Vertical distribution of buoyant Microcystis blooms in a Lagrangian particle tracking model for short-term forecasts in Lake Erie

Abstract

Cyanobacterial harmful algal blooms (CHABs) are a problem in western Lake Erie, and in eutrophic fresh waters worldwide. Western Lake Erie is a large (3000 km2), shallow (8 m mean depth), freshwater system. CHABs occur from July to October, when stratification is intermittent in response to wind and surface heating or cooling (polymictic). Existing forecast models give the present location and extent of CHABs from satellite imagery, then predict two-dimensional (surface) CHAB movement in response to meteorology. In this study, we simulated vertical distribution of buoyant Microcystis colonies, and 3-D advection, using a Lagrangian particle model forced by currents and turbulent diffusivity from the Finite Volume Community Ocean Model (FVCOM). We estimated the frequency distribution of Microcystis colony buoyant velocity from measured size distributions and buoyant velocities. We evaluated several random-walk numerical schemes to efficiently minimize particle accumulation artifacts. We selected the Milstein scheme, with linear interpolation of the diffusivity profile in place of cubic splines, and varied the time step at each particle and step based on the curvature of the local diffusivity profile to ensure that the Visser time step criterion was satisfied. Inclusion of vertical mixing with buoyancy significantly improved model skill statistics compared to an advection-only model, and showed greater skill than a persistence forecast through simulation day 6, in a series of 26 hindcast simulations from 2011. The simulations and in situ observations show the importance of subtle thermal structure, typical of a polymictic lake, along with buoyancy in determining vertical and horizontal distribution of Microcystis.