Inconsistent decadal variations between surface and free tropospheric 1 nitrogen oxides over United States 2 3

Inconsistent decadal variations between surface and free tropospheric 1 nitrogen oxides over United States 2 3 Zhe Jiang, Helen Worden, John R. Worden, Daven K. Henze, Dylan B. A. Jones, Avelino F. 4 Arellano, Emily V. Fischer, Liye Zhu, Kazuyuki Miyazaki , K. Folkert Boersma, Vivienne 5 H. Payne, 6 7 National Center for Atmospheric Research, Boulder, CO, USA 8 Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, USA 9 Department of Mechanical Engineering, University of Colorado, Boulder, CO, USA 10 Department of Physics, University of Toronto, Toronto, ON, Canada 11 Department of Hydrology and Atmospheric Sciences, University of Arizona, Tucson, AZ, USA 12 Department of Atmospheric Science, Colorado State University, Fort Collins, CO, USA 13 Japan Agency for Marine-Earth Science and Technology, Yokohama, Japan 14 Wageningen University, Meteorological and Air Quality department, Wageningen, the 15 Netherlands 16 Royal Netherlands Meteorological Institute, De Bilt, The Netherlands  17 18 19 20


Introduction
Nitrogen oxides play a complex role in tropospheric chemistry and have a strong influence on air quality as precursors in the formation of ozone (O 3 ) and secondary aerosols.Tropospheric NO x is produced through anthropogenic combustion, biomass burning, soil (Jaegle et al., 2005), and lightning emissions (Schumann and Huntrieser, 2007), and is mainly removed by the formation of nitric acid (HNO 3 ).Most NO x is emitted as nitric oxide (NO), however, it is most appropriate to consider the budget of the NO x as a whole, because of the rapid cycling between NO and NO 2 (~ 1 min).Tropospheric NO x has short lifetime, a few hours except in extratropical winter when it increases to 1-2 days (Martin et al. 2003).
Because of the short lifetime, state of the art chemistry/climate models suggest that the direct contribution of long-range transport to tropospheric NO x distribution is limited (e.g.Zhang et al. 2008).However, NO x can also be transported far away from the sources via the formation of long-lived reservoir species, such as peroxyacetyl nitrate (PAN, e.g., Fischer et al., 2014;Jiang et al. 2016a).Models have large uncertainties in PAN abundance (Fischer et al., 2014) and there are also potentially other missing processes in the chemical transport models used to diagnose NO x lifetime and transport.For example, a recent discovery about the rapid cycling of reactive nitrogen in the marine boundary layer (Ye et al. 2016) demonstrates processes that are not represented in modeled NO x transport, and that may help explain existing discrepancies in reactive nitrogen partitioning between models and observations.Due to its critical influence in the troposphere, there are multiple space-based measurements for tropospheric NO 2 that are available from satellites that were launched in the past two decades.These instruments typically measure backscattered solar radiation from which the vertically integrated column abundance of NO 2 is retrieved.The assumption of weak long-range transport allows relatively simple applications of the space-based NO 2 column data to study NO x sources.For example, recent studies (e.g.Reuter et al. 2014;Itahashi et al. 2014;Duncan et al. 2016;Krotkov et al. 2016) assessed the trends of surface NO x emissions by assuming a strong correlation between tropospheric NO 2 columns with local emissions.The tropospheric NO 2 column data are also widely used in inverse modeling analyses to estimate NO x emissions by either scaling the surface NO x emissions with the corresponding ratio of observed over modeled tropospheric NO 2 column (e.g.Lamsal et al. 2011;Mijling et al. 2012;Gu et al. 2014) or through data assimilation techniques with short localization length scales (e.g.Miyazaki et al. 2017).
Since 1990, US regulations have required significant NO x emission reductions over many regions (US Environmental Protection Agency, 2010).The trend of decreasing local US NO x emissions has been confirmed by several studies (e.g.Lamsal et al. 2015;Tong et al. 2015;Kharol et al. 2015;Duncan et al. 2016;Krotkov et al. 2016).In contrast to the decreasing local NO x emissions, recent studies (e.g.Cooper et al. 2010;Verstraeten et al. 2015) have indicated an increase in free tropospheric O 3 over western North America over the past decade.The discrepancy between variations of local NO x emissions and free tropospheric O 3 suggests possible influences from non-local sources, and consequently, provides motivation to re-evaluate the contribution of long-range transport to the free tropospheric NO x distribution.
In this work, we investigate the variation of US tropospheric NO 2 in the past decade to assess the contribution of non-local sources.We will particularly explore the possible answers for the following questions: why there is good agreement between tropospheric NO 2 column and surface measurements over the period of 2005-2008?What is the reason for the appearance of the large and growing divergence at around 2009?What is the impact of the decreasing Chinese NO x emissions since 2013 (Liu et al. 2016)   This paper is organized as follows: in Section 2 we describe the observations and model used in this work.In Section 3 we demonstrate the divergence between the OMI NO 2 column retrievals and surface measurements over the period of 2005-2015 and focus on the evaluation of contributions from various hypotheses that could explain the divergence.Our conclusions follow in Section 4.

