The retrieval of sediment type from airborne electromagnetic data in the central valley of California
Abstract/Contents
- Abstract
- The effective management of groundwater basins worldwide is of immediate concern due to the stresses imposed upon limited groundwater resources by climate change and population growth. Numeric groundwater models are commonly used to model the response of groundwater basins to a variety of management actions and environmental stresses. Groundwater models used in this manner are often created from data acquired during the construction of wells, which can provide detailed information about the subsurface at the locations of the wells, but do not contain any information about the horizontal subsurface heterogeneity in-between the well locations. This lack of information on horizontal heterogeneity impacts the accuracy of model results and can lead to groundwater managers making incorrect management decisions. The goal of this thesis is the development of a complete, repeatable, data-driven workflow for the retrieval of sediment type from airborne electromagnetic (AEM) data. Sediment type can provide the detailed subsurface structural information required for the construction of groundwater models. The AEM method is a geophysical surveying tool that allows for the characterization, in 3D, of subsurface heterogeneity. However, the AEM method is sensitivity only to subsurface resistivity, while information on sediment type and hydraulic properties is what is needed for the construction of groundwater models. The first portion of this thesis arose from the recognition that in an AEM dataset acquired in both the unsaturated and saturated zone, the depth of the top of the saturated zone (TSZ) at the time of data acquisition must be accounted for during the construction of any resistivity-to-sediment type transform. Current methodologies for the construction of resistivity-to-sediment-type transforms often neglect to account for the spatially variable sensitivity of the AEM method and, by basing the built transform on the resistivity model and not the AEM data, the developed transform becomes biased by the data processing and inversion steps used to produce the resistivity model from the AEM data. In this thesis, we developed and tested a methodology for estimating the TSZ from AEM data, using data collected in three survey areas in the Central Valley of California and water-table elevation (WTE) measurements from nearby wells. The methodology was based on the difference in the distribution of resistivity values above and below the TSZ, using the WTE measurements to optimize two components of the general workflow for the estimation of the depth to the TSZ and to estimate the error present in the retrieved estimates. We found that TSZ estimates produced with our methodology under optimal conditions had a level of error that was comparable to the vertical resolution of the AEM method at the depth of the TSZ estimates, allowing us to place the TSZ within the correct layer in the AEM resistivity model. The TSZ estimates produced with this methodology are vital for the next portion of this thesis, the construction of the resistivity-sediment-type transforms. The next major research contribution of this thesis was the development of a Markov Chain Monte Carlo (MCMC) based methodology for the transformation of AEM derived resistivity into sediment type. This methodology was developed and tested using AEM data and well sediment type and resistivity logs from Butte and Glenn Counties in the north of the Central Valley of California and with a synthetic test case. Our methodology accounted for the spatially varying sensitivity of the AEM method by constructing six different transforms that were separated based on the sensitivity of the AEM method. The influence of saturation state was captured by creating one set of transforms for the region above the TSZ and another for below. The transforms constructed with this methodology were used to produce estimates of the probability of each sediment type for every location where AEM data were acquired in the survey area. Through examination of the results from the synthetic test case and after comparing our field results to the sediment type data from nearby well logs, we concluded that our methodology could retrieve the true distribution of resistivity for each sediment type and that it produces transforms that are without many of the issues and limitations common to resistivity-to-sediment-type transforms built upon the AEM method. The final portion of this thesis contains an investigation into how the retrieved sediment type information could be used for the construction of groundwater models based on the Butte and Glenn Counties survey areas. We modified the methodology developed for the estimation of the depth to the TSZ and applied it to the sediment type output by the rock physics transforms to estimate the depths to the boundaries between sediment packages dominated by different sediment types. We also ran geostatistical simulations with the sediment type information as input; the realizations produced were then used to parametrize the small-scale, within layer, variations in hydraulic properties of the groundwater models. Two ensembles of models were built to examine the impact of both large- and small-scale uncertainty on model outputs. The main model output examined was a drawdown curve generated by observing the change with time in the simulated hydraulic head at a location in the center of the model. A comparison model was also built to include a model constructed using a more traditional workflow based on manual interpretation. We found that the results from the traditional model were contained within the results of both ensembles, providing initial validation for the methodology proposed in this final chapter, and that the small-scale uncertainty generated much more variance in the simulated drawdown curves than the large-scale uncertainty. In summary, the research presented in this thesis represents a complete workflow for the data-driven retrieval of sediment type from AEM data with two components that represent significant research advancements: the demonstration of a method for the estimation of the depth to the TSZ from airborne electromagnetic data and the development of a workflow for building resistivity-to-sediment-type transforms that accommodate the spatially varying sensitivity of the underlying method and are independent of the inversion used to produce resistivity from the acquired data. The final portion of the thesis explores a potential methodology for the construction of groundwater models based on layer boundaries estimated from the sediment type probabilities output by the developed resistivity-to-sediment type transforms.
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource. |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2021; ©2021 |
| Publication date | 2021; 2021 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Dewar, Noah Michael |
|---|---|
| Degree supervisor | Knight, Rosemary (rosemary Jane), 1953- |
| Thesis advisor | Knight, Rosemary (rosemary Jane), 1953- |
| Thesis advisor | Caers, Jef |
| Thesis advisor | Harris, Jerry M |
| Thesis advisor | Mukerji, Tapan, 1965- |
| Degree committee member | Caers, Jef |
| Degree committee member | Harris, Jerry M |
| Degree committee member | Mukerji, Tapan, 1965- |
| Associated with | Stanford University, Department of Geophysics |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Noah Dewar. |
|---|---|
| Note | Submitted to the Department of Geophysics. |
| Thesis | Thesis Ph.D. Stanford University 2021. |
| Location | https://purl.stanford.edu/xx035cr5854 |
Access conditions
- Copyright
- © 2021 by Noah Michael Dewar
- License
- This work is licensed under a Creative Commons Attribution Non Commercial 3.0 Unported license (CC BY-NC).
Also listed in
Loading usage metrics...