effects of temperature-dependent no emissions on continental … · 2020. 7. 31. ·...

Atmos Chem Phys 18 2601ndash2614 2018httpsdoiorg105194acp-18-2601-2018copy Author(s) 2018 This work is distributed underthe Creative Commons Attribution 40 License

Effects of temperature-dependent NOx emissions on continentalozone productionPaul S Romer1 Kaitlin C Duffey1 Paul J Wooldridge1 Eric Edgerton2 Karsten Baumann23 Philip A Feiner4David O Miller4 William H Brune4 Abigail R Koss567 Joost A de Gouw67 Pawel K Misztal8Allen H Goldstein89 and Ronald C Cohen110

1Department of Chemistry University of California Berkeley Berkeley CA 94720 USA2Atmospheric Research and Analysis Inc Cary NC 27513 USA3Department of Environmental Sciences and Engineering University of North Carolina at Chapel Hill Chapel HillNC 27599 USA4Department of Meteorology and Atmospheric Science The Pennsylvania State University University Park PA 16802 USA5Chemical Sciences Division NOAA Earth System Research Laboratory (ESRL) Boulder CO 80305 USA6Cooperative Institute for Research in Environmental Sciences University of Colorado Boulder Boulder CO 80309 USA7Department of Chemistry and Biochemistry University of Colorado Boulder Boulder CO 80309 USA8Department of Environmental Science Policy and Management University of California Berkeley BerkeleyCA 94720 USA9Department of Civil and Environmental Engineering University of California Berkeley Berkeley CA 94720 USA10Department of Earth and Planetary Sciences University of California Berkeley Berkeley CA 94720 USA

Correspondence Ronald C Cohen (rccohenberkeleyedu)

Received 21 September 2017 ndash Discussion started 25 September 2017Revised 11 January 2018 ndash Accepted 15 January 2018 ndash Published 22 February 2018

Abstract Surface ozone concentrations are observed to in-crease with rising temperatures but the mechanisms respon-sible for this effect in rural and remote continental regionsremain uncertain Better understanding of the effects of tem-perature on ozone is crucial to understanding global air qual-ity and how it may be affected by climate change We com-bine measurements from a focused ground campaign in sum-mer 2013 with a long-term record from a forested site inthe rural southeastern United States to examine how dailyaverage temperature affects ozone production We find thatchanges to local chemistry are key drivers of increased ozoneconcentrations on hotter days with integrated daily ozoneproduction increasing by 23 ppb Cminus1 Nearly half of this in-crease is attributable to temperature-driven increases in emis-sions of nitrogen oxides (NOx) most likely by soil microbesThe increase of soil NOx emissions with temperature sug-gests that ozone will continue to increase with temperaturein the future even as direct anthropogenic NOx emissionsdecrease dramatically The links between temperature soilNOx and ozone form a positive climate feedback

1 Introduction

Elevated concentrations of tropospheric ozone are an im-portant contributor to anthropogenic radiative forcing andare associated with increased human mortality and decreasedcrop yields (Myhre et al 2013 World Health Organization2005 Booker et al 2009) Observations of increased surfaceozone concentrations on hotter days are widely reported butthe mechanisms driving this relationship are poorly under-stood in regions and climates with low concentrations of ni-trogen oxides (NOx equiv NO+NO2) Understanding the mech-anisms driving these increases is critical to effectively reg-ulating ozone pollution and predicting the effects of globalwarming on air quality

Several previous studies (eg Sillman and Samson 1995Weaver et al 2009 Pusede et al 2014) have used insitu observations and chemical transport models to exam-ine the relationships between ozone (O3) and temperature(T ) Typically observed slopes range from 1 to 6 ppb Cminus1with greater values occurring in more polluted environments(Pusede et al 2015) A few studies have also reported that

Published by Copernicus Publications on behalf of the European Geosciences Union

2602 P S Romer et al Effects of temperature-dependent NOx emissions

this effect is nonlinear and can become significantly lessstrong at the highest temperatures (Steiner et al 2010 Shenet al 2016)

Increased ozone concentrations with temperature in ur-ban areas can be well explained by increased ozone pro-duction caused by greater emissions of volatile organiccompounds (VOCs) and decreased sequestration of NOx inshort-term reservoirs (Jacob and Winner 2009) In contrastthere is little consensus on the mechanisms responsible fortemperature-dependent changes in ozone concentrations inrural and remote environments Arguments in favor of large-scale changes in atmospheric circulation and in favor of lo-cal changes in the chemical production and loss of ozonehave both been presented (Barnes and Fiore 2013 Steineret al 2006) Regional stagnation episodes often associatedwith elevated temperatures allow ozone to accumulate overseveral days and are known to contribute significantly tothe ozone-temperature relationship (Jacob et al 1993) Howvarious temperature-dependent chemical effects interact andtheir relative contributions to ozone production are not wellunderstood outside of polluted environments

Summer daytime ozone concentrations at rural sites inthe United States typically range from 35 to 55 ppb (Cooperet al 2012) sufficient to cause harm to humans crops andthe climate Epidemiological studies and meta-analyses in-vestigating the relationship between ozone and daily mortal-ity have found significant effects in small cities and rural lo-cations with some studies suggesting that increases in ozonemay have a greater effect on daily mortality under less pol-luted conditions (Vedal et al 2003 Ito et al 2005 Atkinsonet al 2012) Studies of crop yield and plant health have tra-ditionally used a threshold of 40 ppb when investigating theeffects of ozone exposure but many crops have been shownto experience reduced yields when exposed to ozone concen-trations as low as 20 ppb (Pleijel et al 2004 Booker et al2009) From a regulatory perspective elevated regional back-ground ozone can strongly exacerbate ozone pollution andthe probability of regulatory exceedances in urban areas suchas Houston (Berlin et al 2013) Understanding the behaviorof O3 in the rural and remote areas that cover the majority ofthe land area of the Earth is therefore crucial for effectivelypredicting and controlling air quality now and in the future

In this paper we use observations from Centreville Al-abama (CTR) a rural site in the southeastern United States(Supplement Fig S1) to investigate how temperature affectsozone production Long-term monitoring from the South-Eastern Aerosol Research and CHaracterization (SEARCH)network shows that ozone increases significantly with tem-perature at this site (Fig 1) despite being in a low-NOx en-vironment where the predicted response of the instantaneousozone production rate to temperature is small (Pusede et al2015) We combine this record with extensive measurementsfrom the Southern Oxidant and Aerosol Study (SOAS) insummer 2013 to explicitly calculate daily integrated ozoneproduction and NOx loss as a function of daily average tem-

Figure 1 The O3ndashtemperature relationship in Centreville Al-abama Daily afternoon (1200ndash1600) average ozone concentrationis shown as a function of temperature over JunendashAugust 2010ndash2014at the SEARCH CTR site The black line and shaded gray regionshow the running median and interquartile range of ozone with tem-perature respectively The red line represents a fit to all daily datapoints

perature We find that changes in local chemistry are impor-tant drivers of the increase in ozone concentrations observedat this site and that increased NOx emissions are responsi-ble for 40 of the temperature-dependent increase in dailyintegrated ozone production We expect similar effects to bepresent in other low-NOx areas with high concentrations ofVOCs where the chemistry of alkyl and multifunctional ni-trates is the majority pathway for permanent NOx loss

2 Chemistry of ozone production and predictedresponse to temperature

Observed O3ndashT relationships are caused by a combinationof chemical changes to the production and loss of O3 andchanges to atmospheric circulation that determine advectionand mixing To begin separating these effects we considerthe chemical production of ozone (PO3) and how it changeswith temperature Temperature-dependent changes in ozoneproduction may be driven directly by temperature or by an-other meteorological parameter that co-varies with tempera-ture such as solar radiation

Ozone is produced in the troposphere when NO is con-verted to NO2 by reaction with HO2 or RO2 in the linkedHOx and NOx cycles (Fig 2a) HO2 and RO2 radicals aregenerated in the HOx cycle when a VOC reacts with OH inthe presence of NOx In one turn of the cycle the VOC isoxidized OH is regenerated and two molecules of O3 areformed The reactions that drive these catalytic cycles for-ward are in constant competition with reactions that removeradicals from the atmosphere terminating the cycles Termi-nation can occur either through the association of two HOx

Atmos Chem Phys 18 2601ndash2614 2018 wwwatmos-chem-physnet1826012018

P S Romer et al Effects of temperature-dependent NOx emissions 2603

Figure 2 The chemistry of ozone production and NOx loss in the troposphere (a) Schematic of the linked NOx and HOx cycles that leadto net ozone production (b) The calculated instantaneous O3 production rate and NOx loss rate as a function of NOx and VOCR with fixedPHOx η and αeff (c) OPE and the fraction of NOx loss that takes place via HNO3 chemistry under the same conditions as (b) (d) Thepercent change in ozone production efficiency caused by chemical changes as a function of NOx

radicals to form inorganic or organic peroxides or throughthe association of HOx and NOx radicals to form nitric acidor an organic nitrate