Tropospheric NO 2 column from OMI
The OMI instrument was launched on NASA's Aura spacecraft.The sensor has a spatial resolution of 13 km x 24 km.OMI provides daily global coverage with measurements of both direct and atmosphere-backscattered sunlight in the ultraviolet-visible range from 270 to 500 nm; the spectral range 405-465 nm is used to retrieve tropospheric NO 2 columns.Two versions of the OMI retrievals (level 2) are used in this work: the NASA (version 3, Krotkov and Veefkind 2006;Bucsela et al. 2013) and DOMINO (version 2, Boersma et al. 2011) retrievals.There are significant differences in the retrieval algorithms of the two products.For example, the a priori NO 2 profiles of the NASA product is based on data from the Global Modeling Initiative (GMI) model with yearly varying emissions, whereas the a priori NO 2 profiles of the DOMINO product is from the Tracer Model 4 (TM4) without interannual variations in emissions.In addition, for the NASA product, the stratospheric contribution to the tropospheric column is estimated from the GMI model simulation.In contrast, for the DOMINO product, the stratospheric contribution is based on the assimilation of OMI data into the TM4 model.Starting in 2007, anomalies were found in OMI data and diagnosed as attenuated measured radiances in certain cross-track positions.This instrument degradation has been referred to as the "row anomaly".In order to ensure the quality and stability of the data, the following filters are applied in our analysis for both OMI products (NASA and DOMINO): After the application of the filters, the number of measurements over the US is about 185,000 per month in 2010.Thus, we expect the uncertainties in the monthly/annual mean NO 2 columns due to random errors are small.The discrepancy between the two OMI products (see Figures 1a-b) is mainly caused by the two different retrieval algorithms.

AQS and NAPS surface in-situ NO 2 concentration
We use daily-averaged in-situ surface NO 2 measurements from the EPA AQS network, and the Environment Canada NAPS network.The AQS/NAPS networks collect ambient air pollution data from monitoring stations located in urban, suburban, and rural areas.In the analysis here, the daily data are averaged to obtain monthly mean concentration at each station.

Flash rate density from Lightning Imaging Sensor (LIS)
LIS is a component of the NASA Tropical Rain Measuring Mission (TRMM).It measures total optical pulses from cloud-to-ground and intracloud lightning flashes during both day and night with global coverage (42.5°S-42.5°N) in the period 1995-2014.Monthly flash rate density (flash/km2) with 2.5°x2.5°resolution is used in this work (Cecil et al. 2006).

