Comparison of Eulerian and Lagrangian transport models for harmful algal bloom forecasts in Lake Erie.

Lake Erie has experienced a re-emergence of cyanobacterial harmful algal blooms (CHABs) since the early 2000s, posing significant socioeconomic and ecological consequences that impact drinking water, human health, fish-eries, tourism, and water quality. As predicting CHAB intensity and spatial distribution is critical to Lake Erie ecosystem management, this study focuses on a comprehensive evaluation of Lagrangian and Eulerian transport models for Lake Erie CHAB forecasts, including 1) a Lagrangian particle model (LPM), 2) an Eulerian tracer model (ETM), and 3) a property-carrying particle model (PCPM) that utilizes the hybrid Eulerian-Lagrangian approach. We evaluated the models' performance against the latest high-resolution satellite product from the European Space Agency's Sentinel-3 OLCI sensor over 24-to 240-h hindcasts for each CHAB occurrence in three consecutive CHAB seasons (2017-2019). We examined the relative contributions of horizontal transport, vertical turbulent mixing, and algal buoyancy on the CHAB inter-and intra-day variability. In the short-term forecast, we emphasize the highly dynamic reaction of currents to weather-scale wind events that are crucial to CHAB transport. While statistical skill assessments show that these three transport models attain comparable levels of hindcast accuracy, we explore the advantages and disadvantages of each model in the context of general bio-physical modeling. In particular, the fact that the ETM and PCPM perform as well as or better than the LPM sets up a promising path to developing more biological realism in future operational forecast models using Eulerian or hybrid approaches.