GEOPHYSICAL RESEARCH LETTERS, VOL. 28, NO.

9, PAGES 1719-1722, MAY 1, 2001

Indirect long-term global radiative cooling

from emissions
Oliver Wild
Frontier Research System for Global Change, Yokohama, 3apan.

Michael J. Prather
Earth System Science,University of California, Irvine

Hjime Akimoto
Frontier Research System for Global Change, Yokohama, 3apan.

Abstract. Anthropogenic emissions of short-lived, chem- emissions have a negative integrated RF. In this case long-
ically reactive gases, such as N Ox and CO, are known to term cooling due to decreases in CH4 and O3 dominates
influence climate by altering the chemistry of the global tro- short-term warming from regional O3 increases.
posphere and thereby the abundance of the greenhousegases
03, CH4 and the HFCs. This study uses the characteristics
of the natural modes of the tropospheric chemical system to 2. Excitation of the Primary Mode
decomposethe greenhouseeffect of NOx and CO emissions
The coupling of chemical processeswith atmospheric
into (i) short-livedmodesinvolvingpredominantlytropo- transport leads to the existence of natural modes in the
spheric03 and (ii) the long-livedmode involvinga global tropospheric system which describe how chemical species
coupled CH4-CO-O3 perturbation. Combining these two
co-vary in space and time through the different chemical
classesof greenhouse perturbations--large, short-lived, re-
environments present in the atmosphere. The modes corre-
gional 03 increasesand smaller, long-lived, global decreases
spond to the eigenstates of the chemical system, and each
in CH4 and O3--we find that most types of anthropogenic
has a characteristic time-scale correspondingto the respec-
NO• emissions lead to a negative radiative forcing and an
tive eigenvalueof the system[Prather,1994, 1996]. In pre-
overall cooling of the earth.
viouswork [Wild and Prather, 2000],we demonstratedthe
structure and behaviour of these global chemical modes us-
1. Introduction ing a troposphericChemicalTransportModel (CTM) and
showedhow the primary natural mode, describing the CH4-
The direct radiativeforcing(RF) of the greenhouse
gases CO-OH coupling, could be excited by regional emissionsof
CH4, O3 and the HFCs has been the subject of numerous
short-lived species. This mode, with a decay time about
studies, and has been reviewed in recent assessmentreports 50% longer than the chemical lifetime of CH4, dominates
[Schimelet al., 1996; Granier and Shine, 1999]. However, the long-term radiative impacts of most tropospheric per-
the chemical lifetime of many of these species in the tropo- turbations.
sphere is strongly influenced by short-lived gaseswith little
These studies are performed using the University of Cal-
or no radiative effects in the wavelength range important
for climate. The 'indirect' climate effects of these reactive
ifornia, Irvine (UCI) CTM describedin earlier work [Wild
and Prather, 2000]. The model is run using a singleyear
specieshave been shown to make a significant contribution
of meteorological fieldsfrom the GISS II' GCM [Rind and
to the total RF [Hauglustaineet al., 1994;Johnsonand Der- Lerner, 1996]at 4ø x 5ø resolution.The CTM resolutionis
went, 1996; Fuglestvedt et al., 1996; Daniel and Solomon,
degraded to 8ø x 10ø to reduce computational costs, and
1998], but detailedquantificationof the effectshas proved the fields are recycled for the multiple-year runs needed
difficult due in part to the wide range of temporal and spa-
here. The model includes a detailed tropospheric chem-
tial scales involved.
istry scheme with hydrocarbon oxidation and the effects of
This paper presents a new method of quantifying these
stratospheric chemistry via first-order rates of change above
effects using the natural modes of the tropospheric chemi-
a tracer-diagnosed tropopause. Emissions data for 1985 are
cal system. The technique allows the full, time-integrated
used, and are described in the previous study.
impacts of emissionschangesto be assessed,including the
The CTM is initialized and run for 10 years with fixed
effects of temporal and spatial variations, but without the
emissionsto provide a control. The perturbation runs are
need for decades-longmodel runs. We apply the technique
initialized likewise and run for 10 years but with emissions
to NO• emissionsfrom a variety of sources,and demonstrate
from particular sources increased by a factor for the first
that currently incremental increasesin anthropogenic NO•
year and returning to the standard emissions for years 2
to 10. The differences between perturbation and control
Copyright2001 by theAmericanGeophysical
Union.
runs are used to integrate the short-lived perturbations to
tropospheric O3 and CH4 and to determine the asymptotic
Papernumber2000GL012573. behaviourof the perturbation (i.e., the amplitudeof the pri-
0094-8276/01/2000GL012573505.00 mary mode). Sincethe primary mode has an exact e-fold
1719
 19448007, 2001, 9, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2000GL012573 by Nat Prov Indonesia, Wiley Online Library on [14/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1720 WILD ET AL.' INDIRECT GLOBAL COOLING FROM NOx EMISSIONS
1.0
it is often masked initially by large 03 perturbations present
0.8
in short-lived modes.
0.6

