Growth rates during coastal and marine new particle formation in western Ireland

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115,, doi: /2010jd014292, 2010 Growth rates during coastal and marine new particle formation in western Ireland Mikael Ehn, 1 Henri Vuollekoski, 1 Tuukka Petäjä, 1 Veli Matti Kerminen, 1,2 Marko Vana, 1,3 Pasi Aalto, 1 Gerrit de Leeuw, 1,2,4 Darius Ceburnis, 5 Regis Dupuy, 5 Colin D. O Dowd, 5 and Markku Kulmala 1 Received 31 March 2010; revised 2 June 2010; accepted 8 June 2010; published 28 September [1] Growth rates of new particles during coastal and marine secondary aerosol particle formation events were studied in western Ireland, both at the Mace Head atmospheric research station and onboard the R/V Celtic Explorer as part of the Marine Aerosol Production project. Strong new particle formation events are frequently detected at Mace Head caused by the emission of precursor gases from exposed seaweed during low tide. Although these events were usually only detected as a mode of particles at a certain size, we were able to link the size of the mode to the growth time of these particles after the initial formation by combining data from several events measured between January 2006 and November 2007 with an air ion spectrometer. Typically, the early growth rates were extremely high, reaching values of several hundred nanometers per hour during the first seconds. The growth rates rapidly decreased and reached values below 1 nm h 1 within 1 h after nucleation. Our results were reproduced with box model calculations. All the obtained growth rates could be explained by the model either by varying the precursor formation time (typically a few seconds) or allowing multiple precursor vapor additions. From the ship borne measurements, we report the first observations of purely open ocean new particle formation detected in this region. In total, four events were detected during this period, with three having a variable continental influence. An estimated average growth rate in marine conditions was 3 nm h 1 for these events. Citation: Ehn, M., et al. (2010), Growth rates during coastal and marine new particle formation in western Ireland, J. Geophys. Res., 115,, doi: /2010jd Introduction [2] Nucleation has been demonstrated to be a significant source of new aerosol particles and, with some reservations, also of cloud condensation nuclei (CCN) in the global atmosphere [e.g., Kulmala et al., 2004; Spracklen et al., 2008; Merikanto et al., 2009; Pierce and Adams, 2009; Kuang et al., 2009]. However, in marine air the importance of nucleation has not been as clearly established. While nucleation appears to be a frequent phenomenon in many coastal areas [Wen et al., 2006; Yoon et al., 2006; O Dowd and de Leeuw, 2007; Modini et al., 2009; Whitehead et al., 2009], in remote marine environments, new particle formation events have only occasionally been observed [Covert 1 Department of Physics, University of Helsinki, Helsinki, Finland. 2 Finnish Meteorological Institute, Climate Change Unit, Helsinki, Finland. 3 Institute of Physics, University of Tartu, Tartu, Estonia. 4 Netherlands Organization for Applied Scientific Research, The Hague, Built Environmental and Geosciences, Utrecht, Netherlands. 5 School of Physics and Centre for Climate and Air Pollution Studies, Environmental Change Institute, National University of Ireland Galway, Galway, Ireland. Copyright 2010 by the American Geophysical Union /10/2010JD et al., 1996; Koponen et al., 2002; Heintzenberg et al., 2004]. There is no direct experimental evidence for marine CCN production associated with nucleation in the marine atmosphere away from the coast [e.g., O Dowd, 2001; Whitehead et al., 2009]. [3] The ability of nucleated particles to form new CCN and thereby their potential effect on climate depends crucially on their growth rate (GR) [Kerminen et al., 2004; Pierce and Adams, 2006]. When nucleation occurs over vast spatial scales, as is the case in so called regional nucleation events typical for continental boundary layers, the value of GR can be determined relatively easily by following the time evolution of the nucleation mode at a single measurement site [e.g., Kulmala and Kerminen, 2008]. The vast majority of regional nucleation events have GR in the range of about 1 10 nm h 1 and with little variability over the course of the event. Substantially larger GR are observed only in coastal areas [Dal Maso et al., 2002] and in highly urbanized megacities [Mönkkönen et al., 2005; Stolzenburg et al., 2005; Iida et al., 2008], where the growing clusters need to grow rapidly to survive the scavenging due to preexisting particles [Kulmala et al., 2005; Kuang et al., 2010]. When emissions of nucleating vapors are restricted to a very small area, the resulting GR can be highly variable, and one 1of11

