LETTERSPUBLISHED ONLINE: 21 JUNE 2009 | DOI: 10.1038/NGEO551

Global ammonia distribution derived from infraredsatellite observationsLieven Clarisse1*, Cathy Clerbaux2, Frank Dentener3, Daniel Hurtmans1 and Pierre-François Coheur1

Global ammonia emissions have more than doubled sincepre-industrial times, largely owing to agricultural intensifica-tion and widespread fertilizer use1. In the atmosphere, ammo-nia accelerates particulate matter formation, thereby reducingair quality. When deposited in nitrogen-limited ecosystems,ammonia can act as a fertilizer. This can lead to biodiversityreductions in terrestrial ecosystems, and algal blooms in aque-ous environments2–8. Despite its ecological significance, thereare large uncertainties in the magnitude of ammonia emissions,mainly owing to a paucity of ground-based observations anda virtual absence of atmospheric measurements3,8–11. Here weuse infrared spectra, obtained by the IASI/MetOp satellite,to map global ammonia concentrations from space over thecourse of 2008. We identify several ammonia hotspots inmiddle–low latitudes across the globe. In general, we find agood qualitative agreement between our satellite measure-ments and simulations made using a global atmospheric chem-istry transport model. However, the satellite data reveal sub-stantially higher concentrations of ammonia north of 30◦ N,compared with model projections. We conclude that ammoniaemissions could have been significantly underestimated in theNorthern Hemisphere, and suggest that satellite monitoringof ammonia from space will improve our understanding of theglobal nitrogen cycle.

Atmospheric NH3 is emitted primarily from livestock wastes(39%), natural sources (19%), volatilization of NH3-based fertiliz-ers (17%), biomass burning (13%), crops (7%) and emissions fromhumans, pets and waste water4 (5%). A large part of the emittedNH3 is quickly removed from the atmosphere by dry depositionclose to the source. The other main sink is the reaction withsulphuric, nitric and hydrochloric acids, leading to the formationof particulate ammonium3,5 (NH+4 ).

Despite regulations in some developed countries2, the totalemission ofNH3 hasmore than doubled from1860 to 1993 andmaydouble again by 2050 (ref. 6). This increase is leading to a cascadeof environmental problems5–8. NH3 accelerates the formation ofparticulate matter12 and can make up a large portion of its finalmass13. In some regions of the USA for instance, ammoniumsulphate makes up 60% of fine particulate PM2.5, and ammoniumnitrate can make up as much as 40% of the total mass14. NH3accounts for almost half of all reactive nitrogen released in theatmosphere, having an important role in the acidification and theeutrophication of our ecosystems5,6.

Uncertainties in the emission budgets of NH3 are as large as50% (refs 4, 10). Apart from missing statistics on fertilizer use andanimal populations, local differences in agricultural practice, suchas fertilization and waste management, become important whenscaling-up to global budgets5,10,11,15. Other uncertainties stem from

1Spectroscopie de l’Atmosphère, Service de Chimie Quantique et Photophysique, Université Libre de Bruxelles (ULB), 1050 Brussels, Belgium, 2UPMC Univ.Paris 06, LATMOS-IPSL; CNRS/INSU, LATMOS-IPSL, 75252 Paris Cedex 05, France, 3European Commission, Joint Research Centre (JRC), I-21027 Ispra,Italy. *e-mail: lclariss@ulb.ac.be.

68° E 72° E 76° E 80° E 84° E

8

6

4

2

0Jan. Feb. Mar. Apr. May Jun. Jul. Aug. Sep. Oct. Nov. Dec. Jan.

a

b

8° N

12° N

16° N

20° N

24° N

28° N

32° N

NH

3 total column (m

g m¬

2)

NH

3 to

tal c

olum

n (m

g m

¬2 )

