Tropospheric ozone, an air pollutant and greenhouse gas, is a hazard to both human health and ecosystems. Health-based ozone standards are regularly exceeded in many regions of the world. Controlling ozone formation is a challenge for scientists and policy makers due to the non-linear dependence of ozone on its precursors, namely nitrogen oxides (NOx) and volatile organic compounds (VOC). Photochemical air quality modeling is one the primary tools used to assess the effectiveness of strategies intended to control ozone formation. Air quality model predictions are sensitive to meteorological and emission inputs, as well as the description of atmospheric chemistry embodied in the chemical mechanism. Typical chemical mechanisms describe the reactants, reaction rates, and product yields for each of hundreds of individual chemical reactions that take place in the atmosphere.
The Statewide Air Pollution Research Center (SAPRC) mechanism developed by Dr. William Carter at UC Riverside is a chemically accurate, but also computationally intensive, example of a chemical mechanism used to describe the dynamics of photochemically-formed air pollutants. This mechanism is widely used for air quality research and planning at urban and regional scales. The first step in my research was to assess the effects of using a new version (SAPRC07) of this mechanism in place of an outdated earlier version (SAPRC99). This assessment was done using an Eulerian photochemical air quality known as the Community Multiscale Air Quality model or CMAQ, applied to the prediction of air quality in Central California under polluted summertime conditions. Domain-wide reductions in predicted ozone were found using the newer mechanism as compared with SAPRC99, with larger ozone reductions colocated spatially and temporally with higher ozone concentrations. Underprediction of peak ozone was already evident using the older (SAPRC99) mechanism, and this bias increased with the new mechanism.
In order to understand the reasons for the changes in predicted ozone, first-order sensitivity coefficients using both mechanisms were calculated with respect to important rate coefficients. Among the more important revisions to the mechanism is a ~20% increase in the rate of the chain-terminating reaction OH. + NO2 → HNO3. This revision in SAPRC07 was a dominant contributor to overall decreases (relative to SAPRC99) of up to 25 ppb in peak predicted ozone concentrations in the San Joaquin Valley. Other significant revisions include the rate coefficient for the reaction O3 + NO → NO2 + O2 and the photolysis rate for the reaction NO2 + hv → NO + O3P, but these changes were found to exert offsetting effects with ozone responses of similar magnitude acting in opposite directions.
A reassessment of ozone sensitivity to changes in precusor emissions shows that the updated mechanism has a response to VOC emissions that is stronger than that of SAPRC99, indicating more ozone reduction potential associated with VOC emission controls. With the exception of the Bay Area, the newer mechanism drives the model predictions more toward a NOx disbenefit regime, where moderate NOx emission reductions (by 20%) lead to local increases in ozone in the near field. An unexpected finding was that use of a condensed and computationally efficient version of SAPRC07 known as CS07A contributed substantially to the decreases in predicted ozone levels mentioned above. For this reason, a less condensed mechanism (SAPRC07A) is used in further research presented in this dissertation, with a resulting increase in computational demand necessary for model simulations. Additionally a recent study (Mollner et al., 2010) recommends a lower value for the reaction rate coefficient for OH. + NO2 → HNO3 that I have implemented as a revision to SAPRC07A.
The many hundreds of individual anthropogenic VOC emitted into the atmosphere are primarily represented in the SAPRC mechanism using lumped species groups. Species are separated into groups based on their reaction mechanism and rate coefficient, and surrogate species are used to represent the reactions of all VOC assigned to each lumped group. The existing surrogate species mixtures defining the chemistry of the lumped groups were developed by Carter based on speciated measurements of ambient hydrocarbon concentrations, but these data were collected in the mid-1980s. Reformulation of fuels and consumer products, and control measures that have altered the total amounts of VOC emitted by various source categories, have substantially changed the concentrations and relative abundances of many individual VOC present in the atmosphere. The next step in my research was to develop updated lumped species definitions using emission inventories and source category-related VOC speciation profiles. The resulting lumped species definitions created in this dissertation are up-to-date and simpler, but still represent 95% of the total moles of species assigned to each lumped group. Most of the inventory-based lumped VOC definitions now align better with ambient VOC measurements compared to the existing lumped species definitions in SAPRC99 and SAPRC07. The updated definitions result in lower rate coefficients for most lumped VOC + OH. reactions, that reflect the effects of emission control measures that have required reformulation to reduce the reactivity of VOC mixtures used in fuels and consumer products. In the air quality model, the new surrogate species definitions result in slightly lower peak ozone concentrations, as much as a 1.8 ppb reduction in houly maximum ozone concentration in some cases.
Consistent with current practice in other mechanisms, the SAPRC mechanism describes the atmospheric oxidation of VOC using temperature-independent parameters that specify branching in peroxy radical + NO reactions between NO to NO2 conversion and organic nitrate-forming channels. Additional temperature-independent parameters are used to describe the fate of alkoxy radicals: reaction with O2, isomerization, and decomposition. In reality, the branching ratios for these pathways depend on temperature. The final major element of my research, needed to reproduce more accurately the spatial, diurnal, and day-to-day differences in air quality that are driven in part by differences in temperature, is development of a new version of the mechanism that captures the missing temperature dependencies of alkoxy and peroxy radical reactions. Organic nitrate formation, which decreases with temperature, competes with NO to NO2 conversion and hence inhibits production of ozone locally. Alkoxy radical decomposition competes with isomerization, resulting in increased aldehyde yields and decreased ketone yields with increasing temperature. This drives the resulting VOC oxidation products toward a more reactive mix at higher temperatures. The VOC oxidation product yields for the existing SAPRC mechanism are all specified for T = 300 K, and do not vary with temperature. Changes in predicted organic nitrate, aldehyde, and ketone concentrations follow as expected with inclusion of these missing temperature dependencies in the mechanism, as do subsequent changes in predicted ozone. Variations in predicted ozone linked to temperature differences are further enhanced with the additional inclusion of temperature-dependent evaporative VOC emissions. However, predicted ozone is not very sensitive to these mechanism changes, the largest changes were in the maximum hourly ozone (up to ~±1.5 pbb).
Temperatures in California are predicted to rise by 1 to 4oC throughout the state of California by 2050 relative to year 2000. In studying the effects of climate change on air quality, it is important to describe correctly the temperature dependence of relevant atmospheric chemistry. The effect of using the temperature-dependent VOC reaction mechanism developed as part of this research is assessed for a future climate change scenario. Compared to previous assessments, use of the revised chemical mechanism indicates a slightly increased potential penalty from climate change with respect to ozone air quality in central California.