Stochastic parameterizations of within-grid concentration variability distribution functions in CMAQ

Stochasticity in photochemical grid models such as Community Multiscale Air Quality (CMAQ) includes inherently unresolved and thus typically ignored within-grid variability. Characteristics of such sub-grid variabilities (SGV) are expected to be highly dependent on the degree of the underlying land surface inhomogeneities, the strength and spatial distribution of emission sources, chemistry, meteorology and of course, the grid size chosen for the simulations. For a number of applications including the exposure assessment studies, resolving the SGV that is inherent to the photochemical grid models is critical to characterizing the neighborhood scale concentrations of air toxics compounds. We present and describe a conceptual framework which introduces stochastic models as a priori distribution functions (DFs) for representing SGV into operational CMAQ simulations. The template involves fitting analytic forms of DFs to the histograms of concentrations from fine scale modeling (or hybrid approaches) of within-grid concentration and subsequently developing practical parameterizations of these analytic variables of the fitted distributions based on each grid cell's land use, meteorological and emissions features. Once determined and parameterized, these DFs can be utilized offline to complement each and all CMAQ coarse grids simulations; and especially for improved means to link air quality simulations to exposure models in urban areas where concentration variations are relatively large. Results of fitted Weibull distributions for 12 km grids based on 1 km fine scale modeling for Delaware are presented to illustrate and contrast (a) the gridded parameterizations and (b) for several pollutants species whose reactivities range from relatively inert, to those that are moderately and highly reactive.