Characterizing Lake Erie HAB dynamics through geostatistical synthesis of multiple sampling programs

Session: 37. - Harmful Algal Blooms (HABs) and their Toxicity: Remote Sensing and Modeling Approaches

Daniel Obenour, NC State University, [email protected]
Shiqi Fang, North Carolina State University, [email protected]
Joseph Guinness, NC State University, [email protected]
Caren Binding, Environment Canada, [email protected]
Thomas Bridgeman, Dept. of Environmental Sciences, University of Toledo, [email protected]
Justin Chaffin, Stone Laboratory, Ohio State University, [email protected]
Mary Anne Evans, USGS, Great Lakes Science Center, [email protected]
Thomas Johengen, CILER, University of Michigan, [email protected]
Richard Stumpf, NOAA, [email protected]
Timothy Wynne, NOAA, [email protected]

Abstract

Harmful algal blooms (HABs) of cyanobacteria have become common in western Lake Erie over the last decade. Accurate quantification of bloom size (total biomass and areal extent) is critical for assessing environmental impacts and for calibrating predictive models.  Remote sensing has been an important tool for assessing bloom size, but it has been limited by sensor resolution, cloud cover, and vertical bloom dynamics.  To provide more direct measures of bloom intensity, several monitoring programs have been performing ship-based sampling, though with limited spatial and temporal coverage.  In this study, we advance HAB characterization through the synthesis of five U.S. and Canadian sampling programs within an innovative space-time geostatistical model.  To address vertical HAB dynamics, we consider sampling method, wind speed, and depth as model covariates.  In addition, we incorporate bathymetry, geographic position, and day-of-year as covariates to explore large-scale trends and reduce estimation uncertainty when sampling is sparse.  Using the model, we map blooms and estimate biomass throughout summers, 2008-2017, and assess factors influencing bloom size and uncertainty.  Results provide unique estimates that can be compared to remotely sensed observations.  We also demonstrate how remote sensing can be incorporated within the model to further refine HAB estimates and reduce uncertainty.

1. Keyword
harmful algal blooms

2. Keyword
Spatial analysis

3. Keyword
modeling