【作者】 马雷；

【导师】 钱家忠；

【作者基本信息】 合肥工业大学 ， 矿物学、岩石学、矿床学， 2013， 博士

【摘要】 基于区域大量的密集分布的钻孔岩性数据，实现多尺度、多精度需求与数据联合驱动的地下水流和溶质运移数值模拟，可为解决区域地下水资源问题、环境问题、生态问题以及地下水灾害问题等提供理论支持。然而，由于含水介质的非均质性，含水层水文地质参数具有空间变异性和尺度效应，导致地下水流和溶质运移模拟的结果往往会与实际有较大的偏离。本文基于较高密度的钻孔岩性数据建立多尺度非均质多孔介质模型，根据建立的多孔介质模型获取水文地质参数场，并将参数场应用于地下水数值模拟中的研究十分有意义。本文主要开展了以下研究：以大量的密集分布钻孔岩性数据为基础，分析了含水介质的非均质性，利用Voronoi图方法实现了钻孔数据的密度和面积分布概况的分析；提出了针对密集分布钻孔数据尺度综合方法；研究了多孔介质的转移概率地质统计随机模拟，提出了侧向转移概率计算方法；分析了侧向转移概率对模拟结果的影响；对比分析了尺度综合虚拟钻孔与原始钻孔建立的转移概率地质统计三维随机模型；基于尺度综合虚拟钻孔和真实钻孔的多尺度多孔介质模拟和确定-随机耦合模拟；基于多孔介质随机模型的水文地质参数提取；分析了基于转移概率地统计模拟提取参数的稳定性与实现个数的关系；并将建立的多孔介质模型与地下水流和溶质运移数值模型结合，应用于地下水数值模拟，从而为区域地下水管理和分析提供支持，得出以下主要结论和成果：（1）基于密集分布钻孔岩性数据，得出了实例研究区含水介质的的两个主方向空间变异相关距离分别为5500m和12000m；提出了对密集分布钻孔岩性数据进行尺度综合生成虚拟钻孔的方法，生成的虚拟钻孔较好地反映了含水介质大尺度上的结构性特征。基于尺度综合虚拟钻孔建立的含水介质随机模型，可以较好的反映含水介质粗尺度上的总体分布特征，在强非均质区域模拟结果表现为连续等效介质。（2）侧向转移概率对模拟结果的影响分析显示，如果侧向与垂向介质平均长度比值估计过小，导致介质分布随机性强，不具有连续性；如果某种介质比值估计过大，则导致该介质水平方向上连续性过强，本文提出的基于垂向分层平面连续地质类型假设的侧向转移概率计算方法，可以有效的根据钻孔数据得到侧向转移率矩阵，并且模拟效果较好。开发了基于GIS的集钻孔等数据管理与转移概率地统计建模与一体的程序。（3）转移概率地统计模拟结果评估显示，在弱非均质或者相对均质区域，模拟结果可以较好的还原区域含水介质的介质类型分布，而在非均质性较强的区域，模拟结果难以还原真实的介质分布；由此，针对大尺度含水介质模拟，提出了基于多尺度钻孔数据转移概率地统计模拟的方法，即耦合弱非均质区的正常尺度综合钻孔和强非均质区更细尺度的尺度综合钻孔进行模拟，使得模拟结果在强非均质区域可以更好地还原含水介质真实分布。（4）基于尺度综合虚拟钻孔和实际钻孔，利用转移概率地统计可以有效地创建多尺度多孔介质模型；将建立的确定性水文地质结构模型与结构体内的随机模型相结合，实现了多孔介质的确定-随机耦合模拟，模拟的多孔介质模型即具有大尺度的结构性又具有小尺度的随机性特征。（5）转移概率地统计模拟各结点的稳定性与钻孔密度和分布，模拟区域的非均质性强弱等因素有关，在弱非均质区域，转移概率地统计的模拟结果趋于稳定较快，每次模拟结果相同的可能性大；而在强非均质区域，模拟结果趋于稳定慢，每次模拟结果随机性强；在数据稀少区域，每次模拟结果的随机性更强，但是随着实现数的增加，各介质所占比例趋于稳定。（6）将多尺度多孔介质模型与大尺度地下水流和多尺度溶质运移数值模拟结合应用的实例显示，大尺度地下水流数值模拟使用基于尺度综合虚拟钻孔进行转移概率地统计模拟提取的渗透系数，小尺度溶质运移模拟使用实际钻孔进行转移概率地统计模拟提取的渗透系数，在地下水数值模拟中是可行的，转移概率地统计生成的条件化的随机模型体现了渗透系数的空间变异性，改变了局部水头分布和水流方向，其模拟结果往往更接近实际情况。本文创新点体现在：（1）创新了基于密集分布钻孔岩性数据进行尺度综合生成虚拟钻孔的方法；（2）提出了基于垂向分层平面连续地质类型假设的侧向转移概率计算方法；（3）基于尺度综合虚拟钻孔和实际钻孔，利用转移概率地统计方法实现了多孔介质多尺度随机模拟。更多还原

