地質與地層模型不確定性之量化及其應用之整合型研究(2/4)

A reasonable geological model plays a critical role in development of geo-resources, protection of geoenvironment, and mitigation of geo-hazards. However, the uncertainty associated with established geological models is often not well characterized or effectively communicated to engineers involved in risk-based engineering design. This study addresses the challenges of quantifying geological model uncertainty and propagating this uncertainty in selected applications of national importance. Specifically, the study focuses on the impact of sedimentary environments on the scale of fluctuation in stochastic Markov random field theory, which is used for modeling subsurface stratigraphic configurations. Four specific applications are considered: (i) evaluation of earthquake site effects (Vs30), (ii) assessment of liquefaction potential, (iii) groundwater anagement, and (iv) failure probability of rock slopes. The Taipei Basin is selected as study region. The main achievements are outlined as follows: (1) In the previous year, a coupled geological and parametric random field modeling tool was developed and used for liquefaction analysis in a small area of the Taipei Basin. In the current year, a larger test area (approximately 20 km2) was selected for liquefaction analysis and calculation of Vs30. The liquefaction and Vs30 maps generated clearly demonstrate the uncertainties and show superiority over traditional methods. (2) Last year, the maximum likelihood estimation was developed to calibrate critical parameters for random field modeling. This year, these parameters were calibrated using borehole data near test area in the Taipei Basin, and the calibrated parameters were employed for random field modeling. Furthermore, an in-depth study was conducted on the directional and scale effect of parameter calibration, revealing that the length of the profile used and the borehole density along the profile have an impact on calibration results. This finding is significant and will have a substantial influence on the global calibration process. (3) In the previous year, the Vs30 map and groundwater flow modeling for the Taipei Basin were produced. These maps have undergone extensive improvements this year, considering both geological and parameter uncertainties asmentioned earlier. The main improvement lies in the identification of the bedrock surface and the top of thegravel formation underlying the Songshan Formation. Additional interpolation boreholes were added throughextrapolation of data to a depth of 30 meters. New data was utilized to develop a transformation function used forpredicting shear wave velocity. These refinements significantly reduce the prediction uncertainty of Vs30 andthe groundwater table. The updated groundwater table has been used for liquefaction and Vs30 mapping this year. (4) In the previous year, the development of a geological model for rock slope stability analysis was carried out through fieldwork, borehole data, and LiDAR data. This year, the focus shifted towards evaluating the influence of uncertainty associated with weak zones on slope failure risk. Additionally, the hydro-mechanical coupling effect on rock slope stability was assessed, revealing that the fracture system dominates hydraulic conductivity and groundwater flow. Various numerical models will be utilized in the coming years to evaluate the influence of geological, ground, and geotechnical model uncertainties on rock slope stability.