Development of a recurrent spatiotemporal deep-learning method coupled with data fusion for correction of hourly ozone forecasts

Ambient ozone (O3) predictions can be very challenging mainly due to the highly nonlinear photochemistry among its precursors, and meteorological conditions and regional transport can further complicate the O3 formation processes. The emission-based chemical transport models (CTM) are broadly used to predict O3 formation, but they may deviate from observations due to input uncertainties such as emissions and meteorological data, in addition to the treatment of O3 nonlinear chemistry. In this study, an innovative recurrent spatiotemporal deep-learning (RSDL) method with model-monitor coupled convolutional recurrent neural networks (ConvRNN) has been developed to improve O3 predictions of CTM. The RSDL method was first used to build the ConvRNN within a 24-h scale to characterize the spatiotemporal relationships between the monitored O3 data and CTM simulations, and then incorporated the recurrent pattern to achieve 72-h multi-site forecasts based on a pilot study over the Pearl River Delta (PRD) region of China. The results showed that the RSDL method predicted O3 with high accuracy over this case study, with an increase of 27.54% in the correlation coefficient (R) average for all sites as well as an increase in R of 0.14-0.21 for all cities compared to CTM. Moreover, the regional distribution of CTM was further improved by the RSDL predictions with the data fusion technique, which greatly reduced the underpredictions of O3 concentrations, particularly in high O3-level areas (concentrations >160 μg/m3), with a 33.55% reduction in the mean absolute error (MAE).

A comprehensive approach combining positive matrix factorization modeling, meteorology, and machine learning for source apportionment of surface ozone precursors: Underlying factors contributing to ozone formation in Houston, Texas

Ozone concentrations in Houston, Texas, are among the highest in the United States, posing significant risks to human health. This study aimed to evaluate the impact of various emissions sources and meteorological factors on ozone formation in Houston from 2017 to 2021 using a comprehensive PMF-SHAP approach. First, we distinguished the unique sources of VOCs in each area and identified differences in the local chemistry that affect ozone production. At the urban station, the primary sources were n_decane, biogenic/industrial/fuel evaporation, oil and gas flaring/production, industrial emissions/evaporation, and ethylene/propylene/aromatics. At the industrial site, the main sources were industrial emissions/evaporation, fuel evaporation, vehicle-related sources, oil and gas flaring/production, biogenic, aromatic, and ethylene and propylene. And then, we performed SHAP analysis to determine the importance and impact of each emissions factor and meteorological variables. Shortwave radiation (SHAP values are ∼5.74 and ∼6.3 for Milby Park and Lynchburg, respectively) and humidity (∼4.87 and ∼4.71, respectively) were the most important variables for both sites. For the urban station, the most important emissions sources were n_decane (∼2.96), industrial emissions/evaporation (∼1.89), and ethylene/propylene/aromatics (∼1.57), while for the industrial site, they were oil and gas flaring/production (∼1.38), ethylene/propylene (∼1.26), and industrial emissions/evaporation (∼0.95). NOx had a negative impact on ozone production at the urban station due to the NOx-rich chemical regime, whereas NOx had positive impacts at the industrial site. The study's findings suggest that the PMF-SHAP approach is efficient, inexpensive, and can be applied to other similar applications to identify factors contributing to ozone-exceedance events. The study's results can be used to develop more effective air quality management strategies for Houston and other cities with high levels of ozone.

Contributions of meteorology to ozone variations: Application of deep learning and the Kolmogorov-Zurbenko filter

From hourly ozone observations obtained from three regions⸻Houston, Dallas, and West Texas⸻we investigated the contributions of meteorology to changes in surface daily maximum 8-h average (MDA8) ozone from 2000 to 2019. We applied a deep convolutional neural network and Shapely additive explanation (SHAP) to examine the complex underlying nonlinearity between variations of surface ozone and meteorological factors. Results of the models showed that between 2000 and 2019, specific humidity (38% and 27%) and temperature (28% and 37%) contributed the most to ozone formation over the Houston and Dallas metropolitan areas, respectively. On the other hand, the results show that solar radiation (50%) strongly impacted ozone variation over West Texas during this time. Using a combination of the Kolmogorov-Zurbenko (KZ) filter and multiple linear regression, we also evaluated the influence of meteorology on ozone and quantified the contributions of meteorological parameters to trends in surface ozone formation. Our findings showed that in Houston and Dallas, meteorology influenced ozone variations to a large extent. The impacts of meteorology on West Texas, however, showed meteorological factors had fewer influences on ozone variabilities from 2000 to 2019. This study showed that SHAP analysis and the KZ approach can investigate the contributions of the meteorological factors on ozone concentrations and help policymakers enact effective ozone mitigation policies.

