Oct 30, 1992 - M. A. SHIPHAM 2, E. V. BROWELL 2, D. J. JACOB ⢠, AND J. A. LOGAN ⢠... haze layers had been converted to peroxyacetyl nitrate (PAN), representing a potential source ... 3A, and Figure 1 shows the corresponding flight paths and loca- ..... 103. 169. 20. 11. 33 33. 577. 1030. 14(4-5 km). 110. 230. 255. 27.
JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 97, NO. D15, PAGES 16,731-16,746, OCTOBER 30, 1992
AtmosphericChemistryin the Arctic andSubarctic' Influenceof Natural Fires, IndustrialEmissions,and StratosphericInputs S.C. WOFSY• , G. W. SACHSE 2, G. L. GREGORY 2, D. R. BLAKE3, J. D. BRADSHAW'*, S. T. SANDHOLM'*, H. B. SlNGHs, J. A. BARRICK 2, R. C. HARRISS 2'6,R. W. TALBOT2'6, M. A. SHIPHAM 2, E. V. BROWELL 2, D. J. JACOB • , AND J. A. LOGAN• Haze layerswith perturbedconcentrations of tracegases,believedto originatefrom tundraand forestwild fires, were observedover extensiveareasof Alaska and Canadain 1988. Enhancementsof CH4, C2H2, C2H6,
C3H8, and C`*H•0 werelinearly correlatedwith CO in haze layers,with mean ratios(mole hydrocarbon/mole CO) of 0.18 (+ 0.04 (1 o)), 0.0019 (+ 0.0001), 0.0055 (+ 0.0002), 0.0008 (+ 0.0001), and 1.2
x10 4 (_+0.2x104), respectively. Enhancements ofNOywerevariable, averaging 0.0056(+ 0.0030) mole NOy/mole CO,whileperturbations of NOxwereverysmall,usually undetectable. At least1/3of theNOyin the hazelayershadbeenconvertedto peroxyacetylnitrate(PAN), representing a potentialsourceof NOx to the global atmosphere; much of the balancewas oxidizedto nitrate (HNO3 and particulate).The compositionof subArctic haze layers was consistentwith aged emissionsfrom smolderingcombustion,exceptfor CH4, which appearsto be partlybiogenic.Inputsfrom the stratosphere andfrom biomassfires contributedmajor fractions
of theNOyin theremotesub-Arctic troposphere. Analysis of aircraftandground dataindicates relatively little influence frommid-latitude industrial NOyin thisregionduringsummer, possibly excepting transport of PAN. Productionof 03 was inefficient in sub-Arctichaze layers, less than 0.1 03 moleculesper molecule of CO, reflectingthe low NOx/CO emissionratiosfrom smolderingcombustion.Mid-latitudepollutionproducedmuch more03, 0.3 - 0.5 03 moleculespermoleculeof CO, a consequence of higherNOx/CO emissionratios.
1. INTRODUCTION
1988], with only a few investigationsin the borealzone [Cofer et al., 1989].
