Chemical Sensitivity Analysis and Uncertainty Analysis of Ozone Production in the Comprehensive Air Quality Model with Extensions Applied to Eastern Texas.

Chemical sensitivity analysis (CSA) is a new probing tool for sampling sensitivities to chemistry parameters during a three-dimensional (3-D) simulation. CSA was applied to rank all of the parameters in the Carbon Bond 6 revision 4 (CB6r4) mechanism and to create an ensemble of six chemical mechanisms representing higher and lower O3 formations than CB6r4. This ensemble of mechanisms was used to estimate the uncertainty from the chemistry in a 3-D simulation and combined with uncertainties from other model inputs obtained from calculations of their sensitivities. The overall uncertainty (1σ) in O3 predictions for eastern Texas was 10-11 ppb in the Gulf of Mexico near Galveston and 7-8 ppb in much of the rest of the domain on the higher O3 days of June 2012. As a percent of the O3 concentration, the uncertainty was more uniform over the domain, 11-14%. Chemistry and emissions make the largest contributions to the O3 uncertainty. Uncertainty in the dry deposition velocities is less important in urban areas and the Gulf, but it is similar in importance to the uncertainty in chemistry and emissions at most other locations. Uncertainty in O3 boundary concentrations is the least important.

Source Apportionment of the Anthropogenic Increment to Ozone, Formaldehyde, and Nitrogen Dioxide by the Path-Integral Method in a 3D Model.

The anthropogenic increment of a species is the difference in concentration between a base-case simulation with all emissions included and a background simulation without the anthropogenic emissions. The Path-Integral Method (PIM) is a new technique that can determine the contributions of individual anthropogenic sources to this increment. The PIM was applied to a simulation of O3 formation in July 2030 in the U.S., using the Comprehensive Air Quality Model with Extensions and assuming advanced controls on light-duty vehicles (LDVs) and other sources. The PIM determines the source contributions by integrating first-order sensitivity coefficients over a range of emissions, a path, from the background case to the base case. There are many potential paths, with each representing a specific emission-control strategy leading to zero anthropogenic emissions, i.e., controlling all sources together versus controlling some source(s) preferentially are different paths. Three paths were considered, and the O3, formaldehyde, and NO2 anthropogenic increments were apportioned to five source categories. At rural and urban sites in the eastern U.S. and for all three paths, point sources typically have the largest contribution to the O3 and NO2 anthropogenic increments, and either LDVs or area sources, the smallest. Results for formaldehyde are more complex.

Comparison of source apportionment and sensitivity analysis in a particulate matter air duality model.

Two efficient methods to study relationships between particulate matter (PM) concentrations and emission sources are compared in the three-dimensional comprehensive air quality model with extensions (CAMx). Particulate source apportionment technology (PSAT) is a tagged species method that apportions concentrations of PM components to their respective primary precursors, e.g., sulfate is apportioned to SOx, nitrate to NOx, etc. The decoupled direct method (DDM) calculates first-order sensitivities of PM concentrations to model inputs. Both tools were applied to two month long (February and July) PM modeling episodes and evaluated against changes in PM concentrations due to various emission reductions. The results show that source contributions calculated by PSAT start to deviate from the actual model responses as indirect effects from limiting reactants or nonprimary precursor emissions become important The DDM first-order sensitivity is useful for determining source contributions only if the model response to input changes is reasonably linear. For secondary inorganic PM, the response is linear for emission reductions of 20% in all cases considered and reasonably linear for reductions of 100% inthe case of on-road mobile sources. The model response for secondary organic aerosols and primary PM remains nearly linear to 100% reductions in anthropogenic emissions.

Implementing the decoupled direct method for sensitivity analysis in a particulate matter air quality model.

The decoupled direct method (DDM) is an efficient and accurate way of performing sensitivity analysis to model inputs. As the impact of atmospheric particulate matter (PM) on human health and visibility became evident, the need to extend the DDM to PM sensitivity has grown. In this work, the DDM is implemented in the three PM modules employed in the Comprehensive Air-quality Model with extensions (CAMx): ISORROPIA for inorganic gas/aerosol partitioning, SOAP for secondary organic gas/aerosol partitioning, and RADM-AQ for aqueous-phase chemistry. The PM modules are complex and the DDM implementation is discussed in detail. Stand-alone tests are performed for each PM module in which first-order sensitivities computed bythe DDM are compared tothe traditional brute-force method (BFM). The DDM sensitivities are shown to be accurate and agree well with the BFM within the linear response range. The SOAP module showed nearly linear response for up to +/-30% changes in concentration inputs. The RADM-AQ module showed moderately nonlinear response in some tests but first-order sensitivities accounted for most of the response for input changes up to +/-20%. ISORROPIA shows greater deviation from linear response than the other PM modules and the near-linear range can be as restricted as +/-10% changes in concentration inputs. Nonlinearity in ISORROPIA results both from the equations that describe thermodynamic equilibrium and the computational approaches within ISORROPIA that are employed to boost efficiency.

