Adjoint sensitivity analysis of ozone nonattainment over the continental United States

Responses of Photochemical Air Pollution in California's San Joaquin Valley to Spatially and Temporally Resolved Changes in Precursor Emissions

Ground-level ozone adversely affects human health and ecosystems. The effectiveness of control programs depends on which precursor(s) are controlled, by how much, and where and when emission reductions occur. We use the adjoint of the Community Multiscale Air Quality model to investigate odd oxygen (O_x ≡ O₃ + NO₂) sensitivities in California's San Joaquin Valley (SJV) to precursor emissions from local and upwind sources. Sensitivities are mapped and disaggregated by hour and day. Taken together, impacts of precursor emissions in the San Francisco Bay area and Sacramento Valley are similar in magnitude to impacts of local SJV emissions. Same-day emission sensitivities are mostly attributable to local sources, with the most influential anthropogenic emissions of VOCs (volatile organic compounds) and NO_x (nitrogen oxides) occurring in the morning (9-11 am) and early afternoon hours (1-3 pm), respectively. For the northernmost SJV receptor, the influence from Sacramento Valley emissions peaks 5-6 h later than Bay area emissions; this difference diminishes for SJV receptors located further downwind. Results show a shift toward more NO_x-sensitive conditions in the afternoon with all but the southernmost receptor shifting from VOC- to NO_x-sensitive conditions. We also evaluate opportunities to control pollution through shifts in precursor emission location and timing.

Optimal estimation of initial concentrations and emission sources with 4D-Var for air pollution prediction in a 2D transport model

Attributing sources of air pollution events by deploying an efficient observational network is an important and interesting problem in air quality control and forecast studies, but it is very challenging. In order to estimate the sensitivities of pollution events to emission sources, a comprehensive framework is built based on a horizontal 2-dimensional transport model and its adjoint in solving this problem. In an analysis of an idealized air pollution event of PM_2.5 over the region of North China, an objective function is defined to optimally estimate the initial concentrations and emission sources through a series of minimization procedures. Results by means of the 4-dimensional variational approach show that, with the optimal initial conditions and emission sources, the model can successfully forecast the pollution event in a few days. The optimal observing network based on sensitivity analysis takes only one third of the cost but greatly retains predictability skill compared to the full-grid observing system, while nearly no predictability skill is detectable if the same number of observational sites is randomly deployed. We evaluate air pollution predictability in the point of focusing on to what degree the root mean square errors between the modeled concentration and the targeted air pollution are limited by the optimal observational network. Results show that air pollution predictability in association with the optimal observational network is limited in the time scales about 6 days. With the high efficiency and in an economic fashion, such a sensitivity-based optimal observing system holds promise for accurately predicting an air pollution event in the targeted area once the adjoint and variational procedure of a realistic atmosphere model including transport and chemical processes is performed.

Sensitivities of Ozone Air Pollution in the Beijing-Tianjin-Hebei Area to Local and Upwind Precursor Emissions Using Adjoint Modeling

Effective mitigation of surface ozone pollution entails detailed knowledge of the contributing precursors' sources. We use the GEOS-Chem adjoint model to analyze the precursors contributing to surface ozone in the Beijing-Tianjin-Hebei area (BTH) of China on days of different ozone pollution severities in June 2019. We find that BTH ozone on heavily polluted days is sensitive to local emissions, as well as to precursors emitted from the provinces south of BTH (Shandong, Henan, and Jiangsu, collectively the SHJ area). Heavy ozone pollution in BTH can be mitigated effectively by reducing NO_x (from industrial processes and transportation), ≥C₃ alkenes (from on-road gasoline vehicles and industrial processes), and xylenes (from paint use) emitted from both BTH and SHJ, as well as by reducing CO (from industrial processes, transportation, and power generation) and ≥C₄ alkanes (from industrial processes, paint and solvent use, and on-road gasoline vehicles) emissions from SHJ. In addition, reduction of NO_x, xylene, and ≥C₃ alkene emissions within BTH would effectively decrease the number of BTH ozone-exceedance days. Our analysis pinpoint the key areas and activities for locally and regionally coordinated emission control efforts to improve surface ozone air quality in BTH.

