Characterization of geological uncertainties from limited boreholes using copula-based coupled Markov chains for underground construction

Fan Wang, Heng Li, Gang Li, Zheng-Jun You, Elton J. ChenUnderground 2023 · DOI

Abstract This paper presents an efficient methodology for quantifying stratigraphic uncertainties and modeling geological formations using sparse borehole data. By integrating copula functions into the Coupled Markov Chain (CMC) framework, the authors provide an analytical expression for transition probability matrices (TPMs), reducing complex geological transitions to a few estimable parameters. Furthermore, a random path sampling method is introduced to eliminate directional bias in simulations. Validated against real-world data from Wuhan, China, the method demonstrates high predictive accuracy and computational efficiency, offering a robust tool for underground construction risk management.

Introduction: The Sparse Data Challenge

In underground construction, the precise mapping of soil and rock strata is paramount for structural safety and cost control. However, geological information is typically derived from boreholes, which provide high-resolution vertical data but are spatially sparse due to high drilling costs. This leaves the horizontal space between boreholes as a significant source of uncertainty. Traditional interpolation and simulation methods, such as standard Coupled Markov Chains (CMC), often struggle with statistical instability when data is limited or produce "directional effects"—artificial tilting of geological layers caused by the sequential order of the simulation algorithm.

Methodology: Copulas and Random Path Sampling

The researchers introduced two major innovations to overcome the limitations of existing stratigraphic modeling techniques.

1. Copula-based Transition Probability Matrices

Instead of calculating every possible transition between soil types through direct counting—which requires dense datasets—the authors employed copulas. These mathematical functions describe the dependence between random variables, allowing complex TPMs to be represented by a small set of unknown parameters. This parsimonious approach enables the use of Maximum Likelihood Estimation (MLE) to fit the model accurately even when borehole data is sparse.

2. Eliminating Directional Bias

To solve the "directional effect" problem, the authors moved away from standard top-down, left-to-right sequential filling. Instead, they proposed a random path sampling scheme. In this method, the soil type at any unknown location is inferred by considering its nearest known neighbors in all four cardinal directions simultaneously.

Schematic of the random path sampling method where the soil type at location x is inferred from its neighbors in four cardinal directions (A, B, C, D).

By applying Pickard’s theorem and Bayes' rule, the model ensures that the conditional probability of a soil type at a specific point remains mathematically consistent regardless of the simulation order, effectively removing artificial inclination from the resulting geological maps.

Case Study: Wuhan Construction Site

The model was rigorously tested using real-world borehole data from a construction site in Wuhan, China. The study area covered a cross-section 206 meters long and approximately 60 meters deep.

Geological cross-section of the construction site in Wuhan, China, showing the five primary soil strata and the locations of the six test boreholes.

The site contains five distinct geological categories: miscellaneous fill, clay, fine sand, coarse sand, and weathered rock. The researchers used a "leave-one-out" validation approach, building the model with five boreholes and testing its ability to predict the strata of the sixth, hidden borehole.

Key Results and Insights

  • Improved Accuracy: The copula-based approach consistently outperformed traditional methods. For instance, in predicting the layers of Borehole #3, accuracy improved from 81.7% (using standard Walther’s Law methods) to 90.0%.
  • Computational Efficiency: The analytical nature of the model allows for rapid simulation. Generating 500 distinct geological realizations took less than one minute on a standard computer, making it practical for real-time engineering applications.
  • Realistic Geological Features: The simulations successfully captured "mutual inclusions," such as small pockets of fine sand within larger coarse sand layers, which are characteristic of complex deltaic deposits.
  • Uncertainty Quantification: By utilizing information entropy, the method produces "uncertainty heat maps." These maps allow engineers to identify specific zones—typically at the boundaries between soil types—where geological confidence is lowest and additional caution is required.

Conclusion

By combining the statistical parsimony of copulas with a non-directional random path sampling scheme, this research provides a statistically sound and computationally efficient framework for geological characterization. The method effectively bridges the gap between sparse borehole data and the need for high-fidelity underground mapping, significantly reducing the risk of unforeseen geological conditions in civil engineering projects.

Summary updated

Abstract

This paper proposes an efficient method for quantifying the stratigraphic uncertainties and modeling the geological formations based on boreholes. Two Markov chains are used to describe the soil transitions along different directions, and the transition probability matrices (TPMs) of the Markov chains are analytically expressed by copulas. This copula expression is efficient since it can represent a large TPM by a few unknown parameters. Due to the analytical expression of the TPMs, the likelihood function of the Markov chain model is given in an explicit form. The estimation of the TPMs is then re-casted as a multi-objective constrained optimization problem that aims to maximize the likelihoods of two independent Markov chains subject to a set of parameter constraints. Unlike the method which determines the TPMs by counting the number of transitions between soil types, the proposed method is more statistically sound. Moreover, a random path sampling method is presented to avoid the directional effect problem in simulations. The soil type at a location is inferred from its nearest known neighbors along the cardinal directions. A general form of the conditional probability, based on Pickard's theorem and Bayes rule, is presented for the soil type generation. The proposed stratigraphic characterization and simulation method is applied to real borehole data collected from a construction site in Wuhan, China. It is illustrated that the proposed method is accurate in prediction and does not show an inclination during simulation.
BibTeX
@inproceedings{wang2023characterization,
  author = {Fan Wang and Heng Li and Gang Li and Zheng-Jun You and Elton J. Chen},
  title = {{Characterization of geological uncertainties from limited boreholes using copula-based coupled Markov chains for underground construction}},
  booktitle = {Underground},
  year = {2023},
  doi = {10.1016/j.undsp.2023.09.009},
}
Figure