underground water level change. The classification and regression tree model has been constructed by subset to forecast the ground water level. As a result, the predictive result of the test sample is consistent with the actually measured value, the mean absolute error and the relative error is 0.28 m and 1.15% respectively.