2 should perform measurements at several locations downwind of that area to get an idea on the magnitude and behavior of the GR of the nucleated particles. Such measurements are extremely rare and usually require the employment of an air craft [e.g., O Dowd et al., 2007]. [4] At the Mace Head atmospheric research station on the west coast of Ireland, new particle formation events occur on more than half of the days; these events typically occur at low tide and preferentially in the presence of intensive solar radiation [O Dowd et al., 2002; Yoon et al., 2006; O Dowd and de Leeuw, 2007]. Only a small fraction of these events display a banana shape that, in the particle number size distribution versus time series plot, was characterized by a steady modal diameter growth. This is typical for regional nucleation events observed in many continental locations [Vana et al., 2008; Kulmala and Kerminen, 2008]. Instead, the events observed in Mace Head are characterized with apple, bump, or mixed shapes indicative of relatively local origin of the vapors responsible for the nucleation and initial growth of the particles [Vana et al., 2008]. The apple and bump shapes correspond to situations where no growth is observed during the majority of the event, with a distinct mode visible for the apple events, and a mode still merged with the background small ions for the bump events. The most probable source area for the nucleating/condensing vapors around Mace Head is the coast line, in which seaweed is exposed to sunlight during low tide. The seaweed is believed to emit iodocarbons that readily photolyze to low vapor pressure iodine compounds that produce new particles [Mäkelä et al., 2002]. [5] In this study we investigate the growth of nucleated particles in marine air by analyzing measurements made at a coastal zone and over the nearby open ocean. As regards the coastal zone, our analysis will be based on a concept termed growth time, which is essentially the time that it takes for particles nucleated over a small area to reach a fixed measurement site. The analysis is supplemented by contrasting our findings with those made over the open ocean as well as with box model calculations using the parameterization by Vuollekoski et al. [2009]. 2. Materials and Methods [6] Ion and particle size distributions were measured at the Mace Head atmospheric research station on the west coast of Ireland between January 2006 and November Particle size distributions were measured during an intensive campaign on board the R/V Celtic Explorer during June July 2006 off the western coast of Ireland, as part of the European Union project MAP (Marine Aerosol Production). The instruments on the ship were located in a container on the fore deck Instrumentation [7] Particle number size distributions in the dry diameter range from 3 to 500 nm were measured in Mace Head with a twin scanning mobility particle sizer (SMPS) as described by Yoon et al. [2006], with modification of the measured size ranges of the two systems to 3 20 nm and nm, respectively, and on the R/V Celtic Explorer with a twindifferential mobility particle sizer (DMPS) [Hoppel, 1978; Aalto et al., 2001]. Both instruments use differential mobility analyzers (DMA) to select particles of desired size, which are subsequently counted by a condensation particle counter (CPC). In both cases, twin refers to instruments consisting of two DMA + CPC setups, measuring different parts of the size distribution, which are combined in the inversion to obtain the number size distribution for the whole size range. Data for the SMPS in Mace Head are available for 2006 and for part of The DMPS on the ship measured between 11 June and 5 July 2006, except from noon 19 June to 24 June, due to a storm that cut the power to the container. [8] The ion size distribution in Mace Head was measured with an air ion spectrometer (AIS) [Mirme et al., 2007]. The AIS classifies ambient ions according to their electrical mobility, in a size range of roughly 1 40 nm in mobility diameter. It consists of two columns similar to DMAs, with the main difference that the classifying voltage is not scanned to pass through particles of a certain size for subsequent detection. Instead, in the AIS the voltages are fixed, and the outer cylinder is lined with electrometers. On the basis of their attenuation in the electric field, ions of different sizes (electrical mobilities) impact on different stages (electrometers) of the column. The ions thus produce a current in each electrometer depending on their ambient concentration Analysis of New Particle Formation Events and Growth Time [9] The AIS spectra for each day were inspected visually, and if there was a period with clear nucleation forming a distinct mode of particles, this day was selected for further analysis. For each new particle formation event, we selected the time when the observed mode appeared to be stable in time. A 15 min average size distribution around this time was then plotted, and an average diameter of the mode was estimated from the number size distribution. This was also done manually, as the shape of the modes varied greatly between days, and we wanted to avoid an automatic fitting procedure. An example of an event, with the selected time and particle size, is shown in Figure 1. [10] The coast line around Mace Head acts as a strong source of precursor vapors for new particle formation during low tide conditions [O Dowd et al., 2002; Yoon et al., 2006; O Dowd and de Leeuw, 2007]. Motivated by this observation, we calculated a so called growth time, which is the time that an air parcel reaching Mace Head has traveled after passing over the coast line. The initial formation of the particles was assumed to occur at the coast line. In practice, we determined the position of an arriving air mass 1 h before it reached Mace Head by using the Hysplit model [Draxler and Rolph, 2003; Rolph, 2003]. The time resolution is limited by the fact that Hysplit trajectories can be computed with only 1 h intervals. Thus, the above mentioned points that only events that were stable over an extended period of time were selected become important. We then assumed that the air parcel followed a straight line at a constant velocity from this point to Mace Head (Figure 2a). Finally, we calculated a distance from Mace Head to the coast along this line, and from this distance and from the air parcel velocity (i.e., wind speed), we obtained an estimate for the growth time as the time it took to transport particles freshly formed at the 2of11