Transboundary transport of ozone pollution to a US border region: A case study of Yuma

High concentrations of ground-level ozone affect human health, plants, and animals. Reducing ozone pollution in rural regions, where local emissions are already low, poses challenge. We use meteorological back-trajectories, air quality model sensitivity analysis, and satellite remote sensing data to investigate the ozone sources in Yuma, Arizona and find strong international influences from Northern Mexico on 12 out of 16 ozone exceedance days. We find that such exceedances could not be mitigated by reducing emissions in Arizona; complete removal of state emissions would reduce the maximum daily 8-h average (MDA8) ozone in Yuma by only 0.7% on exceeding days. In contrast, emissions in Mexico are estimated to contribute to 11% of the ozone during these exceedances, and their reduction would reduce MDA8 ozone in Yuma to below the standard. Using satellite-based remote sensing measurements, we find that emissions of nitrogen oxides (NOx, a key photochemical precursor of ozone) increase slightly in Mexico from 2005 to 2016, opposite to decreases shown in the bottom-up inventory. In comparison, a decrease of NOx emissions in the US and meteorological factors lead to an overall of summer mean and annual MDA8 ozone in Yuma (by ∼1-4% and ∼3%, respectively). Analysis of meteorological back-trajectories also shows similar transboundary transport of ozone at the US-Mexico border in California and New Mexico, where strong influences from Northern Mexico coincide with 11 out of 17 and 6 out of 8 ozone exceedances. 2020 is the final year of the U.S.-Mexico Border 2020 Program, which aimed to reduce pollution at border regions of the US and Mexico. Our results indicate the importance of sustaining a substantial cooperative program to improve air quality at the border area.

Evaluating a Space-Based Indicator of Surface Ozone-NO x -VOC Sensitivity Over Midlatitude Source Regions and Application to Decadal Trends

Determining effective strategies for mitigating surface ozone (O₃) pollution requires knowledge of the relative ambient concentrations of its precursors, NO _x , and VOCs. The space-based tropospheric column ratio of formaldehyde to NO₂ (FNR) has been used as an indicator to identify NO _x -limited versus NO _x -saturated O₃ formation regimes. Quantitative use of this indicator ratio is subject to three major uncertainties: (1) the split between NO _x -limited and NO _x -saturated conditions may shift in space and time, (2) the ratio of the vertically integrated column may not represent the near-surface environment, and (3) satellite products contain errors. We use the GEOS-Chem global chemical transport model to evaluate the quantitative utility of FNR observed from the Ozone Monitoring Instrument over three northern midlatitude source regions. We find that FNR in the model surface layer is a robust predictor of the simulated near-surface O₃ production regime. Extending this surface-based predictor to a column-based FNR requires accounting for differences in the HCHO and NO₂ vertical profiles. We compare four combinations of two OMI HCHO and NO₂ retrievals with modeled FNR. The spatial and temporal correlations between the modeled and satellite-derived FNR vary with the choice of NO₂ product, while the mean offset depends on the choice of HCHO product. Space-based FNR indicates that the spring transition to NO _x -limited regimes has shifted at least a month earlier over major cities (e.g., New York, London, and Seoul) between 2005 and 2015. This increase in NO _x sensitivity implies that NO _x emission controls will improve O₃ air quality more now than it would have a decade ago.

Impacts of Foreign, Domestic, and State-Level Emissions on Ozone-Induced Vegetation Loss in the United States

Exposure to elevated levels of ozone leads to yield reduction in agricultural crops and biomass loss in trees. Here, we quantify the impact of ozone pollution on two major U.S. crops, wheat and soybean, and two ozone-sensitive tree species, ponderosa pine and quaking aspen, using simulations with the GEOS-Chem model for 2010. Using previously established exposure-response functions, we estimate nationwide relative yield reductions of 4.9% for wheat and 6.7% for soybean, and relative biomass loss of 2.5% and 2.9% for ponderosa pine and aspen seedlings, respectively. Adjoint model sensitivities are used to estimate the impact of emissions sources from different locations, species, and sectors. We find that the nationwide relative loss in each vegetation type is influenced most by domestic anthropogenic NOx (>75%). Long-range transport from foreign sources is small relative to domestic influences. More than half of the anthropogenic NOx responsible for vegetation damage originates from outside the states where the damage occurs. Texas and Missouri are the highest contributors to the nationwide loss of wheat and soybean, respectively. California "exports" ozone damage for all types of vegetation studied, due to its location, high share of anthropogenic NOx, and a relatively low share of vegetation.