Drivers of 2013-2020 ozone trends in the Sichuan Basin, China: Impacts of meteorology and precursor emission changes

The Sichuan Basin (SCB) of China is known for excessive ozone (O₃) pollution owing to high anthropogenic emissions combined with terrain-induced poor ventilation and weak wind fields against the surrounding mountains. While O₃ pollution has emerged as a prominent concern in southwestern China yet variations in O₃ levels during 2013-2020 are still unclear and the dominant factor in explaining the long-term O₃ trend throughout the SCB remains elusive due to uncertainties in emission inventory and variability associated with meteorological conditions. Here, we use extensive basin-wide ambient measurements to examine the spatial pattern and trend of O₃ and leverage OMI and TROPOMI satellites in conjunction with MEIC emission inventory to track emission changes. Sensitivity simulations are conducted by using WRF-CMAQ model to investigate the impacts of meteorological variability and emission changes on O₃ changes over 2013-2020. O₃ concentrations exhibit obvious interannual increases during 2013-2019 and a slight decrease in 2020. Both decreases in the MEIC emission inventory (-2.9% yr^-1) and OMI NO₂ column density (-3.1% yr^-1) reflects the declining trend in NO_x emissions over 2013-2020, while anthropogenic VOCs were not adequately regulated during 2013-2017, which explained the majority of deteriorated O₃ pollution from 2013 to 2017. Furthermore, attribution analysis based on CMAQ simulations indicate that the unexpected aggravated O₃ levels in 2019 is not only modulated by disproportional reductions in VOCs and NO_x emissions, but also associated with unfavorable meteorological conditions featured by profound heatwaves and frequent stagnant conditions. In 2020, the abnormal meteorological conditions in May leads to substantial increase of O₃ by 26.8 μg m^-3 as compared to May 2019, while the considerable enhancement was fully offset by low O₃ levels over the whole period which attributes to substantial emission reductions. This study reveals the long-term trend of O₃ levels and precursor emissions and highlights the effects of meteorological variability and emission changes on O₃ pollution over the SCB, with strong implications for designing effective O₃ control measures.

Ambient volatile organic compounds in tropical environments: Potential sources, composition and impacts - A review

Volatile organic compounds (VOCs) are widely recognized to affect the environment and human health. This review provides a comprehensive presentation of the types and levels of VOCs, their sources and potential effects on human health and the environment based on past and current observations made at tropical sites. Isoprene was found to be the dominant biogenic VOC in the tropics. Tropical broad leaf evergreen trees are the main emitters of isoprene, making up more than 70% of the total emissions. The VOCs found in the tropical remote marine atmosphere included isoprene (>100 ppt), dimethyl sulfide (≤100 ppt) and halocarbons, i.e. bromoform (≤8.4 ppt), dibromomethane (≤2.7 ppt) and dibromochloromethane (≤1.6 ppt). VOCs such as benzene, toluene, ethylbenzene and xylene (BTEX) are the most monitored anthropogenic VOCs and are present mainly due to motor vehicles emissions. Additionally, biomass burning contributes to anthropogenic VOCs, especially high molecular weight VOCs, e.g. methanol and acetonitrile. The relative contributions of VOC species to ozone are determined through the level of the Ozone Formation Potential (OFP) of different species. Emissions of VOCs (e.g. very short-lived halogenated gases) in the tropics are capable of contributing to stratospheric ozone depletion. BTEX has been identified as the main types of VOCs that are associated with the cancer risk in urban areas in tropical regions. Finally, future studies related to VOCs in the tropics and their associated health risks are needed to address these concerns.