The balance between propagating and terminating reac-tions causes PO3 to be a non-linear function of the NOx andVOC reactivity (VOCR) as well as the production rate ofHOx radicals (PHOx) The largest source of HOx radicalsin the summertime is the photolysis of O3 followed by reac-tion with water vapor to produce OH additional sources in-clude the photolysis of formaldehyde and peroxides ozonol-ysis of alkenes and isomerization pathways in the oxidationof isoprene and other VOCs To understand the response ofozone production to changes in chemistry we use a simpli-

fied framework based on the balance of HOx radical produc-tion and loss (Farmer et al 2011)

Under moderate or high NOx conditions the primary lossprocess of HO2 and RO2 radicals is reaction with NO and theconcentration of OH radicals can be expressed as a quadraticequation To modify this approach to work under low-NOxconditions reactions between HOx radicals must also be in-cluded leading to a set of four algebraic equations that can besolved numerically (details given in Appendix A) Figure 2bshows the calculated rate of ozone production as a functionof NOx at two different VOC reactivities Depending on at-mospheric conditions the ozone production rate can either beNOx-limited where additional NOx causes PO3 to increase

wwwatmos-chem-physnet1826012018 Atmos Chem Phys 18 2601ndash2614 2018


or NOx-saturated where additional NOx suppresses ozoneformation

When considering day-to-day variations the total amountof ozone produced over the course of a day (

intPO3) is a more

representative metric than the instantaneous ozone produc-tion rate Total daily ozone production depends on all of thefactors that affect PO3 as well as their diurnal evolutionIn places where ozone production is NOx-limited changesto chemistry with temperature that affect the NOx loss rate(LNOx) can affect

intPO3 by changing the amount of NOx

available for photochemistry later in the day (Hirsch et al1996)

Permanent NOx loss occurs through two primary path-ways in the troposphere the association of OH and NO2 toform HNO3 and through the chemistry of alkyl and mul-tifunctional nitrates (6RONO2) These organic nitrates areformed as a minor channel of the RO2+NO reaction withthe alkyl nitrate branching ratio αi ranging from near zero forsmall hydrocarbons to over 020 for monoterpenes and long-chain alkanes (Perring et al 2013) The overall alkyl nitratebranching ratio αeff represents the reactivity-weighted aver-age of αi for all VOCs While some fraction of 6RONO2quickly recycles NOx to the atmosphere a significant frac-tion η permanently removes NOx through deposition andhydrolysis (eg Browne et al 2013) Romer et al (2016)determined that η = 055 during SOAS and was controlledprimarily by the hydrolysis of isoprene hydroxy-nitrates Be-cause the hydrolysis rate is set primarily by the distributionof nitrate isomers which does not change appreciably withtemperature we assume that η is constant with temperaturein this study (Hu et al 2011 Peeters et al 2014) Deposi-tion is only a minor loss process for 6RONO2 therefore anychanges in the deposition rate with temperature will have atmost a minor effect on η

NOx also has several temporary sinks that can sequesterNOx most importantly peroxy acyl nitrate (PAN) In thesummertime in the southeastern United States the lifetime ofPAN is typically 1ndash2 h too short to act as a permanent sink ofNOx Past studies in forested regions have found remarkablylittle variation in PAN with temperature due to compensat-ing changes in both its production and loss (eg LaFranchiet al 2009) As a result the formation or destruction of PANdoes not contribute significantly to net ozone production orNOx loss and we do not include it in these calculations

The ozone production efficiency (OPE equiv PO3LNOx)represents the number of ozone molecules formed permolecule of NOx consumed and directly links the ozone andNOx budgets Because OPE accounts for changes in bothPO3 and LNOx the temperature response of OPE capturesfeedbacks in ozone production chemistry that PO3 alonedoes not

As the concentration of NOx decreases and VOCR in-creases the fraction of NOx loss that takes place via HNO3chemistry decreases and the OPE increases (Fig 2c) The rel-ative importance of HNO3 and RONO2 chemistry determines

the relationship between PO3 andLNOx When HNO3 is themost important NOx loss pathway O3 production and NOxloss occur through separate channels O3 production occurswhen OH reacts with a VOC generating RO2 and HO2 rad-icals NOx loss primarily occurs when OH reacts with NO2Although these channels are linked by a shared dependenceon OH the relative importance of these pathways can varyFor example under these conditions an increase in VOCRwill cause NOx loss to decrease ozone production to in-crease and OPE to increase (Fig 2bndashc)

In contrast when RONO2 chemistry dominates NOx lossozone production and NOx loss are intrinsically linked bytheir shared dependence on the RO2+NO reaction This re-action produces O3 in its main channel and consumes NOxin the minor channel that forms organic nitrates with theratio between these two channels set by αeff Under theseconditions changes to the chemistry that do not affect αeffhave a minimal effect on OPE (Fig 2d) and the OPE can beconsidered to be unvarying with temperature An increase inVOCR or a decrease in NOx will affect both NOx loss andozone production equally because both processes are depen-dent on the same set of reactions Because of this change inbehavior from variable OPE to fixed OPE the drivers of theO3ndashT relationship are expected to be categorically differentin areas where RONO2 chemistry dominates NOx loss Asa result the effects that cause O3 to increase with tempera-ture in urban and other polluted regions where HNO3 chem-istry dominates NOx loss are unlikely to apply in areas withlow concentrations of NOx and high concentrations of reac-tive VOCs where RONO2 chemistry is most important Inthese areas more NOx must be oxidized in order to producemore O3

3 Observed response of ozone production totemperature

31 Measurements during SOAS

The theoretical results presented in Fig 2 can be compared tothe observed behavior during SOAS (SOAS Science Team2013) Measurements during SOAS have been described indetail elsewhere (eg Hidy et al 2014 Romer et al 2016Feiner et al 2016) and are summarized below The primaryground site for SOAS was co-located with the CTR site of theSEARCH network (3290289 N 8724968W) in a clear-ing surrounded by a dense mixed forest (Hansen et al 2003)Direct anthropogenic emissions of NOx near this site are es-timated to be low and predominantly from mobile sources(Hidy et al 2014) Figure S1 shows the location of the CTRsite relative to major population centers in the region Mea-surements taken as part of the SEARCH network were lo-cated on a 10 m tower approximately 100 m away from theforest edge while the other measurements from the SOAScampaign used in this analysis were located on a 20 m walk-



up tower at the edge of the forest Species measured on boththe SOAS walk-up tower and the SEARCH platform werewell correlated with each other indicating that similar airmasses were sampled at both locations

Several chemical and meteorological measurements usedin this study including NOx O3 total reactive nitrogen(NOy) and temperature were collected by Atmospheric Re-search and Analysis (ARA) as part of SEARCH (Hidy et al2014) NO was measured using the chemiluminescent reac-tion of NO with excess ozone NO2 was measured basedon the same principle using blue LED photolysis to con-vert NO2 to NO The photolytic conversion of NO2 to NOis nearly 100 efficient and does not affect higher oxides ofnitrogen (Ryerson et al 2000) Ozone was measured usinga commercially available ozone analyzer (Thermo-Scientific49i)

During the SOAS campaign NO2 total peroxy ni-trates (6PNs) and total alkyl and multifunctional nitrates(6RONO2) were measured via thermal dissociation laser-induced fluorescence as described by Day et al (2002) AnNO chemiluminescence instrument located on the walk-uptower provided additional measurements of NO co-locatedwith the other SOAS measurements (Min et al 2014)

HOx radicals were measured with the Penn State Ground-based Tropospheric Hydrogen Oxides Sensor (GTHOS)which uses laser-induced fluorescence to measure OH(Faloona et al 2004) HO2 was also measured in this in-strument by adding NO to convert HO2 to OH C3F6 wasperiodically added to the sampling inlet to quantify the inter-ference from internally generated OH (Feiner et al 2016)Measurements of total OH reactivity (OHR equiv inverse OHlifetime) were made by sampling ambient air injecting OHand letting the mixture react for a variable period of time Theslope of the OH signal vs reaction time provides a top-downmeasure of OHR (Mao et al 2009)