Passive tracer simulation using GEOS-Chem model
The GEOS-Chem global chemical transport model (CTM) [www.geos-chem.org] is driven by assimilated meteorological fields (MERRA) from the NASA Goddard Earth Observing System at the Global Modeling and data Assimilation Office.We use version v9-01-03 of GEOS-Chem at a horizontal resolution of 4°x5°.Bertram et al. (2013) indicated the dominant role of long-lived reservoir species in the transpacific transport of reactive nitrogen using aircraft measurements from the INTEX-B campaign.Although the lifetime of tropospheric NO x is short, the lifetime of longlived reservoir species is much longer, for example, the lifetime of free tropospheric PAN is about 1 month.In order to assess the effects of physical transport processes on the long-range transport of reactive nitrogen, we performed a "passive" tracer simulation, with a constant and uniform timescale for loss of 15 days (i.e.360 hours) over the period of 2005-2015 following the approach of Jiang et al. (2016b).The global a priori surface NO x emissions (anthropogenic, biomass burning and soil emissions) are fixed at 2005 level.For each time step (one hour), the tropospheric NO 2 is calculated by:  # $ =  # $&'  &'/*+, .The lightning NO x emissions are not included in the simulation.The 15-day lifetime was selected to provide an approximation for the variation of free tropospheric NO x via the formation and transport of long-lived reservoir species, due to changes in meteorology.Although actual lifetimes of long-lived reservoir species will vary, we found that 15-days was a reasonable compromise to understand the influence of decadal-scale variability on long-range transport patterns.

Results and Discussion
Figures 1a-b show the variations of mean tropospheric NO 2 columns from OMI (NASA and DOMINO products) over the US and east China, respectively.Although there is a significant bias in the magnitude of tropospheric NO 2 column between two OMI products, indicating the influence of different retrieval algorithms, this bias should not affect the trend analysis, as demonstrated by the consistent interannual variations between the two data products.Figure 1c shows percent changes, relative to 2009, of the annual mean tropospheric NO 2 columns over the US, and of the total US NO x emissions (anthropogenic + biomass burning) from the US EPA (https://www.epa.gov/air-emissions-inventories/air-pollutant-emissions-trends-data).There is good agreement between the changes in the OMI retrievals and the emissions estimates in the 2005-2009 period: the annual slope of the EPA's estimates is -6.4%±0.03%(slope of linear regression ± uncertainty of slope), and the annual slopes of the two sets of OMI retrievals are -6.8%±1.1% (NASA) and -8.0%±0.8%(DOMINO).Conversely, we find a large, growing separation in the 2009-2015 period: the annual slope of the EPA's estimates is -4.6%±0.03%,whereas the annual slopes of OMI retrievals are -0.5%±0.6%(NASA) and 1.6%±1.1% (DOMINO).Figures 1d-e show the percent changes in the seasonal mean tropospheric NO 2 columns from OMI retrievals, and in the EPA's estimates (annual mean).The divergence between the seasonal NO 2 columns and the emissions is similar to that shown in Figure 1c, suggesting there is no obvious seasonal dependence.
Our intention here is to understand the possible reasons for the divergence between observed changes in NO 2 vs. changes expected from NO x emissions.Figure 2 depicts the potential hypotheses that could explain the divergence: We note that our analysis based on surface in-situ measurements may not provide sufficient representation for emissions from oil and gas exploration and production.However, though potentially important locally, these activities only contribute about 5% to total US NO x emissions based on EPA's estimates.Therefore, we do not expect significant contributions to the overall changes in tropospheric NO 2 from these sources.
Similarly, we expect limited contributions from other sources which are not included in EPA's inventory.The contribution from aircraft NO x is only about 3% of total US NO x emissions (Skowron et al. 2014) discrepancies between the OMI retrieveals and the EPA's emission estimates lack a clear seasonal dependence (see Figures 1d-e), we expect negligible contributions from soil and lightning NO x emissions to the enhanced tropospheric NO 2 in the period of 2009-2015.Furthermore, Figure 4 indicates that the flash rate density over North America from LIS is uncorrelated with the observed NO 2 variation.