The Arctic and sub-Arcticregionsof Alaska, Canada,and Greenlandrepresent a vastwilderness with extremelylow levels of humanactivity,oneof thelargestsuchlandareasremaining in the world. Anthropogenic emissions are negligibleovermostof theregion,exceptfor oil operations on theNorthSlope[Blakeet al., thisissue].Atmospheric composition is regulated mainlyby naturalprocesses andby long-rangetransportof pollution.Natural influences includestratosphere-troposphere exchange[Gregory et al., thisissue;Browellet al., thisissue],tundraandforestwild
The presentpaperinvestigates layerswith enhanced concentrationsof tracegasesintercepted by theNASA Electraaircraftover AlaskaduringtheArcticBoundary LayerExpedition (ABLE 3A) ' in July-August1988. The summerof 1988 wasnotablyhot and dry overAlaskaandthehazelayersarebelievedto haveoriginated fromnaturalfiresthatwerewidespread in theregion[Shipham et al., this issue]. Data from the haze layersare examinedto define primaryemissionfactorsfrom borealwild fires and to delincatethe courseof chemicalaging. The chemicalsignatures of
fires,anduptakeof reactive chemical species byvegetation [Jacob thesehazelayersarefoundtoberemarkably consistent withemiset al., thisissue].Anthropogenic pollutants havebeenobserved at sionsfrom smoldering combustion observedin the laboratory particularlyhigh levelsin late winterandspring,during the "Arc- [Lobertet al., 1991], and notablydifferentfrom emissionsfrom
tichaze"period[Rahn,1981; RahnandMcCaffrey,1980;Barrie flamingcombustion. andHoff, 1985;Hansenet al., 1989]. We alsoassess therelativeimportances of naturalandanthropoNaturalfiresoccurthroughout theborealzoneduringsummer,geniesources in regulating tracegasconcentrations oversouthern representing a potentially dominant sourceof hydrocarbons, NO,,, Alaska. Analysisof datafor background air indicates thatinput andparticulates. Mostpreviousstudiesof fireshavefocussed on fromthestratosphere provideda dominantsourcefor O3 [Gregory mid-latitudes or ontropicalburning[e.g.,Hegget al., 1990;Seiler et al., thisissue].We arguethatthestratosphere provideda signi-
andCrutzen, 1980,Andreae et al., 1988;Crutzen et al., 1985; ticantsource forNO• andthatnatural fireswerealsoimportant. Ward and Hardy, 1991;Greenberget al., 1984;Coferet al., Long-range transport of pollutionfrommid-latitudes mayhaveaffectedvertical distributionsof C2-C4 alkanesand CO but could
notbedetected unambiguously forNO• or 03. •Division of Applied Science andDepartment of EarthandPlanetary Science,HarvardUniversity,Cambridge,Massachusetts.
2. SLrMMERTtME HAZ• LAYERS IN TI-m SUB-ARCTIC
2NASALangley Research Center, Hampton, Virginia. 3Department ofChemistry, University ofCalifornia atIrvine. Plate 1 showslidar imagesfrom severalflightsduringABLE `*School of EarthandAtmospheric Science, Georgia Institute ofTechnolo3A, and Figure 1 showsthe corresponding flight pathsandlocagy, Atlanta.
SNASA Ames Research Center, Moffett Field, California.
tions offiresonthedayofFlights 14and20/21.Flight14,on
6Present address: Complex Systems Research Center, University ofNew July26,1988(Plate la), shows anextensive hazelayerbetween 2 Hampshire, Durham.
Copyright1992 by the AmericanGeophysical Union. Papernumber921D0•22. 0148-0227/92/92J-DO0622505.00
and3kinaltitude,asindicated by darkareasin theaerosol image. This layer was samplednear2-km and againat 4-km altitudein a verticalprofile at point 2 (see Figure 2). Weak ozoneenhancement may have been associatedwith the aerosollayer (for example, examinethe lidar datanear Point 1). A fire coveringseveral squarekilometerswas burning about 100 km to the north, and a numberof very large fires were burningto the east;visibility had 16,731
b PM
SPIRAL
ABI.E--3A
FLIGHT
AEROSOL
14-
ABLE
7-26-88
1242
FLIGHT
AEROSOL
DISTRIBUTION
21
8--
4-88
DISTRIBUTION
RELATIVEAEROSOL SCATTERING
P,m•TrV• •oso•. •ooo
SPIRAI•
3A
o
4oo0
L
1250 1256 1•00 1305 LT I i
-
•ooo
4ooo
i
6ooo
i
1634
uooo
I
1640
I,
i
•ooco
I
_
1645 I
_
LT _
_
4
3
-
60.1•
61.11
162.09
61.29
161.91
OZONE
161.21
61.26
161.31
-
{, i61.9•
WLON.
81.85
62.13
N LAT.
161•30
160_90
160_55
w LON
DISTRIBUTION OZONE
OZONE MIX•G 0
1õ
1242
0
61.51
l)I•T!•.IIHJ'I'!ON
RATIO, PPBV
30
4•
60
75
1305
1250 125•
LT
I,
- FTM
16:t4
! B40
t
{,
1645
l/r
&,
-
5
4
3
2
!
-
'-
1
o
60.B3
t, H•..O9
61.11
,
61.•9
I
I
161.91
161.21
61.26
,I ,,, 161.31
({1.11
N lAT.
I 161.94
W ION.