A wide range of VOCs were measured during SOAS us-ing gas chromatography-mass spectrometry (GC-MS) Sam-ples were collected in a liquid-nitrogen cooled trap forfive minutes then transferred by heating onto an analyticalcolumn and detected using an electron-impact quadrupolemass-spectrometer (Gilman et al 2010) This system is ableto quantify a wide range of compounds including alkanesalkenes aromatics isoprene and multiple monoterpenes ata time resolution of 30 min Methyl vinyl ketone (MVK) andmethacrolein (MACR) were measured individually by GC-MS and their sum was also measured using a proton transferreaction mass spectrometer (PTR-MS) (Kaser et al 2013)The calculated rates of ozone production and NOx loss donot change significantly depending on which measurementis used

32 Calculation ofint

P O3 and effects of temperature

During the SOAS campaign afternoon concentrations ofNOx averaged 03 ppb and concentrations of isoprene 55 ppb

Figure 3 Observed dependence of dailyintPO3 (a)

intLO3 (b) andint

LNOx (c) on daily afternoon average temperature during SOASEach point shows the afternoon average temperature and integratedproduction or loss for a single day Black lines show a least squaresfit to all points shaded areas show the 90 confidence limits of thefit calculated via bootstrap sampling

(Fig S2) 6RONO2 chemistry was responsible for overthree-quarters of the permanent NOx loss (Romer et al2016) Daily average afternoon (1200ndash1600) ozone concen-trations increased with daily average afternoon temperatureduring SOAS (23 plusmn 1 ppb Cminus1) This trend is greater thanthe long-term trend reported by the SEARCH network butthe difference is not statistically significant

Measurements of NO NO2 OH HO2 and a wide rangeof VOCs (Table S1) were used to calculate the steady-stateconcentrations of RO2 radicals using the Master ChemicalMechanism v331 run in a MATLAB framework (Jenkinet al 2015 Wolfe et al 2016) Before 24 June HO2 mea-surements are not available and steady-state concentrationsof both HO2 and RO2 were calculated Input species weretaken to 30 min averages and the model was run until radi-cal concentrations reached steady state Top-down measure-ments of OHR were used to include the contribution to ozoneproduction from unmeasured VOCs

To understand the day-to-day variation of ozone chem-istry the calculated ozone production rate was integratedfrom 0600 to 1600 for each of the 24 days during the cam-paign period with greater than 75 data coverage of all inputspecies When plotted against daily average afternoon tem-perature

intPO3 is seen to increase strongly with tempera-

ture (23plusmn 06 ppb Cminus1 Fig 3a) The change inintPO3 with

temperature demonstrates that local chemistry is an impor-



tant contributor to the observed O3ndashT relationship howeverthe observed O3ndashT trend also includes the effects of chem-ical loss advection entrainment and multi-day buildup onoverall O3 concentration (eg Baumann et al 2000)

While elevated temperatures are associated with enhancedproduction of ozone they are also associated with increasedchemical loss The chemical loss of ozone occurs via threemain pathways in this region photolysis followed by reac-tion with H2O reaction with HO2 and reaction with VOCs(Frost et al 1998) The loss of O3 was calculated for eachof these pathways and then integrated over the course ofthe day to determine total daily ozone loss (

intLO3) Chem-

ical loss of ozone is found to increase with temperature(11 plusmn 03 ppb Cminus1 Fig 3b) but much less than the chemi-cal production

The difference between the trend in the net chemical pro-duction and loss of O3 and the trend in ozone concentra-tion gives a rough estimate of how non-chemical processescontribute to the ozonendashtemperature relationship We calcu-late that non-chemical processes cause O3 to increase by1 plusmn 12 ppb Cminus1 This approach does not take into accountthe interactions between chemical and non-chemical effectssuch as how changes to advection and mixing may impactconcentrations of VOCs NOx and other reactants Althoughthe large uncertainty does not allow for quantitative analysisqualitatively chemical and non-chemical processes are bothfound to be important contributors to the ozonendashtemperaturerelationship Other approaches such as chemical transportmodels that can more directly investigate and control spe-cific physical processes are likely to be better suited to cal-culating the contribution of non-chemical processes to theozonendashtemperature relationship (eg Fu et al 2015)

Using the same calculated radical concentrations the rateof NOx loss was calculated as the rate of direct HNO3production plus the fraction η of alkyl nitrate productionthat leads to permanent NOx loss Figure 3c shows the in-crease in

intLNOx with temperature for the SOAS campaign

(005 plusmn 001 ppb Cminus1) As expected from the importanceof RONO2 chemistry to NOx loss

intLNOx and

intPO3 are

tightly correlated (r2= 090) and OPE is high (OPE aver-

age 45 plusmn 3 ppbppbminus1) and is effectively constant with tem-perature (calculated trend 02plusmn 06 Cminus1) Therefore the in-crease in

intPO3 with temperature is not caused by more effi-

cient production of ozone while the same amount of NOx isconsumed

OPE can also be estimated from the ratio of odd oxy-gen (Ox equiv O3+NO2) to NOx oxidation products (NOz equivNOy minusNOx) (Trainer et al 1993) The afternoon ratio ofOx to NOz during SOAS varied from 43 to 67 (interquar-tile range) slightly higher than the average ratio of

intPO3 toint

LNOx However since the Ox to NOz ratio includes the ef-fects of chemical loss and transport which the ratio of

intPO3

tointLNOx does not these two values are not expected to be

equivalent particularly in non-polluted areas

Figure 4 Afternoon average concentrations of NOx (a) 6PNs (b)and NOy (c) at the CTR site as a function of daily average afternoontemperature during SOAS

The trend inintPO3 with temperature is robust and extends

beyond the short temporal window of the SOAS campaignAlthough long-term measurements of HOx and VOCs are notavailable the ozone production rate can be estimated fromSEARCH measurements using the deviation of NO and NO2from photostationary state (Eq 1) (Baumann et al 2000Pusede et al 2015)

PO3 = jNO2 [NO2] minus kNO+O3 [NO][O3] (1)

The NO2 photolysis rate was parameterized as a quadraticfunction of total solar radiation (Trebs et al 2009) Us-ing this method and scaling the result to match the val-ues calculated using steady-state RO2 concentrations duringSOAS we find that

intPO3 increased by 23 plusmn 08 ppb Cminus1

during JunendashAugust 2010ndash2014 (Fig S3) Without scalingthe long-term trend in

intPO3 with temperature is 40 plusmn

05 ppb Cminus1 Based on the long-term SEARCH record wedo not find evidence that the relationship between tempera-ture and ozone concentration or ozone production changessignificantly at the highest temperatures (the top 5 of ob-servations) This agrees broadly with Shen et al (2016) whofound ozone suppression at extreme temperatures to be un-common in the southeastern United States



Figure 5 Concentrations of NOx relative to their concentration at1600 the day before over JunendashAugust 2010ndash2014 at the CTR siteThe thick lines and shaded areas show the hourly mean and 90 confidence interval of the mean for cool and warm days

4 Drivers of increased ozone production

While the increase in ozone production is accompanied byan observed increase in ozone concentration the increase inNOx loss is not accompanied by a significant decrease inNOx concentration (minus0002 plusmn 001 ppb Cminus1 Fig 4a) Forthis to occur NOx must have a source that increases withtemperature to compensate for its increased loss One pos-sible explanation is that the increased thermal decomposi-tion rate of peroxy nitrates (6PNs) causes less NOx to be se-questered in these short-term reservoirs This is not the caseduring SOAS The increased decomposition rate of peroxynitrates is counteracted by an increase in their productionrate such that the average concentration of total peroxy ni-trates shows no decrease with temperature (Fig 4b)

More generally increased transformations from NOx ox-idation products back into NOx cannot explain the observa-tions The concentration of NOy increases significantly withtemperature (Fig 4c) Because NOy includes NOx as wellas all of its reservoirs and sinks changes in the transforma-tion rates between NOx and its oxidation products cannot ex-plain the increase of NOy with temperature There must bea source of NOy not just of NOx that increases with tem-perature

Data from the SEARCH network indicate that the increasein NOy with temperature observed during SOAS is primar-ily a local effect Measurements from June to August 2010ndash2014 show a consistent increase of NOy with temperature atthe two rural monitoring sites in the network but total NOydecreases with temperature at the four urban and suburbansites (Table S2) The increase in NOy with temperature can-not therefore be explained by regional meteorological effectssince those would lead to similar relationships between NOyand temperature across the southeastern United States

Measurements at night and in the early morning be-fore significant photochemistry has occurred show a strongtemperature-dependent increase of NOx over the course ofthe night Because surface wind speeds are low at night and

the increase in NOx during the night is not accompanied bylarge increases in NOx oxidation products the increase inNOx must be caused by emissions local to the CTR site

