Author

Jason Hamel

Abstract

Chemical leachates from landfills can turn into dangerous hazards if they are not identified and properly disposed. Currently, these chemicals are identified in the laboratory with samples hand collected from contaminated sites. This is economical for small sites but quickly gets expensive and time consuming for larger facilities that can cover hundreds of acres or multiple sites. This study examines the feasibility of using new hyperspectral detectors and computer processing to autonomously identify the presence of these leachates and greatly simplifying the monitoring process for these large sites. The effect of even low concentrations of many chemicals requires the identification of leachates. In most cases, the spectral signature is extremely subtle and cannot be directly detected from the background soil spectra. The approach used here will identify secondary effects of the leachates on surrounding features such as vegetation stress or effects on soil moisture that might indicate contamination. This study is exploring the detectability of various states of vegetation health and soil moistures resulting from these contaminants by processing the spectra with several spectral matching algorithms. The ability to classify various health levels will determine if this monitoring method has useful applications in the field. With the levels of health and soil moisture used in this research, the linear spectral unmixing (LSU) and orthogonal subspace projection (OSP) algorithms performed the best. They not only could identify the major constituents of mixed pixels but also generated fractions maps listing the amount of material with a specific level of health or moisture. Since their classification results were very accurate in identifying the materials with simple thresholding for the OSP and LSU algorithms, the material fractions were also included in evaluating the performance of LSU and OSP algorithms.

Publication Date

1999

Document Type

Thesis

Advisor

Not listed.

Comments

Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works in February 2014. senior project.

Campus

RIT – Main Campus

Share

COinS