0

1.25

2.50

3.75

5.00

6.25

7.50

Figure 1 |NH3 total columns retrieved with IASI over India and Pakistan.a, The IASI orbit on 13 May 2008 (morning), where each circle indicates anIASI pixel of 12 km diameter at nadir, increasing on both sides of the swath.The colour represents the retrieved concentration. b, Seasonal variation ofthe NH3 columns (mg m−2) measured near Ludhiana (31◦ N, 76◦ E) inNorth India. The solid line depicts the moving thirty day average, and thegrey area represents the associated temporal 1σ standard deviation aroundthis average.

the dependence of NH3 volatilization on temperature, humidity,wind velocity and soil acidity10,11 and uncertainties related to theemissions of natural sources and biomass burning, for whichexperimental data show large scatter9,10. Another contributingfactor is the limited spatial representativeness of ground-basedmeasurements, as NH3 is very variable in time and space, with atropospheric lifetime up to a couple of hours3,9. Complementingground-based measurements, satellite observations of NH3 would

NATURE GEOSCIENCE | VOL 2 | JULY 2009 | www.nature.com/naturegeoscience 479© 2009 Macmillan Publishers Limited. All rights reserved.

LETTERS NATURE GEOSCIENCE DOI: 10.1038/NGEO551

0

0.17

0.33

0.50

0.67

0.83

>1.00N

H3 total colum

n (mg m

¬2)

NH

3 total column (m

g m¬

2)

NH

3 total column (m

g m¬

2)

0

0.33

0.67

1.00

1.33

1.67

>2.00

<¬2.00

¬1.33

¬0.67

0

0.33

0.67

>1.00c

12

34 5

67

8

9

10

1112

13

1415

161718

19

20

21 22

2324

25

26 27

28

1. Inland Empire, Imperial Valley, Valley of the Sun2. Central Valley (Sacramento, San Joaquin Valley)3. Snake River Plain, Bear River Valley4. Yakima Valley5. Missouri Plateau6. Platte River Valley, Northwest Iowa Plains7. Central and Southern High Plains Aquifer8. Comarca Lagunera, Los Altos de Jalisco9. Central-West Brazil10. Central and North Argentina11. Ebro Valley, Region of Murcia, Province of Toledo12. Po Valley13. Belgium, the Netherlands, North Germany14. Kura-Araks Lowland, Azerbaijan

15. Garagum Canal, Khatlon, Uzbekistan16. Fergana Valley, Tashkent, Shymkent oblast17. North of Tian Shan Mountain Range, Dzungaria 18. Tarim Basin19. Indo-Gangetic Plain, Deccan Plateau, West Burma20. Yinchuan, Lanzhou, Wuwei21. South Siberia, East Mongolia, Inner Mongolia22. West Heilongjiang and Jilin province23. North China Plain 24. North African coast25. Nile Delta26. Nigeria27. Sudan, Ethiopia28. Angola, Zambia, Zimbabwe, Mozambique, Congo

¬60°

¬30°

0°

30°

60°

45°0° 90° 135°

a

0°

30°

60°

¬60°

¬30°

¬135° ¬90° ¬45° 45°0° 90° 135°

b

¬135° ¬90° ¬45°

1234 5

67

8

9

10

1112

13

1415

161718

19

202122

232425

26 27

28

0°

30°

60°

¬60°

¬30°

¬135° ¬90° ¬45° 45°0° 90° 135°

Figure 2 |Annual averaged retrieved and modelled NH3 columns, and their difference. a, Yearly average total columns of NH3 in 2008 retrieved fromIASI measurements on a 0.25◦ by 0.25◦ grid. b, Modelled columns for the year 2000 using the TM5 model on a 2◦ by 3◦ latitude–longitude grid9.c, Difference between measured and modelled columns.

be a welcome addition, although we are not aware of anysuccessful previous attempts to retrieve global distributions ofNH3 by satellite.

A few local observations of NH3 from space were reportedrecently from the tropospheric emission spectrometer16 (TES) usinga subset of features between 960 and 972 cm−1 and applyinga residual fitting method to extract quantitative concentrationestimates. Despite high spectral resolution, the poor geographicalcoverage of TES does not allow the retrieval of NH3 on a dailyglobal scale. Designed primarily for meteorology, the InfraredAtmospheric Sounding Interferometer17,18 (IASI) has a coarserspectral resolution, but has the advantage over TES of havingan exceptional geographical sampling and low noise over a widespectral range of the thermal infrared. Here we present and discuss

the first high-spatial-resolution global NH3 measurements acquiredwith IASI over a full year.

