5:15 PM - 7:15 PM
[MIS09-P11] Monitoring groundwater dynamics with using self-potential tomography
In conventional research of the landslide prediction, researchers have considered relationships between slope failure caused by rainfall and hydrological and mechanical phenomena such as incline and ground displacement that occur under slope before landslide . Monitoring this kind of phenomena requires the observation of changes that occur in the groundwater and soil inside the slope. Then, in Chiba university, the self- potential (SP) method has investigated, which has the advantage that the observation of these changes can be performed over a wide range simply and the cost can be reduced. The SP method is a method measuring the potential generated spontaneously caused by fluctuations of charges due to groundwater flows using electrodes installed in the ground. From the results of laboratory experiments to date, it has been confirmed that there is a relationship between water flows and soil layer displacement and SP fluctuation. In sandbox experiment, two-dimensional tomography of charge density and pressure head was created from the observed natural potential values, and the electrokinetic phenomenon was verified by two-dimensional simulation of groundwater flow. In these two-dimensional analyses, the water flow is assumed to be homogeneous across the depth of the sandbox. However, the two-dimensional approximation resulted in a large error between the simulated and measured values of the self-potential. Then, three-dimensional simulation and self-potential tomography algorithm have developed.
Then, in this study, we demonstrated the necessity and effectiveness of three-dimensional analysis and models in a rectangular sandbox experiment measuring 200×20×60 cm (width×depth×height). Specifically, actual data from the sandbox experiment was compared with self-potential generation simulations using forward problems of 2D and 3D flow models, and groundwater flow estimation from self-potentials using 2D and 3D self-potential tomography was compared with the model and experimental data. In the water injection experiment, the sandbox was laid with homogeneous sand, and water was injected from the lower part of the sandbox, simulating the groundwater flow. Parameters related to the sand of the soil layer, such as the permeability coefficient and the electro-kinetic coefficient, were measured in another experiment and set to be constant in the sandbox.
The root mean square (RMS) of the difference between the simulated and observed self-potential values in the water injection experiment was smaller in 3D than in 2D, which was a better result. Next, the difference between the length of the flow velocity of the pressure head tomography obtained by inverse analysis of the SP actual values was smaller in 3D than in 2D compared to the simulation results. It was also found that the flow velocity vector of the 3D reconstruction results had a spread in the depth direction. This shows that the assumption of laminar flow in the 2D approximation is not valid, and it can be said that 3D analysis is essential. Application to multi-layer structures and actual slopes is a topic for the future. Details will be announced at the lecture.
Then, in this study, we demonstrated the necessity and effectiveness of three-dimensional analysis and models in a rectangular sandbox experiment measuring 200×20×60 cm (width×depth×height). Specifically, actual data from the sandbox experiment was compared with self-potential generation simulations using forward problems of 2D and 3D flow models, and groundwater flow estimation from self-potentials using 2D and 3D self-potential tomography was compared with the model and experimental data. In the water injection experiment, the sandbox was laid with homogeneous sand, and water was injected from the lower part of the sandbox, simulating the groundwater flow. Parameters related to the sand of the soil layer, such as the permeability coefficient and the electro-kinetic coefficient, were measured in another experiment and set to be constant in the sandbox.
The root mean square (RMS) of the difference between the simulated and observed self-potential values in the water injection experiment was smaller in 3D than in 2D, which was a better result. Next, the difference between the length of the flow velocity of the pressure head tomography obtained by inverse analysis of the SP actual values was smaller in 3D than in 2D compared to the simulation results. It was also found that the flow velocity vector of the 3D reconstruction results had a spread in the depth direction. This shows that the assumption of laminar flow in the 2D approximation is not valid, and it can be said that 3D analysis is essential. Application to multi-layer structures and actual slopes is a topic for the future. Details will be announced at the lecture.