0.4

0.2
4. NOx perturbation studies
0.0 In Figure I we show the evolution of the perturbation
-0.2 in global mass of 03, CH4 and CO for a one-year, 0.5 TgN
-0.4 increment in NOx emissionsfrom global fossil fuel sources.
-0.6
The extra NOx leads to additional 03 formation in the re-

-0.8
gion of emissionsand to increased OH, exciting the primary
mode by increasing oxidation of CH4 and CO. Immediately
-1.0
0 1 2 3 4 5 following the period of additional NO• emissions,the 03 and
Year of Run CO perturbations associatedwith short-lived modes decay
rapidly with e-fold times of 2-3 months, but the CH4 per-
Figure 1. Evolution of the global mass perturbations of 03, turbation decays with an e-fold time of 14.2 years. Two
CO and CH4 due to the additional emission of 0.5 TgN of global years later, the 03 perturbation changessign, as the global
fossil fuel NO for one year.
decrease associated with the primary mode dominates the
short-term regional increases.
The evolution of the integrated RF due to these changes
decay(with annuallyrecyclingmeteorology),the long-term in 03 and CH4 is shown in Figure 2. While the initial RF
integration can be analytically evaluated from the product
driven by 03 formation is positive, long-term cooling due to
of the amplitude and the e-fold time, saving much com-
the primary mode gradually reduces this until, after about
putational effort over a steady-state CTM calculation that
24 years, the integrated RF becomesnegative and tends to
would take several decadesto achieve even 10% accuracy.
a total net of-0.54 mW m-2. The net integratedperturba-
The incremental NOx emissionsexcite the primary mode in
tions and RFs due to NO• emissionsfrom a variety of sources
a negative sense;CH4 decreasesare paralleled by an increase are shown in Table 1. Industrial and fossil fuel sources have
in OH and a decreasein Oa, CO, NO• and peroxy radicals
been subdivided by region to highlight differencesin chemi-
[Wild and Prather, 2000]. After the transient,regionalef- cal and meteorological environment; while all perturbations
fects have decayed, perturbations of all species decay with
follow a pattern similar to that shown in Figure 1, their
the timescale of the primary mode, 14.2 years in this model.
magnitudes differ greatly, dependent on the 03 production
Although we follow the decay of a pulse, the integral of a
efficiency of the added NO• and on the additional OH for-
single year's emissionsis equivalent to the steady state that
mation. For aircraft emissions, predominantly in the up-
would accumulate from continuous emissions.
per troposphere at northern mid-latitudes, the additional
O3 dominates the CH4 removal and leads to net radiative
3. Estimates of Radiative Forcing warming. The effects are proportional to, but slightly lower
The direct RF for CH4 and 03 perturbations in the than, those derived in the recent IPCC Aircraft Assessment
[Penneret al., 1999](-15 mW m-2 for CH4,+24 mW m-2
current atmosphere is based on previous studies: +0.37
et al., 1996]for a con- for 03). For lightning emissions,principally in tropical re-
mW m-2 ppb-z for CH4 [Schimel
gions, the perturbation to OH and hence CH4 is larger per
stant mixing ratio perturbation through the troposphere,
and +42 mW m-2 per DobsonUnit changein columntro- unit NO• emissionand dominates the warming due to 03.
pospheric03 [D. Hauglustaine,personalcommunication, This reduces the likelihood of positive climate feedbacksbe-
2000]. The RF per unit 03 dependson region and sea- tween03 and lightning[Toumi et al., 1996].
son [Fuglestvedtet al., 1996] and particularly on altitude For all surface emissionsstudied here, the primary mode
[Lacis et al., 1990]. While the short-livedperturbations dominates the RF, with a small net cooling in each case.
here are regionally driven, the changes in 03 tend to be
dispersed through the mid-troposphere in mid-latitudes and
hence may be represented by the mean factor above. The .•, 2.5
long-lived, primary-mode perturbations are more globally E 2.0
uniform, and hence this simple approach to scaling the RF
to the change in total tropospheric 03 provides a suitable • 1.5
estimate. We adopt an uncertainty of 20%, based on the ß:- 1.0
range of forcingsfrom a recentmodel intercomparison[Ra- o
LL 0.5
maswamyet al., 2001], to accountfor this effect.
We calculate the time-integrated forcing for the primary .• o.o
mode from the amplitude of excitation and the decay time. n' -0.5
For a 1.0 Tg perturbation in CH4, 99.3% of the oxidation
is accountedfor by the primary mode [Wild and Prather, • -1.0
2000],and hencethe integratedperturbationis 14 Tg-yr of • -1.5
0 10 20 30 40 50
CH4 with an RF of 1.9 mW m-2 yr. The 03 perturbation Year of Run
tied to this excitation of the primary mode is 0.016 Tg-yr of
03 with an integratedRF of about0.87mW m-2 yr, 45%of Figure 2. Evolution of the integrated radiative forcing due
the magnitude of the CH4 RF. This long-lived 03 perturba- to changes in CH4 and 03 caused by the additional emission of
tion occurs whenever the primary mode is excited, although 0.5 TgN of global fossil fuel NO for one year.
 19448007, 2001, 9, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2000GL012573 by Nat Prov Indonesia, Wiley Online Library on [14/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
WILD ET AL' INDIRECT GLOBAL COOLING FROM NOx EMISSIONS 1721