The consistent increase of NOx over the course of thenight can be used to quantitatively measure the local NOxemissions rate Figure 5 shows the temperature-dependent in-crease of NOx relative to the concentration of NOx at 1600the day before separating the effects of the previous dayfrom the nighttime increase Measurements from June to Au-gust 2010ndash2014 from the CTR SEARCH network site areused to obtain more representative statistics The average rateof NOx increase during the night is 0095 ppbhminus1 To ac-count for the chemical removal of NOx the cumulative lossof NOx during the night was added to the observations Dur-ing SOAS the nighttime loss of NOx occurred almost ex-clusively through the reaction of NO2 with O3 to form NO3which then reacted with a VOC to form an organic nitrate(Ayres et al 2015) N2O5 chemistry made a negligible con-tribution to total NOx loss The loss rate of NOx during thenight was therefore calculated as the rate of reaction of NO2with O3 In this form the rate of increase of the adjusted NOxconcentrations (NOlowastx) is equal to the local NOx emission rateThe emission rate of NOx and its temperature dependencewere calculated by a linear regression following the form ofEq (2) where the adjusted concentration of NOx dependsboth on time (H is hours after 1600) and temperature (T )

NOlowastx = (αT +β)H + b (2)

In this regression the fitted parameter α represents the in-crease of NOx emissions with temperature and the averagevalue of αT +β provides an estimated NOx emission rate

Because emissions are localized to the surface the ef-fective depth of the nighttime boundary layer must also beaccounted for which we estimate to be 150 m This agreeswell with the derived mixing heights from daily 0500 sondelaunches at the Birmingham (BHM) airport (Durre and Yin2008) and past estimates of the nocturnal boundary layerheight (eg Liu and Liang 2010 VandenBoer et al 2013)while it is significantly lower than the average ceilometer-reported 0500 boundary layer height of 400 m during SOAS(Fig S4)

After accounting for these factors the NOx emissions rateis calculated to be 74 ppt m sminus1 or 42 ng N mminus2 sminus1 Basedon the change in slope with temperature the emissions rate isestimated to increase by 04 ppt m sminus1 Cminus1 The rise in NOxemissions with temperature over 24 h agrees to within theuncertainty with the increase of daily

intLNOx with tempera-

ture sufficient to explain why afternoon NOx concentrationsare not observed to decrease with temperature even as theirloss rate increases

The inferred local NOx source bears all the hallmarks ofsoil microbial emissions (SNOx ) Soil microbes emit NOx asa byproduct of both nitrification and denitrification and therate of NOx emissions strongly correlates with microbial ac-tivity in soil (Pilegaard 2013) The inferred NOx source is



Figure 6 Decomposed effects of ozone and temperature The topbar shows the model-calculated

intPO3-T trend all other bars show

how theintPO3-T slope changes when the temperature dependence

of each factor is removed

active during day and night increases strongly with temper-ature and is present in a rural area with low anthropogenicemissions The only plausible source of NOx that matchesall of these constraints is soil microbial emissions near theSOAS site Soil NOx emissions also depend on the watercontent and nitrogen availability neither of which is gener-ally limiting in the southeastern United States (eg Hickmanet al 2010) The most likely anthropogenic sources of NOxat this location are mobile sources which are not thoughtto change significantly with temperature (Singh and Sloan2006) and therefore cannot explain the results of Fig 5

To calculate how the increase in NOx emissions affectsozone production we use the same chemical framework fromFig 2 For each half-hour period the average value of the in-put parameters and their temperature dependence during theSOAS campaign were calculated (Fig S5) The diurnal cy-cle and trend with temperature of all model inputs were thenused to calculate total daily ozone production as a functionof temperature (Fig S6) By altering whether the tempera-ture dependence for each parameter is included the overalltrend in

intPO3 can be decomposed into individual compo-

nents (Fig 6) The effect of increased NOx emissions wascalculated by fixing the trend in NOx with temperature tomatch the trend in

intLNOx We find that the increase of NOx

emissions with temperature accounts for 40 of the increasein

intPO3 with temperature or approximately 09 ppb Cminus1

The other 60 is primarily caused by the increase of PHOxwith temperature The increase in PHOx with temperature ismost likely caused by changes in solar radiation which iswell correlated with the total PHOx rate (Fig S7a) and in-creases strongly with temperature In contrast water vapor isnot correlated with total PHOx (Fig S7b) Although VOCRincreases strongly with temperature the RONO2-dominatedNOx chemistry causes neither the ozone production rate nor

the NOx loss rate to be sensitive to this increase leading tothe minimal effect of VOCR on

intPO3

5 Conclusions

Changes in NOx emissions with temperature have an out-sized effect when considering the impacts of ozone on hu-man health and climate At the CTR site and other areaswhere OPE does not vary with temperature the total amountof ozone produced on weekly or monthly timescales is di-rectly proportional to the amount of available NOx Whilefaster oxidation on hotter days causes more ozone to be pro-duced without changes in NOx emissions there would bean associated decrease in ozone production on subsequentdays because the NOx necessary for ozone production wouldbe depleted In contrast increased NOx emissions can causeweekly or monthly average ozone concentrations to increasewith temperature Change in long-term average ozone con-centrations is often more important to the ozone climate feed-back and human health than day-to-day variation The mech-anisms described here are likely to be active in all areas withlow concentrations of NOx and high concentrations of re-active VOCs Only regions where RONO2 chemistry is thedominant pathway for NOx loss have effectively constantOPE with temperature but the effect of soil NOx emissionson ozone production is widespread

Past direct measurements of soil NOx using soil cham-bers have found enormous variability both between sites andwithin different plots in the same field Pilegaard et al (2006)found variability of a factor of over 100 between soil NOxemissions in different European forests Within the south-eastern United States direct measurements at forested siteshave reported emissions rates ranging from 01 to 10 ng Nmminus2 sminus1 (Williams and Fehsenfeld 1991 Thornton et al1997 Hickman et al 2010) Besides temperature the mostimportant variables affecting soil NOx emissions are typi-cally nitrogen availability and soil water content as well asplant cover and soil pH (Pilegaard 2013) In very wet en-vironments soil microbes typically emit N2O or N2 insteadof NOx and in arid environments soil emissions of HONOcan be equal to or larger than soil NOx emissions (Oswaldet al 2013) Although conditions at the CTR site are toowet and acidic for soil HONO emissions to be significantin environments where soil HONO emissions are large theywould likely have an even greater effect on ozone productionby acting as a source of both NOx and HOx radicals

The variability between sites and the interaction betweenseveral biotic and abiotic factors make it difficult to applyregional or model estimates of soil NOx emissions to a par-ticular location Our approach from this study using obser-vations of the nighttime atmosphere to determine the NOxemissions rate helps span the gap between soil chambers andthe regional atmosphere Although soil NOx emissions de-pend on several environmental factors process-driven mod-



els predict that the response of soil NOx emissions to globalwarming will be driven primarily by the increase in temper-ature (Kesik et al 2006)

While soil NOx emissions have been known and studiedfor decades the impact of soil NOx emissions on ozone fromnon-agricultural regions was often found to be insignificantcompared to anthropogenic sources (eg Davidson et al1998) Years of declining anthropogenic NOx emissions inthe United States and recent higher estimates for forest soilNOx emissions (eg Hickman et al 2010) mean that thisis no longer the case Non-agricultural soil NOx emissionsmay now account for nearly a third of total NOx emissionsin the summertime in the southeastern United States (Traviset al 2016) and have significant effects on regional ozoneproduction

The rise in ozone production caused by increased NOxemissions on hotter days established here suggests that therelationship between ozone and temperature will be positiveunder a wider range of conditions than previously thoughtThis includes 1 the pre-industrial atmosphere 2 present dayrural continental locations and 3 future scenarios with dra-matically reduced anthropogenic NOx emissions

1 In pre-industrial times semi-quantitative measurementsof ozone show significantly lower concentrations ofozone than are currently observed in rural and re-mote regions or generally predicted by global models(Cooper et al 2014) While alkyl nitrate chemistry es-tablishes an upper limit to the ozone production effi-ciency under low-NOx conditions the significant con-tribution of SNOx to ozone production makes reconcilingthe semi-quantitative measurements with model predic-tions more difficult and suggests that natural emissionsof NOx in pre-industrial models may be over-estimated(Mickley et al 2001)

1 In the present day effective ozone regulation espe-cially on hot days requires taking the effect of SNOx

into account Because these emissions are distributedover broad areas and are not directly anthropogenicthey present additional challenges to air quality man-agement Indirect approaches such as changes to fertil-izer application practices have the potential to signifi-cantly reduce SNOx from agricultural regions (Oikawaet al 2015) Decreases in direct anthropogenic NOxemissions may also lead to a decrease in SNOx by re-ducing the amount of nitrogen available to the ecosys-tem (Pilegaard 2013)