IASI is a Fourier transform spectrometer onboard the mete-orological platform MetOp-A, launched at the end of 2006. Theplatform circles in a polar sun-synchronous orbit around the earth.IASI operates in nadir mode (vertically downward, within 48.3◦ ofthe normal of the Earth’s surface) with a pixel footprint of 12 kmalong the satellite track, and provides global coverage twice a day byscanning along a swath of 2,200 km off-nadir. The overpass timesare 9:30 and 21:30 mean local solar time. Figure 1 illustrates IASImeasurements over land (India and Pakistan) for one particularmorning orbit. The spectrometer measures the infrared radiationemitted by the Earth’s surface and atmosphere in the spectralrange 645–2,760 cm−1 without gaps. It has a spectral resolution

480 NATURE GEOSCIENCE | VOL 2 | JULY 2009 | www.nature.com/naturegeoscience

© 2009 Macmillan Publishers Limited. All rights reserved.

NATURE GEOSCIENCE DOI: 10.1038/NGEO551 LETTERS

6° E 7° E 8° E 9° E 10° E 11° E 12° E 13° E 14° E

44° E 48° E 52° E 56° E 60° E 64° E 68° E 72° E

118° W 117° W 116° W 115° W 114° W 113° W 112° W 110° W111° W

46° 0’ N

45° 6’ N

45° 2’ N

44° 8’ N

44° 4’ N

74° 0’ N

73° 5’ N

73° 0’ N

72° 5’ N

72° 0’ N

71° 5’ N

71° 0’ N

70° 5’ N

70° 0’ N

44° 4’ N

44° 0’ N

43° 6’ N

43° 2’ N

42° 8’ N

42° 4’ N

0 0.17 0.33 0.50 0.67 0.83 >1.00

NH3 total columns (mg m¬2)

Po Valley¬Italy

Fergana Valley¬Uzbekistan

Snake River Valley¬United States

Figure 3 |Annual averaged NH3 columns over three agricultural valleys. NH3 concentrations derived from IASI observations above the Po Valley, theFergana Valley and the Snake River Valley. These correspond to the hotspots labelled 12, 16 and 3 in Fig. 2a. The aerial background photographs are © 2008Google—Imagery © 2008 Terrametrics (http://maps.google.com/).

of 0.5 cm−1 apodized and low noise levels (∼0.2K at 950 cm−1and 280K), which allows for the retrieval of a series of atmo-spheric key species17,18.

NH3absorbs infrared radiation in the ν2 vibrational band around950 cm−1 (750–1,200 cm−1), from which concentrations can be

retrieved using inverse methods (see theMethods section). Figure 1shows IASI-retrieved NH3 total column amounts above India andPakistan on 13 May 2008 and the 30-day moving average for2008 near Ludhiana in northern India. The increased columnsbetween April and August probably result from the temperature

NATURE GEOSCIENCE | VOL 2 | JULY 2009 | www.nature.com/naturegeoscience 481© 2009 Macmillan Publishers Limited. All rights reserved.

LETTERS NATURE GEOSCIENCE DOI: 10.1038/NGEO551

dependence of NH3 volatilization and intensive fertilization19.Figure 1 demonstrates the exceptional spatial resolution at whichNH3 can be monitored from IASI.

