LETTERS PUBLISHED ONLINE: 28 APRIL 2013 | DOI: 10.1038/NGEO1800 Warming-induced increase in aerosol number concentration likely to moderate climate change Pauli Paasonen1,2 *† , Ari Asmi1† , Tuukka Petäjä1 , Maija K. Kajos1 , Mikko Äijälä1 , Heikki Junninen1 , Thomas Holst3 , Jonathan P. D. Abbatt4 , Almut Arneth5 , Wolfram Birmili6 , Hugo Denier van der Gon7 , Amar Hamed8 , András Hoffer9 , Lauri Laakso10,11 , Ari Laaksonen8,10 , W. Richard Leaitch12 , Christian Plass-Dülmer13 , Sara C. Pryor14 , Petri Räisänen10 , Erik Swietlicki15 , Alfred Wiedensohler6 , Douglas R. Worsnop1,8,10,16 , Veli-Matti Kerminen1 and Markku Kulmala1 Atmospheric aerosol particles influence the climate system directly by scattering and absorbing solar radiation, and indirectly by acting as cloud condensation nuclei1–4 . Apart from black carbon aerosol, aerosols cause a negative radiative forcing at the top of the atmosphere and substantially mitigate the warming caused by greenhouse gases1 . In the future, tightening of controls on anthropogenic aerosol and precursor vapour emissions to achieve higher air quality may weaken this beneficial effect5–7 . Natural aerosols, too, might affect future warming2,3,8,9 . Here we analyse long-term observations of concentrations and compositions of aerosol particles and their biogenic precursor vapours in continental mid- and high-latitude environments. We use measurements of particle number size distribution together with boundary layer heights derived from reanalysis data to show that the boundary layer burden of cloud condensation nuclei increases exponentially with temperature. Our results confirm a negative feedback mechanism between the continental biosphere, aerosols and climate: aerosol cooling effects are strengthened by rising biogenic organic vapour emissions in response to warming, which in turn enhance condensation on particles and their growth to the size of cloud condensation nuclei. This natural growth mechanism produces roughly 50% of particles at the size of cloud condensation nuclei across Europe. We conclude that biosphere–atmosphere interactions are crucial for aerosol climate effects and can significantly influence the effects of anthropogenic aerosol emission controls, both on climate and air quality. Climate feedback mechanisms are phenomena that originate from climate warming and either amplify or moderate warming. Common to practically all the proposed climate feedback mechanisms involving natural aerosols is that the estimation of their magnitude, or even existence, is based on a very limited amount of observational evidence. The feedback mechanism focused on in this study can be summarized as follows9 : (i) increasing temperatures increase emissions of biogenic volatile organic vapours (BVOCs), (ii) after rapid atmospheric oxidation these vapours condense on aerosol particles, thus enhancing particle growth, (iii) the number concentration of particles large enough to act as cloud condensation nuclei (CCN) increases, (iv) the cloud droplet number concentration and thus cloud albedo increases, which decreases the fraction of solar radiation penetrating the atmosphere and thus suppresses the climate warming. The validity of stages (i), (ii) and (iv) is commonly accepted1,10–15 , although the quantitative impact remains uncertain. Stage (iii), linking atmospheric CCN and organic vapour concentrations, has been demonstrated in modelling and conceptual studies16,17 , but the corresponding observations are scarce and reported only in local or regional scale studies14,18 . The long-term average number concentration of CCN-sized aerosol particles has been found to correlate with the time the air mass has spent over forests in Scandinavia14 . This suggests that the condensation of biogenic organic vapours affects not only the total particle mass but also the number concentrations of CCN-sized particles, indicated also in a case study at Whistler Mountain in North America18 . Although the above-mentioned experimental studies suggest a connection between condensable organic vapours and CCN number concentration, they do not demonstrate increasing CCN concentrations with increasing