Regional Differences of Primary Meteorological Factors Impacting O3 Variability in South Korea

Surface ozone (O3) is a harmful pollutant and effective strategies must be developed for its reduction. In this study, the impact of meteorological factors on the annual O3 variability for South Korea were analyzed. In addition, the regional differences of meteorological factors in six air quality regions in South Korea (Seoul Metropolitan Area, SMA; Central region, CN; Honam, HN; Yeongnam, YN; Gangwon, GW; Jeju, JJ) were compared. The analysis of ground observation data from 2001 to 2017 revealed that the long-term variability of O3 concentration in South Korea continuously increased since 2001, and the upward trend in 2010 to 2017 (Period 2, PRD2) was 29.8% higher than that in 2001 to 2009 (Period 1, PRD1). This was because the meteorological conditions during PRD2 became relatively favorable for high O3 concentrations compared to conditions during PRD1. In particular, the increase in the solar radiation (SR) and maximum temperature (TMAX) and the decrease in the precipitation (PRCP) and wind speed (WS) of South Korea in PRD2 were identified as the main causes for the rise in O3 concentrations. When meteorological factors and O3 variability were compared among the six air quality regions in South Korea during PRD1 and PRD2, significant differences were observed. This indicated that different meteorological changes occurred in South Korea after 2010 due to the different topographical characteristics of each region; thus, O3 variability also changed differently in each region. Interestingly, for the regions with almost similar meteorological changes after 2010, the O3 concentration changed differently depending on the difference in the distribution of emissions. These results indicate that the O3–meteorology relationship shows spatiotemporal differences depending on the topographical and emission distribution characteristics of each area and implies that it is necessary to fully consider such differences for efficient O3 reduction.

Uncertainty associated with ambient ozone metrics in epidemiologic studies and risk assessments

Epidemiologic studies relating ambient ozone concentrations to adverse health outcomes have typically relied on spatial averages of concentrations from nearby monitoring stations, referred to as "composite monitors." This practice reflects the assumption that ambient ozone concentrations within an urban area are spatially homogenous. We tested the validity of this assumption by comparing ozone data measured at individual monitoring sites within selected US urban areas to their respective composite monitor time series. We first characterized the temporal correlation between the composite monitor and individual monitors in each area. Next, we analyzed the heteroskedasticity of each relationship. Finally, we compared the distribution of concentrations measured at individual monitors to the composite monitor distribution. Individual monitors showed high correlation with the composite monitor over much of the range of ambient ozone concentrations, though correlations were lower at higher concentrations. The variance between individual monitors and the composite monitor increased as a function of concentration in nearly all the urban areas. Finally, we observed statistical bias in the composite monitor concentrations at the high end of the distribution. The degree to which these results introduce uncertainty into studies that utilize composite monitors depends on the contributions of peak ozone concentrations to reported health effect associations.

Future-year ozone prediction for the United States using updated models and inputs

The relationship between emission reductions and changes in ozone can be studied using photochemical grid models. These models are updated with new information as it becomes available. The primary objective of this study was to update the previous Collet et al. studies by using the most up-to-date (at the time the study was done) modeling emission tools, inventories, and meteorology available to conduct ozone source attribution and sensitivity studies. Results show future-year, 2030, design values for 8-hr ozone concentrations were lower than base-year values, 2011. The ozone source attribution results for selected cities showed that boundary conditions were the dominant contributors to ozone concentrations at the western U.S. locations, and were important for many of the eastern U.S.

LOCATIONS

Point sources were generally more important in the eastern United States than in the western United States. The contributions of on-road mobile emissions were less than 5 ppb at a majority of the cities selected for analysis. The higher-order decoupled direct method (HDDM) results showed that in most of the locations selected for analysis, NOx emission reductions were more effective than VOC emission reductions in reducing ozone levels. The source attribution results from this study provide useful information on the important source categories and provide some initial guidance on future emission reduction strategies.

IMPLICATIONS

The relationship between emission reductions and changes in ozone can be studied using photochemical grid models, which are updated with new available information. This study was to update the previous Collet et al. studies by using the most current, at the time the study was done, models and inventory to conduct ozone source attribution and sensitivity studies. The source attribution results from this study provide useful information on the important source categories and provide some initial guidance on future emission reduction strategies.

Assessing Temporal and Spatial Patterns of Observed and Predicted Ozone in Multiple Urban Areas

BACKGROUND

Ambient monitoring data show spatial gradients in ozone (O3) across urban areas. Nitrogen oxide (NOx) emissions reductions will likely alter these gradients. Epidemiological studies often use exposure surrogates that may not fully account for the impacts of spatially and temporally changing concentrations on population exposure.

OBJECTIVES

We examined the impact of large NOx decreases on spatial and temporal O3 patterns and the implications on exposure.

