| Uncertainty is one of the inherent natures of the whole world, and exists in all scientific subjects. It has been a focus to analyze and assess the uncertainties in groundwater modeling. In the recent decade, hydrogeologists have paid increasing attention to Multiple Model Analysis (MMA), making it more and more popular in modeling practices.MMA was applied to study the uncertainties due to the permeability of media including the lithologic character, the random distribution of hydraulic conductivity, and the complicated structure of fractured media. First, we studied a common problem about groundwater seepage to parallel ditches. According to the uncertainty of lithology distribution, a serial of models were constructed to predict the elevation of water table and calculate its confidence interval. The practicablity of MMA was verified by the correspondence between water level variation and lithologic heterogeneity.Then we evaluated the scale effect of grid on solute transport simulation on the basis of a transit well flow. The errors of sampling and interpolation were also taken into account in the process of generation of the random hydraulic conductivity field. The characteristics of observed concentration at three points and the difference of contaminant plume patterns were compared. Results show that the grid scale has the most influence on simulation, the sampling error next, and the kriging interpolation method the least. Additionally, this study pointed out the corrected Akaike's Information Criterion (AICc) is deficient to evaluate the uncertainty of grid scale as well as the randomicity of hydraulic conductivity.Third, three-dimensional steady state flow models under five different scenarios were established within MODFLOW, to predict the water inflow in tunnels of Jinpingâ…¡hydropower station. We discussed the parameterization uncertainty caused by conceptualization of complex fracture structure. UCODE was employed to optimize parameters, the models selected by MMA forecasted water inflows in all tunnels before and after waterproof protections, and the results are 40.35 and 31.19 m~3/s, respectively.Besides the permeability of media, uncertainties of hydrogeological conditions need to be studied. After thoroughly illustrating the uncertainties of conceptualization for karst groundwater system in Taiyuan area, we applied the distribuated hydrogeological model MODFLOW to simulate the attenuation process of discharge rate of Jinci and Lancun springs from 1956 to 1994. Both climate change and human activities are the main factors concerned.Understandings obtained including: 1) it is better to take the East Mountain area into consideration to be an entire problem domain. 2) there is no so-called variable boundary between Jinci and Lancun groundwater subsystems. 3) uncertaity resulted from different degrees of depth-decay of hydraulic conductivity is greater than that caused by various recharge patterns. 4) Jinci springs are more sensitive to changes of interannual rainfall; both the reduction of precipitation and the increase of groundwater exploitation caused the springs stopped flow.It is often neglected by modelers that the uncertainty of grid scale plays important role to numerical simulation. We conducted experiments to explore the methods to assess this uncertainty. Different sizes of grid cell with respect to the same hydrogeological problem can result in resolution variation of hyaulic conductivity, and lead to other design changes related to elements such as wells, rivers, and observations etc. MMA results using AICc show that the traditional information criterion is not suitable for evaluation of grid scale effect. Therefore, we proposed a new criterion named Grid Information Criterion (GIC), and proved it is an improvement of AICc by a numerical test.
|
PDF Full Text Request