Elsevier

Computers & Geosciences

Volume 39, February 2012, Pages 111-119
Computers & Geosciences

3D exploratory analysis of descriptive lithology records using regular expressions

https://doi.org/10.1016/j.cageo.2011.06.018Get rights and content

Abstract

This paper presents an interactive approach for analyzing a database of descriptive lithology records to locate a specific lithology feature in three-dimensional space. The method uses a regular expression to search individual lithology records, assigning a match score to indicate the relative strength of a search result. Spatial analysis of the resulting match scores generates a three-dimensional representation of the search results, which indicates the likely locations of the lithology feature. The method enables direct analysis of lithology descriptions, which are often characterized by inconsistencies in terminology, accuracy, level of detail, and spatial distribution. Using regular expressions, the approach circumvents the need for manual interpretation and classification of lithology records. The method was applied to a case study area in Western Australia to delineate the extent of three lithology features (clay, calcareous sediments, and iron-rich sediments). The approach is generally applicable for layered lithology features that are regularly documented in a lithology database.

Highlights

► An interactive approach to analyze a lithology database. ► Enables direct analysis of lithology descriptions. ► Circumvents manual interpretation and classification of lithology records. ► Combines regular expression pattern matching and spatial analysis techniques. ► Identifies likely locations for a specific lithology feature.

Introduction

Lithology databases are regularly utilized in geological and hydrogeological modeling, often requiring a subjective and manual process to interpret and classify the descriptive data. This paper presents an interactive method developed to explore descriptive lithology records and build a three-dimensional (3D) representation of a lithology feature.

Lithology descriptions are commonly recorded for discrete depth intervals during borehole drilling. Lithology databases provide a primary resource for subsurface geological interpretation, although the data are often difficult to utilize as there may be inconsistencies in the terminology used to describe a core, the level of descriptive detail, features that are mentioned or omitted, the vertical resolution of the logging, and the spatial (horizontal) coverage of drilling locations. These inconsistencies arise for a variety of reasons: lithology data are often qualitative, and the information recorded is dependent on the experience, skill, and prior knowledge of the person recording the logs; lithology databases often contain data collected over a period of many years and generated by various groups working on drilling programs, employing different drilling equipment and with different aims and objectives; and the total drilling depth may also influence the level of detail given for individual log intervals.

Lithology descriptions are usually interpreted and classified before being used for further analysis or geological modeling. The initial interpretations that are recorded in existing lithology databases can be of limited use if the lithology classification is not relevant for the intended study, or the interpretation is inconsistent at the required scale. The usual solution is for an expert to reinterpret the logs, which may be a suitable option for a small study area but is a very time-intensive option when a large number of records require classification, particularly when there is a need to undertake a consistent spatial classification among many neighboring drill holes. Automated methods of reclassification are difficult due to the descriptive and inconsistent nature of the data.

Technological improvements have considerably enhanced our ability to characterize subsurface conditions; however, the modeling and visualization process remains a challenge (Turner, 2006). Research into representation and querying of geological features includes the development of a 3D geoscience information system with query functionality and data management tools (Apel, 2006), and the development of data structures to represent 3D geological objects (Gong et al., 2004, Wu, 2004, Xue et al., 2004). Various attempts have been made to streamline the process of developing and maintaining geological models. Brandel et al. (2005) present a knowledge-driven approach for automatically building a topologically and geologically consistent geological model. Wu et al. (2005) propose a stepwise refinement method that integrates data from multiple sources and simulates geological structures to build a 3D geological model. Kaufmann and Martin (2008) outline a methodology for processing data from numerous sources to build a 3D geological model. Smirnoff et al. (2008) propose the use of a support vector machine to automate the building of 3D geological models. Sharpe et al. (2007) and Logan et al. (2006) describe the use of a semi-automated expert system, which could utilize manual interpretations to classify a series of numerous lower-quality records. However, these approaches generally require the data to be classified into a predetermined set of lithology classes before the method can be applied.

The objective of this study was to delineate lithology features using existing lithology data, without the need to manually interpret individual lithology descriptions. This paper describes an exploratory method that directly analyzes plain text lithology descriptions to locate a specific lithology feature in 3D. The method description is followed by a case study, where three types of lithology were spatially delineated.

Section snippets

Method and implementation

The developed method combined regular expression pattern matching and spatial analysis techniques. Regular expressions are a formal and concise text pattern notation for describing and parsing text (Friedl, 2006). A familiar regular expression analogy might be the wildcard character “⁎”, which is commonly used in computing to form a match with any other characters (e.g., in a file name). However, regular expressions can describe far more detailed patterns for text matching, and provide options

Case study

The case study area, located 20 km south-east of Perth, Western Australia (WA) (Fig. 1), is a peri-urban zone containing various environmentally sensitive areas, protected wetlands, and important groundwater and surface water resources for the Perth metropolitan area. The catchment has been undergoing rapid urbanization, accompanied with the introduction of a decentralized nonpotable water supply scheme from local groundwater resources. This requires the identification of sustainable groundwater

Results

The lithology analysis produced plausible results for each of the three lithology classes included in the case study. A clay feature was identified at the base of the Darling Scarp on the eastern edge of the case study area. Fig. 2 plots the clay match score points, and the corresponding interpolated surface for elevations 15, 20, 25, and 30 m AHD for the entire case study area. At lower elevations within the study area, the clay layer is more extensive. Fig. 3 shows the interpolated clay match

Discussion

The case study produced results that were consistent with the current understanding of the geology in the area, and presented a 3D representation of the three lithology features. This was achieved through direct analysis of descriptive lithology records, which avoided the need for manual classification of the available lithology data. In the case study, the method identified zones within the Superficial Aquifer that might have an impact on a proposed nonpotable groundwater supply scheme,

Conclusions

The presented method provides an interactive tool for the exploration of a large lithology database containing inconsistent lithology descriptions at variable spatial resolutions. Since the method requires the user to provide suitable regular expression patterns, an understanding of the geological characteristics of a study area is required. The definition of the patterns generally follows from some expectation of what lithology features are present in a study area and how those features are

Acknowledgments

The authors acknowledge the support of the Government of Western Australia, which provided funding through the Water Foundation program, and the CSIRO Water for a Healthy Country National Research Flagship. The authors also thank the WA Department of Water, WA Department of Industry and Resources, WA Department of Land Information, and Geoscience Australia for the provision of the data that were used for the reported analysis.

References (23)

  • ESRI, 2010. What is ArcPy,...
  • Cited by (0)

    View full text