With 1,280,000 IASI spectra per day, sophisticated inversemethods are computationally demanding and have a significantfailure rate for weak absorbers. We have developed a methodfor calculating NH3 distributions over a longer period of time,with improved sensitivity. It starts by calculating the brightnesstemperature difference, between the IASI channel at 867.75 cm−1and two window channels at 861.25 and 873.50 cm−1. The channelat 867.75 cm−1 is sensitive to a manifold of five individual lines ofNH3 and almost not affected by other species. We have calculatedthis brightness temperature difference (with a correction factorfor the viewing angle) for all of the observed cloud-free spectrafrom 2008. The sensitivity of thermal nadir measurements near thesurface is intimately related to the thermal contrast between thesurface and the first atmospheric layers18. Only the IASI morningorbit, characterized by significant thermal contrast in most places,was therefore used for further calculations. Even though dailyobservations are possible near the largest sources (Fig. 1), becauseof the low emission height and short lifetime, the NH3 signature isoften within the instrumental noise of 0.2 K and emerges over otherregions only when averaging over time. Monthly averages werecalculated on a 0.25◦ by 0.25◦ grid from the daily values. Residualinstrumental noise andweak interferences from absorption featuresof other species can be estimated from the monthly averagedbrightness temperature differences over oceans, where we expectNH3 concentrations to be too low to be observed from space9,10.(According to the TM5 model, the average total column valuesabove oceans are of the order of 0.01mgm−2, corresponding tobrightness temperature differences lower than 0.001K. Parts of theIndian Ocean were excluded for the calculation of the residuals,as they showed elevated concentrations owing to transport fromIndia.) This remaining noise was found to depend mainly on thelatitude, ranging from below 0.08 K at the poles to below 0.05 Karound the Equator, and was subtracted from the monthly average.The resulting brightness temperature differences were convertedto total columns assuming linearity and using the conversionfactor 15± 7.5mgm−2 K−1 (see the Methods section for detailsand validation). The global average total columns for 2008 areshown in Fig. 2a with columns ranging from about 0.15 up to3mgm−2. A total of 28 hotspots with columns above 0.5mgm−2have been identified.

In parts of the world, NH3 emissions are linked to biomassburning. In 2008, there were large mid-latitude fires in SouthSiberia and Inner Mongolia from April to May20 and in therespective dry seasons of South America and East, West and SouthAfrica, which could explain parts of the observed distributions (seealso Supplementary Fig. S1). Virtually all of the other hotspotsindicated on the map are above the agricultural regions of NorthAmerica, Europe and Asia. We observe the largest NH3 columnsin agricultural valleys. Surrounded by mountains, these areas showstagnant weather allowing NH3 concentrations to build up. A well-known example is the San Joaquin Valley in California, infamousfor its poor air quality21. Also in non-mountainous regions, weobserve NH3 columns well above background levels, for instanceover the Nile Delta and large parts of Western Europe (Belgium,the Netherlands and Germany).

Figure 2b shows total columns estimated for the year 2000using the global atmospheric chemistry transport model TM5(refs 4, 5, 9). Comparison with Fig. 2a shows a relatively goodagreement in the location of the different NH3 hotspots. On aglobal quantitative level, the modelled NH3 columns are on averagehigher than the measured ones, as best seen from Fig. 2c. In someplaces such as southeast Asia, the observations do not show anyelevated values. As the high concentrations were confirmed by in

situ measurements at the surface22, this difference suggests poorsensitivity of the IASI measurements in the boundary layer inthis region. The overall underestimation in the measurements isprobably due to a combination of factors. An obvious reason is thatsatellite measurements have an inherent detection limit, resultingin NH3 signatures below the noise threshold (see the discussionabove) and underestimation of concentrations in periods of lowemission. For relatively colder areas, such as West Europe, thisis also linked to limited thermal contrast, which in turn limitsthe sensitivity of nadir sounders to surface concentrations. Themorning orbit of IASI is around 09:30 local mean solar time,which might not be representative of the daily average NH3concentration, given its short lifetime. This is particularly truefor agricultural biomass burning, which peaks in the afternoon.A final reason is a possible overall overestimation in the modeldue to uncertainties in inventories, NH3 profiles and lifetime. Itshould be emphasized that current models have large differencesamongst them, depending on the specific grid size, time steppingand emission inventory that was used. Also note that the mismatchbetween the year of observation and the model data is probablyresponsible for some difference in the annual averaged columnsbecause of differences in annual emissions and meteorology. Toachieve a better quantitative understanding, future work shouldinclude an extensive sensitivity analysis, validation and carefulassimilation of measured data into models.