2 In the future because soil NOx emissions lead to theformation of ozone itself an important greenhouse gasthe increase of soil NOx emissions with temperaturerepresents a positive climate feedback and an addi-tional link between changes to the nitrogen cycle andthe environment The effects of increased ozone pol-lution to plants including reduced photosynthesis andslower growth have the potential to alter the carboncycle on a regional scale (Heagle 1989 Booker et al2009) Soil NOx emissions therefore represent an addi-tional link between the nitrogen and carbon cycles thatshould be included when considering the consequencesof a warming world

Data availability Measurements from the SOAS campaign areavailable at httpsesrlnoaagovcsdgroupscsd7measurements2013senexGroundDataDownload (SOAS Science Team 2013)



Appendix A

A1 Analytical P O3 model

To conceptually understand O3 production and NOx losswe use a simplified framework similar to that described byFarmer et al (2011) This framework uses fixed values of to-tal organic reactivity (VOCR) alkyl nitrate branching ratio αand loss efficiency η NOx and HOx radical production rate(PHOx)

Since HOx radicals are highly reactive it is a valid as-sumption under nearly all NOx concentrations that HOx rad-icals are in steady state and that PHOx is equal to the grossHOx loss rate (Eq A1)

PHOx = kOH+NO2 [OH][NO2]

+α middot kRO2+NO[RO2][NO] + 2kHO2+HO2 [HO2][HO2]

+ 2kRO2+HO2 [RO2][HO2] + 2kRO2+RO2 [RO2][RO2] (A1)

Individual HOx radicals (OH HO2 and RO2) can also beassumed to be in steady state such that their production andloss are equal Under low-NOx conditions the reactions thatinitiate and terminate the HOx cycle must be included aswell as the cycling rate We further constrain the model byrequiring that the concentration of HO2 and RO2 radicalsbe equal This constraint is satisfied by introducing an addi-tional parameter c which allows PHOx to produce both HO2and OH radicals in a varying ratio These constraints providea system of four equations that can be solved numerically(Eqs A2ndashA5)

[OH] =kHO2+NO[HO2][NO] + c middotPHOx

VOCR+ kOH+NO2 [NO2](A2)

[RO2] =

[OH] middotVOCRkRO2+NO[NO] + kRO2+HO2 [HO2] + 2kRO2+RO2 [RO2]

(A3)

[HO2] =

(1minusα)kRO2+NO[RO2][NO] + (1minus c)PHOxkHO2+NO[NO] + 2kHO2+HO2 [HO2] + kHO2+RO2 [RO2]

(A4)

[HO2] = [RO2] (A5)

For the calculations in Fig 2 the values of VOCR α andPHOx were fixed at 18 sminus1 006 and 115times107 moleccmminus3

sminus1 Rate constants are taken from the IUPAC chemi-cal kinetics database assuming that all RO2 radicals reactwith the kinetics of CH3CH2O2 (Atkinson et al 2006)The system of equations was solved numerically using thevpasolve function in MATLAB subject to the constraintsthat [OH][HO2] and [RO2] are positive and c is between 0and 1

The resulting concentrations of HOx radicals can be usedto calculate the rates of ozone production and NOx loss usingEqs (A6)ndash(A9)

PO3 = (1minusα)kRO2+NO[RO2][NO] + kHO2+NO[HO2][NO] (A6)

PHNO3 = kOH+NO2 [OH][NO2] (A7)P6RONO2 = α middot kRO2+NO[RO2][NO] (A8)LNOx = PHNO3+ η middotP6RONO2 (A9)

A2 Decomposition of the O3-temperature relationship

The simplified HOx model described above was used to de-compose the contribution of different parameters to the in-crease of

intPO3 with temperature Peroxy nitrates are not in-

cluded in this model but because there is no significant trendin 6PNs with temperature their absence does not affect theresults To validate that this model gave accurate

intPO3 re-

sults it was first run using inputs based on measured valuesfor each half-hour period

ndash Model inputs of NOx were taken directly from measure-ments of NO and NO2

ndash VOCR was calculated as the measured OHR minus thereactivity of species that do not form RO2 radicals (egCO NO2)

ndash PHOx was calculated as equal to the measured rate ofHOx loss using Eq (A1) and measured HOx radicalconcentrations

ndash αeff was calculated as the reactivity-weighted average ofαi for all measured VOCs

The comparison ofintPO3 calculated from the full data

set and that from the steady-state HOx model is shownin Fig S6a The two calculations are well-correlated witha slope close to one showing that the steady-state HOxmodel can accurately reproduce ozone production at this lo-cation

To use this model to explore how ozone productionchanges with temperature the diurnal cycle and trend in tem-perature of each of these inputs was calculated Because theresponse to temperature is different at different times of daythe trend with temperature was calculated independently foreach half-hour bin and is shown in Fig S5 These trendswere used to construct temperature-dependent diurnal cyclesof each of the parameters which were then used as inputs tothe model at a range of daily average afternoon temperaturesfrom 24 to 32 C Figure S6b shows that

intPO3 calculated

this way has a very similar trend with temperature as that us-ing the full data set although it cannot capture day-to-dayvariability not caused by temperature The nonlinear shapeof the trend with temperature is caused primarily by the im-posed exponential increase of PHOx with temperature Us-ing a linear or quadratic increase of PHOx with temperaturechanges the shape of the increase but does not significantlyaffect the overall

intPO3-T slope



Competing interests The authors declare that they have no conflictof interest

Supplement The supplement related to this article is availableonline at httpsdoiorg105194acp-18-2601-2018-supplement

Acknowledgements Financial and logistical support for SOASwas provided by the NSF the Earth Observing Laboratory at theNational Center for Atmospheric Research (operated by NSF) thepersonnel at Atmospheric Research and Analysis and the ElectricPower Research Institute We are grateful to Kevin Olson andLi Zhang for assistance with measurements during SOAS Fundingfor the SEARCH network was provided by Southern CompanyServices and the Electric Power Research Institute The Berkeleyauthors acknowledge the support of the NOAA Office of GlobalPrograms grant NA13OAR4310067 and NSF grant AGS-1352972

Edited by Andreas HofzumahausReviewed by two anonymous referees

References

Atkinson R Baulch D L Cox R A Crowley J N Hamp-son R F Hynes R G Jenkin M E Rossi M J Troe Jand IUPAC Subcommittee Evaluated kinetic and photochemi-cal data for atmospheric chemistry Volume II ndash gas phase re-actions of organic species Atmos Chem Phys 6 3625ndash4055httpsdoiorg105194acp-6-3625-2006 2006

Atkinson R W Yu D Armstrong B G Pattenden S Wilkin-son P Doherty R M Heal M R and Anderson H RConcentrationndashResponse Function for Ozone and Daily Mortal-ity Results from Five Urban and Five Rural UK PopulationsEnviron Health Persp 120 1411ndash1417 2012

Ayres B R Allen H M Draper D C Brown S S Wild RJ Jimenez J L Day D A Campuzano-Jost P Hu W deGouw J Koss A Cohen R C Duffey K C Romer P Bau-mann K Edgerton E Takahama S Thornton J A Lee BH Lopez-Hilfiker F D Mohr C Wennberg P O NguyenT B Teng A Goldstein A H Olson K and Fry J LOrganic nitrate aerosol formation via NO3+ biogenic volatileorganic compounds in the southeastern United States AtmosChem Phys 15 13377ndash13392 httpsdoiorg105194acp-15-13377-2015 2015

Barnes E A and Fiore A M Surface ozone variability and the jetposition Implications for projecting future air quality GeophysRes Lett 40 2839ndash2844 httpsdoiorg101002grl504112013

Baumann K Williams E J Angevine W M Roberts J MNorton R B Frost G J Fehsenfeld F C Springston S RBertman S B and Hartsell B Ozone production and trans-port near Nashville Tennessee Results from the 1994 studyat New Hendersonville J Geophys Res 105 9137ndash9153httpsdoiorg1010291999JD901017 2000

Berlin S R Langford A O Estes M Dong M and Par-rish D D Magnitude Decadal Changes and Impact of Re-gional Background Ozone Transported into the Greater Hous-

ton Texas Area Environ Sci Technol 47 13985ndash13992httpsdoiorg101021es4037644 2013

Booker F Muntifering R McGrath M Burkey K De-coteau D Fiscus E Manning W Krupa S Chappelka Aand Grantz D The Ozone Component of Global Change Po-tential Effects on Agricultural and Horticultural Plant YieldProduct Quality and Interactions with Invasive Species J In-tegr Plant Biol 51 337ndash351 httpsdoiorg101111j1744-7909200800805x 2009

Browne E C Min K-E Wooldridge P J Apel E Blake DR Brune W H Cantrell C A Cubison M J Diskin G SJimenez J L Weinheimer A J Wennberg P O Wisthaler Aand Cohen R C Observations of total RONO2 over the borealforest NOx sinks and HNO3 sources Atmos Chem Phys 134543ndash4562 httpsdoiorg105194acp-13-4543-2013 2013