temperature. To investigate this dependence, we analysed data from eleven continental measurement stations, from clean semi-Arctic to polluted agricultural areas (Fig. 1, more information in Supplementary Methods). Number concentrations of particles with dry diameters larger than 100 nm (N100 ) were used as a proxy of CCN number concentrations19–21 to analyse how this quantity is associated with air temperature and concentrations of precursor vapours in both the gas and aerosol phase. The functional dependence of N100 on air temperature is depicted in Fig. 2a. The lines show the median concentrations in each 5 ◦ C temperature class at each site, and the shading 1 Department of Physics, University of Helsinki, FI-00014, Finland, 2 International Institute for Applied Systems Analysis, A-2361 Laxenburg, Austria, for Geo Biosphere Science, Lund University, S-22362 Lund, Sweden, 4 Department of Chemistry, University of Toronto, Ontario, M5S 3H6, Canada, 5 Karlsruhe Institute of Technology, Institute of Meteorology and Climate Research/Atmospheric Environmental Research, 82467 Garmisch-Partenkirchen, Germany, 6 Leibniz Institute for Tropospheric Research, 04318 Leipzig, Germany, 7 TNO Built Environment and Geosciences, 3584 CB Utrecht, The Netherlands, 8 Department of Applied Physics, University of Eastern Finland, 70210 Kuopio, Finland, 9 MTA-PE Air Chemistry Research Group, 8200 Veszprém, Hungary, 10 Finnish Meteorological Institute (FMI), FI-00101 Helsinki, Finland, 11 School of Physical and Chemical Sciences, North-West University, Potchefstroom 2520, South Africa, 12 Science and Technology Branch, Environment Canada, Toronto, Ontario, M3H 5T4, Canada, 13 Deutscher Wetterdienst, Meteorologisches Observatorium Hohenpeissenberg, 82383 Hohenpeissenberg, Germany, 14 Department of Geological Sciences, Indiana University, Bloomington, Indiana 47405, USA, 15 Division of Nuclear Physics, Lund University, S-221 00 Lund, Sweden, 16 Aerodyne Research, Inc., Billerica, Massachusetts 01821-3976, USA. † These authors contributed equally to this work. *e-mail: [email protected]. 3 Centre NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2013 Macmillan Publishers Limited. All rights reserved. 1 NATURE GEOSCIENCE DOI: 10.1038/NGEO1800 LETTERS Hohenpeissenberg (DE) Egbert (CA) 1998¬2000, Vavihill (SE) 2007¬2008 2007¬2008 2008 .. .. Hyytiala (FI) 1996¬2010 .. .. Varrio (FI) 1996¬2008 Melpitz (DE) 2003¬2008 Yakutsk (RU) 2009 (summer) 1 2 3 4 5 6 Woodland Forests 60° N 60° W Morgan Monroe (US) 2007¬2008 30° W San Pietro Capofiume (IT) 2002¬2006 10° S 30° E 10 Grassland 11 Cropland 12 Urban/built K-Puszta (HU) 2006¬2008 Botsalano (ZA) 2006¬2007 30° S Shrublands 7 8 9 0 Water Figure 1 | Measurement locations and data periods. The colours and numbers indicate the primary land use as described in ref. 28. 2 a N100 concentration (cm¬3) 10,000 Yakutsk .. .. Varrio .. .. Hyytiala Egbert Botsalano Vavihill 1,000 100 10 ¬40 Hohenpeiss. Morgan M. Melpitz K-Puszta S.Pietr.C. ¬30 ¬20 ¬10 0 10 20 30 40 20 30 40 Air temperature (°C) b 1 × 1013 1 × 1012 1 × 1011 1 × 1010 1 × 109 ¬40 Variability of anthopogenic primary emission inventory burdens Mixed boundary layer burden B100 (m¬2) denotes the 95% range of data from four example stations from distinctly different environments (95% areas for the other sites are shown in Supplementary Fig. S1). The number concentrations N100 exhibited clear increases with increasing temperatures (p-values below 0.01 for 8 sites out of 11, see Supplementary Methods and Table S1), especially in colder and cleaner environments (Värriö, Hyytiälä and Yakutsk). By contrast, the most polluted and the warmest environments (K-Puszta, San Pietro Capofiume and Botsalano) showed either constant or slightly decreasing N100 with increasing temperature. Notably, at T > 0 ◦ C there was a common temperature-dependent minimum of N100 , approximately ◦ at N100 = 101+0.05T / C cm−3 (dashed line in Fig. 2a). Our aim was to study the temperature dependence of the sources of CCN-sized particles. This cannot be done directly by investigating their measured concentrations, because the height of the boundary layer (BL), within which the aerosols are