METHODS

We used a photochemical model to estimate O3 response to large NOx reductions. We derived time series of 2006-2008 O3 concentrations consistent with 50% and 75% NOx emissions reduction scenarios in three urban areas (Atlanta, Philadelphia, and Chicago) at each monitor location and spatially interpolated O3 to census-tract centroids.

RESULTS

We predicted that low O3 concentrations would increase and high O3 concentrations would decrease in response to NOx reductions within an urban area. O3 increases occurred across larger areas for the seasonal mean metric than for the regulatory metric (annual 4th highest daily 8-hr maximum) and were located only in urban core areas. O3 always decreased outside the urban core (e.g., at locations of maximum local ozone concentration) for both metrics and decreased within the urban core in some instances. NOx reductions led to more uniform spatial gradients and diurnal and seasonal patterns and caused seasonal peaks in midrange O3 concentrations to shift from midsummer to earlier in the year.

CONCLUSIONS

These changes have implications for how O3 exposure may change in response to NOx reductions and are informative for the design of future epidemiology studies and risk assessments.

CITATION

Simon H, Wells B, Baker KR, Hubbell B. 2016. Assessing temporal and spatial patterns of observed and predicted ozone in multiple urban areas. Environ Health Perspect 124:1443-1452; http://dx.doi.org/10.1289/EHP190.

Reforestation as a novel abatement and compliance measure for ground-level ozone

High ambient ozone (O3) concentrations are a widespread and persistent problem globally. Although studies have documented the role of forests in removing O3 and one of its precursors, nitrogen dioxide (NO2), the cost effectiveness of using peri-urban reforestation for O3 abatement purposes has not been examined. We develop a methodology that uses available air quality and meteorological data and simplified forest structure growth-mortality and dry deposition models to assess the performance of reforestation for O3 precursor abatement. We apply this methodology to identify the cost-effective design for a hypothetical 405-ha, peri-urban reforestation project in the Houston-Galveston-Brazoria O3 nonattainment area in Texas. The project would remove an estimated 310 tons of (t) O3 and 58 t NO2 total over 30 y. Given its location in a nitrogen oxide (NOx)-limited area, and using the range of Houston area O3 production efficiencies to convert forest O3 removal to its NOx equivalent, this is equivalent to 127-209 t of the regulated NOx. The cost of reforestation per ton of NOx abated compares favorably to that of additional conventional controls if no land costs are incurred, especially if carbon offsets are generated. Purchasing agricultural lands for reforestation removes this cost advantage, but this problem could be overcome through cost-share opportunities that exist due to the public and conservation benefits of reforestation. Our findings suggest that peri-urban reforestation should be considered in O3 control efforts in Houston, other US nonattainment areas, and areas with O3 pollution problems in other countries, wherever O3 formation is predominantly NOx limited.

Assessment of a regulatory model's performance relative to large spatial heterogeneity in observed ozone in Houston, Texas

In Houston, some of the highest measured 8-hr ozone (O3) peaks are characterized by sudden increases in observed concentrations of at least 40 ppb in 1 hr or 60 ppb in 2 hr. Measurements show that these large hourly changes appear at only a few monitors and span a narrow geographic area, suggesting a spatially heterogeneous field of O3 concentrations. This study assessed whether a regulatory air quality model (AQM) can simulate this observed behavior. The AQM did not reproduce the magnitude or location of some of the highest observed hourly O3 changes, and it also failed to capture the limited spatial extent. On days with measured large hourly changes in O3 concentrations, the AQM predicted high O3 over large regions of Houston, resulting in overpredictions at several monitors. This analysis shows that the model can make high O3, but on these days the predicted spatial field suggests that the model had a different cause. Some observed large hourly changes in O3 concentrations have been linked to random releases of industrial volatile organic compounds (VOCs). In the AQM emission inventory, there are several emission events when an industrial point source increases VOC emissions in excess of 10,000 mol/hr. One instance increased predicted downwind O3 concentrations up to 25 ppb. These results show that the modeling system is responsive to a large VOC release, but the timing and location of the release, and meteorological conditions, are critical requirements. Attainment of the O3 standard requires the use of observational data and AQM predictions. If the large observed hourly changes are indicative of a separate cause of high O3, then the model may not include that cause, which might result in regulators enacting control strategies that could be ineffective.