Time dependent OMI retrieval errors (H2)
The quality of the OMI retrievals has been evaluated with surface in-situ measurements.Lamsal et al. (2015) reported that the correlation between OMI NO 2 tropospheric columns (NASA) and the AQS surface in-situ NO 2 measurements was 0.68 for the period 2005-2010.Hoek et al.

Non-local sources (H3)
We have demonstrated that hypotheses H1 and H2 are not likely the dominant factors, which leaves hypothesis H3 (non-local sources) as possible important contributors.Figures 5g-h Oscillation (Trenberth 1997).To the best of our understanding, ENSO is the dominant climate phenomenon linked to extreme weather conditions globally (Cai et al. 2015), and it also exerts a major influence on the interannual variability of O 3 in the troposphere (Doherty et al., 2006).
Following Jiang et al. (2016b), we conducted an analysis using an idealized passive tracer to assess the possible influences of transport patterns.We performed a GEOS-Chem model simulation for tropospheric NO 2 over the period of 2005-2015 with an NO 2 -like tracer with a constant 15-day lifetime and fixed (2005 level) surface NO x emissions.The passive tracer simulation with constant lifetime avoids the possible influences from uncertainties in the modeled nonlinear NO x chemistry, particularly, the conversion between NO x and its longer-lived reservior species.
Figures 6a-c show that even with emissions held constant, interannual variations in transport produce differences in NO 2 (or the passive tracer) columns over the eastern Pacific.
Based on the passive tracer simulation, transpacific transport decreased over the period of 2005-2008 (Figure 5a).During this four-year period, declining transport efficiency appears to have offset the increase of Asian emissions, resulting in insignificant changes of tropospheric NO 2 columns Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2017-382Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 6 June 2017 c Author(s) 2017.CC BY 3.0 License.
Figure 3a shows the differences of mean surface NO 2 concentrations, as measured by the AQS and NAPS network, from 2009-2010 to 2014-2015.These time periods were chosen to determine changes in surface NO 2 concentrations over the period of 2009-2015 with sufficient

( 2015 )
indicated that the correlation between the OMI NO 2 tropospheric columns (DOMINO) and surface in-situ measurements in the Netherlands was 0.74 at 2007.The stability of our analysis based on OMI retrievals (NASA and DOMINO) is ensured by the strict quality filters; these ensure that changes in OMI sampling due to detector problems (e.g.row anomaly) do not affect our conclusions.

