Forward and Inverse Analysis of Chemical Transport Models

Citation

Henze, Daven Ker (2007) Forward and Inverse Analysis of Chemical Transport Models. Dissertation (Ph.D.), California Institute of Technology. doi:10.7907/5647-T777. https://resolver.caltech.edu/CaltechETD:etd-05312007-133239

Abstract

Assessing the discrepancy between modeled and observed distributions of aerosols is a persistent problem on many scales. Tools for analyzing the evolution of aerosol size distributions using the adjoint method are presented in idealized box model calculations. The ability to recover information about aerosol growth rates and initial size distributions is assessed given a range of simulated observations of evolving systems. While such tools alone could facilitate analysis of chamber measurements, improving estimates of aerosol sources on regional and global scales requires explicit consideration of many additional chemical and physical processes that govern secondary formation of atmospheric aerosols from emissions of gas-phase precursors. The adjoint of the global chemical transport model GEOS-Chem is derived, affording detailed analysis of the relationship between gas-phase aerosol precursor emissions (SOx, NOx, and NH3) and the subsequent distributions of sulfate - ammonium - nitrate aerosol. Assimilation of surface measurements of sulfate and nitrate aerosol is shown to provide valuable constraints on emissions of ammonia. Adjoint sensitivities are used to propose strategies for air quality control, suggesting, for example, that reduction of SOx emissions in the summer and NH3 emissions in the winter would most effectively reduce non-attainment of aerosol air quality standards. The ability of this model to estimate global distributions of carbonaceous aerosol is also addressed. Based on new yield data from environmental chamber studies, mechanisms for incorporating the dependence of secondary organic aerosol (SOA) formation on NOx concentrations are developed for use in global models. When NOx levels are appropriately accounted for, it is demonstrated that sources such as isoprene and aromatics, previously neglected as sources of aerosol in global models, significantly contribute to predicted SOA burdens downwind of polluted areas (owing to benzene and toluene) and in the free troposphere (owing to isoprene).

Thesis Files