Cooper O R Gao R-S Tarasick D Leblanc T andSweeney C Long-term ozone trends at rural ozone monitor-ing sites across the United States 1990ndash2010 J Geophys ResAtmos 117 D22307 httpsdoiorg1010292012JD0182612012

Cooper O R Parrish D D Ziemke J Balashov N V Cu-peiro M Galbally I E Gilge S Horowitz L Jensen N RLamarque J-F Naik V Oltmans S J Schwab J Shin-dell D T Thompson A M Thouret V Wang Y andZbinden R M Global distribution and trends of tropo-spheric ozone An observation-based review Elem Sci Anth2 000029 httpsdoiorg1012952journalelementa0000292014

Davidson E A Potter C S Schlesinger P andKlooster S A Model estimates of regional nitric oxideemissions from soils of the Southeastern United StatesEcol Appl 8 748ndash759 httpsdoiorg1018901051-0761(1998)008[0748MEORNO]20CO2 1998

Day D A Wooldridge P J Dillon M B Thornton J Aand Cohen R C A thermal dissociation laser-induced fluo-rescence instrument for in situ detection of NO2 peroxy ni-trates alkyl nitrates and HNO3 J Geophys Res 107 4046httpsdoiorg1010292001JD000779 2002

Durre I and Yin X Enhanced radiosonde data for studiesof vertical structure B Am Meteorol Soc 89 1257ndash1262httpsdoiorg1011752008BAMS26031 2008

Faloona I C Tan D Lesher R L Hazen N L Frame C LSimpas J B Harder H Martinez M Di Carlo PRen X and Brune W H A Laser-induced FluorescenceInstrument for Detecting Tropospheric OH and HO2 Char-acteristics and Calibration J Atmos Chem 47 139ndash167httpsdoiorg101023BJOCH0000021036531850e 2004

Farmer D K Perring A E Wooldridge P J Blake D R BakerA Meinardi S Huey L G Tanner D Vargas O and CohenR C Impact of organic nitrates on urban ozone production At-mos Chem Phys 11 4085ndash4094 httpsdoiorg105194acp-11-4085-2011 2011

Feiner P A Brune W H Miller D O Zhang L Cohen R CRomer P S Goldstein A H Keutsch F N Skog K MWennberg P O Nguyen T B Teng A P DeGouw JKoss A Wild R J Brown S S Guenther A Edger-ton E Baumann K and Fry J L Testing Atmospheric Ox-idation in an Alabama Forest J Atmos Sci 73 4699ndash4710httpsdoiorg101175JAS-D-16-00441 2016



Frost G J Trainer M Allwine G Buhr M P Calvert J GCantrell C A Fehsenfeld F C Goldan P D Herwehe JHuumlbler G Kuster W C Martin R McMillen R TMontzka S A Norton R B Parrish D D Ridley B AShetter R E Walega J G Watkins B A Westberg H Hand Williams E J Photochemical ozone production in the ruralsoutheastern United States during the 1990 Rural Oxidants in theSouthern Environment (ROSE) program J Geophys Res 10322491ndash22508 httpsdoiorg10102998JD00881 1998

Fu T-M Zheng Y Paulot F Mao J and Yantosca R M Pos-itive but variable sensitivity of August surface ozone to large-scale warming in the southeast United States Nat Clim Change5 454ndash458 httpsdoiorg101038nclimate2567 2015

Gilman J B Burkhart J F Lerner B M Williams E J KusterW C Goldan P D Murphy P C Warneke C Fowler CMontzka S A Miller B R Miller L Oltmans S J RyersonT B Cooper O R Stohl A and de Gouw J A Ozone vari-ability and halogen oxidation within the Arctic and sub-Arcticspringtime boundary layer Atmos Chem Phys 10 10223ndash10236 httpsdoiorg105194acp-10-10223-2010 2010

Hansen D A Edgerton E S Hartsell B E Jansen J JKandasamy N Hidy G M and Blanchard C L TheSoutheastern Aerosol Research and Characterization StudyPart 1 ndash Overview J Air Waste Manage 53 1460ndash1471httpsdoiorg10108010473289200310466318 2003

Heagle A S Ozone and Crop Yield Annu Rev Phytopathol 27397ndash423 httpsdoiorg101146annurevpy270901890021451989

Hickman J E Wu S Mickley L J and Lerdau M T Kudzu(Pueraria montana) invasion doubles emissions of nitric oxideand increases ozone pollution Proc Natl Acad Sci USA 10710115ndash10119 httpsdoiorg101073pnas0912279107 2010

Hidy G M Blanchard C L Baumann K Edgerton E Tanen-baum S Shaw S Knipping E Tombach I Jansen J andWalters J Chemical climatology of the southeastern UnitedStates 1999ndash2013 Atmos Chem Phys 14 11893ndash11914httpsdoiorg105194acp-14-11893-2014 2014

Hirsch A I Munger J W Jacob D J Horowitz L Wand Goldstein A H Seasonal variation of the ozoneproduction efficiency per unit NOx at Harvard For-est Massachusetts J Geophys Res 101 12659ndash12666httpsdoiorg10102996JD00557 1996

Hu K S Darer A I and Elrod M J Thermodynamicsand kinetics of the hydrolysis of atmospherically relevantorganonitrates and organosulfates Atmos Chem Phys 118307ndash8320 httpsdoiorg105194acp-11-8307-2011 2011

Ito K De Leon S F and Lippmann M Associa-tions Between Ozone and Daily Mortality Analy-sis and Meta-Analysis Epidemiology 16 446ndash457httpsdoiorg10109701ede0000165821901147f 2005

Jacob D J and Winner D A Effect of climatechange on air quality Atmos Environ 43 51ndash63httpsdoiorg101016jatmosenv200809051 2009

Jacob D J Logan J A Gardner G M Yevich R M Spi-vakovsky C M Wofsy S C Sillman S and Prather M JFactors regulating ozone over the United States and its exportto the global atmosphere J Geophys Res 98 14817ndash14826httpsdoiorg10102998JD01224 1993

Jenkin M E Young J C and Rickard A R The MCMv331 degradation scheme for isoprene Atmos Chem Phys15 11433ndash11459 httpsdoiorg105194acp-15-11433-20152015

Kaser L Karl T Schnitzhofer R Graus M Herdlinger-BlattI S DiGangi J P Sive B Turnipseed A Hornbrook RS Zheng W Flocke F M Guenther A Keutsch F NApel E and Hansel A Comparison of different real timeVOC measurement techniques in a ponderosa pine forest At-mos Chem Phys 13 2893ndash2906 httpsdoiorg105194acp-13-2893-2013 2013

Kesik M Bruumlggemann N Forkel R Kiese R Knoche RLi C Seufert G Simpson D and Butterbach-Bahl KFuture scenarios of N2O and NO emissions from Eu-ropean forest soils J Geophys Res 111 G02018httpsdoiorg1010292005JG000115 2006

LaFranchi B W Wolfe G M Thornton J A Harrold S ABrowne E C Min K E Wooldridge P J Gilman J BKuster W C Goldan P D de Gouw J A McKay M Gold-stein A H Ren X Mao J and Cohen R C Closing theperoxy acetyl nitrate budget observations of acyl peroxy nitrates(PAN PPN and MPAN) during BEARPEX 2007 Atmos ChemPhys 9 7623ndash7641 httpsdoiorg105194acp-9-7623-20092009

Liu S and Liang X-Z Observed Diurnal Cycle Climatology ofPlanetary Boundary Layer Height J Climate 23 5790ndash5809httpsdoiorg1011752010JCLI35521 2010

Mao J Ren X Brune W H Olson J R Crawford J H FriedA Huey L G Cohen R C Heikes B Singh H B BlakeD R Sachse G W Diskin G S Hall S R and Shetter RE Airborne measurement of OH reactivity during INTEX-BAtmos Chem Phys 9 163ndash173 httpsdoiorg105194acp-9-163-2009 2009

Mickley L J Jacob D J and Rind D Uncertainty in prein-dustrial abundance of tropospheric ozone Implications for ra-diative forcing calculations J Geophys Res 106 3389ndash3399httpsdoiorg1010292000JD900594 2001

Min K-E Pusede S E Browne E C LaFranchi B W andCohen R C Eddy covariance fluxes and vertical concentra-tion gradient measurements of NO and NO2 over a ponderosapine ecosystem observational evidence for within-canopy chem-ical removal of NOx Atmos Chem Phys 14 5495ndash5512httpsdoiorg105194acp-14-5495-2014 2014

