Master of Science (MS), Bowling Green State University, 2020, Geology
Western Kentucky coal mines have long disposed of coal slurry by dumping the material into waterbodies located on property. This fine-grained material contains high amounts of sulfur, iron, and other heavy metals, placing nearby waterways and biota at risk for contamination. This study proposes the implementation of WorldView-3 imagery, reflectance spectroscopy, chemical composition analyses, and multiple Neural Networks to establish a prediction model that would map the extent and concentration of total organic carbon, arsenic, lead, and chromium throughout these slurry deposits. This method of chemometric analysis has proven effective in the determination and prediction of heavy metals but has yet to be applied to WorldView-3 imagery or coal slurry deposits. Worldview-3 imagery provides significantly higher spatial and spectral resolution than most other spaceborne-sensors, as well as allows for a < 1-day return time. Hyperspectral-based predictions of Total Organic Carbon, arsenic, chromium, and lead resulted in R2 values of 0.95, 0.90, 0.77, and 0.75, respectively. WorldView-3 based predictions resulted in Overall Accuracies of 84%, 79%, 70%, and 75%, respectively. This very high resolution (VHR) remote sensing is vital for monitoring complex ecosystems and mapping those substances which pose a risk to soil, biota, and human health, such as coal slurry. By successfully predicting these constituents, coal mines will have a better understanding of contamination extent and can more effectively conduct remediation efforts.
Committee: Anita Simic Milas Dr. (Advisor); Angélica Vázquez-Ortega Dr. (Committee Member)
Subjects: Geological; Remote Sensing; Soil Sciences