High-resolution mapping of sources contributing to urban air pollution using adjoint sensitivity analysis: benzene and diesel black carbon

The adjoint of the Community Multiscale Air Quality (CMAQ) model at 1 km horizontal resolution is used to map emissions that contribute to ambient concentrations of benzene and diesel black carbon (BC) in the San Francisco Bay area. Model responses of interest include population-weighted average concentrations for three highly polluted receptor areas and the entire air basin. We consider both summer (July) and winter (December) conditions. We introduce a novel approach to evaluate adjoint sensitivity calculations that complements existing methods. Adjoint sensitivities to emissions are found to be accurate to within a few percent, except at some locations associated with large sensitivities to emissions. Sensitivity of model responses to emissions is larger in winter, reflecting weaker atmospheric transport and mixing. The contribution of sources located within each receptor area to the same receptor's air pollution burden increases from 38-74% in summer to 56-85% in winter. The contribution of local sources is higher for diesel BC (62-85%) than for benzene (38-71%), reflecting the difference in these pollutants' atmospheric lifetimes. Morning (6-9am) and afternoon (4-7 pm) commuting-related emissions dominate region-wide benzene levels in winter (14 and 25% of the total response, respectively). In contrast, afternoon rush hour emissions do not contribute significantly in summer. Similar morning and afternoon peaks in sensitivity to emissions are observed for the BC response; these peaks are shifted toward midday because most diesel truck traffic occurs during off-peak hours.

Differences between magnitudes and health impacts of BC emissions across the United States using 12 km scale seasonal source apportionment

Recent assessments have analyzed the health impacts of PM2.5 from emissions from different locations and sectors using simplified or reduced-form air quality models. Here we present an alternative approach using the adjoint of the Community Multiscale Air Quality (CMAQ) model, which provides source-receptor relationships at highly resolved sectoral, spatial, and temporal scales. While damage resulting from anthropogenic emissions of BC is strongly correlated with population and premature death, we found little correlation between damage and emission magnitude, suggesting that controls on the largest emissions may not be the most efficient means of reducing damage resulting from anthropogenic BC emissions. Rather, the best proxy for locations with damaging BC emissions is locations where premature deaths occur. Onroad diesel and nonroad vehicle emissions are the largest contributors to premature deaths attributed to exposure to BC, while onroad gasoline emissions cause the highest deaths per amount emitted. Emissions in fall and winter contribute to more premature deaths (and more per amount emitted) than emissions in spring and summer. Overall, these results show the value of the high-resolution source attribution for determining the locations, seasons, and sectors for which BC emission controls have the most effective health benefits.

Sources of C₂-C₄ alkenes, the most important ozone nonmethane hydrocarbon precursors in the Pearl River Delta region

Surface ozone is becoming an increasing concern in China's megacities such as the urban centers located in the highly industrialized and densely populated Pearl River Delta (PRD) region, where previous studies suggested that ozone production is sensitive to VOC emissions with alkenes being important precursors. However, little was known about sources of alkenes. Here we present our monitoring of ambient volatile organic compounds at four representative urban, suburban and rural sites in the PRD region during November-December 2009, which experienced frequent ozone episodes. C2-C4 alkenes, whose total mixing ratios were 11-20% of non-methane hydrocarbons (NMHCs) quantified, accounted for 38-64% of ozone formation potentials (OFPs) and 30-50% of the total hydroxyl radical (OH) reactivity by NMHCs. Ethylene was the most abundant alkene, accounting for 8-15% in total mixing ratios of NMHCs and contributed 25-46% of OFPs. Correlations between C2-C4 alkenes and typical source tracers suggested that ethylene might be largely related to vehicle exhausts and industry activities, while propene and butenes were much more LPG-related. Positive Matrix Factorization (PMF) confirmed that vehicle exhaust and liquefied petroleum gas (LPG) were two major sources that altogether accounted for 52-62%, 58-77%, 73-83%, 68-79% and 73-84% for ethylene, propene, 1-butene, trans-2-butene and cis-2-butene, respectively. Vehicle exhausts alone contributed 32-49% ethylene and 35-41% propene. Industry activities contributed 13-23% ethylene and 7-20% propene. LPG instead contributed the most to butenes (38-65%) and substantially to propene (23-36%). Extensive tests confirmed high fractions of propene and butenes in LPG then used in Guangzhou and in LPG combustion plumes; therefore, limiting alkene contents in LPG would benefit regional ozone control.

