Supp Figure S1. Construction of emission data for the Detroit 4km domain.Left: baseline ground-level SO2 emissions, comprising of gridded surface emissions and elevated inline point sources (not shown). Right: idealized SO2 emissions. Note the different scales between the two panels.
Supp Figure S2. Comparison of contributions of added idealized precursor emissions to dry depositions as calculated by ISAM versus zero-out in the form of scattered density plots. Density is defined by the percentage of the total sample (36 columns × 45 rows × 1 layer × 24 hours × 31 days) at any dry deposition result. Samples are all hourly data in January 2005 on 4km Detroit domain. Results are shown for (a) I+J mode elemental carbon from contributions of idealized primary EC emissions; (b) SO2, (c) sulfuric acid, (d) I+J mode sulfate from contribution of idealized SO2 emissions;(e) NO+NO2, (f) nitric acid, (g) I+J mode nitrate from contributions of idealized NOx emissions; and(h) ammonia gas, and (j) I+J mode ammonium from contributions of idealized ammonia gas emissions.
Supp Figure S3. Comparison of contributions of added idealized precursor emissions to wet depositions as calculated by ISAM versus zero-out in the form of scattered density plots. Density is defined by the percentage of the total sample (36 columns × 45 rows × 1 layer × 24 hours × 31 days) at any wet deposition result. Samples are hourly data in January 2005 on 4km Detroit domain. Results are shown for (a) I+J mode elemental carbon from contributions of idealized primary EC emissions; (b) sulfate from contributions of idealized SO2 emissions; (c) nitrate from contributions of idealized NOx emissions; and (d) ammonium from contributions of idealized ammonia gas emissions.
Supp Figure S4. Comparison of I+J mode sulfate, at different combinations ofthe five CMAQ aqueous sulfate production pathways, between one scenario representing the sum of contributions from separately zeroed-out idealized SO2 emissions and contributions from separately zeroed-out standard emissions (vertical axis), and another scenario with both types of emissions (horizontal axis).Results are shown for January 2, 2005 on the 4 km Detroit domainby turning (Top left) all five pathways off; (top right) only H2O2on; (middle left) only O3on; (middle right) only catalytic Mn2+-Fe3+on; (bottom left) only MHP on; and(bottom right) only PAA on.
Supp Figure S5. Comparison of contributions to aerosol nitrate from boundary conditions (BCON) in the U.S. domain averaged over January 2010. Results are shown for (Left) ISAM contribution; and (right) zero-out contribution.
Supp Figure S6. Month-to-month variations of emission sector contributionsto four aerosol species concentrations for the continental U.S. 36 km domain average during 2010. Sector contributions are shown for (top left) EC, (top right) sulfate, (bottom left) ammonium, and (bottom right) nitrate. The colors in the legend represent: light blue for initial conditions (ICON), orange for boundary conditions (BCON), blue for emissions of unspecified sources (OTHR), yellow for emissions of agricultural source (AG), green for emissions of on-road mobile source (ONROAD), and red for emissions of electricity generation units (EGU). Circles joined by line segments indicate bulk concentrations of the species in each panel, calculated from the host model.
Concluding remarks
Evaluating source attribution methodology such as ISAM is challenging. Attributing emission sources to the resulting pollutant concentrations (source apportionment) is quite different from estimating a concentration change by perturbing those emissions (source sensitivity), since the balance of chemical mass and associated ions changes between the two simulations of the zero-out approach. Still, zero-out simulations can provide a qualitative comparison with source-apportioned mass by exhibiting generally similar spatial distributions and concentration levels for most pollutants. The two approaches are not meant to provide identical results, but complement each other by addressing two different but related questions: “what impact can a source have on the model predictions (source apportionment)” and “how are the model predictions changed if emissions from contributors changed (zero-out)”. In the context of emissions controls, source apportionment tools suggest possible major sources contributing to the current pollutant level, while sensitivity studies, including the deployment of the zero-out method, provide scenarios that would be caused by perturbing those major sources.