Robust conclusions can be drawn already for places where themeasured columns exceed the modelled ones. These are foundin the Northern Hemisphere above 30◦. A remarkable example isCentral Asia (labels 15–18 in Fig. 2a). The largest columns thereare observed from May to August in the area that includes thewhole agricultural region irrigated by the AmuDarya, the Syr Daryaand their tributaries (the Aral Sea Basin, labels 15–16). The area isknown for its intensive cotton and wheat production, which usesvast amounts of inorganic fertilizer23. Elevated columns are alsofound north of the Tian Shan mountain range, in the Dzungariaregion and in the Tarim Basin (labels 17–18), which are highlypolluted areas with large amounts of ammonium-rich particulate24.The difference between observed and modelled distributions inCentral Asia points to an underestimation of NH3 emissions incurrent inventories. In parts of North America and Europe, we alsoobserve elevated columns of NH3, with the largest differences aboveagricultural valleys that the model does not reproduce. Examplesinclude the previously mentioned San Joaquin Valley, the Po andEbro valleys (Europe) and the Snake River Valley (United States).Figure 3 shows a detailed overlay of the global map onto aerialphotographs of three agricultural valleys.

Here, we have presented global NH3 integrated concentrationsretrieved from satellite measurements. The IASI instrument nevertargeted NH3 as a product for the mission, and hence theinstrumental specifications, especially the spectral resolution andradiometric noise, were not optimized accordingly. Despite this,we have derived global concentration maps on a daily basis usingthe IASI exceptional spatial resolution. These first results opennew and unexpected perspectives for fostering from space thesurveillance of air quality and the monitoring of troposphericchemistry. Annual and long-term trends in NH3 concentrationand emissions will be available given the planned 15 years ofIASI operation on the suite of MetOp satellites (and follow-upmissions), enhancing both the scientific and societal outputs ofEarth observation programmes.

MethodsCalibrated radiance spectra (Level 1C data) measured by the infrared spectrometerIASI onboard the meteorological platform MetOp-A were received throughthe EUMETCast near-real-time data distribution service. For this study, wealso used for each observation the ancillary meteorological products (Level2 data): they consist of pressure, temperature and humidity vertical profiles;

482 NATURE GEOSCIENCE | VOL 2 | JULY 2009 | www.nature.com/naturegeoscience

© 2009 Macmillan Publishers Limited. All rights reserved.

NATURE GEOSCIENCE DOI: 10.1038/NGEO551 LETTERSsurface temperature and cloud coverage. Surface emissivity was calculatedas the average from the 12 channels from the Moderate Resolution ImagingSpectroradiometer/Terra climatology.

Brightness temperatures were calculated from the radiance spectra innear real time after removing cloudy spectra as well as spectra with missingLevel 2 data. The brightness temperature difference between a NH3-sensitivechannel at 867.75 cm−1 and two window channels at 861.25 and 873.50 cm−1was systematically calculated for each observation and the resulting values werestored for further processing. Brightness temperature differences up to 3K inbiomass-burning plumes were seen on individual observations. Monthly averagesof these brightness temperature differences were then calculated and noisy valueswere removed as described in the text.

To map the brightness temperature differences computed globally intocolumn distributions of NH3, full radiative transfer model simulations andinverse retrievals were carried out on selected regions and periods. Starting froman a priori atmospheric state, inverse methods iterate a calculated spectrum,minimizing the difference with the observed spectrum. The retrieval is successfulwhen the calculated spectrum converges with a residual below the spectralnoise. Here, the retrievals were carried out using the optimal estimationinversion method25,26 implemented in the line-by-line radiative transfer modelAtmosphit26. The retrieval range was set at 940–969 cm−1, around the centreof the ν2 band of NH3, where the strongest absorption features are observed,also including one of the three windows used in earlier work with the TESinstrument16. A priori information of the atmospheric state (surface temperature,pressure and temperature profile) was taken from the Level 2 data. The primaryinterfering molecules in the retrieval range are H2O and CO2, and these werefitted together with NH3. A priori profiles of NH3 were taken from the 1976 USStandard Atmosphere model27. H2O was fitted in 2-km-thick partial columnsfrom the surface to 20 km, and total columns were fitted for CO2 and NH3 byscaling the prior profiles.

