Plant Spectroscopy

During the summer of 2023, I worked with an R.E.U. team on a National Science Foundation grant at California State University Long Beach. We were given the opportunity to develop our own unique projects around geography and environmental science. Being the only physics student on the team, I wanted to somehow branch the three fields. My project was formed around my idea that we could apply astronomy techniques used to identify seemingly minute differences in stars to current plant species classification programs.

The following research focused on utilizing a spectrometer mounted on a drone (sUAS) to develop a plant species spectral library. This library served as training data for a machine learning program I built through the ENVI software, which enabled it to identify specific tree species. We first used the sUAS's to capture a 10-band orthomosaic image of a Riparian Area at a conservation research station located within the Sierra Nevada foothills. We then spent 4 weeks camping at the research station while taking spectral data of different plant species. After returning to the campus, I built a supervised classification system in the ENVI 5.7 software that analyzed the spectral data against the orthomosaic image, allowing for the computer to identify differences in species that can not be seen with the human eye. Although I was not able to get a fully classified map of different species, I successfully created a method for future researchers to identify invasive species, improve image classification software, and support extensive vegetation surveys.