Project Info

Maximizing the Value of Hyperspectral Data

Eileen Martin
Katha Pfaff

Project Goals and Description:

Knowledge of the deposit mineralogy and physical and mechanical properties of rock units is critical at many stages of project development from early exploration to mining and remediation. Hyperspectral core scanning currently represents the method of choice to determine the alteration mineralogy of ore deposits and their host rocks. The most common hyperspectral sensor covers the short-wave infrared (SWIR) regions of the electromagnetic spectrum. However, within the SWIR wavelength range, traditional methods of spectrum matching are inefficient to correctly identify and quantify common minerals such as garnet, olivine, feldspar, quartz, oxides, and sulfides. The aim of this project is to automate mineral identification in a fast, accurate, and efficient way by finding functional relations between hyperspectral and quantitative automated mineralogy data using advanced machine learning techniques. Preliminary results show that we can successfully apply deep learning (CNN) to predict the most abundant mineral in each data acquisition point. Prediction of the modal abundances of minerals were around 95% and spatial distribution of minerals was maintained. It is the goal of this IMURF to build upon this proof of concept and advance our ability to interpret hyperspectral data efficiently and with high accuracy.
The PIs (Martin and Pfaff) have been working together within the Center to Advance the Science of Exploration to Reclamation in Mining (CASERM). The successful students will be mentored by Dr. Martin (joint appointment in Geophysics and Applied Math and Statistics) and Dr. Pfaff (Geology and Geological Engineering). PI Pfaff will introduce the students to mineral and materials characterization, including a background in hyperspectral methods and applications as well as SEM-based automated mineralogy for ground-truthing. Dr. Pfaff will supervise the undergraduate research students in how to use the hyperspectral sample/core scanner (including data acquisition, interpretation and exporting) and the automated mineralogy system. PI Martin will supervise and guide students in data masking, image processing, data co-registration, and machine learning techniques to advance hyperspectral data interpretation.

More Information:

Grand Challenge: Engineer the tools of scientific discovery.

Primary Contacts:

Eileen Martin, | Katha Pfaff,

Student Preparation


The students should have taken a course in digital signal processing, a course using programming for spatial or temporal data analysis, and either an introductory geology course or a keen interest in geology, in the mining industry, or applied data science.




The students will become familiar with state-of-the-art mineral and materials and characterizing to characterize the mineralogy of the subsurface. The research facilities are equipped with a sample preparation facility, multiple microscopes, a field-emission SEM, quantitative automated mineralogy systems, hand-held XRF and hyperspectral systems, and XRF- and hyperspectral-core scanning systems. The students will be able to prepare and analyze sample material and interpret data, and put them into context. Hyperspectral methods are widely used in a large variety of industries (geology, mining, food, environmental, disaster), given the successful students a clear advantage on the job market or graduate school applications over their peers. The students will also gain in-depth expertise in computational data analysis, specifically through machine learning and inverse problems applied to geoscience data.


The undergraduate research project (Fall 2023 and Spring 2024) will be subdivided into three general Milestones with discrete goals: a) August through October 2023 - Learning: - Working in a mineral and materials characterization laboratory, including but not limited to safety training. - Hyperspectral scanning and SEM-based automated mineralogy training and data acquisition. - Introduction to project specific computational data analysis and software tutorials - Organizational skills and preparatory work to ensure success and high-quality data generation and literature review. b) November 2023 through February 2024 – Learning/Analysis: - Continued literature review and achievement of information literacy, data analysis and critical thinking. - Training of effective communication skills. - SEM-based automated mineralogy and hyperspectral (core) scanning, data analysis and data co-registration - Computational data analysis and inverse modeling on co-registered datasets c) March through May 2024 – Analysis/Synthesis/Presentation: - Data interpretation, critical thinking during data analysis and quantitative reasoning skills. - Data presentation – quantitative reasoning and scientific reasoning - Effective communication skills The undergraduate research students will be working closely with PIs Martin and Pfaff and graduate student assistants. The students will participate in weekly research group meetings (in geophysics and in geology to gain near-to-peer mentors among graduate students in both fields), meet with the faculty mentors and present at quarterly meetings with stakeholders (CASERM research sponsors from industry and government). The graduate student assistants will train the undergraduate research students in sample preparation and PI Pfaff will train the undergraduate research students in hyperspectral core scanning and data analysis. PI Martin will supervise and train the students in computational data analysis and inverse problems. The undergraduate research students will participate in discussions and present their own findings to their co-mentors once per month and at the end of Fall 2023 and Spring 2024 to the research team.

Preferred Student Status

Share This