Plate 1. LIDAR imagesof aerosolextinctionat 1 pm (upperpanels)and ozonemixing ratio (lowerpanels)for (a) Flight 14 (July 26, 1988, and (b) Flight 21 (August4, 1988), showinghazelayersin the regionnearBethel,
õl .,'s4
61 .rib
6Z. 13
t
•
t
161.',C7
160.90
160.f,5
N LAT.
W !•)N
Alaska.(•) DataforFlight33(August 17,1988)cover thecoastal transect from Portland,Maine, to WallopsIsland,Virginia.
been reducedby smokeduringthe previousfew days in Bethel 'mettime Arctictroposphere observations relatedto N. Oydis•ibu[Shiphamet al., thisissue].The high aerosolburdensuggests that tion andpartitioning:ABLE 3A, submittedto Journalof Geophya biomassfire was the main source;howeverthe origin cannotbe sical Research,1991) for experimentaldetail) was observed(see uniquelytraced. ElevatedNO, (see S.T. Sandholmet al., Sum- Figure2 andTable 1), indicatingrelativelyrecentemissions,and
WOFSY ET AL.: INFLUENCESON SUBARCTICATMOSPHERICCHEMISTRY
PORTI•ND
-
ABI•--3A
LANGLEY
FLIGHT AEROSOL
•71•
16,733
33
8-17-88
DISTRIBUTION
•T:ROSO],, SCATTERINC
o
4oo0
!
!
.
!
IL
•J..
1029 1030 1040 1050 I ..... I J ,,, I , I•00
1110 , •
1120
LT
-
4-2.47
41.66
41.11
40.78
5
40.19
a9.68
39•0
38.5.q
N LAT.
73.16
73.89
74.64
7•.17
W LON.
I 70.61
70.99
71.70
72:.51
OZONE OZO• o
15
I,..
1009 L,
42.44 !
70.61
,
DISTRIBUTION IdTX•G 3O
L
•TIO,
PPBV
4õ
!
80
L
i,
1020 1030 1040 1050 11•0 I
41.66
L,
I
,!....
75 .._J
1110 1120 I,•
,
LT
I
41.11
40.7fi
40.19
39.68
39.20
38.53
71.69
72:.52
73.16
73.8g
74•64
75.17
N LAT.
{
70.99
Plate 1 (continued)
the town of Bethel (population4000, 30 - 100 km distant)may Back trajectories passedovernumerousandextensivefires idenfihavecontributed. fled in satelliteimages200 - 1000 km to the east and northeast On Flights20 and 21 (August3, 1988) a haze layer was ob- (seeFigures34 and 35 in Shiphamet al. [this issue]. Enhance-
servedbetween3 and 4.Skm altitude(Plate lb and Figure3). mentsof CO andC2 hydrocarbons weresimilaron Flights14 and
16,734
WOFSY ET AL.: INFLUENCESON SUBARCTICATMOSPHERICCHEMISTRY
-
3
_
'--
•
_ Bethel
(a)
i
I
I
I
I
-166
-164
-162
-160
-158
(b)
,
-166
-164
-162
-160
-158
Longitude
Longitude
Fig. 1. Flight tracksfor (a) Flight 14 and 0a) Flight 20/21 near Bethel, Alaska, and (c) Flight 33 alongthe eastcoastof the United States. The numberscorrespondto pointsin the LIDAR imagesin Plate 1. The locationsof spiralsare indicatedby D (first spiral,descending) andU (second spiral,ascending).The tower siteis denotedby X; locationsof major active fires by crosses.The arrow denotesdirectionof motionof air parcels from trajectorycalculationsfor Flights20/21 [Shiphamet al., this issue], for the 300 K level (closeto thehazelayeraltitude).
20/21, but NO was not perturbedand NOy was only slightly
elevated onFlights 20/21.
