OES Fall Seminar - Megan Coffer, ORISE Fellow, EPA
- 10/27/2022 3:00 PM EST - 4:00 PM EST
- Oceanography & Physical Sciences Building - 200
- Megan Coffer, a postdoctoral ORISE fellow at the U.S. Environmental Protection Agency (EPA) Office of Research and Development, will speak. Please join us for a reception in OCNPA 404 after the seminar.
The Ocean & Earth Sciences departmental seminar informs students, staff, faculty and the University community about recent issues. For more than 40 years, the series has fostered connection among current and future professionals. Seminars are held most Thursdays at 3 p.m. during the fall and spring semesters.
For information, email Professor Tal Ezer at email@example.com.
Eyes in the sky lend support for water quality monitoring across inland and coastal environments
Satellite remote sensing can complement field observations by offering improved spatial and temporal resolutions, collecting imagery more frequently and over larger areas. Here, imagery from several satellites is used to monitor water quality in the United States as indicated by both cyanobacterial occurrence and seagrass extent. Cyanobacterial blooms act as a direct measure of water quality while seagrass coverage is often highly responsive to the conditions of its environment. Imagery from a publicly available satellite sensor, the European Space Agency's Ocean and Land Colour Imager, was used to assess the temporal frequency and occurrence of cyanobacterial blooms at over 2,000 inland lakes across the United States. A subset of this dataset was also used to assess the prevalence of cyanobacterial blooms at nearly 700 drinking water intakes nationwide. Generally, cyanobacterial occurrence followed well-known ecological patterns, with some exceptions. Additionally, imagery from two commercial, high spatial resolution satellite sensors, Maxar's WorldView-2 and WorldView-3 platforms, was used to map seagrass extent at twelve coastal sites across the United States, representing geographically, ecologically, and climatically diverse regions. First, a reproducible processing regime was developed to transform imagery from basic products, as delivered from Maxar, into analysis-ready data usable for various scientific applications. Next, a deep convolutional neural network was used to classify imagery into four general classes: land, seagrass, no seagrass, and no data. Satellite classification performed best in areas of dense, continuous seagrass compared to areas of sparse, discontinuous seagrass and provided a suitable spatial representation of seagrass distribution within each study area.