Given the relatively low absorption contribution of NH3, far fromsaturation, we expect linear correspondence between the brightness temperaturedifference and the retrieved total column. This was confirmed using forwardmodelling simulations for a range of NH3 concentrations up to 60mgm−2 (seeSupplementary Fig. S2a), with, however, a small departure from linearity for thehighest concentrations.

As an illustration of the mapping procedure between brightness temperaturedifferences and total columns of NH3, Supplementary Fig. S2b shows thecorrelation plot for spectra measured above the Baikal region on 12 May20. Linearregression for that case gives a slope of 20.02mgm−2 K−1. This value varies withlocation and time period because of the sensitivity dependence of the infraredobservations to the emissivity and temperature of the surface, the temperatureprofile of the atmosphere or the aerosol content. Overall regression analysis on awide selection of regions at different times of the year showed 1K to correspond to15±7.5mgm−2 of NH3 with a confidence interval of 80%. This is the conversionfactor that we used to calculate the global distributions from the averaged brightnesstemperature differences.

Received 4 February 2009; accepted 19 May 2009;published online 21 June 2009

References1. Galloway, J. N. et al. The nitrogen cascade. BioScience 53, 341–353 (2003).2. Sutton, M. A., Reis, S. & Baker, S. M. H. (eds) Atmospheric Ammonia

(Springer, 2009).3. Asman, W. A., Sutton, M. A. & Schjørring, J. K. Ammonia: Emission,

atmospheric transport and deposition. New Phytol. 139, 27–48 (1998).4. Galloway, J. N. et al. Nitrogen cycles: Past, present and future. Biogeochemistry

70, 153–226 (2004).5. Sutton, M. A., Erisman, J. W., Dentener, F. & Möller, D. Ammonia in

the environment: From ancient times to the present. Environ. Pollut. 156,583–604 (2008).

6. Krupa, S. V. Effects of atmospheric ammonia (NH3) on terrestrial vegetation:A review. Environ. Pollut. 124, 179–221 (2003).

7. Erisman, J. W., Bleeker, A., Galloway, J. & Sutton, M. S. Reduced nitrogen inecology and the environment. Environ. Pollut. 150, 140–149 (2007).

8. Galloway, J. N. et al. Transformation of the nitrogen cycle: Recent trends,questions, and potential solutions. Science 320, 889–892 (2008).

9. Dentener, F. J. & Crutzen, P. J. A three-dimensional model of the globalammonia cycle. J. Atmos. Chem. 19, 331–369 (1994).

10. Bouwman, A. F. et al. A global high-resolution emission inventory forammonia. Glob. Biogeochem. Cycles 11, 561–587 (1997).

11. Matthews, E. Nitrogenous fertilizers: Global distribution of consumption andassociated emissions of nitrous oxide and ammonia. Glob. Biogeochem. Cycles8, 411–439 (1994).

12. Larsen, L., Roth, B., Van Dingenen, R. & Raes, F. Photolytic aerosol formationin SO2–HNO2–H2O–air mixtures, with and without NH3. J. Aerosol. Sci. 28,S719–S720 (1997).

13. Anderson, N., Strader, R. & Davidson, C. Airborne reduced nitrogen:Ammonia emissions from agriculture and other sources. Environ. Int. 29,277–286 (2003).

14. Malm, W. C., Schichtel, B. A., Pitchford, M. L., Ashbaugh, L. L. & Eldred, R. A.Spatial and monthly trends in speciated fine particle concentration in theUnited States. J. Geophys. Res. 109, D03306 (2004).

15. Sutton, M. A. et al. Challenges in quantifying biosphere–atmosphere exchangeof nitrogen species. Environ. Pollut. 150, 125–139 (2007).

16. Beer, R. et al. First satellite observations of lower tropospheric ammonia andmethanol. Geophys. Res. Lett. 35, L09801 (2008).