Myhre G D Shindell D T Breacuteon F-M Collins W Fu-glestvedt J Huang J Koch D Lamarque J-F Lee D Men-doza B Nakajima T Robock A Stephens G Takemura Tand Zhang H Anthropogenic and natural radiative forcing inClimate Change 2013 The Physical Science Basis Contributionof Working Group I to the Fifth Assessment Report of the Inter-governmental Panel on Climate Change edited by Stocker T FQin D Plattner G-K Tignor M Allen S K Boschung JNauels A Xia Y Bex V and Midgley P M 659ndash740 Cam-bridge University Press Cambridge UK 2013

Oikawa P Y Ge C Wang J Eberwein J R Liang L LAllsman L A Grantz D A and Jenerette G D Unusu-ally high soil nitrogen oxide emissions influence air quality ina high-temperature agricultural region Nat Commun 6 8753httpsdoiorg101038ncomms9753 2015



Oswald R Behrendt T Ermel M Wu D Su H Cheng YBreuninger C Moravek A Mougin E Delon C Lou-bet B Pommerening-Roumlser A Soumlrgel M Poumlschl U Hoff-mann T Andreae M O Meixner F X and Trebs IHONO Emissions from Soil Bacteria as a Major Sourceof Atmospheric Reactive Nitrogen Science 341 1233ndash1235httpsdoiorg101126science1242266 2013

Peeters J Muumlller J-F Stavrakou T and Nguyen V SHydroxyl Radical Recycling in Isoprene Oxidation Drivenby Hydrogen Bonding and Hydrogen Tunneling The Up-graded LIM1 Mechanism J Phys Chem A 118 8625ndash8643httpsdoiorg101021jp5033146 2014

Perring A E Pusede S E and Cohen R C An ObservationalPerspective on the Atmospheric Impacts of Alkyl and Multifunc-tional Nitrates on Ozone and Secondary Organic Aerosol ChemRev 113 5848ndash5870 httpsdoiorg101021cr300520x 2013

Pilegaard K Processes regulating nitric oxide emissionsfrom soils Philos T Roy Soc B 368 20130126httpsdoiorg101098rstb20130126 2013

Pilegaard K Skiba U Ambus P Beier C Bruumlggemann NButterbach-Bahl K Dick J Dorsey J Duyzer J GallagherM Gasche R Horvath L Kitzler B Leip A Pihlatie MK Rosenkranz P Seufert G Vesala T Westrate H andZechmeister-Boltenstern S Factors controlling regional differ-ences in forest soil emission of nitrogen oxides (NO and N2O)Biogeosciences 3 651ndash661 httpsdoiorg105194bg-3-651-2006 2006

Pleijel H Danielsson H Ojanperauml K De Temmerman LHoumlgy P Badiani M and Karlsson P E Relation-ships between ozone exposure and yield loss in Europeanwheat and potatomdasha comparison of concentration- and flux-based exposure indices Atmos Environ 38 2259ndash2269httpsdoiorg101016jatmosenv200309076 2004

Pusede S E Gentner D R Wooldridge P J Browne E CRollins A W Min K-E Russell A R Thomas J ZhangL Brune W H Henry S B DiGangi J P Keutsch FN Harrold S A Thornton J A Beaver M R St ClairJ M Wennberg P O Sanders J Ren X VandenBoer TC Markovic M Z Guha A Weber R Goldstein A Hand Cohen R C On the temperature dependence of organicreactivity nitrogen oxides ozone production and the impactof emission controls in San Joaquin Valley California AtmosChem Phys 14 3373ndash3395 httpsdoiorg105194acp-14-3373-2014 2014

Pusede S E Steiner A L and Cohen R C Temper-ature and Recent Trends in the Chemistry of Conti-nental Surface Ozone Chem Rev 115 3898ndash3918httpsdoiorg101021cr5006815 2015

Romer P S Duffey K C Wooldridge P J Allen H M AyresB R Brown S S Brune W H Crounse J D de GouwJ Draper D C Feiner P A Fry J L Goldstein A HKoss A Misztal P K Nguyen T B Olson K Teng A PWennberg P O Wild R J Zhang L and Cohen R C Thelifetime of nitrogen oxides in an isoprene-dominated forest At-mos Chem Phys 16 7623ndash7637 httpsdoiorg105194acp-16-7623-2016 2016

Ryerson T B Williams E J and Fehsenfeld F CAn efficient photolysis system for fast-response NO2 mea-

surements J Geophys Res Atmos 105 26447ndash26461httpsdoiorg1010292000JD900389 2000

Shen L Mickley L J and Gilleland E Impact of increasing heatwaves on US ozone episodes in the 2050s Results from a multi-model analysis using extreme value theory Geophys Res Lett43 4017ndash4025 httpsdoiorg1010022016GL068432 2016

Sillman S and Samson P J Impact of temperature onoxidant photochemistry in urban polluted rural and re-mote environments J Geophys Res 100 11497ndash11508httpsdoiorg10102994JD02146 1995

Singh R B and Sloan J J A high-resolution NOx emission fac-tor model for North American motor vehicles Atmos Environ40 5214ndash5223 httpsdoiorg101016jatmosenv2006040122006

SOAS Science Team SOAS 2013 Centreville Site Data NOAAhttpsesrlnoaagovcsdgroupscsd7measurements2013senexGroundDataDownload (last access 17 June 2017) 2013

Steiner A L Tonse S Cohen R C Goldstein A H andHarley R A Influence of future climate and emissions on re-gional air quality in California J Geophys Res 111 D18303httpsdoiorg1010292005JD006935 2006

Steiner A L Davis A J Sillman S Owen R C Micha-lak A M and Fiore A M Observed suppression of ozoneformation at extremely high temperatures due to chemical andbiophysical feedbacks Proc Natl Acad Sci USA 107 19685ndash19690 httpsdoiorg101073pnas1008336107 2010

Thornton F C Pier P A and Valente R J NO emissions fromsoils in the southeastern United States J Geophys Res 10221189ndash21195 httpsdoiorg10102997JD01567 1997

Trainer M Parrish D D Buhr M P Norton R B Fehsen-feld F C Anlauf K G Bottenheim J W Tang Y ZWiebe H A Roberts J M Tanner R L Newman L Bow-ersox V C Meagher J F Olszyna K J Rodgers M OWang T Berresheim H Demerjian K L and Roy-chowdhury U K Correlation of ozone with NOy in pho-tochemically aged air J Geophys Res 98 2917ndash2925httpsdoiorg10102992JD01910 1993

Travis K R Jacob D J Fisher J A Kim P S Marais E AZhu L Yu K Miller C C Yantosca R M Sulprizio MP Thompson A M Wennberg P O Crounse J D St ClairJ M Cohen R C Laughner J L Dibb J E Hall S RUllmann K Wolfe G M Pollack I B Peischl J Neuman JA and Zhou X Why do models overestimate surface ozone inthe Southeast United States Atmos Chem Phys 16 13561ndash13577 httpsdoiorg105194acp-16-13561-2016 2016

Trebs I Bohn B Ammann C Rummel U Blumthaler MKoumlnigstedt R Meixner F X Fan S and Andreae MO Relationship between the NO2 photolysis frequency andthe solar global irradiance Atmos Meas Tech 2 725ndash739httpsdoiorg105194amt-2-725-2009 2009

VandenBoer T C Brown S S Murphy J G Keene W CYoung C J Pszenny A A P Kim S Warneke Cde Gouw J A Maben J R Wagner N L Riedel T P Thorn-ton J A Wolfe D E Dubeacute W P Oumlztuumlrk F Brock C AGrossberg N Lefer B Lerner B Middlebrook A M andRoberts J M Understanding the role of the ground surface inHONO vertical structure High resolution vertical profiles dur-ing NACHTT-11 J Geophys Res Atmos 118 10155ndash10171httpsdoiorg101002jgrd50721 2013



Vedal S Brauer M White R and Petkau J Air Pollution andDaily Mortality in a City with Low Levels of Pollution EnvironHealth Persp 111 45ndash51 httpsdoiorg1014288102207332003

Weaver C P Liang X-Z Zhu J Adams P J Amar P AviseJ Caughey M Chen J Cohen R C Cooter E Dawson JP Gilliam R Gilliland A Goldstein A H Grambsch AGrano D Guenther A Gustafson W I Harley R A HeS Hemming B Hogrefe C Huang H-C Hunt S W Ja-cob D Kinney P L Kunkel K Lamarque J-F Lamb BLarkin N K Leung L R Liao K-J Lin J-T Lynn B HManomaiphiboon K Mass C McKenzie D Mickley L JOrsquoNeill S M Nolte C Pandis S N Racherla P N Rosen-zweig C Russell A G Salatheacute E Steiner A L TagarisE Tao Z Tonse S Wiedinmyer C Williams A WinnerD A Woo J-H Wu S and Wuebbles D J A PreliminarySynthesis of Modeled Climate Change Impacts on US RegionalOzone Concentrations B Am Meteorol Soc 90 1843ndash1863httpsdoiorg1011752009BAMS25681 2009