--2
Table 1. Integratedperturbationsto CH4 and tropospheric03 (Tg-yr) and their estimatedradiative forcing(RF: W m yr) due
to a 0.5 TgN increase in emissionsfor one year.

ACH4 AO3 AO3 (Tg-yr)a R.F. (mW m-2 yr)

Perturbation Total Total Short-lived Primary ACH4 AO3 Net

(Tg-yr) (Tg-yr) Modes Mode
Aircraft NO -67.1 3.79 4.91 - 1.11 -9.1 14.6 5.4
Lightning NO -87.8 2.37 3.83 -1.46 -11.9 9.1 -2.8
BiomassBurning NO -24.1 0.53 0.93 -0.40 -3.3 2.0 -1.2
Industrial NO -14.2 0.36 0.59 -0.23 -1.9 1.4 -0.5
Europe -5.2 0.08 0.17 -0.09 -0.7 0.3 -0.4
N.America -8.9 0.20 0.34 -0.15 - 1.2 0.8 -0.4
-- E.Asia - 15.4 0.50 0.76 -0.26 -2.1 1.9 -0.2
Tropics -39.6 1.26 1.92 -0.66 -5.4 4.9 -0.5
Industrial NO and CO b -6.8 0.68 0.79 -0.11 -0.9 2.6 1.7

aThe 03 perturbation is split into contributions from the short-lived modes and the long-lived primary mode.
blncludes
alsoCO emissions
(10 Tg CO for oneyear).

