Integrated Simulation Modeling Approach for Investigating Pore Water Pressure Induced Landslides

17 Soil pore water pressure analysis is crucial for understanding landslide initiation and prediction. 18 However, field-scale transient pore water pressure measurements are complex. This study investigates 19 the integrated application of simulation models (HYDRUS-2D/3D and GeoStudio–Slope/W) to analyze 20 pore water pressure-induced landslides. The proposed methodology is illustrated and validated using a 21 case study (landslide in India, 2018). Model simulated pore water pressure was correlated with the 22 stability of hillslope, and simulation results were found to be co-aligned with the actual landslide that 23 occurred in 2018. Simulations were carried out for natural and modified hill slope geometry in the study 24 area. The volume of water in the hill slope, temporal and spatial evolution of pore water pressure, and 25 factor of safety were analysed. Results indicated higher stability in natural hillslope (factor of safety of 26 1.243) compared to modified hill slope (factor of safety of 0.946) despite a higher pore water pressure 27 in the natural hillslope. The study demonstrates the integrated applicability of the physics-based models 28 in analyzing the stability of hill slopes under varying pore water pressure and hill slope geometry and its 29 accuracy in predicting future landslides. 30


GeoStudio-Slope/W model 139
GeoStudio-Slope/W model is developed based on the general limit equilibrium (GLE) formulation 140 (Fredlund & Krahn, 1977). This formulation is based on two factors of safety equations; (a) the factor 141 of safety with respect to moment equilibrium and (b) the factor of safety with respect to horizontal force 142 equilibrium (Spencer, 1967). Where f(x) is the function describing the distribution of internal forces, λ is the percentage of the function 146 used, E is the interslice normal force, and X is the interslice shear force. 147

Integration of HYDRUS-2D/3D with GeoStudio-Slope/W model 148
The integrated application of two different models was performed to utilize the advantages of HYDRUS-149 2D/3D in its accurate estimation of the pore water pressure with the stability analysis capabilities of the 150 GeoStudio-Slope/W module. HYDRUS-2D/3D solves the water flow in the soil using a finite element 151 formulation. A finite element mesh was generated in the soil domain of the hillslope by dividing the 152 flow region into quadrilateral or triangular elements (Fig. 5). Once the water flow simulations were 153 carried out using HYDRUS-2D/3D, the pore water pressure at the nodes that form the corners of the 154 elements was extracted for discrete time intervals. The time variable pore water pressure distribution 155 was mapped into the GeoStudio-Slope/W model corresponding to the discrete-time intervals and spatial 156 locations. The slope stability analysis in the GeoStudio-Slope/W model was then carried out based on 157 these pore pressure distributions. 158

Model setup 159
The input data required for simulations using HYDRUS-2D/3D and GeoStudio-Slope/W were (a) cross-160 sectional details of the hill slope, (b) soil physical and hydraulic properties, and (c) initial and boundary 161 conditions. The cross-sectional details of the hill slope and soil properties in the study area were obtained 162 based on the field investigation to examine the causes of repeated extreme heavy rainfall events, 163 subsequent floods, and landslides in Kerala (Kerala Planning Board, 2019; Choudhury et al., 2019). The 164 geometry of the hill slope in Case1 (with road cut) and Case 2 (without road cut) is shown in Fig. 2. The 165 average angle of elevation is 24.8˚. The maximum depth of the soil (shown in yellow color in Fig. 2) 166 above the rock (shown in grey color in Fig. 2)  Pore connectivity parameter, l 0.5 Shape parameters, α and n α = 1.37 m -1 , n = 1.4027

179
These parameters were obtained using the neural network prediction of soil hydraulic properties using 180 the Rosetta Lite V.1.1 (Schaap et al., 2001). The specific weight of water was considered as 10 KN/m 3 . 181

