Prepared for California Air Resources Board Prepared

A condensed version of the SAPRC-07 mechanism (Carter, 2008) was implemented in a 3-D air quality model and applied to the Los Angeles area as part of the present review. The CIT airshed model was used; see Martien et al. (2003 and 2006) for description of model input data. The summer 1997 emission inventory was used here. Martien et al. used an extended version of SAPRC-99 that had many individual organic species treated explicitly so that incremental reactivities could be studied; here a more standard version of SAPRC-99 was used, with 5 lumped alkanes, ethene, isoprene, 2 lumped anthropogenic alkenes, lumped terpenes, and 2 lumped aromatics. To help ensure that any differences in model results were due to differences in the descriptions of atmospheric chemistry, the dry deposition calculations were turned off, and secondary aerosol formation was likewise disabled in these 3-D model test runs. The version of SAPRC-07 used was CS07A without chlorine chemistry. This mechanism uses the same operator species as in the earlier SAPRC-99, and is compatible with existing mechanism preparation software. Airshed model outputs using CS07A were compared against analogous predictions obtained using the older SAPRC-99 mechanism. The emissions in both model runs were identical, except that ALK1, ALK2 and ALK3 were lumped together (with reactivity weighting) in SAPRC-07, and likewise ALK4 and ALK5 were combined (unweighted sum). Also acetone (ACET) and methyl ethyl ketone (MEK) emissions in SAPRC-99 were combined with reactivity weighting and represented using PROD2 in SAPRC-07. The summary table below provides some further details on relevant mechanisms. # of reactions (no chlorine chemistry) 139 221 210 # of species (excluding constant) 47 74 74 In this test case, airshed model run time decreased by 30% relative to SAPRC-99 when condensed SAPRC-07 (CS07A) was used. This will speed up 3-D air quality models that currently use SAPRC-99: large numbers of computational cells are involved and the numerical integration of stiff systems of differential equations describing atmospheric chemistry can be a major contributor to the overall computational burden. The speed-up of 30% reported here is in reasonable agreement with the decrease in the number of active chemical species being tracked in CS07A relative to SAPRC-99, as shown above. There is a similar reduction in the number of chemical reactions listed as well. Note that these test runs were deliberately simplified by turning off secondary 1 aerosol and dry deposition calculations; the speed-up in air …