【Abstract】 Numerical simulation of groundwater flow and solute transport based on relative high densitydrilling lithology data of a region, which is jointly driven by multi-scale, different accuracyrequirements and data, can provide theoretical support for solving issues of regional groundwaterresources, environmental problems, ecological problems, and groundwater disasters. However, dueto the heterogeneity of aquifer medium, hydrogeological parameters of aquifer arespatial variabilityand scale effect, which result in there is a large deviation between groundwater flow and solutetransport simulation results with the actual situation. Therefore, it is very meaningful thatestablishing multi-scale heterogeneous porous medium model based on high density drillinglithology data to obtain hydrogeological parameters field, and applying parameter field to numericalsimulation of groundwater.This paper carried out the following researches. The heterogeneity of aquifer medium wasanalysed based on high density drillings lithology data. Voronoi diagram method was employed toanalyze the density and distribution of drilling data. A scale generalization approach for high densityboreholes data was proposed. The stochastic simulation of transition probabilities geostatistics wasresearched in this paper, and a method for computing lateral transition probabilities was proposed.The relationship of lateral transition probabilities and simulation result was analyzed, and theestablished three-dimensional stochastic models with scale generalization virtual boreholes andoriginal boreholes were compared. Simulation of multiscale porous medium model based on scalegeneralization virtual drillings and real drillings and deterministic-stochastic coupled simulationwere realized. Hydrogeological parameters were extracted from stochastic model of porous medium.The relationship of the stability of the extracted parameters with realization number in transitionprobabilities geostatistical simulation was analyzed. Porous medium model with numerical model ofgroundwater flow and solute transport was combined, which can provide support for regionalgroundwater management and multi-scale numerical simulation of groundwater. The mainconclusions and results are as follows:(1) Based on the high density drilling lithology data, the spatial correlation distance of the twomain directions for the instance study area of the aquifer were obtained; A scale generalizationapproach for high density boreholes data was proposed to generate virtual drillings, and thegenerated virtual drillings can better reflect the large-scale structural characteristics of aquifermedium. Stochastic model based on scale generalization virtual drillings can shows the overalldistribution characteristics of aquifer medium, and the simulation result in strong heterogeneousregion are shown as continuous equivalent material. (2) The analysis of lateral transition probabilities shows if the ratio of mean lens length of thelateral to vertical is estimated too small, it will lead to a strong random materials distribution in thesimulation result, which is uncontinuity. If the estimated ratio is too big, it will lead to too strongcontinuity materials in horizontal direction.The proposed method based on vertical stratification andthe hypothesis of continuous geological type in plane for computing lateral transition probabilitiescan effectively obtain lateral transition rate matrix, and can get better simulation result. A programbased on GIS which can manage drilling data and conduct transition probabilities geostatistics wasdeveloped.(3) The simulation result of transition probabilities geostatistics was assessed, which shows themodel can better restore materials distribution of aquifer medium in weak heterogeneous orhomogeneous area, but difficult to restore the true materials distribution in strong heterogeneousregion. Thus the simulation method based on multi-scale virtual boreholes data was proposed, whichis combining the normal scale generalization boreholes in weak heterogeneous area and finer scalegeneralization boreholes in strong heterogeneous area to simulate.The simulation results can betterrestore the actual distribution of aquifer medium in a strong heterogeneous region.(4) Multi-scale porous medium model based on scale generalization virtual drillings and actualdrillings can effectively established with transition probabilities geostatistics. Hydrogeologicalstructure model and the stochastic models in the structures were combined to establishdeterministic-stochastic porous medium model. The final model has large-scale structuralcharacteristic and small-scale random characteristic.(5) The stabilities of nodes in transition probabilities geostatistical model are related withdrillings’ density and distribution, the strength of heterogeneity of the simulation area, and otherfactors. In weak heterogeneity area the result of realizations with increasing number is faster to tendto be stable, and the probability of the same result in the same node in different realization is higher,but slower to tend to be stable and random of that in strong heterogenous area. There is a morerandomness simulation results in the area of sparse data. However, the proportion of each material ata node tends to be stable with the increase in realization number.(6) The instances of combining multi-scale porous medium model with groundwater flow andsolute transport numerical simulation show it is feasible that numerical simulation of large-scalegroundwater flow uses hydraulic conductivity values extracted from transition probabilitiesgeostatistical model based on scale generalization virtual boreholes, and small-scale solute transportsimulation uses hydraulic conductivity values extracted from the model based on actual boreholes.The conditional stochastic model reflects the spatial variability of hydraulic conductivity, which maychange the local head distribution and flow direction, and make the simulation results often closer tothe actual situation. This innovations are reflected in:(1) proposing scale generalization method for generatingvirtual boreholes based on high density boreholes lithology data;(2) proposing the method forcomputing lateral transition probabilities based on vertical stratification and the hypothesis ofcontinuous geological type in plane;(3) realizing multi-scale porous medium simulation withtransition probabilities geostatistics based on scale generalization virtual drillings and actualdrillings.更多还原