efficiently mixed, is also a strong function of temperature22 . We calculated the columnar BL aerosol number burden (B100 ), which is the number of >100 nm particles within a presumed well-mixed BL column, by multiplying the measured concentration with the boundary layer height (BLH) computed from the NCEP/DOE AMIP-II Reanalysis 2 database (Supplementary Methods). Although the BLH obtained from the reanalysis data is only an approximation, it has been shown to give reasonable estimates of the real BLH, including its seasonal and diurnal variation22 . B100 increased with increasing air temperature at all the sites over the observed temperature range (Fig. 2b and Supplementary Fig. S1) with a high level of statistical significance (p-value below 0.001 at all the sites at T > 5 ◦ C, see Supplementary Methods and Table S1). However, the temperature response was less evident at colder temperatures. At low temperatures, B100 at the four representative European stations agreed with the anthropogenic primary emission burdens (depicted as bars on the left hand side of Fig. 2b) calculated using a number emission inventory for primary anthropogenic particles23 (Supplementary Methods). The point at which a clear temperature dependence becomes evident varied, depending on anthropogenic emissions, clearly dividing B100 into regionally varying temperatureindependent and uniform temperature-dependent regimes. The uniform temperature dependency followed roughly the logto-linear least squares fit for burdens at T > 15 ◦ C, B100 = ◦ 1010.82+0.05T / C m−2 (dashed line in Fig. 2b). In addition to the 3-hour averages presented in Fig. 2, the temperature dependency was clear for the daily averages. Our budget estimation, based on the anthropogenic emission inventory and yearly temperature profiles, suggests that almost 50% of CCN-sized particles over the European continental area were formed through the temperature-dependent mechanism (Supplementary Methods). The marked increase in B100 with increasing temperatures is assumed to be due to the growth of smaller particles via secondary ¬30 ¬20 ¬10 0 10 Air temperature (°C) Figure 2 | The relationship between air temperature and the number concentration and burden of CCN-sized aerosol particles. a,b, The panels present 3-hour-mean number concentration (N100 ) (a) and boundary layer number burden (B100 ) (b) of CCN-sized particles as functions of temperature. Median values for each temperature are shown for all stations (lines) and the densest distribution of observations for four stations (shaded areas). The dotted line in a approximates the lower boundary of concentrations and the dotted line in b approximates the fit to median burdens at T > 15 ◦ C. The bars on the left in b show the calculated ranges of anthropogenic primary emission burdens in the regions of the highlighted stations (thick lines). aerosol formation (that is, condensation of vapours) associated with biological activity. Nevertheless, it might, in principle, also be related to reduced particle wet deposition by precipitation, as warm and dry conditions often coincide, or to increased primary particle emissions. In our data sets, however, precipitation had no NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2013 Macmillan Publishers Limited. All rights reserved. N100 (cm¬3) 104 N100 (cm¬3) NATURE GEOSCIENCE DOI: 10.1038/NGEO1800 LETTERS significant impact on the relationship between temperature and B100 (Supplementary Fig. S2). Emissions related to wildfires or biological activities (for example, spores and pollen) are the only primary aerosol sources expected to be positively correlated with the air temperature. However, an increase in wildfire emissions does not cause as pronounced an increase in median number concentrations as is evident in our observations24 , and primary biological emissions contribute minimally to continental total particle number concentrations25 . Thus, we postulate that the ambient temperature exerts a control over the concentration of CCN-sized particles by limiting the particle growth through biogenic secondary organic aerosol (SOA) formation11,12 . This inference is supported by aerosol particle