Three-Dimensional Stochastic Modeling Method Of Geological Entities Based On Multiple-point Geostatistics

The three-dimensional(3D)visualization of geological entities and structures can provide a more realistic and intuitive description for subsurface geological phenomena and structures.It has been a crucial technology for promoting the construction of “Glass Earth” and “Smart City”.With the development of 3D geological modeling and the in-depth application in various geological fields,the requirements for 3D geological models have shifted from qualitative description to quantitative evaluation.The stochastic modeling methods by using multiple-point geostatistics(MPS)have been an important branch for the automatic construction and simulation of 3D geological models.The development of its basic theory and the various algorithms offer technical supports for 3D reconstruction and quantitative analysis of anisotropic and complex subsurface structures.For a real 3D application,however,a complete and reliable 3D training image is very difficult to be obtained,especially for the Quaternary sedimentary system with loose,uneven,thin,and multi-fragmented structures.In addition,the realizations based on regular grid cannot realistically reflect the configurations and distribution characteristics of geological structures.In summary,the main challenges of 3D geological modeling include: first,the rapid,dynamic and automatic reconstruction of anisotropic underground geological structures;second,the integrated modeling and characterization of geological structures and complicated attributes;third,whether the obtained models can support quantitative analysis,evaluation and visual decision.Quaternary loose sediments have the greatest impact on human activities.The majority of Chinese megalopolis is located on the Quaternary sedimentary strata of the plains or plateaus.Moreover,various geological and environmental issues are related to the spatial distribution and dynamic processes of the shallow geological structures in the Quaternary.Consequently,3D geological modeling of the Quaternary sedimentary system is the groundwork for the development and utilization of underground space and the quantitative evaluation of geological environment.Therefore,there are very important theoretical and practical significances for the study on 3D geological stochastic modeling methods for the Quaternary sedimentary system,such as proposing new strategies,approaches and models,as well as building the entire solution from data processing,modeling methods,to visualization.In order to overcome the above problems and challenges,a series of improved methods and an integrated 3D stochastic modeling workflow based on multiple-point geostatistics are proposed.To address the the problem of unbalanced distribution of actual geological survey data,the improved methods of simulation path,neighborhood and post-processing in multi-point geostatistical simulations are studied emphatically.To overcome the difficulties in obtaining 3D training images and the highly dependent on stationarity assumption,a method for automatic reconstruction of 3D geological models based on low-dimensional data is developed.In order to achieve the integrated modeling of geological structures and attributes,the automatic generation of the fine 3D voxel model is studied according to the constraint of the geological structures.On this basis,the verification of the proposed methods is carried out by taking the modeling of the complex Quaternary sedimentary structures in Minjiang Estuary as an example.Several comprehensive experiments and real 3D applications verify the performance and effectiveness of the proposed methods.Moreover,the proposed methods have good portability and extensibility,and they can be applied into other similar fields.This work is able to provide guidances and technical supports for the modeling of the Quaternary sedimentary environment and even other extensive 3D geological modeling practices such as resource exploration and development,geological environment evaluation,etc.In a nutshell,the main innovations of this dissertation include the following points.(1)Against the characteristics of the Quaternary sedimentary system and the distribution features of the dataset,a series of improved strategies are proposed so that the MPS methods can be adapted to the 3D modeling requirements of complex Quaternary sedimentary structures.(2)A novel method for reconstructing 3D complex heterogeneous structures based on a local search strategy is proposed in which 3D geological models can be constructed fast and automatically from low dimensional data making 3-D training images unnecessary.(3)A new gridding approach of 3D geological models is presented where the constraints of complex geological structures are considered.The integrated modeling and visualization of geological structures and attributes are achieved on the basis of the outputs of MPS simulations.Moreover,a detailed workflow of MPS-based 3D stochastic modeling is customized.