基于 GIS 的三维隐伏矿体成矿预测是矿产预测评价领域研究的热点问题，而构建三维数字矿床模型则是解决该问题的重要手段之一。目前，在构建三维数字矿床模型方面，仍有大量问题亟待解决，特别是基于三维空间分析构建找矿标志的三维模型方面尤为突出。三维缓冲区分析是三维空间分析的基本内容。目前在成矿预测研究中，通常使用地质对象的栅格模型表达其缓冲区，但其存在耗时、不精确、可视化效果相对较差的缺点。本文提出缓冲区分析方法，具有快速、精确、矢量化的特点，由地质对象的离散化入手，通过改进的带符号的欧氏距离变换方法和精确的光滑曲面重构算法，最终获得使用以矢量表达的地质对象的三维缓冲区(带)。基于多个矿区实际的三维地质模型，从三维可视化、优度与等距性分析等方面开展两种模型的对比实验，结果表明，矢量缓冲区分析在三维可视化、计算精度及其算法扩展性方面比栅格缓冲区模型更加适合三维隐伏矿体的定位与定量预测。%The way that constructs geometry of geological objects and carries out 3D prospectivity research is one of the important and available methods to find concealed mineralization. However, there are currently plenty of problems in 3D modeling for geological objects and quantitative assessment. The difficulty especially lies in constructing 3D geological objects based on calculations of spatial analysis, which are related to mineralization. 3D Buffer Analysis is one of the methods among them. In plenty of 3D prospectivity researches, geologists apply block model of geological objects to express results of buffer analysis. Disadvantages of the way are time-consuming, inexact, and poor visualization effects compared with triangle mess. Meanwhile, the grid mess will produce negative influence on the result of quantitative assessment. This paper proposed a method that is with rapid, exact and triangulated characteristics. Specially, the method serves as a surface model of geological objects. And it includes 4 steps, i.e., voxelization of geological objects, revising 3D Euclidean Distance Transform and calculating signed distance field, extracting surfaces from the field, and buffer surface construction based on DSI algorithm. Meanwhile, this paper applies buffer analysis to 3D prospectivity and carries out contrasting tests in aspects of 3D visualization effect, requirement and goodness of fit by using 3D models that are from different study areas and represent different kinds of geological situations. The result shows that buffer expressed by triangle mess is better due to its block model and advantages in quantitative assessment.
展开▼