For CH4, the long-term integrated RF per increment in an- In contrast to NO•, emissionsof CO reduce OH concen-
nualNOx emissions
rangesfrom-1.4 to -10.8 mW m-2 per trations and therefore excite the primary mode in a positive
TgN/yr, similarto the -5 mW m-2 foundby Fugtestvedtsense. As NO• and CO have similar combustion sources,
et at., [1996]from step increasesin continuousemissions. we consider the impact of coupled emissionsby pairing the
For 03, the long-term integrated RF ranges from +0.6 to 0.5 TgN/yr increasein globalindustrialNO• sourceswith a
+9.7 mW m-2 per TgN/yr. Previous2-D modelstudies proportionalincreasein CO emissions(10 TgCO/yr). The
have suggested
+3.1 [Fugtestvedt
et at., 1996]and +1.4 to primary mode is still excited in a negative sense but with
+6.4 mW m-2 [Johnson
andDerwent,1996],with strong about half the amplitude of that due to increased NO• emis-
regionalvariations. [Fugtestvedt
et at., 1999]find CH4 and sions alone. This highlights the greater sensitivity of OH to
O3RF of-5.0 and+3.5 rnWm-2 perTgN/yr, respectively,NOx rather than to CO emissions near the surface. Addi-
from US emissions.We find lower but proportional forcings tional O3 formation attributable to CO emissions leads to
of-2.4 and +1.6 mW m -2 for fossil fuel emissionsover about a third more O3 in the short-lived modes, and drives
a similar region. The relationship between the transient the net RF positivefor this coupledemission(shownwith a
responsein O3 associated with short-lived modes and the crossin Figure 3).
long-term perturbation in CH4 associatedwith the primary
mode is shown in Figure 3. With the exception of aircraft 5. Conclusions
emissions(open circle), all scenariosfor N Ox increasesfall
below the zero-forcing line, indicating a net negative inte- The identification of RF caused by emission of short-
grated forcing. lived, chemically reactive gases with both short-lived and
long-lived chemical modes is extremely useful. This sepa-
ration of the problem can greatly reduce the computational
0
effort for tropospheric chemical modeling, enabling a wide
ß Sudace NO range of sensitivity studies such as the examples here. Fur-
[] LightningNO ther, it provides both insight and a clear quantitative sepa-
0 Aircraft NO
ration of RF that is regional(short-livedmodes)from that

•.• xIndustrial
NC+CO which is global (the long-livedCH4-CO-O3 mode). Since
these two different patterns of radiative forcing may not can-
• -lO cel, even with opposite signs, it may become important to
E
diagnose them separately.
,_

Anthropogenic su•rfaceemissionsof NO• produce a net

E -15 RF, due to changes in O3 and CH4, that is consistently
o

, negative but can vary by an order of magnitude depend-

, i
ing on region. Combined industrial emissions of CO and
-20
5 10 15 20 N O•, however, yield a positive RF. Thus decisions to con-
RFfromShort-lived
Modes/mWm-2yr trol global tropospheric 03 and hence greenhousewarming
by cutting NO• emissionsalone would produce the opposite
Figure 3. A scatter plot of the integrated radiative forcing effect when the long-term, global changesto both CH4 and
(RF) from CH4 and 03 in the primary mode against that of 03 are considered.
03 in short-lived modes for model runs with additional one-year
This study has some clear inadequacies. For one, the
emissionsof 0.5 TgN from a variety of sources(see Table 1).
treatment of urban emissions in the CTM does not include
Above the zero-forcing line, the warming effects of short-term 03
increases dominate; below, radiative cooling due to the primary the effects of non-linearities in small-scale plumes where
mode dominates. Error bars of 20ø-/0are indicated for the RF from NO• concentrations may be sufficiently high to suppress 03
03. production or to remove NOx rapidly. It is not clear in these
 19448007, 2001, 9, Downloaded from https://agupubs.onlinelibrary.wiley.com/doi/10.1029/2000GL012573 by Nat Prov Indonesia, Wiley Online Library on [14/09/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
1722 WILD ET AL.: INDIRECT GLOBAL COOLING FROM NOx EMISSIONS

