Large Scale Parameter Sweeps Procedures for developing condensed aerosol schemes for climate models

Aerosol physical and chemical process models involve multiple physical parameters, chemical pathways, multiple components and microphysical processes. Representation of aerosols in climate models has proved to be particularly challenging as a result. The increased computational burden as a result of including aerosol processes models in all its details has not yet been justified. A need exists for developing approximate approaches to develop and represent condensed forms of aerosol formation, portioning between various chemical compositions and the effect of atmospheric parameters on these processes. In addition, there is a need to develop a condensed chemical mechanism for represent secondary organic aerosols from all the currently hypothesized organic gas precursor chemical pathways.

Here we have implemented a large-scale parameter sweep procedure on a modern high performance computing facility using a 1-D atmospheric chemistry, and aerosol chemistry-physics model. The chemistry schemes implemented include RACM, CB4, and SAPRC-99 for the chemistry of volatile organic compounds; JPL for inorganic chemistry; SORGAM for secondary organic aerosols; FAST J for photolysis rates; and ISOROPIA for thermodynamic equilibrium calculations for inorganic aerosols. In addition, this 1 D model includes a PBL dynamics model based on k epsilon theory, achieving a 2.5 level closure for the Reynolds stress in calculating turbulence in the PBL. The PBL is very finely resolved in this 1 D model, which has been used to investigate fast reactions and interactions with PBL turbulence [1]. The coupling of this model with the climate model radiation code CRM allows us to investigate radiative feedbacks from selected aerosols for direct calculation of radiative forcing.

The parameter sweep procedure uses a workflow management tool known as SWIFT in combination with a latin hypercube sampling procedure implemented for a multi-parameter aerosol system describing the NH3-NO3-SO4 inorganic system. Results from these parameter sweeps, ranging over a 1000 to 100,000 simulations and covering a wide range of physical, chemical and input parameters to this aerosol processes in the 1-D model will be presented. Methods for developing a condensed parametric representation of this process for climate model applications will be discussed.