3 Figure 1. An example of an apple type event measured with the AIS. The mode remained stable for several hours, and the parameters used for further analysis from this day were time = 14:38:24 and size = 5.36 nm. coast line to the Mace Head station. The analysis was performed for all new particle formation events. As the main reason for a mode of new particles not being stable is a change in meteorological conditions, such as changing wind direction during the event, no bias is expected from only analyzing events where the mode was stable over a longer period. [11] Figure 2b shows a more detailed view of the coast line around Mace Head. This coast line was the one used for the calculations, and it is thus obvious that no islands or coast lines further from Mace Head were taken into account. If the trajectory passed over the coast line more than once, as is the case if air masses came from the south east, the growth time was calculated from the last pass. Additionally, there is a narrow sector of wind directions ( 80 ) where a straight line trajectory will not cross a coast line at all, and these days were discarded from further analysis. The axes in Figure 2b are in kilometers and show that the furthest distance an air mass can travel on land in this analysis is roughly 3 km Model Description [12] Simulations of nucleation and subsequent growth of the new particles were performed with the UHMA model (University of Helsinki Multicomponent Aerosol model) [Korhonen et al., 2004], which is a size segregated box model. The model incorporates all basic aerosol dynamical processes: nucleation, coagulation, condensation/evaporation, and dry deposition. To extend the model into coastal and marine environments, it was modified to include iodine dynamics and a dilution scheme. More details, as well as the model parameterization also used in this study, have been given in the study by Vuollekoski et al. [2009]. [13] In this study the model was used to simulate the formation and growth of new particles originating at the coast line. The preexisting background particle distribution was set to match a typical distribution measured in Mace Head. New particles were formed via activation of clusters containing an iodine compound. The particles were allowed to form and grow for a duration equal to the growth time, and the final particle size distribution was output for further analysis. As Figure 2. An illustration (a) on how an air mass is assumed to approach Mace Head (red cross). The blue cross displays the position of an air mass 1 h before reaching Mace Head, and the dashed line depicts the assumed back trajectory. (b) Also shown is the more detailed map of the local geography around the research station, which was used for the growth time calculations. 3of11

4 Figure 3. Mean dry diameter of the nucleated particles as a function of growth time grouped into four wind direction (WD) sectors, where WD was determined as the direction during the last hour before arriving at Mace Head as indicated from the HYSPLIT data (see section 2.2). Each data point corresponds to an individual new particle formation event observed at Mace Head. The symbols are colored according to the concentration of ions in the size range of 3 20 nm during the events. the variable parameter, we used the production time of nucleating iodine vapor as well as that of a nonvolatile organic vapor. The model was also modified to allow for two passages over the coast line by enabling another burst of iodine vapors and limiting the dilution to new particles only. 3. Results and Discussion 3.1. Coastal Nucleation Observed in Mace Head [14] The analysis will concentrate on the apple type nucleation events, similar to the one in Figure 1, for which the nucleating precursor vapor source is expected to be the most localized. Furthermore, to exclude events that may have occurred away from the coast, we restricted our analysis to events where the mode of new particles was detected within a time period from 2 h before to 4 h after the lowest tide Mean Size and Growth Rate of Nucleated Particles [15] Figure 3 shows the relation between the mean size of nucleated particles and their growth time at Mace Head. A clear difference between the wind direction (WD) sectors can be seen, the probable explanation being the local geography (see Figure 2) and associated emissions. Some parts of the coast line will emit more precursors than others, depending on the size and type of exposed tidal region. For wind directions between 270 and 360, most of the arriving air parcels have very