chemical composition measurements at Hyytiälä during 2008–200923 and at Egbert in 200713 (Fig. 3, insert). At both locations, the number concentration N100 is directly proportional to the mass concentration of organic compounds in particles with diameters between approximately 70 and 800 nm (mOA ), which explains 73% of the variability in N100 . The formation of SOA occurs via condensation of the oxidation products of BVOCs, for example monoterpenes, sesquiterpenes and/or isoprene11 . We investigated the relative importance of temperature-dependent precursor emissions and atmospheric oxidation chemistry in CCN formation by comparing the relationships between B100 , temperature and radiation (Supplementary Fig. S3). Our analysis shows a clear dominance of temperature (and thus precursor emissions) over radiation (and thus photochemical oxidation), which is consistent with a recent modelling study26 . Figure 3 shows close to direct proportionality between N100 and monoterpene concentration ([MT]) at the two sites with simultaneous Egbert .. .. Hyytiala Yakutsk Hohenpeissenberg 103 102 10¬1 100 101 mOA (µg m¬3) 103 4) 0) 102 : pe Slo 84 0. (0 .70 1.0 e: op Sl 108 5 .6 4 1.0 (0 e: op Sl 7) 1.0 3 1.0 9 .8 (0 1.0 109 1010 1011 [MT] (cm¬3) Figure 3 | Concentration N100 as a function of gas phase monoterpene concentration ([MT]) and organic aerosol mass mOA (inset) at T > 5 ◦ C. The data for N100 versus [MT] cover 21 months at Hyytiälä and 15 months at Hohenpeissenberg from all the seasons, and two months at Yakutsk between May and August. The lines present bivariate regression fits, 95% confidence intervals are given in parenthesis (Supplementary Information). The data for N100 versus mOA (inset) cover four measurement campaigns (one month each) during spring, summer and autumn at Hyytiälä and one month in spring and summer at Egbert. The line in the insert represents direct proportionality. a Emission Oxidation Particle formation and growth Cloud activation SOA BVOC CCN Gas Aerosol Cloud albedo Air temperature Biosphere Radiation balance Biogenic effects b Moderately polluted Polluted background Morgan M. Hohenpeis. Vavihill Melpitz Botsalano S.Pietr.C. K-Puszta .. .. Hyytiala .. .. Varrio Egbert Yakutsk (only summer) Clean background Annual mean Tdaily > 5 °C ¬0.8 ¬0.6 ¬0.4 ¬0.2 0.0 0.2 Mean Mean local cloud albedo feedback (W m¬2 K¬1) Range of mean local direct effect feedbacks from the same stations Figure 4 | Schematic and estimated regional strengths of the proposed climate feedback mechanism. a, Proposed feedback mechanism, in which the biosphere reacts to warming climate by emitting more organic compounds, leading to enhanced growth of aerosols and to an increase in the cloud droplet concentrations, thus diminishing the net radiative forcing9 . b, Calculated approximate local annual and growing-season mean (the latter determined as the daily mean temperature >5 ◦ C) cloud albedo effect feedbacks for the stations. Green stations are located in different kinds of forest environments, orange in polluted forest or savannah environments and pink in croplands or built environments. BVOC: Biogenic volatile organic compounds; SOA: secondary organic aerosols; CCN: cloud condensation nuclei. NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2013 Macmillan Publishers Limited. All rights reserved. 3 NATURE GEOSCIENCE DOI: 10.1038/NGEO1800 LETTERS A' A High primary emissions d2 d1 B B' Low primary emissions Primary emissions dominate ns o Bi n ge e ic at in io iss m d om e Reduction from high to low aerosol emissions Log of aerosol number burden (dp > 100 nm) Climate warming (exaggerated) 0 Air temperature (°C) Figure 5 | Schematic summarizing the implications of the temperature dependence of CCN-sized particle burdens (B100 ). The thick black line indicates the natural aerosol background. If a similar warming happens in two regions A and B (the pink and yellow areas show idealized distributions of B100 ), the aerosol concentrations in less polluted region B will increase (B0 , light yellow area) and reduce the warming (maximum temperatures in B’ are lower than in A0 , light pink area) through the discussed feedback