cases that the OH effect is also proportionally suppressed. Lacis, A. A., D. J. Wuebbles, and J. A. Logan, Radiative forcing
Uncertainties such as these can be evaluated by comparing of climate by changesin the vertical distribution of ozone, J.
the global CTM with carefully designed, high-resolution re- Geophys. Res., 95, 9971-9981, 1990.
Penner, J. E., D. H. Lister, D. J. (]riggs, D. J. Dokken, and M.
gional models. Second, the inhomogeneity of the short-term
McFarland (Eds.), Aviation and the GlobalAtmosphere,373
O3 increasesadds uncertainty in estimating its RF that is pp., Cambridge Univ. Press, New York, 1999.
estimated here to be about +20%. Also, we have assumed Prather, M. J., Lifetimes and eigenstates in atmospheric chem-
that RF is additive in terms of climate impact and hence istry, Geophys. Res. Left., 21, 801-804, 1994.
that tropospheric O3 heating cancels CH4 cooling. These Prather, M. J., Time scales in atmospheric chemistry: Theory,
latter assumptionscan only be tested in a climate model. GWPs for CH4 and CO, and runaway growth, Geophys.Res.
Left., 23, 2597-2600, 1996.
Ramaswamy, Vo, et al., Radiative forcing of climate change, in
Acknowledgments. The authors wish to thank the At- Climate Change 2001: The Scientific Basis, Cambridge Uni-
mospheric Chemistry Program of the NSF and the NASA Atmo- versity Press, Cambridge, U.K., 2001.
sphericChemistryModelingand AnalysisProgram(ACMAP) for Rind, D., and J. Lerner, Use of on-line tracers as a diagnostic
their support of this research. tool in general circulation model development: 1. Horizontal
and vertical transport in the troposphere, J. Geophys. Res.,
References 101, 12,667-12,683, 1996.
Schimel, D., et al., Radiative forcing of climate change, in Climate
Daniel, J. S., and S. Solomon, On the climate forcing of carbon Change 1995: The Science of Climate Change, edited by J.
monoxide, J. Geophys. Res., 103, 13,249-13,260, 1998. T. Houghton et al., Cambridge University Press, Cambridge,
Fuglestvedt, J. S., I. S. A. Isaksen, and W.-C. Wang, Estimates U.K., 1996.
of indirect global warming potentials for CH4, CO and NOx, Toumi, R., J. D. Haigh, and K. S. Law, A tropospheric ozone-
Climatic Change, $J, 405-437, 1996. lightning climate feedback, Geophys. Res. Lett., 23, 1037-
Fuglestvedt, J. S., T. K. Berntsen, I. S. A. Isaksen, H. Mao, X.-Z. 1040, 1996.
Liang, and W.-C. Wang, Climatic forcing of nitrogen oxides Wild. O., and M. J. Prather, Excitation of the primary tropo-
through changes in tropospheric ozone and methane; global spheric chemical mode in a global three-dimensional model, J.
3-D model studies, Atmos. Env., 33, 961-977, 1999. Geophys. Res., 105, 24,647-24,660, 2000.
Granier, C., and K. P. Shine, Climate effects of ozone and halo-
carbon changes, in Scientific Assessment of Ozone Depletion: H. Akimoto and O. Wild, Frontier Research System for
1998, WMO Global Ozone Research and Monitoring Project,
Global Change, 3173-25 Showa-machi, Kanazawa-ku, Yokohama,
Report 44, 1999.
Kanagawa236-0001,Japan. (e-maih akimoto@frontier.esto.or.jp;
Hauglustaine, D. A., C. Granier, G. P. Brasseur, and G. Megie,
oliver@frontier.esto.or.j
p)
The importance of atmospheric chemistry in the calculation of
M. J. Prather, Earth System Science,University of California,
radiative forcing on the climate system, J. Geophys. Res., 99,
Irvine, CA 92697. (e-mail: mprather@uci.edu)
1173-1186, 1994.
Johnson, C. E., and R. G. Derwent, Relative radiative forcing con-
sequencesof global emissionsof hydrocarbons, carbon monox-
ide and NOx from human activities estimated with a zonally-
averaged two-dimensional model, Climatic Change, $J, 439- (ReceivedOctober 20, 2000; revisedFebruary 7, 2001;
462, 1996. acceptedFebruary 9, 2001.)