Download the full-sized PDF of Hyperspectral imaging for the characterization of Athabasca oil sands coreDownload the full-sized PDF



Permanent link (DOI):


Export to: EndNote  |  Zotero  |  Mendeley


This file is in the following communities:

Graduate Studies and Research, Faculty of


This file is in the following collections:

Theses and Dissertations

Hyperspectral imaging for the characterization of Athabasca oil sands core Open Access


Other title
hyperspectral imaging
oil sands
total bitumen content estimation
sedimentological feature enhancement
automated lithological mapping
Type of item
Degree grantor
University of Alberta
Author or creator
Speta, Michelle A
Supervisor and department
Rivard, Benoit (Earth and Atmospheric Sciences)
Examining committee member and department
Gingras, Murray (Earth and Atmospheric Sciences)
Lipsett, Michael (Mechanical Engineering)
Department of Earth and Atmospheric Sciences

Date accepted
Graduation date
Doctor of Philosophy
Degree level
The Athabasca oil sands of northeastern Alberta, Canada, are one of the largest accumulations of crude bitumen in the world. Drill core sampling is the principal method for investigating subsurface geology in the oil sands industry. Cores are logged to record sedimentological characteristics and sub-sampled for total bitumen content (TBC) determination. However, these processes are time- and labour-intensive, and in the case of TBC analysis, destructive to the core. Hyperspectral imaging is a remote sensing technique that combines reflectance spectroscopy with digital imaging. This study investigates the application of hyperspectral imaging for the characterization of Athabasca oil sands drill core with three specific objectives: 1) spectral determination of TBC in core samples, 2) visual enhancement of sedimentological features in oil-saturated sediments, and 3) automated classification of lithological units in core imagery. Two spectral models for the determination of TBC were tested on four suites of fresh core and one suite of dry core from different locations and depths in the Athabasca deposit. The models produce greyscale images that show the spatial distribution of oil saturation at a per-pixel scale (~1 mm). For all cores and both models, spectral TBC results were highly correlated with Dean-Stark data (R² = 0.94-0.99). The margin of error in the spectral predictions for three of the fresh cores was comparable to that of Dean-Stark analysis (±1.5 wt %). A fourth fresh core and the dry core had higher margins of error (±2.5-3.3 wt %) due to some instances of overestimation in high grade samples (>10 wt %). Surface roughness was identified as a possible source of error in the spectral TBC estimates. Logging oil sands core can be challenging because sedimentary and biogenic features are often difficult to see in the bitumen-saturated sediment. Shortwave infrared (SWIR; 1.0-2.5 µm) hyperspectral imagery, using the three-band combination of 2.16 µm, 2.20 µm and 2.35 µm, was found to greatly enhance the visibility of sedimentological features in massive-appearing oil sand. This combination of wavelength bands produced near-natural colour images that in many cases revealed features completely invisible to the unaided eye. A spatial resolution of at least 0.25 mm/pixel is required for accurate trace fossil identification, but most sedimentary structures can be accurately identified even at lower resolutions (1.2-1.5 mm/pixel). Enhanced visibility of features is due to variable reflectance that for large-scale sedimentary structures (>1 cm) is attributed primarily to changes in grain size and bitumen content. Variable reflectance across smaller-scale (<1 cm) sedimentary structures or trace fossils is mainly attributed to differences in clay content. Finally, SWIR and longwave infrared (LWIR; 8-12 µm) imagery of oil sand core was investigated for the automated mapping of five common rock types (oil sand, barren sand, siltstone, mudstone, and siderite) using the spectral angle mapper (SAM) classification technique. Detailed maps that are consistent with visual inspection of the core were successfully produced with both datasets. In cases of disagreement, the SWIR imagery was more accurate for mapping the spatial distribution of oil sand and the LWIR imagery was more accurate for the discrimination of barren sand, siltstone and mudstone. Improvements in the mapping methodology, such as using a spectral TBC model to map oil sands at different bitumen saturation thresholds, or using spectral subsets of the image data for SAM input, enhanced the mapping results. The combined results of this thesis demonstrate the immense potential of hyperspectral imaging for facilitating the processes of routine oil sands core logging and bitumen content analysis.
This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for the purpose of private, scholarly or scientific research. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.
Citation for previous publication
Speta, M., Rivard, B., Feng, J., Lipsett, M., Gingras, M., 2015, Hyperspectral imaging for the determination of bitumen content in Athabasca oil sands core samples, AAPG Bulletin, 99(7), 1245-1259.

File Details

Date Uploaded
Date Modified
Audit Status
Audits have not yet been run on this file.
File format: pdf (PDF/A)
Mime type: application/pdf
File size: 13282355
Last modified: 2016:11:16 14:00:03-07:00
Filename: Speta_Michelle_A_201604_PhD.pdf
Original checksum: 0a839355a865ad78243a36f5c83fc886
Activity of users you follow
User Activity Date