Figure 5
Figure 5 shows the annual slopes of tropospheric NO 2 columns from OMI (NASA and DOMINO) over the period of 2005-2015.Both OMI products (NASA and DOMINO), with various a priori models (GMI and TM4) and algorithms, show consistent variations over the northern Pacific Ocean and the western US: insignificant changes in the period 2005-2008 (Figure 5a-b), positive changes in the period 2009-2012 (Figure 5c-d) and insignificant changes in the period 2013-2015 (Figure 5e-f).Using the Berkeley High-Resolution (BEHR) NO 2 product for OMI, Russell et al. (2012) obtained similar positive change over the western US with the Weather Research Forecasting Chemistry model (WRF-Chem) as an a priori model in the period 2005-2011.The consistency among the various data products suggests that the variations in the retrieved show the annual slope of tropospheric NO 2 columns (percent base) from OMI (NASA and DOMINO) over the period of 2009-2015.Our analysis demonstrates a significant positive change over the northern Pacific Ocean during this time period, and an insignificant but positive change over the western US, suggesting possible contributions from transpacific transport to tropospheric NO 2 over the western US.Over the period of 2005-2008, the lack of change in tropospheric NO 2 columns over the northern Pacific Ocean (Figure 5a-b) indicates the dominant role of local sources to the decrease of US tropospheric NO 2 in this period.Conversely, the increase in tropospheric NO 2 columns over northern Pacific Ocean over the period of 2009-2012 (Figure 5c-d) is consistent Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2017-382Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 6 June 2017 c Author(s) 2017.CC BY 3.0 License.with the appearance of a discrepancy between the OMI retrievals and EPA's emission estimates (Figure 1c).Accompanying with the observed decrease of Chinese NO 2 emissions (Figure 1b), no significant change is observed over the northern Pacific Ocean over the period of 2013-2015 (Figure 5e-f).Decadal climate variability has non-negligible influences on tropospheric compositions by affecting the physical and chemical processes.For example, Lin et al. (2014) indicated that transpacific transport of O 3 is modulated by decadal variability of El Niño-Southern Oscillation (ENSO).El Niño is defined as the appearance of anomalously warm water off northern Peru and Ecuador in December.The atmospheric component tied to El Niño is called the Southern Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2017-382Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 6 June 2017 c Author(s) 2017.CC BY 3.0 License.over the northern Pacific Ocean (Figures 5a-b), and consequently, good agreement between the OMI retrievals and the EPA's emission estimates (Figure 1c).The efficiency of transpacific transport is more stable over the period of 2009-2012 (Figure 6b), which allows stronger transpacific transport of rising Asian emissions.It leads to positive changes of tropospheric NO 2 columns over the northern Pacific Ocean (Figure 5c-d), and consequent growing discrepancy between the OMI retrievals and the EPA's emission estimates (Figure 1c).Increasing efficiency in transpacific transport over the period of 2013-2015 (Figure 6c) counteracts the decrease of Chinese NO 2 (Figure 1b), again resulting in no change in the tropospheric NO 2 column over northern Pacific Ocean (Figure 5e-f) and relatively flat changes in US tropospheric NO 2 column.

Figure
Figure6dshows the comparison between regional mean of passive tracer columns over the northern Pacific Ocean with the NOAA Niño 3.4 index.There is strong correlation between transpacific transport and ENSO: the transpacific transport is stronger in El Niño years and weaker in La Niña years, demonstrating strong influence of decadal climate variability on the transpacific transport.
Figure 1.(a-b): monthly mean (dash lines) and annual mean (solid lines) tropospheric NO 2 column over contiguous United States and East China from OMI (NASA and DOMINO) products; (c-e): percent changes of annual mean (c) and seasonal mean (d-e) tropospheric NO 2 column over the US from the OMI and EPA's emission estimates, normalized at 2009.Figure 2. Schematic figure showing the sources of tropospheric NO x .Figure 3. (a) difference of mean NO 2 concentrations of surface in-situ measurements (AQS and NAPS stations) from 2009-2010 to 2014-2015; Blue (red) means decrease (increase) of NO 2 concentrations.(b) same as panel a, but averaged with 4°x5° resolution.Figure 4. Flash rate density (1x10 -3 flash/km 2 /month) over North America (15°N-42.5°N,130°W-60°W) from Lightning Imaging Sensor (LIS).Figure 5. (a-f) annual slope of tropospheric NO 2 column (unit 1x10 15 molec/cm 2 ) from OMI (NASA and DOMINO products); (g-h) same as panels a-f, with percent (%) as unit.Figure 6. (a-c) Annual slope of passive tracer column (percent base).The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.The lightning NO x emissions are not included in the simulation.(d) Blue line: regional mean (box in panel a) of passive tracer column, normalized by the 11-year mean (2005-2015).The black line shows the NOAA Niño 3.4 index.Figure 7. Seasonal mean passive tracer column (2005-2015), with 10 16 molec cm -2 .The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.It should be noticed that the actual transpacific transport of reactive nitrogen in fall and winter is stronger than panels c-d due to the decrease of temperature.
Figure 1.(a-b): monthly mean (dash lines) and annual mean (solid lines) tropospheric NO 2 column over contiguous United States and East China from OMI (NASA and DOMINO) products; (c-e): percent changes of annual mean (c) and seasonal mean (d-e) tropospheric NO 2 column over the US from the OMI and EPA's emission estimates, normalized at 2009.Figure 2. Schematic figure showing the sources of tropospheric NO x .Figure 3. (a) difference of mean NO 2 concentrations of surface in-situ measurements (AQS and NAPS stations) from 2009-2010 to 2014-2015; Blue (red) means decrease (increase) of NO 2 concentrations.(b) same as panel a, but averaged with 4°x5° resolution.Figure 4. Flash rate density (1x10 -3 flash/km 2 /month) over North America (15°N-42.5°N,130°W-60°W) from Lightning Imaging Sensor (LIS).Figure 5. (a-f) annual slope of tropospheric NO 2 column (unit 1x10 15 molec/cm 2 ) from OMI (NASA and DOMINO products); (g-h) same as panels a-f, with percent (%) as unit.Figure 6. (a-c) Annual slope of passive tracer column (percent base).The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.The lightning NO x emissions are not included in the simulation.(d) Blue line: regional mean (box in panel a) of passive tracer column, normalized by the 11-year mean (2005-2015).The black line shows the NOAA Niño 3.4 index.Figure 7. Seasonal mean passive tracer column (2005-2015), with 10 16 molec cm -2 .The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.It should be noticed that the actual transpacific transport of reactive nitrogen in fall and winter is stronger than panels c-d due to the decrease of temperature.

