Abstract

Harmful algal blooms (HABs) and nuisance algal blooms (NABs) are a worldwide phenomenon with implications for human health and safety. HABs occur when algae or bacteria grow in high enough densities to harm animals and humans. A primary component of harmful algal blooms is cyanobacteria, which are aquatic, photosynthesizing microorganisms that produce toxins at high concentrations. Cyanobacterial biomass has increased worldwide in recent decades, raising concern about the future of fresh- and marine-water systems in a changing climate. Understanding the patterns and conditions of past algal blooms can provide useful insights for managing future blooms. Remote sensing can enhance our understanding of the spatiotemporal distribution of HABs and NABs. We used radiometrically corrected images from the USGS Landsat Collections available in the Google Earth Engine for cloud processing. In 2016, the USGS calibrated the sensors of Landsat 4, 5, 7, and 8 to create a continuous collection of satellite images from 1984 to present. We use this 34-year dataset to expand the historical record of algal blooms at our study site and to understand factors relating to the spatiotemporal patterns of these blooms. We applied three models, including the Floating Algae Index (FAI), the Normalized Difference Vegetation Index (NDVI), and one developed with in situ chlorophyll-a (chl-a) data, to 398 images masked for cloud cover and lake elevation taken from 34 growing seasons (April – October). We found that the Normalized Difference Water Index (NDWI) used to separate water and land pixels fails under algal bloom conditions, whereas a modified NDWI does not. We also performed an emerging hot spot analysis in ArcGIS using the chlorophyll-a, NDVI, and FAI predictions from the surface reflectance values of the images. Our analysis indicates that the Provo Bay and parts of the eastern shoreline of Utah Lake have had algal blooms for 30 out of the 34 years included in this study, rendering them enduring hot spots. The remainder of the lake is a cold spot, showing clusters of low mean chl-a, NDVI, and FAI values over time. The overall trend of mean NDVI and lake surface area over this 34-year dataset is decreasing, whereas lake water temperature is increasing. This study develops a method for analyzing algal blooms over multiple decades and provides useful information for the management and prediction of future blooms.

Degree

College and Department

Life Sciences; Plant and Wildlife Sciences

Rights

http://lib.byu.edu/about/copyright/