Williams E J and Fehsenfeld F C Measurement of soil nitrogenoxide emissions at three North American ecosystems J Geo-phys Res 96 1033ndash1042 httpsdoiorg10102990JD019031991

Wolfe G M Marvin M R Roberts S J Travis K R and LiaoJ The Framework for 0-D Atmospheric Modeling (F0AM) v31Geosci Model Dev 9 3309ndash3319 httpsdoiorg105194gmd-9-3309-2016 2016

World Health Organization Air Quality Guidelines Global Update2005 Particulate Matter Ozone Nitrogen Dioxide and SulphurDioxide WHO Regional Office for Europe Copenhagen Den-mark 2005


Abstract

Introduction

Chemistry of ozone production and predicted response to temperature

Observed response of ozone production to temperature

Measurements during SOAS

Calculation of PO3 and effects of temperature

Drivers of increased ozone production

Conclusions

Data availability

Appendix A

Appendix A1 Analytical PO3 model

Appendix A2 Decomposition of the O3-temperature relationship

Competing interests

Supplement

Acknowledgements

References


this effect is nonlinear and can become significantly lessstrong at the highest temperatures (Steiner et al 2010 Shenet al 2016)

Increased ozone concentrations with temperature in ur-ban areas can be well explained by increased ozone pro-duction caused by greater emissions of volatile organiccompounds (VOCs) and decreased sequestration of NOx inshort-term reservoirs (Jacob and Winner 2009) In contrastthere is little consensus on the mechanisms responsible fortemperature-dependent changes in ozone concentrations inrural and remote environments Arguments in favor of large-scale changes in atmospheric circulation and in favor of lo-cal changes in the chemical production and loss of ozonehave both been presented (Barnes and Fiore 2013 Steineret al 2006) Regional stagnation episodes often associatedwith elevated temperatures allow ozone to accumulate overseveral days and are known to contribute significantly tothe ozone-temperature relationship (Jacob et al 1993) Howvarious temperature-dependent chemical effects interact andtheir relative contributions to ozone production are not wellunderstood outside of polluted environments

Summer daytime ozone concentrations at rural sites inthe United States typically range from 35 to 55 ppb (Cooperet al 2012) sufficient to cause harm to humans crops andthe climate Epidemiological studies and meta-analyses in-vestigating the relationship between ozone and daily mortal-ity have found significant effects in small cities and rural lo-cations with some studies suggesting that increases in ozonemay have a greater effect on daily mortality under less pol-luted conditions (Vedal et al 2003 Ito et al 2005 Atkinsonet al 2012) Studies of crop yield and plant health have tra-ditionally used a threshold of 40 ppb when investigating theeffects of ozone exposure but many crops have been shownto experience reduced yields when exposed to ozone concen-trations as low as 20 ppb (Pleijel et al 2004 Booker et al2009) From a regulatory perspective elevated regional back-ground ozone can strongly exacerbate ozone pollution andthe probability of regulatory exceedances in urban areas suchas Houston (Berlin et al 2013) Understanding the behaviorof O3 in the rural and remote areas that cover the majority ofthe land area of the Earth is therefore crucial for effectivelypredicting and controlling air quality now and in the future

In this paper we use observations from Centreville Al-abama (CTR) a rural site in the southeastern United States(Supplement Fig S1) to investigate how temperature affectsozone production Long-term monitoring from the South-Eastern Aerosol Research and CHaracterization (SEARCH)network shows that ozone increases significantly with tem-perature at this site (Fig 1) despite being in a low-NOx en-vironment where the predicted response of the instantaneousozone production rate to temperature is small (Pusede et al2015) We combine this record with extensive measurementsfrom the Southern Oxidant and Aerosol Study (SOAS) insummer 2013 to explicitly calculate daily integrated ozoneproduction and NOx loss as a function of daily average tem-

Figure 1 The O3ndashtemperature relationship in Centreville Al-abama Daily afternoon (1200ndash1600) average ozone concentrationis shown as a function of temperature over JunendashAugust 2010ndash2014at the SEARCH CTR site The black line and shaded gray regionshow the running median and interquartile range of ozone with tem-perature respectively The red line represents a fit to all daily datapoints

perature We find that changes in local chemistry are impor-tant drivers of the increase in ozone concentrations observedat this site and that increased NOx emissions are responsi-ble for 40 of the temperature-dependent increase in dailyintegrated ozone production We expect similar effects to bepresent in other low-NOx areas with high concentrations ofVOCs where the chemistry of alkyl and multifunctional ni-trates is the majority pathway for permanent NOx loss

2 Chemistry of ozone production and predictedresponse to temperature

Observed O3ndashT relationships are caused by a combinationof chemical changes to the production and loss of O3 andchanges to atmospheric circulation that determine advectionand mixing To begin separating these effects we considerthe chemical production of ozone (PO3) and how it changeswith temperature Temperature-dependent changes in ozoneproduction may be driven directly by temperature or by an-other meteorological parameter that co-varies with tempera-ture such as solar radiation

Ozone is produced in the troposphere when NO is con-verted to NO2 by reaction with HO2 or RO2 in the linkedHOx and NOx cycles (Fig 2a) HO2 and RO2 radicals aregenerated in the HOx cycle when a VOC reacts with OH inthe presence of NOx In one turn of the cycle the VOC isoxidized OH is regenerated and two molecules of O3 areformed The reactions that drive these catalytic cycles for-ward are in constant competition with reactions that removeradicals from the atmosphere terminating the cycles Termi-nation can occur either through the association of two HOx












intPO3) is a more
























intLO3 (b) andint












intLO3) Chem-






intLNOx and

intPO3 are






intPO3 toint


intPO3












































intPO3

5 Conclusions
















Appendix A










[RO2] =


(A3)

[HO2] =


(A4)

[HO2] = [RO2] (A5)










intPO3 re-









intPO3 calculated


intPO3-T slope







References










































































Abstract

Introduction






Conclusions

Data availability

Appendix A



Competing interests

Supplement

Acknowledgements

References











intPO3) is a more
























intLO3 (b) andint












intLO3) Chem-






intLNOx and

intPO3 are






intPO3 toint


intPO3












































intPO3

5 Conclusions
















Appendix A










[RO2] =


(A3)

[HO2] =


(A4)

[HO2] = [RO2] (A5)










intPO3 re-









intPO3 calculated


intPO3-T slope







References










































































Abstract

Introduction






Conclusions

Data availability

Appendix A



Competing interests

Supplement

Acknowledgements

References











intLO3 (b) andint












intLO3) Chem-






intLNOx and

intPO3 are






intPO3 toint


intPO3












































intPO3

5 Conclusions
















Appendix A










[RO2] =


(A3)

[HO2] =


(A4)

[HO2] = [RO2] (A5)










intPO3 re-









intPO3 calculated


intPO3-T slope







References










































































Abstract

Introduction






Conclusions

Data availability

Appendix A



Competing interests

Supplement

Acknowledgements

References




intLO3) Chem-






intLNOx and

intPO3 are






intPO3 toint


intPO3












































intPO3

5 Conclusions
















Appendix A










[RO2] =


(A3)

[HO2] =


(A4)

[HO2] = [RO2] (A5)










intPO3 re-









intPO3 calculated


intPO3-T slope







References










































































Abstract

Introduction






Conclusions

Data availability

Appendix A



Competing interests

Supplement

Acknowledgements

References
































intPO3

5 Conclusions
















Appendix A










[RO2] =


(A3)

[HO2] =


(A4)

[HO2] = [RO2] (A5)










intPO3 re-









intPO3 calculated


intPO3-T slope







References










































































Abstract

Introduction






Conclusions

Data availability

Appendix A



Competing interests

Supplement

Acknowledgements

References












Appendix A










[RO2] =


(A3)

[HO2] =


(A4)

[HO2] = [RO2] (A5)










intPO3 re-









intPO3 calculated


intPO3-T slope







References










































































Abstract

Introduction






Conclusions

Data availability

Appendix A



Competing interests

Supplement

Acknowledgements

References






References










































































Abstract

Introduction






Conclusions

Data availability

Appendix A



Competing interests

Supplement

Acknowledgements

References
























































Abstract

Introduction






Conclusions

Data availability

Appendix A



Competing interests

Supplement

Acknowledgements

References

effects of temperature-dependent no emissions on continental … · 2020. 7. 31. ·...

Documents