mechanism. If primary aerosol emissions are reduced similarly in cooler (d1 ) and warmer (d2 ) environments (initial and final primary emission burdens depicted with dashed lines), the resulting reduction in B100 will be smaller in the warmer environment (red arrows). long-term particle number and MT measurements (Hyytiälä and Hohenpeissenberg) and in data from a shorter campaign at Yakutsk (for proportionality analysis see Supplementary Methods). Less uniform proportionality was observed between N100 and isoprene concentrations (Supplementary Fig. S4). The linear dependency between [MT] and N100 , together with the similarity of the exponential temperature dependencies of [MT], mOA and N100 (Supplementary Methods, Fig. S5 and Table S2), as well as other evidence presented above, indicates BVOC emissions as the foremost factor driving the observed temperature dependency of N100 and B100 . Assuming the observed temperature dependence will remain in a warming climate, our findings indicate the existence of a negative aerosol–climate feedback mechanism in the continental biosphere9 (Figs 4a and 5). This feedback is conceptually similar to the widely investigated CLAW hypothesis27 related to marine sulphur emissions: in both cases, warming temperatures increase biogenic aerosol precursor emissions and subsequently enhance cloud albedo and negative climate forcing. The CLAW feedback originated mainly from theoretical reasoning and is difficult to validate8 . In contrast, the climate feedback considered here is strongly supported by direct observations. An order-of-magnitude estimate of the strength of the investigated aerosol–climate feedback can be calculated based on the change in direct and the first indirect (cloud albedo) aerosol effects. The cloud albedo effect was estimated directly from changes in measured concentrations N100 with respect to changes in temperature, whereas the direct effect was calculated based on the changes in the total volume of measured particle populations and in modelled BLH with respect to changes in temperature (Supplementary Methods and Fig. S5). Our calculations resulted in regional mean annual feedbacks of up to −0.3 W m−2 K−1 and demonstrated a clear dominance of the cloud albedo effect over the direct effect (Fig. 4b and Supplementary Table S3). The negative 4 feedbacks were strongest at the most northern and remote sites. At the most polluted sites of this study the calculated feedback values were positive (that is, enhancement of warming) because N100 decreased with increasing temperature at T < 15 ◦ C (Fig. 2a). This decrease seems to originate mainly from the more efficient dilution of anthropogenic emissions in deeper BL at warmer temperatures, because in B100 the corresponding decreases are either absent or far less prominent (Fig. 2b). Notably, during the growing season (T > 5 ◦ C) the negative feedback values are roughly doubled, whereas the positive values remain constant (Fig. 4b and Supplementary Table S3). As forests (excluding rainforests) and croplands cover about 10% of the total global area28 , we estimate that the overall order of magnitude of the feedback in the global atmosphere is around −0.01 W m−2 K−1 (Supplementary Methods). For more exact estimates, the temperature dependence of N100 needs to be studied further, especially in tropics, subtropics and isoprene-rich environments. As well as providing a negative feedback mechanism, the biogenic contribution of B100 has distinct consequences for the outcomes of anthropogenic emission abatement policies, potentially reducing the expected climate warming impact of the lowered anthropogenic emissions5–7 . According to our analysis, the formation of biogenic SOA sets a minimum level for B100 and thus maintains part of the aerosol cooling effect regardless of the reductions in anthropogenic aerosol and precursor emissions (Fig. 5). Because the biogenic contribution of B100 increases with temperature, the reductions in anthropogenic emissions will have a lesser impact on CCN concentrations in warmer regions. Our results indicate that the biogenic continental aerosol attenuates air temperature changes via a negative organic