Attainment vs exposure: ozone metric responses to source-specific NOx controls using adjoint sensitivity analysis

We establish linkages between sources of NOx emissions and two types of national ozone metrics in Canada and the U.S. using the adjoint of an air quality model. We define an attainment-based metric using probabilistic design values (PDVs) exceeding 65 ppb to represent polluted regions and define an exposure-based metric as the premature mortality count related to short-term ozone exposure, both in Canada and the U.S. Our results reveal differences in both temporally averaged and day-specific influences of NOx emission controls across source locations. We find NOx emission reductions in California and the eastern U.S. to be most effective for reducing attainment- and exposure-based metrics, amounting to a total reduction of 6500 ppb in PDVs and 613 deaths/season nationally from a 10% reduction in NOx emissions from those source locations. While source controls in the remainder of the western U.S. are beneficial at reducing nonattainment, these reductions are less influential on ozone mortality. We also find that while exposure-based metrics are sensitive to daily emission reductions, much of the reduction in PDVs arises from controlling emissions on only a fraction of simulation days. We further illustrate the dependency of adjoint estimates of emission influences on the choice of averaging period as a follow-up to previous work.

Optimal ozone reduction policy design using adjoint-based NOx marginal damage information

Despite substantial reductions in nitrogen oxide (NOx) emissions in the United States, the success of emission control programs in optimal ozone reduction is disputable because they do not consider the spatial and temporal differences in health and environmental damages caused by NOx emissions. This shortcoming in the current U.S. NOx control policy is explored, and various methodologies for identifying optimal NOx emission control strategies are evaluated. The proposed approach combines an optimization platform with an adjoint (or backward) sensitivity analysis model and is able to examine the environmental performance of the current cap-and-trade policy and two damage-based emissions-differentiated policies. Using the proposed methodology, a 2007 case study of 218 U.S. electricity generation units participating in the NOx trading program is examined. The results indicate that inclusion of damage information can significantly enhance public health performance of an economic instrument. The net benefit under the policy that minimizes the social cost (i.e., health costs plus abatement costs) is six times larger than that of an exchange rate cap-and-trade policy.

Ozone sensitivity to its precursor emissions in northeastern Mexico for a summer air pollution episode

A summer episode was modeled to address the expected response of ambient air O3 to hypothetical emission control scenarios in northeastern Mexico, and in particular in the Monterrey Metropolitan Area (MMA). This region is of interest because the MMA holds one of the worst air quality problems in the country and levels of air pollutants in the rest of northeastern Mexico are starting to be a concern. The MM5-SMOKE-CMAQ platform was used to conduct the numerical experiments. Twenty-four control scenarios were evaluated, combining the level of emission controls of O3 precursors (NO(x) and volatile organic compounds [VOCs]) from 0% to 50%. For the MMA, VOC-only controls result in the best option to reduce O3 concentrations, though the benefit is limited to the urban core. This same strategy results in negligible benefits for the rest of northeastern Mexico. NO(x) controls result in an increase in O3 concentration within the MMA of up to 20 ppbv and a decrease at downwind locations of up to 11 ppbv, with respect to the base-case scenario. Indicator ratios were also used to probe for NO(x)-sensitive and VOC-sensitive areas. Locations with an important influence of NO(x) point sources (i.e., Monclova and Nava/Acuña) are quite sensitive to changes in NO(x) emissions. Border cities in the Rio Bravo/Grande Valley tend to be marginally NO(x)-sensitive. Overall, the MMA seems to be dominated by a VOC-sensitive regime, while the rest of the region would tend to have a NO(x)-sensitive response. The results obtained serve to expand the current knowledge on the chemical regimes that dominate this region (VOC- or NO(x)-sensitive), and thus could help guide public policies related to emission regional control strategies.

