PNNL

Abstract

A fully linear advection algorithm is developed to improve the representation of aerosols and clouds in atmospheric models. Linear advection schemes satisfy superposition and thereby preserve tracer correlations during the advection of aerosol and cloud particle populations in Eulerian models. Nevertheless, the basic linear scheme, employing zerothorder finite differencing, is rarely used in coarse gridded models on account of excessive numerical diffusion. Higher-order finite difference schemes have been developed and are in widespread use, but these present new problems, such as the need to introduce nonlinear corrections like filling and flux limitation to avoid negative concentrations, spurious oscillations, and other non-physical effects. Generally successful at reducing numerical diffusion during the advection of individual tracers, e.g. particle number or mass, these models fail to preserve even the simplest correlations between interrelated tracers. Important attributes of aerosol and cloud populations - radial moments of particle size distributions, molecular precursors related through chemical equilibria, aerosol mixing state, and cloud phase are thereby poorly represented. We introduce a new scheme, minVAR, that is both non-diffusive and non-dispersive, satisfies linear superposition, fully preserves tracer set correlations, and is capable of tracking sub-grid information at arbitrarily fine scales with high computational efficiency.