Carbon monoxide, measured continuously bythedifferential
absorptionCO measurement(DACOM) insmament[Harrisset al., this issue],providesthe most sensitiveindicatorfor combustion, to which otherconcentrations may be ratioedto obtainemission factors. CorrelationsbetweenCO and C2H6, andbetweenCO and C2H2, were remarkablyuniform for the haze layers. Figure 4 showslinear regressionsfor compositedata from Flights 14, 20 and 21 (20 grabsamplesanalyzedfor hydrocarbons [Blakeet al.,
thisissue]),givingr2= 0.97 for both,i.e., lineardependence on
(c)
CO couldaccountfor 97% of the varianceobservedfor C2H2 and
C2H6. Propanewasmorevariablethanacetylene andethane,rela-
tive to CO, but a significant correlation (r2=0.82)wasstillobtainedin the compositedata set (Figure4c) and for individual haze layers(seeTable 1). The uniformityof hydrocarbon/CO ratiosin Flights14 and20/21 arguesstronglyfor a similaroriginfor hazelayersencountered on theseflights. Concentrations
-76
I
I
!
-74
-72
.70
Longitude
of butane were not correlated with CO in the
composite set,howeverconsistent correlations, with similarproportionalitycoefficients, werefoundin individuallayers(Table1). The variableresultslikely reflectthe difficultyin makingmeasurementsat very low concentrations, andatmospheric lossescould alsoplay a role. The lifetimefor C4Hx0is only a few hoursin the
daytime,andlayersmorethana day old mightlosethe signature of primaryemissions. We define emissionfactorsfrom faresby focussingon haze layers with well-definedboundaries,believedto representfare plumes. Primary emissionratios are preservedwithin the haze
WOFSYET AL.' INFLUENCESON SUBARCTICATMOSPHERIC CHEMISTRY
16,735
7OOO
(a) 50
100
150
680
1700
17'20
CO
1740
1760
10
1780
20
CH4
30
40
50 0
200
600
NO
1000
1400
60
80
100
120
NO),
6500
6000
55OO
5O0O
4500
4000
3500 3000
2500
2000
i........................................................ 1500
_
.,..,
',
(b)
o
1 ooo
!,
60
i
,
,I
80
I
i
i
!00
i
120
i
i
140
i
i
720
!
!
780
i
I
i
840
C2H•
i
i
J
i
900
C2H(•
i
i
960
i
i
i
!
1020
60
,
i
80
I
i
100
i
i
120
i
i
!40
i
i .
160
!
180
C.•H•
Fig. 2. Verticalprofilesfor tracegaseson Flight 14, at Point U (spiral2) in Plate la. Data for CO, CH4 and03 represem10-s
averages; dataforNO andNOyarel-rainaverages, anddatafornonmethane hydrocarbons represem grabsamples.
layer as cleanair is entrained.The uniformratiosobtainedfor C2H2andC2H6support thevalidityof thisframework. Table1 summarizes observations of tracegasconcentrations in hazelayersencountered in theBethelregion(Flights14, 20, and
21) and over the Bering Sea (Flight 23). Linear correlations between tracegasesandCO werederivedfromtheslopeof theregression of Ai againstACO, whereA denotes theexcessof i overa background obtained by linearinterpolation betweenaltitudelim-
16,736
WOFSY El' AL.' INFLUENCESON SUBARCTICATMOSPHERICCHEMISTRY
7000
60O0
(
5OOO
•000
20O0
1000
(a) i
0
,
lOO
50
0
,
150
20 1700
,
t
1740
CO
i
i
182o
1780
i ....
186( o
lO
i
,
20
CH4
,l
,
J
ß
,
30
40
50
0
200
NO
i
,
L
1000
20
i
i
i
40
60
•0
NO•
,i
100
O:•
6500
6000
5500
5000
4500
•
4000
E (:D
---"
3500
3000
25O0
20O0
1500
1000
500
,
40
80
120
160
200
,
241600 700
i
,
800
,
!
900
,
,
i
1050
,
J J J
1200 70
90
110 130 150 170 19 10
C.•H•
15
20
25
30
35
40
45
C4H•o
Fig. 3. (a andb) VerticalprofilesonFlight21 at Point3 (descending) in Platelb. (c andd) ProfilesonFlight21 at point2 (ascending)in Plate 1b.
its for thepolluted layerdefinedby theCO enhancement. For of r2 for0 3 wererelatively low,andratiosAO3/ACO werevarieachintercepted layer,resultsfor twoprofileswereaveraged (des- able.