17. Clerbaux, C. et al. The IASI/MetOp I Mission: First observations and highlightsof its potential contribution to GMES. COSPAR Inf. Bul. 2007, 19–24 (2007).

18. Clerbaux, C. et al. Monitoring of atmospheric composition using thethermal infrared IASI/MetOp sounder. Atmos. Chem. Phys. Discuss. 9,8307–8339 (2009).

19. Streets, D. G. et al. An inventory of gaseous and primary aerosol emissions inAsia in the year 2000. J. Geophys. Res. 108, 8809 (2003).

20. Coheur, P.-F., Clarisse, L., Turquety, S., Hurtmans, D. & Clerbaux, C.IASI measurements of reactive trace species in biomass burning plumes.Atmos. Chem. Phys. Discuss. 9, 8757–8789 (2009).

21. Battye, W., Aneja, V. P. & Roelle, P. A. Evaluation and improvement ofammonia emissions inventories. Atmos. Environ. 37, 3873–3883 (2003).

22. Carmichael, G. R. et al. Measurements of sulfur dioxide, ozone and ammoniaconcentrations in Asia, Africa, and South America using passive samplers.Atmos. Environ. 37, 1293–1308 (2003).

23. Scheer, C., Wassmann, R., Kienzler, K., Ibragimov, N. & Eschanov, R. Nitrousoxide emissions from fertilized, irrigated cotton (Gossypium hirsutum L) in theAral Sea Basin, Uzbekistan: Influence of nitrogen applications and irrigationpractices. Soil Biol. Biochem. 40, 290–301 (2008).

24. Li, J. et al. Characteristics and sources of air-borne particulate in Urumqi,China, the upstream area of Asia dust. Atmos. Environ. 42, 776–787 (2008).

25. Rodgers, C. Inverse Methods for Atmospheric Sounding: Theory and Practice(World Scientific, 2000).

26. Coheur, P.-F. Retrieval and characterization of ozone vertical profiles from athermal infrared nadir sounder. J. Geophys. Res. 110, D24303 (2005).

27. US Standard Atmosphere, 1976 (US Government Printing Office, 1976).

AcknowledgementsIASI has been developed and built under the responsibility of the Centre Nationald’Etudes Spatiales (CNES, France). It is flown onboard the MetOp satellites as part ofthe EUMETSAT Polar System. The IASI L1 data are received through the EUMETCastnear-real-time data distribution service. L.C. and P.-F.C. are respectively a ScientificResearch Worker (Collaborateur Scientifique) and Research Associate (ChercheurQualifié) with F.R.S.-FNRS. C.C. is grateful to CNES for scientific collaboration andfinancial support. The research in Belgium was financially supported by the F.R.S.-FNRS(M.I.S. nF.4511.08), the Belgian State Federal Office for Scientific, Technical and CulturalAffairs and the European Space Agency (ESA-Prodex arrangements C90-327). Financialsupport by the ‘Actions de Recherche Concertées’ (Communauté Française de Belgique)is also acknowledged.Wewould like to thankM.VanDamme for his assistance.

Author contributionsL.C. obtained the first NH3 global distributions, carried out the retrievals and theanalyses, drafted the manuscript and prepared the figures. C.C. had an active role inthe development of the IASI mission for the atmospheric composition aspects, andsupervised the project. F.D. developed themodel simulations and provided interpretationof the results. D.H. was responsible for the development of forward and inverse models.P.-F.C. made the first observations of NH3 from IASI, designed the filter and supervisedthe project. All authors contributed actively to the discussions of the results andpreparation of the manuscript.

Additional informationSupplementary information accompanies this paper on www.nature.com/naturegeoscience.Reprints and permissions information is available online at http://npg.nature.com/reprintsandpermissions. Correspondence and requests for materials should beaddressed to L.C.

NATURE GEOSCIENCE | VOL 2 | JULY 2009 | www.nature.com/naturegeoscience 483© 2009 Macmillan Publishers Limited. All rights reserved.