Modeling weekday/weekend ozone differences in the Los Angeles region for 1997.

Numerous studies of ambient ozone (O3) in the Los Angeles (LA) area have found both increases and decreases in elevated O3 levels on weekends, depending on location and year. Since the mid-1990s, average daily maximum O3 levels have been higher on weekends than on weekdays throughout most of the area. We used the Comprehensive Air-Quality Model with extensions to investigate causes of weekday/weekend O3 differences in the LA area for August 3-7, 1997, from the Southern California Ozone Study. Weekday/weekend emission changes were estimated, because explicit weekend inventories are not yet available from regulatory agencies. Changes to on-road motor vehicle (MV) emissions were derived from observed weekday/weekend traffic differences. The estimated changes in MV emissions of nitrogen oxides (NOx) were a 5% increase on Friday, a 27% decrease on Saturday, and a 37% decrease on Sunday, relative to Monday-Thursday levels. The corresponding changes in MV volatile organic carbon (VOC) emissions were an 8% increase on Friday, an 8% decrease on Saturday, and a 15% decrease on Sunday. Modeling these MV emissions changes explained the observed weekend O3 effect very well. Furthermore, changes to the mass of MV NOx emissions were the main contributor to O3 differences rather than changes to the timing of MV emissions. Ozone increases on weekends were caused by NOx emission decreases, because O3 formation is strongly VOC-limited throughout most of the LA area.

Comparison of source apportionment and source sensitivity of ozone in a three-dimensional air quality model.

The ozone source apportionment technology (OSAT) estimates the contributions of different sources to ozone concentrations using a set of tracers for NOx, total VOCs, and ozone and an indicator that ascribes instantaneous ozone production to NOx or VOCs. These source contributions were compared to first-order sensitivities obtained by the decoupled direct method (DDM) for a three-dimensional simulation of an ozone episode in the Lake Michigan region. The cut-point for the OSAT indicator between VOC- and NOx-sensitive ozone production agrees well with the DDM sensitivities to VOC and NOx. In a ranking of the most important contributors to ozone concentrations >80 ppb, the OSAT and DDM results agreed on four of the top five contributors on average. The spatial distributions of the sensitivities and source contributions are similar, and the OSAT and DDM results for ozone >80 ppb correlate well. However, the source contributions ascribe substantially less relative importance to anthropogenic emissions and greater relative importance to the boundary concentrations than do the sensitivities. In regions where NOx inhibits ozone formation and the sensitivity is negative, the source contribution is small and positive. For the same subdivision of the emissions, the OSAT is 14 times faster than the DDM, but the DDM has greater flexibility in defining which emissions to include and generates results for species other than ozone. The first-order sensitivities explain, on average, 70% of the ozone concentrations.

The decoupled direct method for sensitivity analysis in a three-dimensional air quality model--implementation, accuracy, and efficiency.

The decoupled direct method (DDM) has been implemented in a three-dimensional (3D) air quality model in order to calculate first-order sensitivities with respect to emissions and initial and boundary concentrations. This required deriving new equations for the sensitivities from the equations of the hybrid chemistry solver and the nonlinear advection algorithm in the model. The sensitivities for the chemistry and advection steps were tested in box-model and rotating-hill simulations, respectively. The complete model was then applied to an ozone episode of the Lake Michigan region during July 7-13, 1995. The DDM was found to be highly accurate for calculating the sensitivity of the 3D model. The sensitivities obtained by perturbing the inputs (brute-force method) converged toward the DDM sensitivities, as the brute-force perturbations became small. Ozone changes predicted with the DDM sensitivities were also compared to actual changes obtained from simulations with reduced inputs. For 40% reductions in volatile organic compound and/or NOx emissions,the predicted changes correlate highly with the actual changes and are directionally correct for nearly all grid cells in the modeling domain. However, the magnitude of the predicted changes is 10-20% smaller than the actual changes on average. Agreement between predicted and actual ozone changes is better for 40% reductions in initial or boundary concentrations. Calculating one sensitivity by the DDM is up to 2.5 times faster than calculating the concentrations alone.