cending andascending spirals). Relationships between CO andnonmethane hydrocarbons were Examples of regressions against ACOaregivenin Figure5for remarkablyconsistent with laboratorydata for smoldering thevertical profileat Pt.D onFlight14 (Platela). Valuesof r2 combustion of biomass material[CrutzenandAndreae,1990; for hydrocarbons andNO• typicallyexceeded 0.7 andin many Coferet al., 1989;Lobertet al., 1991]. In fact,observed ratios cases, werelargerthan0.9. Smallenhancements of NO wereob- fell within10%of laboratory meansfor C2H2/COandC2H6/CO. servedfor Flights14 and23, butnonefor Flights20/21. Values Laboratory datafor flamingcombustion showmorethan3 times
WOFSY ET AL.: INFLUENCESON SUBARCTiCATMOSPHERICCHEMISTRY
16,737
F__.3ooo
2000
(c) !
50
lOO
15o
20 17oo
1740
1780
1820
186
0
10
3O
40
50
0
2O0
NO
CH4
CO
20
600
1000
20
40
60
80
1O0
12C
03
NOy
5000
4500
4000
, ß ß
350O ß
':.
ß ß
25oo
2000 1500
lOOO
;. ,,
500 ß
ß
0
i 40
60
80
100
130
C2H2
160
650
750
850
950
• 1050
i 11!60
80
C2H6
100
120
140
C•H8
160
18( 5
10
15
20
25
30
35
40
45
C4H]o
Fig. 3. (continued)
higher emissionsof C2H2, and 3 times lower of C2H6, relative to very low in the Arctic, and the small yields of NOy from tundra CO [Lobertet al., 1991]. fires cannevertheless representa significantsource.
Arctichazelayerscontained lessNOythanobserved in associa- Enhancements of ozonearesmallin thehazelayers,evennegation with fires in the Amazon or at mid-latitudes(seeTable la), tive in some,reflectingthe low NO,, emissionstypicalof smolderconsistent with a dominantrole for smoldering combustion.Most ing fires [Jacobet al., thisissue]. Significantpositivecorrelation
NOy frombiomassfiresevolvesduringflamingcombustion, by between03 andCO wasobserved onlyin a layerwithdetectable
oxidation of fuelnitrogen [Lobert eta/., 1991].Arcticvegetationenhancement of NOx andrelatively highANO•/ACO(Flight14). is notablylow in nitrogen[Chapinand Shaver,1985]. The vari- The small valuesfor AO3/ACOin Arctic haze layers,about0.1 anceof NO• ratiosto CO suggests variablecontributions from (Table la), may be contrastedwith valuesaveraging0.4 in
smallareasof flamingcombustion. Background levelsof NO• are urban/industrial pollution (Table2 andFigure7).
16,738
WOFSY ET AL.' INFLUENCESON SUBARCTICATMOSPHERICCHEMISTRY 1400
350
1300 300
1200
250
11oo
200
-• 1 ooo
150 900
100
800
5O
700
(b) 0 80
i
I
!
I
I
1O0
120
140
160
1ao
600 200
80
!
I
I
I
I
100
120
140
160
180
200
co (ppb)
CO (ppb)
Fig. 4a. (a) Relationship betweenCO (ppb)andC2H2(ppt)obtained from Fig. 45. Sameasfor Figure4a, for C2H6. Regression hasslope0.0055(+
composited datafor hazelayersfromFlights14 (triangle), 20 (diamond)0.002)mole/mole, r2 = 0.96. and 21 (square). The regressionline shown,with slope0.0021 (+ 0.03)
moleC2H2permoleCO,givesr2=0.97. 200
800
700 180
600
160
500 140
=
400
120 300
100 200
8O
100
60
I 80
1O0
120
140
C0 (ppb)
I
,1
160
180
, 200
0 80
I
I
I
I
I
1O0
120
140
160
180
200
CO (ppb)
Fig. 4c.Sameasfor Figure4•, for C3H8. Regression hasslope0.0010(+ Fig. 4d. Sameasfor Figure4a, for CnH•0. Regression is not statistically 0.002)mole/mole, r2 = 0.82. valid.
WOFSY ET AL.: INFLUENCESON SUBARCTICATMOSPHERICCHEMISTRY
16,739
TABLE la. Enhancement Ratiosin BiomassBurningandPollutionHumes
C2H2
C2H6
C3H8
C4Hlo CH4
03
PAN
14(Bethel) 20(Bethel)
Flight
0.0084 0.0003 0.0036