Probabilistic Evaluation of Geoscientific Hypotheses With Geophysical Data: Application to Electrical Resistivity Imaging of a Fractured Bedrock Zone. Issue 9 (17th September 2021)

• Record Type: Journal Article
• Authors: Miltenberger, Alex; Uhlemann, Sebastian; Mukerji, Tapan; Williams, Ken; Dafflon, Baptiste; Wang, Lijing; Wainwright, Haruko
• Is Part Of: Journal of geophysical research. Volume 126:Issue 9(2021)
• Page Start: n/a
• Page End: n/a
• Publication Date: 2021-09-17
• Subjects: electrical resistivity -- fractures -- groundwater -- hypothesis testing -- uncertainty quantification; Geomagnetism -- Periodicals; Geochemistry -- Periodicals; Geophysics -- Periodicals; Earth sciences -- Periodicals; 551.1
• DOI: 10.1029/2021JB021767
• Languages: English
• ISSNs: 2169-9313

Abstract

Abstract: As climate changes and populations grow, groundwater sustainability is becoming increasingly important. Hydrogeologic models, which are based on a conceptual understanding of the subsurface, are crucial tools for informing decisions. Conceptual models of the subsurface incorporate knowledge of geological processes, and, frequently, observations from geophysical data into a common subsurface parameterization where the parameters may still be uncertain. Many methods exist to test how different conceptual subsurface parameterizations compare to geophysical data, but a frequent problem in hydrogeologic model development occurs when multiple geological phenomena could explain a single subsurface parameterization. In this work, we present a framework for testing geological hypotheses in conditions where a geological feature is observed in geophysical data, but its physical characteristics are uncertain. The framework uses Popper‐Bayes methods developed in previous studies, and is applied to study a fractured bedrock zone in a mountainous watershed in southwest Colorado. First, we propose six hypotheses based on the geological history of the watershed. Then, using the proposed Popper‐Bayes approach, we demonstrate that two of the hypotheses are inconsistent with the electrical resistivity tomography data. Finally, we discuss the importance of the prior model, and in what other scenarios the framework can be applied to. Plain Language Summary: Groundwater is a major source of fresh water throughout the world, yet much is still unknown about where and how water moves underground. One big reason for this is that it is hard to map out underground flow paths from the surface. One technique to map underground flow paths is to send electrical signals deep into the ground and measure how the ground reacts to the signals. In this work, we show how these measurements can be used to understand the geologic history of an area, which in turn helps to identify underground flow pathways. The data interpretation method we propose in this paper considers multiple potential causes for the formation of underground flow pathways. Then, we rank the scenarios by how consistent they are with the measured data. We demonstrate this method in the Elk Mountains of Central Colorado near a zone of highly fractured bedrock, which water easily flows through compared to the surrounding rock. Through the proposed method we show that two of the six potential causes for the creation of the fracture zone are inconsistent with the electrical data collected in the field. Key Points: Machine learning is integrated into the Popper‐Bayes framework to test conceptual geologic hypotheses with geophysical data The approach is demonstrated using electrical resistivity tomography (ERT) data from a mountainous watershed to study the geological origins of a zone of highly fractured shale Two of six hypotheses are falsified by the ERT data, showing the utility and limitations of ERT data for imaging this fractured bedrock zone