Figure 2 .
Figure 1.(a-b): monthly mean (dash lines) and annual mean (solid lines) tropospheric NO 2 column over contiguous United States and East China from OMI (NASA and DOMINO) products; (c-e): percent changes of annual mean (c) and seasonal mean (d-e) tropospheric NO 2 column over the US from the OMI and EPA's emission estimates, normalized at 2009.Figure 2. Schematic figure showing the sources of tropospheric NO x .Figure 3. (a) difference of mean NO 2 concentrations of surface in-situ measurements (AQS and NAPS stations) from 2009-2010 to 2014-2015; Blue (red) means decrease (increase) of NO 2 concentrations.(b) same as panel a, but averaged with 4°x5° resolution.Figure 4. Flash rate density (1x10 -3 flash/km 2 /month) over North America (15°N-42.5°N,130°W-60°W) from Lightning Imaging Sensor (LIS).Figure 5. (a-f) annual slope of tropospheric NO 2 column (unit 1x10 15 molec/cm 2 ) from OMI (NASA and DOMINO products); (g-h) same as panels a-f, with percent (%) as unit.Figure 6. (a-c) Annual slope of passive tracer column (percent base).The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.The lightning NO x emissions are not included in the simulation.(d) Blue line: regional mean (box in panel a) of passive tracer column, normalized by the 11-year mean (2005-2015).The black line shows the NOAA Niño 3.4 index.Figure 7. Seasonal mean passive tracer column (2005-2015), with 10 16 molec cm -2 .The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.It should be noticed that the actual transpacific transport of reactive nitrogen in fall and winter is stronger than panels c-d due to the decrease of temperature.

Figure 4 .
Figure 1.(a-b): monthly mean (dash lines) and annual mean (solid lines) tropospheric NO 2 column over contiguous United States and East China from OMI (NASA and DOMINO) products; (c-e): percent changes of annual mean (c) and seasonal mean (d-e) tropospheric NO 2 column over the US from the OMI and EPA's emission estimates, normalized at 2009.Figure 2. Schematic figure showing the sources of tropospheric NO x .Figure 3. (a) difference of mean NO 2 concentrations of surface in-situ measurements (AQS and NAPS stations) from 2009-2010 to 2014-2015; Blue (red) means decrease (increase) of NO 2 concentrations.(b) same as panel a, but averaged with 4°x5° resolution.Figure 4. Flash rate density (1x10 -3 flash/km 2 /month) over North America (15°N-42.5°N,130°W-60°W) from Lightning Imaging Sensor (LIS).Figure 5. (a-f) annual slope of tropospheric NO 2 column (unit 1x10 15 molec/cm 2 ) from OMI (NASA and DOMINO products); (g-h) same as panels a-f, with percent (%) as unit.Figure 6. (a-c) Annual slope of passive tracer column (percent base).The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.The lightning NO x emissions are not included in the simulation.(d) Blue line: regional mean (box in panel a) of passive tracer column, normalized by the 11-year mean (2005-2015).The black line shows the NOAA Niño 3.4 index.Figure 7. Seasonal mean passive tracer column (2005-2015), with 10 16 molec cm -2 .The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.It should be noticed that the actual transpacific transport of reactive nitrogen in fall and winter is stronger than panels c-d due to the decrease of temperature.

