| Uncertainty is widely existed everywhere in the world. It is commonly happened in the hydrogeology study field. Especially to the modeler, uncertainty can be met every single step in the modeling process. It is very important to analyze the uncertainty of groundwater numerical simulation. In this study, three critical simulation problems are represented to be studied. They are sample effectiveness, parameter field with trend and build the complex regional model in Daxing overlapping area and the uncertainty of the model. In this paper, the number of samples and the distribution of samples are studied to analyze the sample effectiveness. Monte Carlo method is used to study the influence of trended parameter field to the numerical simulation.In the sample effectiveness study, a synthetic case with three different rock types has been created. They are coarse sand with hydraulic conductivity 40m/d, mid-coarse sand with hydraulic conductivity 30m/d and fine sand with hydraulic conductivity 20m/d. The synthetic case is 50 rows and 50 columns; the cell size is 100m×100m. By analyzed the sythentic case, we found that in the beginning of the sample increasement, the simulation error was decreased rapidly. When the samples reached 300, the decreasement of simulation error was slow down. After over 1000, the decreasement of simulation error presented a stable state. The sample locations can also make an obvious impact on the lithology distribution simulation. The uniform random sampling can achieve an ideal simulation result. The mean error is about 5.4%. Nearest Neighbor Index (NNI) is an indicator to present the heterogeneity of samples. With the increasment of the NNI, the error will be decrease. After NNI is greater than 0.9, the error will be less than 10%.In this study, we compared the influences on the simulation of stochastic hydraulic conductivity field and trended stochastic hydraulic conductivity field. The trended stochastic hydraulic conductivity field can reduce half of the simulation error compared to the one without trend. Of course, the number of samples and the set of trend parameter played a very important role in the simulation process. With the increasement of samples, the improvement of simulation result gained by the trended hydraulic conductivity field would be decreased. And after the trended process on the hydraulic conductivity field, Ucode can be used for improving the accuracy of the simulation and make the sensitivity analysis.Daxing overlapping karst water area has a very complex geologic structure. In this study, we built a five layers model for the Daxing overlapping karst water flow system. The model is well reflected the practical situation of the study area. It can be used to evaluate and predict the change of the local groundwater resources. After created the model, we also analyzed two critical uncertainty problems in Daxing overlapping area.By the uncertainty analysis, we found that the Nanyuan-Tongxian fault is a water resistant fault. And Yongding River fault is water-blocking at the boundary, but is water conductive in the vertical. Trended hydraulic conductivity field has made a remarkable improvement of water level fitting in the area with coarse sand and gravel, but made a tiny different in the area filled with fine sand and clay. The reason is fine sand has a narrow change interval of hydraulic conductivity and the coarse sand and gravel has a wide change interval of hydraulic conductivity. It can be seen that the trended hydraulic conductivity field is suitable for the coarse sand and gravel area when the fitting value is the water level. For the trended hydraulic conductivity field influence on the solute model, it still neends some further studies. |
PDF Full Text Request