aerosol–climate feedback mechanism. However, at more polluted locations the levels of anthropogenic aerosol emissions exceed, at least for most parts of the year, the natural production of CCN-sized particles, thus inducing an anthropogenic cooling effect and partly surpassing the biogenic feedback. Estimating the full scale of the effects of this feedback requires understanding of the sources, processes and evolution of natural and anthropogenic organic aerosols in the atmosphere12,29,30 , and of the processes taking place in the biosphere3 . This understanding is based on long-term, comprehensive, multi-platform observations23 . In estimates of the anthropogenic climate change and the effects of aerosol-related air quality directives, the response of the biosphere has to be considered. Received 1 October 2012; accepted 14 March 2013; published online 28 April 2013 References 1. Forster, P. et al. in Climate Change 2007: The Physical Science Basis (eds Solomon, S. et al.) 129–234 (Cambridge Univ. Press, 2007). 2. Carslaw, K. S. et al. A review of natural aerosol interactions and feedbacks within the Earth system. Atmos. Chem. Phys. 10, 1701–1737 (2010). 3. Arneth, A. et al. Terrestrial biogeochemical feedbacks in the climate system. Nature Geosci. 3, 525–532 (2010). 4. Mahowald, N. Aerosol indirect effects on biogeochemical cycles and climate. Science 334, 794–796 (2011). 5. Arneth, A., Unger, N., Kulmala, M. & Andreae, M. O. Clean the air, heat the planet. Science 326, 672–673 (2009). 6. Mickley, L. J., Leibensperger, E. M., Jacob, D. J. & Rind, D. Regional warming from aerosol removal over the United States: Results from a transient 2010–2050 climate simulation. Atmos. Environ. 46, 545–553 (2012). 7. Makkonen, R. et al. Air pollution control and decreasing new particle formation lead to strong climate warming. Atmos. Chem. Phys. 12, 1515–1524 (2012). 8. Quinn, P. K. & Bates, T. S. The case against climate regulation via oceanic phytoplankton sulphur emissions. Nature 480, 51–56 (2011). 9. Kulmala, M. et al. A new feedback mechanism linking forests, aerosols, and climate. Atmos. Chem. Phys. 4, 557–562 (2004). 10. Guenther, A. B. et al. A global model for natural volatile organic compound emissions. J. Geophys. Res. 100, 8873–8892 (1995). 11. Hallquist, M. et al. The formation, properties and impact of secondary organic aerosol: Current and emerging issues. Atmos. Chem. Phys. 9, 5155–5236 (2009). NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2013 Macmillan Publishers Limited. All rights reserved. NATURE GEOSCIENCE DOI: 10.1038/NGEO1800 12. Jimenez, J. L. et al. Evolution of organic aerosols in the atmosphere. Science 326, 1525–1529 (2009). 13. Leaitch, W. R. et al. Temperature response of the submicron organic aerosol from temperate forests. Atmos. Environ. 45, 6696–6704 (2011). 14. Tunved, P. et al. High natural aerosol loading over boreal forests. Science 312, 261–263 (2006). 15. Lee, A. K. Y. et al. Characterization of aerosol and cloud water at a mountain site during WACS 2010: Secondary organic aerosol formation through oxidative cloud processing. Atmos. Chem. Phys. 12, 7103–7116 (2012). 16. Riipinen, I. et al. Organic condensation: A vital link connecting aerosol formation to cloud condensation nuclei (CCN) concentrations. Atmos. Chem. Phys. 11, 3865–3878 (2011). 17. Riipinen, I. et al. The contribution of organics to atmospheric nanoparticle growth. Nature Geosci. 5, 453–458 (2012). 18. Pierce, J. R. et al. Nucleation and condensational growth to CCN sizes during a sustained pristine biogenic SOA event in a forested mountain valley. Atmos. Chem. Phys. 12, 3147–3163 (2012). 19. Dusek, U. et al. Size matters more than chemistry for cloud-nucleating ability of aerosol particles. Science 312, 1375–1378 (2006). 20. Clarke, A & Kapustin, V. Hemispheric aerosol vertical profiles: Anthropogenic impacts on optical depth and cloud nuclei. Science 329, 1488–1492 (2010). 21. Sihto, S-L. et al. Seasonal variation of CCN concentrations and aerosol activation properties in boreal forest. Atmos. Chem. Phys. 11, 13269–13285 (2011). 22. Seidel, D. J. et al. Climatology of the planetary boundary layer over the continental United States and Europe. J. Geophys. Res. 117, D17106 (2012). 