Sparse Aerosol Models Beyond the Quadrature Method of Moments

摘要

This study examines a class of sparse aerosol models derived from linear programming (LP). The widely used quadrature method of moments (QMOM) is shown to fall into this class. Here it is shown how other sparse aerosol models can be constructed, which are not based on moments of the particle size distribution. The new methods enable one to bound atmospheric aerosol physical and optical properties using arbitrary combinations of model parameters and measurements. Rigorous upper and lower bounds, e.g. on the number of aerosol particles that can activate to form cloud droplets, can be obtained this way from measurement constraints that may include total particle number concentration and size distribution moments. The new LP-based methods allow a much wider range of aerosol properties, such as light backscatter or extinction coefficient, which are not easily connected to particle size moments, to also be assimilated into a list of constraints. Finally, it is shown that many of these more general aerosol properties can be tracked directly in an aerosol dynamics simulation, using SAMs, in much the same way that moments are tracked directly in the QMOM.