Date of Award
5-2024
Culminating Project Type
Thesis
Styleguide
apa
Degree Name
Geography - Geographic Information Science: M.S.
Department
Geography and Planning
College
School of Public Affairs
First Advisor
Dr. Feilin Lai
Second Advisor
Dr. Mikhail Blinnikov
Third Advisor
Dr. Sarah Gibson
Creative Commons License
This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 4.0 License.
Keywords and Subject Headings
Plutonic rock, granite detection, image classification, sentinel-2, machine learning
Abstract
Plutonic rocks contain valuable mineral ores like iron, gold, and copper. Mapping plutonic rocks is important for mineral resource exploration, geological hazard assessment, land use planning, sustainable resource management, and decision-making for zoning, construction, environmental protection, and scientific research. Remote sensing provides a cost-efficient and time-efficient solution for identifying various land cover and land use types from satellite imagery with much-reduced in-field labor and logistic costs. This capability has been enhanced with the integration of Machine Learning (ML), a part of Artificial Intelligence. However, limited research has been conducted on the application of remote sensing for plutonic rock mapping in the United States, particularly the performance assessment of various ML algorithms tailored to this specific task. Regarding this research gap, this study aims to explore and evaluate the performance of several well-established ML algorithms for mapping plutonic rock alongside other major land cover types in a complex urban environment, utilizing Sentinel-2 imagery. The primary study area for training and evaluating the ML models is the city of Waite Park, MN. Additionally, we assessed the transferability of these models to the city of Rockville, MN, both featuring intricate urban landscapes with active quarry sites for plutonic rock production. Our findings reveal key insights: (1) Plutonic rock exhibits distinctive spectral responses on multi-spectral imagery; (2) ML algorithms prove effective in mapping plutonic rock in a complex urban environment, displaying varying overall accuracies and class-specific accuracies for plutonic rock; (3) the established ML models demonstrate a considerable level of transferability to a new study city, with potential for further improvement.
Recommended Citation
APENTENG, HENRY, "Mapping Plutonic Rocks in a Complex Urban Environment based on Sentinel-2 Imagery: A Comparison of Machine Learning Algorithms." (2024). Culminating Projects in Geography and Planning. 17.
https://repository.stcloudstate.edu/gp_etds/17