Initial and boundary condition 182
In both cases (Case 1 and Case 2), a hydrostatic pressure head distribution was considered in the soil 183 domain at the beginning of the simulation. The left boundary and the bottom of the soil layer (or the top 184 of the rock layer) were considered to have a no-flow boundary. A seepage boundary was given at the 185 extreme right slope of the domain for a depth of 2 m (Fig. 3). In the seepage boundary, when the node 186 next to seepage face becomes saturated, water is immediately removed by overland flow, which in 187 HYDRUS-2D/3D is considered to be removed from the system. corresponding to the rainfall event on that day. On 9 th August 2018, 253.6 mm/day rainfall was received 220 in this area. A sudden increase in volume of water in the domain was observed from 8 th to 9 th August 221 2018 for both cases (Fig. 6). Though the rainfall on 10 th , 11 th , and 12 th August was less than the rainfall 222 on the 9 th August, volume of water in soil did not show a considerable decrease. This indicates that the 223 volume of water added to soil due to the rainfall on 9 th August 2018 was drained at a slow rate in the hill 224 slope. This was mainly because of lateritic soil in this region with clay and slit particles which retain The factor of safety (FoS) is a crucial indicator of slope stability and is defined as the ratio of the resisting 255 force to the driving force along a failure surface. An FoS equal to or greater than one represents that the 256 slope is stable, and a value less than one represents likely failure. The FoS of the hill slope corresponding 257 to the pore water pressure distribution (determined using HYDRUS-2D/3D) in the soil in August 2018 258 was simulated using the GeoStudio-Slope/W. 259  sections 3.1 and 3.2). It was on this day 268 the actual landslide occurred in this location (a picture of the landslide is shown in Fig. 1). The weight 269 of the soil, geometry of the hill slope, and the moisture in the soil in Case 2 were such that the resisting 270 moment in the slide was greater than the activating moments, which resulted in FoS>1. This shows that 271 the natural slope of the hill was able to prevent rainfall-induced landslides in the study area. Figure 9 shows the slip surface with minimum FoS for Case 1 (FoS= 0.946) and Case 2 (FoS= 1.243)   Case 1 and 2. 288 Figure 10 shows the pore water pressure distribution in the soil for Case 1 and 2 on 16 th August 2018 289 simulated using HYDRUS-2D/3D. The hill slope in Case 2 was subjected to larger pore water pressure 290 than Case 1, with a maximum of 23 KPa and 31 KPa for Case 1 and 2, respectively. Figure 11 shows 291 the pore water pressure along the slip surface from 14 th August 2018 to 19 th August 2018 for Case 1 292 and 2. For both cases, pore water pressure increased from 14 th to its maximum at 16 th and then decreased. 293 The pore water pressure is depended on the amount of saturation in the soil. Larger pore water pressure 294 was observed in Case 2 (14.5 KPa at a distance of 7 m) compared to Case 1(12.8 KPa at 6.8 m). It was 295 observed that Case 2 was more stable compared to Case 1 even when the pore water pressure along the 296 slip surface was more in Case 2. This shows that the geometry of the hill slope plays a predominant role Several slope strengthening measures can be adopted to prevent the slope from failing (e.g., anchors and 304 piles, geosynthetic reinforcement, sheet pile walls, etc.). In this study, one of the strengthening measures 305 was studied to improve the slope stability in Case 1 (hillslope with road cut). Model simulations were 306 carried out to analyze the slope stability using a nail reinforcement (Fig. 12). This method of 307 reinforcement is generally used for strengthening the natural slope. Soil nails are included in GeoStudio-308 SLOPE/W by defining the pull-out resistance, representing the amount of stress mobilized per unit area 309 at the interface between the nail and soil. Table 3 shows the nail specifications used in this case study. 310 Table 3. Specification of the nail reinforcement  Figure 13 shows the FoS of the slip surface in Case 1 (with road cut) and the case with road cut and nail 318 reinforcement corresponding to the rainfall in August 2018. It was observed that the FoS has improved 319 after incorporating the nail reinforcement throughout the month. The lowest FoS when there is no 320 reinforcement was observed as 0.946, and the lowest FoS after incorporation of the nail reinforcement 321 was found to be 1.524 on 16 th August 2018. This demonstrates that strengthening measures can be 322 incorporated to improve the stability of this hillslope, and this can be analyzed using the integrated 323 modeling approach. A detailed investigation can be carried out to optimize the strengthening measure, 324 its design, and the related parameters. 325

Conclusions 326
In the context of a large number of landslides worldwide, it is essential to investigate the potential 327 solutions for its mitigation. This requires analysis of the landslide triggering mechanisms. Though 328 several triggering factors exist that independently and combinedly act upon a hill slope, the current study 329 focuses on slope stability analysis based on rainfall-induced pore water pressure in the soil, which is one 330 of the significant triggering mechanisms. A methodology for integrating existing models (HYDRUS-331 2D/3D and GeoStudio-Slope/W) for simulating pore water pressure-induced landslides was developed. 332 As a case study to illustrate the methodology, a hill slope in Munnar, India, was investigated for its 333 stability corresponding to the ERE during August 2018. The stability analysis considered the pore water 334 pressure distribution in the soil corresponding to the daily variation in the rainfall in the hill slope. The 335 volume of water in the hill slope, temporal and spatial evolution of pore water pressure and factor of 336 safety were analyzed and correlated with the actual landslide that occurred in the study area. It was 337 observed that the slope was stable (with FoS equal to 1.243 ) when there was no road cut in the natural 338 slope of the hill, whereas the slope failed on 16 th August 2018 in the case with road cut (with FoS equal 339 to 0.946). The integrated application of the simulation models (HYDRUS-2D/3D and GeoStudio-340 Slope/W) effectively predicted the landslide that occurred in the study area on 16 th August 2018. The integrated model application also helped analyze the importance of the hill slope geometry in resisting 342 forces that drive the initiation of a slide. Though the pore water pressure was found to be more in Case 343 2 (without road cut) compared to Case 1 (with road cut), it was Case 1 that failed compared to Case 2. 344 A similar simulation modeling approach can be utilized for predicting landslides by anticipating extreme 345 rainfall conditions. The study also demonstrated the analysis of one of the strengthening measures (nail 346 reinforcement) for improving slope stability in Case 1 (hillslope with road cut) using the integrated 347 modeling approach. 348