Using satellite observations of tropospheric NO2 columns to infer trends in US NOx emissions: the importance of accounting for the NO2 background Rachel Silvern, Daniel Jacob, Melissa Sulprizio, Katie Travis, Eloise Marais, Ron Cohen, Josh Laughner, Joanna Joiner, Sunny Choi, Loretta Mickley with support from NASA ACMAP/Aura and EPA ACE

Satellite observations of tropospheric NO2 columns as top-down constraints on NOx emissions OMI NO2, summer 2005 Russell et al. , ACP 2012

OMI observations of tropospheric NO2 columns show fast decrease of US NOx emissions, 2005-2011 OMI NO2, summer 2011 OMI NO2, summer 2005 30% decrease in NOx emissions from 2005 to 2011 Russell et al., ACP 2012

US NOx emissions have continued to decrease according to EPA… but OMI tropospheric NO2 column observations suggest otherwise! Mobile sources Industry NOx emissions, Mt a-1 EPA National Emission Inventory (NEI): 53% sustained decrease of NOx emissions over 2005-2017 EPA, 2018 OMI NO2 columns over CONUS, 2005-2015: flat after 2009 Jiang et al., PNAS 2018 % change relative to 2009

Hypothesis #1: EPA NOx emission trends are wrong Mobile sources in EPA inventory are too high, so maybe trends are wrong as well? NASA SEAC4RS aircraft campaign, Aug Sep 2013 Observed GEOS-Chem with EPA NOx emissions with EPA NOx emissions/2 nitrate wet deposition fluxes, Aug-Sep 2013 Travis et al., ACP 2016 Circles: NADP observations Contours: GEOS-Chem with EPA emissions/2 model bias of -7%

Hypothesis #2: US NOx emissions are now so low that OMI mostly sees background OMI NO2 sensitivity vs. altitude Mean SEAC4RS NO2 profile over Southeast NOAA (Ryerson) Berkeley (Cohen) GEOS-Chem Boundary layer below 2 km accounts for only 20-35% of tropospheric NO2 column as seen by OMI Travis et al. , ACP 2016; Silvern et al., GRL 2018

Use GEOS-Chem to interpret trends from different data sets GEOS-Chem NOx emissions (Tg N a-1) 2005-2017 GEOS-Chem simulation 0.5ox0.625o (50 km) resolution EPA – 40% NO2 column, 1015 cm-2 NO2 column % change relative to 2005 OMI (NASA retrieval) GEOS-Chem GEOS-Chem driven by EPA emission trend does not reproduce flat OMI trend after 2009 GEOS-Chem NOx tropospheric column lifetime increases by 10% from 2005 to 2017

Examining 2005-2017 trends in surface data Annual mean trends for CONUS AQS NO2* (urban) SEARCH NO2 (2 rural sites) NADP wet deposition GEOS-Chem matches the observed relative trends: - surface NO2: most sensitive to anthropogenic emissions - nitrate wet deposition: more sensitive to background US anthropogenic sources contribute only 50% of nitrate wet deposition in 2005, 31% in 2017

Comparing 2005-2017 relative trends for the different quantities Fractional change relative to 2005 Nitrate wet deposition OMI NO2 column SEARCH NO2 AQS NO2 EPA emissions We deduce that: - EPA 2005-2017 trend in NOx emissions is correct - OMI NO2 columns have strong background influence - GEOS-Chem underestimates background NO2 above boundary layer

Separating OMI trends by urban vs. rural, winter vs. summer Trends in NO2 columns relative to 2005 (%) urban winter urban summer rural winter rural summer OMI shows: - steady decrease in urban winter where background is relatively low - no trend in rural summer where background is relatively high

NO2 vertical profiles in SEAC4RS GEOS-Chem underestimates free tropospheric NO2 in winter, upper tropospheric NO2 in summer OMI cloud-sliced free tropospheric NO2 (6-10 km): Sungyeon Choi and Joanna Joiner, NASA NO2 vertical profiles in SEAC4RS (Aug-Sep 2013) GEOS-Chem NOAA (Ryerson) Berkeley (Cohen) GEOS-Chem Marais et al. [ACP 2018] Could the high NO2 observed in upper troposphere extend into the lower stratosphere?

Ozone trends show further evidence of steady decline in US NOx emissions 95th percentile of summer MDA8 ozone (ppb) for each site, averaged over CONUS Ozone would not be expected to continue decreasing if NOx emissions had flattened

Conclusions Surface observations of NO2 and ozone confirm steady decrease of US NOx emissions as reported by the EPA National Emission Inventory Nitrate wet deposition trends are also consistent with inventory but are weaker because of background influence OMI NO2 trends are similar to wet deposition: large background influence from free/upper tropospheric NO2 Better understanding of this free/upper tropospheric NO2 is needed to interpret satellite NO2 tropospheric column data in terms of NOx emissions