Inferred sea spray generation functions in the marine atmospheric boundary layer using an Eulerian-Lagrangian model

Prediction of the temporal change and spatial distribution of sea spray aerosol particles is critical to understand their effect on momentum and energy exchange with the atmosphere, their role as cloud condensation nuclei, and their importance to atmospheric radiative transfer. A key component of this is to assess the total amount of aerosol particles generated along the sea surface with regard to different sizes and atmospheric conditions. In practice, field instruments that sample aerosol particles in the atmosphere have been used to estimate sea spray generation functions (SSGF) and infer effective fluxes within the surface layer. However, field observations are limited in that they cannot sample in full temporal and spatial resolution. As a result, numerical simulations are used to simulate the marine atmospheric boundary layer (MABL) in an idealized manner to understand the uncertainty associated with limited time and restricted spatial sampling. Large eddy simulations (LES) with Lagrangian particles are utilized in order to understand the uncertainty in measured fluxes of aerosol particles. From directly simulated flux measurements, the bias in sampling is found to be strongly attributed to the existence of large coherent turbulent structures. These large structures are responsible for the lack of convergence to the true vertical flux at any height, where uncertainty worsens within the mixed layer. Using eddy correlations, uncertainty is further quantified with regard to aerosol particle sampling in the streamwise or spanwise direction relative to persistent flow structures. In order to obtain a vertical flux measurement of aerosol particles with minimal sampling uncertainty, it is advised to sample in the spanwise direction due to these large-scale motions. In this work, we will show the uncertainty of inferred surface fluxes to that of the true simulated flux with respect to different spatial and temporal averaging, direction-based sampling, and theoretical profiles.