Improving NO(x) cap-and-trade system with adjoint-based emission exchange rates

Cap-and-trade programs have proven to be effective instruments for achieving environmental goals while incurring minimum cost. The nature of the pollutant, however, affects the design of these programs. NO(x), an ozone precursor, is a nonuniformly mixed pollutant with a short atmospheric lifetime. NO(x) cap-and-trade programs in the U.S. are successful in reducing total NO(x) emissions but may result in suboptimal environmental performance because location-specific ozone formation potentials are neglected. In this paper, the current NO(x) cap-and-trade system is contrasted to a hypothetical NO(x) trading policy with sensitivity-based exchange rates. Location-specific exchange rates, calculated through adjoint sensitivity analysis, are combined with constrained optimization for prediction of NO(x) emissions trading behavior and post-trade ozone concentrations. The current and proposed policies are examined in a case study for 218 coal-fired power plants that participated in the NO(x) Budget Trading Program in 2007. We find that better environmental performance at negligibly higher system-wide abatement cost can be achieved through inclusion of emission exchange rates. Exposure-based exchange rates result in better environmental performance than those based on concentrations.

Bayesian analysis of a reduced-form air quality model

Numerical air quality models are being used for assessing emission control strategies for improving ambient pollution levels across the globe. This paper applies probabilistic modeling to evaluate the effectiveness of emission reduction scenarios aimed at lowering ground-level ozone concentrations. A Bayesian hierarchical model is used to combine air quality model output and monitoring data in order to characterize the impact of emissions reductions while accounting for different degrees of uncertainty in the modeled emissions inputs. The probabilistic model predictions are weighted based on population density in order to better quantify the societal benefits/disbenefits of four hypothetical emission reduction scenarios in which domain-wide NO(x) emissions from various sectors are reduced individually and then simultaneously. Cross validation analysis shows the statistical model performs well compared to observed ozone levels. Accounting for the variability and uncertainty in the emissions and atmospheric systems being modeled is shown to impact how emission reduction scenarios would be ranked, compared to standard methodology.

Efficient probabilistic estimates of surface ozone concentration using an ensemble of model configurations and direct sensitivity calculations

Because all models are a simplification of the phenomenon they aim to represent, it is often more useful to estimate the probability of an event rather than a single "best" model result. Previous air quality ensemble approaches have used computationally expensive simulations of separately developed modeling systems. We present an efficient method to generate ensembles with hundreds of members based on several structural configurations of a single air quality modeling system. We use the Decoupled Direct Method in three dimensions to directly calculate how ozone concentrations change as a result of changes in input parameters. The modeled probability estimate is compared to observations and is shown to have a high level of skill and improved resolution and sharpness. This approach can help resolve the practical limits of incorporating uncertainty estimation into deterministic air quality management modeling applications.

The adjoint of CMAQ

An adjoint model for the internationally used Community Multiscale Air Quality (CMAQ) modeling platform of the U.S. EPA is developed. The adjoint version for CMAQ (CMAQ-ADJ) provides the user community with forward (decoupled direct method or DDM) and backward (adjoint) sensitivity analysis capabilities. Current implementation is for gas-phase processes. Discrete adjoints are implemented for all processes with the exception of horizontal advection, for which, because of inherent discontinuities in the advection scheme, the continuous approach is superior. The adjoint of chemistry is constructed by interfacing CMAQ with the kinetic pre-processor, which provides for increased flexibility in the choice of chemical solver and facilitates the implementation of new chemical mechanisms. The adjoint implementation is evaluated both on a process-by-process basis and for the full model. In general, adjoint results show good agreement with brute-force and DDM sensitivities. As expected for a continuous adjoint implementation in a nonlinear scheme, the agreement is not perfect for horizontal transport. Sensitivities of various air quality, public health, and environmental metrics with respect to emissions are calculated using the adjoint method. In order to show applicability to regional climate studies, as an example, the sensitivities of these metrics with respect to local temperatures are calculated.