Figure 5 .Figure 1 .
Figure 1.(a-b): monthly mean (dash lines) and annual mean (solid lines) tropospheric NO 2 column over contiguous United States and East China from OMI (NASA and DOMINO) products; (c-e): percent changes of annual mean (c) and seasonal mean (d-e) tropospheric NO 2 column over the US from the OMI and EPA's emission estimates, normalized at 2009.Figure 2. Schematic figure showing the sources of tropospheric NO x .Figure 3. (a) difference of mean NO 2 concentrations of surface in-situ measurements (AQS and NAPS stations) from 2009-2010 to 2014-2015; Blue (red) means decrease (increase) of NO 2 concentrations.(b) same as panel a, but averaged with 4°x5° resolution.Figure 4. Flash rate density (1x10 -3 flash/km 2 /month) over North America (15°N-42.5°N,130°W-60°W) from Lightning Imaging Sensor (LIS).Figure 5. (a-f) annual slope of tropospheric NO 2 column (unit 1x10 15 molec/cm 2 ) from OMI (NASA and DOMINO products); (g-h) same as panels a-f, with percent (%) as unit.Figure 6. (a-c) Annual slope of passive tracer column (percent base).The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.The lightning NO x emissions are not included in the simulation.(d) Blue line: regional mean (box in panel a) of passive tracer column, normalized by the 11-year mean (2005-2015).The black line shows the NOAA Niño 3.4 index.Figure 7. Seasonal mean passive tracer column (2005-2015), with 10 16 molec cm -2 .The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.It should be noticed that the actual transpacific transport of reactive nitrogen in fall and winter is stronger than panels c-d due to the decrease of temperature.

Figure 2 .Figure 3 .
Figure 2. Schematic figure showing the sources of tropospheric NO x .

Figure 6 .
Figure 6.(a-c) Annual slope of passive tracer column (percent base).The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.The lightning NO x emissions are not included in the simulation.(d) Blue line: regional mean (box in panel a) of passive tracer column, normalized by the 11-year mean (2005-2015).The black line shows the NOAA Niño 3.4 index.

Figure 7 .
Figure 7. Seasonal mean passive tracer column (2005-2015), with 10 16 molec cm -2 .The passive tracer is simulated with GEOS-Chem model with constant 15-day lifetime.The surface NO x emissions are fixed at 2005 level.It should be noticed that the actual transpacific transport of reactive nitrogen in fall and winter is stronger than panels c-d due to the decrease of temperature.
on North America?To evaluate these critical questions, multiple data sets and model are used in this work, including remotely sensed NO 2 column measurements from Ozone Monitoring Instrument (OMI, NASA and DOMINO products), in-situ surface NO 2 measurements from the Environmental Protection Agency (EPA) Air Quality System (Hudman et al. 2010)s account for up to 40% of the tropospheric NO 2 columns in summer over US rural areas(Hudman et al. 2010), but only contribute a few percent of the US annual mean tropospheric NO 2 column.Similarly, NO x production by lightning is stronger in summer, with an estimated annual contribution of 15% to the total emissions.Because the Atmos.Chem.Phys.Discuss., https://doi.org/10.5194/acp-2017-382Manuscript under review for journal Atmos.Chem.Phys.Discussion started: 6 June 2017 c Author(s) 2017.CC BY 3.0 License.