定量地层学以生物地层学原理为核心, 将地层学信息定量化, 运用数学模型进行地层对比, 作为传统地层学对比的重要补充。传统的生物地层学根据经验选取标准化石来建立生物地层序列, 但同时也丢失了大量的化石信息。定量地层学则能够利用所有的化石信息, 在传统生物地层学的基础上得到更高分辨率的地层对比结果。现阶段定量地层对比常用的方法主要有图形对比法、约束最优化法和单元组合法三种, 但对于这三种方法各自适应的数据情况方面, 目前成果较少。本文对这三种方法的原理进行了简要的介绍和分析, 并建立数据模型, 从标准化石的可对比性、数据集的物种总数、剖面间共有物种和单延限分子的比例四个方面对三种方法的适用条件进行了讨论。其中图形对比法更适用于单延限分子较少、标准化石可对比性强的数据集。单元组合法对数据集中物种间相互关系的复杂程度较为敏感, 共有物种较多的数据集有更好的对比结果, 但数据集中物种总数的增加会对其产生一定的负面影响, 需要依靠进一步的人工调整。约束最优化法则对数据集各方面的优化均有响应, 对物种总数较大的数据集有较好的对比结果, 且剖面间共有物种占比越大, 对比结果越理想。

Quantitative stratigraphy, based on the principles of biostratigraphy and mathematical models, is a new research field compared with traditional stratigraphy. Traditional stratigraphic researstratigraphic framework and to make correlations among different sections. This method, in which only the index fossils are emphasized while other fossils are ignored, relies largely on personal experiences. On the other hand, quantitative stratigraphy aims to make full use of the stratigraphic information by translating all fossil records into structured data and reconstruct a composite time scale with mathematical tools, such as regression, graph theory and randomized algorithms. Quantitative methods have remarkably improved the resolution of stratigraphic correlation, which is critical for understanding geological events that span a relatively short time interval and a wide geographic range. At present, there are three commonly used quantitative methods, including Graphic Correlation, Constrained Optimization (CONOP) and Unitary Association Method (UAM). In this study we introduce the fundamentals of the three methods and evaluate the possible factors that may influence correlation outcomes using mathematical models. Four factors, including the distribution of index fossils among the sections, the number of species studied, the proportions of species occurring in more than one section and the proportion of singletons are considered. Our results show that Graphic Correlation is highly dependent on the isochronous biostratigraphic events such as first appearance data of index fossils. This method is reliable when the biostratigraphic timeframe is well-established, but not suitable for the situation that sections have different regional time scales. UAM responses significantly to the proportion of common species and singletons, and shows high bias and constrained power of resolution, based on its algorithm. In contrast, CONOP has the highest applicability and obtains a relatively stable outcome with different factors considered.

国家自然科学基金(41830323)、地质调查项目(DD20190009)和中国科学院先导 B 项目(XDB26000000)联合支持