Cyanobacteria Abundance Modeling: Development and Assessment of Season- Ahead Forecasts To Improve Beach Management on Lake Mendota

Abstract

As eutrophication and the associated dominance of cyanobacteria continue to increase across inland waterbodies, local officials are seeking novel methods to proactively manage water resources. Presently, however, there is no long-term cyanobacteria forecast that can provide advance warning of a potential threat in the upcoming season for beach managers. In this study, a statistical model is developed utilizing local and global scale season-ahead hydro-climatic predictors to evaluate the potential for informative cyanobacteria biomass forecasts across the June-August (JJA) season to aid in beach management. Model skill is quite strong in comparison to JJA cyanobacteria observations (R2=0.52, RPSS=0.64). Cyanobacteria predictions are in turn used to inform categorical forecasts of JJA beach days closed, specifically the probability of normal (0-2 days closed) and above normal (> 2 days closed). Forecast categories and observed categories tend to correlate strongly, with 70% to 90% of predictions falling into the correct category across the six beaches modeled, demonstrating encouraging prediction skill. To assess the potential application and value of the forecasts developed, local stakeholders, including non-profit organizations, University of Wisconsin – Madison affiliates, and government agencies that are involved in various aspects of water resources research and management, were asked to participate in a qualitative study consisting of an online survey and focus group discussion session. From both the survey and the focus group discussion session we identified an expressed interest from stakeholders for integrating season-ahead cyanobacteria and beach closing forecasts into current beach management practices, particularly for the purposing of improving public safety and increasing public awareness. However, there were alternative stakeholder visions of the value of the forecasts, likely due to the absence of real world examples. Nonetheless, we conclude that there is clear untapped potential for the application of season-ahead forecasts to local beach management.