23. Kulmala, M. et al. General overview: European integrated project on aerosol cloud climate and air quality interactions (EUCAARI)—integrating aerosol research from nano to global scales. Atmos. Chem. Phys. 11, 13061–13143 (2011). 24. Barnaba, F., Angelini, F., Curci, G. & Gobbi, G. P. An important fingerprint of wildfires on the European aerosol load. Atmos. Chem. Phys. 11, 10487–10501 (2011). 25. Després, V. R. et al. Primary biological aerosol particle in the atmosphere: A review. Tellus B 64, 015598 (2012). 26. Day, M. C. & Pandis, S. N. Predicted changes in summertime organic aerosol concentrations due to increased temperatures. Atmos. Environ. 45, 6546–6556 (2011). 27. Charlson, R. J., Lovelock, J. E., Andreae, M. O. & Warren, S. G. Oceanic phytoplankton, atmospheric sulphur, cloud albedo and climate. Nature 326, 655–661 (1987). 28. Hansen, M. C., DeFries, R. S., Townshend, J. R. G. & Sohlberg, R. Global land cover classification at 1 km resolution using a decision tree classifier. Int. J. Remote Sens. 21, 1331–1365 (2000). 29. Spracklen, D. V. et al. Aerosol mass spectrometer constraint on the global secondary organic aerosol budget. Atmos. Chem. Phys 11, 12109–12136 (2011). LETTERS 30. Stevens, B. & Feingold, G. Untangling aerosol effects on clouds and precipitation in a buffered system. Nature 461, 607–613 (2009). Acknowledgements This work was funded by European Research Council (ATMNUCLE, 227463), Academy of Finland (Center of Excellence Program project 1118615; projects 11750, 127210, 132640 and 139656), European Commission sixth Framework program (EUCAARI, contract no 036833-2; EUSAAR, contract no 026140), European Commission seventh Framework program (ACTRIS, contract no 262254; PEGASOS, contract no 265148), Maj and Tor Nessling Foundation (projects 2010143, 2011200, 2012443 and 2013325) and the Otto A. Malm foundation. For providing data, we acknowledge the Physical Sciences Division, Earth System Research Laboratory, NOAA, Boulder, Colorado (for the boundary layer height data acquired from their web site at http://www.esrl.noaa.gov/ psd/), the Hungarian Meteorological Service, Gerald Spindler, Pasi P. Aalto and Harald Flentje. We thank Ville Vakkari for help in quality assurance of Botsalano data and students and staff of North-West University (Mafikeng Campus), RSA, for weekly maintenance. The various measurements were supported by the German Federal Ministry for the Environment (FE 370343200), Environment Canada, NSERC, CFCAS-CAFC, the National Science Foundation (#0544745 and supplement, and 1102309), the Swedish Research Council (projects 2007-3745, 2007-4619 and 2010-4683) and the Swedish Research Council for Environment, Agricultural Sciences and Spatial Planning (Formas, project 2009-615). Author contributions P.P. had the original idea; P.P. and A. Asmi made the data analysis; A. Asmi made the figures; M.K.K., T.H. and C.P-D. made the monoterpene concentration measurements and the data pre-analysis; M.Ä., H.J. and J.P.D.A. made the aerosol mass concentration measurements; A. Asmi, V-M.K., A.L. and P.R. made the calculations of the feedback strength; A. Arneth, W.B., A. Hamed, A. Hoffer, T.H., L.L., W.R.L., S.C.P., E.S., A.W. and T.P. provided the particle number size distribution and meteorological data; H.D.v.d.G. provided the anthropogenic emission inventory data; T.P., D.R.W. and M.K. supervised the analysis and writing; P.P., A. Asmi and V-M.K. wrote the paper; All authors contributed with their comments to the paper. Additional information Supplementary information is available in the online version of the paper. Reprints and permissions information is available online at www.nature.com/reprints. The processed data for this manuscript can be found in the ACTRIS database (http://actris.nilu.no/ Content/Products/). Correspondence and requests for materials should be addressed to P.P. Competing financial interests The authors declare no competing financial interests. NATURE GEOSCIENCE | ADVANCE ONLINE PUBLICATION | www.nature.com/naturegeoscience © 2013 Macmillan Publishers Limited. All rights reserved. 5
© Copyright 2026 Paperzz