likely passed over the coast line only once, in addition to which the air mass transport has been almost perpendicular to the coast line, except close to the WD of 360. Growth times in this sector were found to be shorter than a few minutes, and the size of the nucleated particles is strongly correlated with growth time. The longest growth times and largest particles were seen in the direction south to east (WD: ). Air parcels approaching Mace Head from this direction had very probably either passed over the coast line several times or moved along a coast line (see Figure 2a) with a potentially long exposure time to the emitted precursors. The longer an air parcel spends above the coastal area, the more nucleating precursor vapors it is expected to contain [McFiggans et al., 2004; Sellegri et al., 2005], and these vapors will subsequently contribute either to the formation of more new particles or to the growth of existing particles due to condensation [O Dowd et al., 2004; Vaattovaara et al., 2006]. The concentration of ions during the events did not show any consistent relation with growth time, but it tended to be higher when the particles were larger. The role of ions will be returned to in section [16] In Figure 4, we have combined all measurement data together and grouped them in the following way: First, we discarded the events for which the measured air masses had possibly traveled over a coast line more than once or moved along a coast line. This was done by removing all events corresponding to wind directions , , and (see Figure 2b). Second, we discarded the events for which SMPS data were not available. Finally, to remove any artifacts resulting from our analysis or instrumental problems, we discarded the events for which the peak size of the sub 20 nm particle population measured by the SMPS was not within 2.5 nm of that measured by the AIS. In Figure 4 the removed events are plotted in gray. The remaining 18 nucleation events are plotted in color. Also plotted is the logarithmic fit to these data and three additional data points taken from the air craft measurements by O Dowd et al. [2007]. The logarithmic fit provides an estimate for the growth rate of the nucleated particles since the last contact of the air parcel with the source area at the coast line (Table 1). When this contact has been very recent, the growth rates may be extremely high, reaching several hundreds of nanometers per hour. An hour after contact with 4of11

5 Figure 4. The mean size of the nucleated particles as a function of growth time for all wind sectors as in Figure 3. Different symbol colors indicate local wind direction, and the black crosses represent the data taken from air craft measurements by O Dowd et al. [2007] in the vicinity of Mace Head. Gray markers are events that were removed from the analysis (see text), the yellow line is a logarithmic fit to the colored data points, and the black lines represent results obtained from the box model simulations with varying precursor vapor production time, and one or two passes over a coast line. the coast the growth rates have decreased to values below 1nmh 1. [17] To get further insight into the growth behavior of nucleated particles, we conducted a series of box model simulations using the parameterization by Vuollekoski et al. [2009]. In the model, new particles are formed and grown in a diluting plume downwind of a strong, local source of chemically reactive vapors. The model dynamics involve iodine vapors emitted by the source and sulfuric acid present in the marine atmosphere. Sulfuric acid at Mace Head has in earlier work been shown to be too low to alone explain the observed growth of the particles and contributes substantially to the growth only after it has decreased to about 1 nm h 1 [O Dowd et al., 2002; Berresheim et al., 2002]. In addition to these vapors, the nuclei growth can be assisted by other low volatile vapors emitted by the smallarea source, such as organic vapors. [18] With the available nucleation mechanisms, our simulations revealed that the logarithmic fit line corresponding to observations can be reproduced if (1) the effective duration of production of nucleating vapors is very short, of the order of 1 s, corresponding to a very narrow source area such as a coast line with exposed seaweed, and (2) the source also emits significant concentrations of vapors that do not actively participate in nucleation but contribute to the nuclei growth. The model simulation of the particle diameter as a function of growth time with a 1 s production time of precursor vapors is in good agreement with the fit to the experimental data (Figure 4, black dotted line). [19] We performed further simulations to explore the effects of (1) a longer production time for nucleating vapors, which corresponds to the transport of air along a coast line, or over a wide tidal region, (2) passage over two coast lines, and corresponding source areas. The dash dotted black line in Figure 4 corresponds to a 2 s production time, and as anticipated, the particles grew faster and reached larger sizes. The dashed black line depicts the case where the air mass passes over two tidal areas. In this case, the particles first nucleate, grow, and dilute for a time t. Then, a second pass over a coast line is modeled, with 1 s of precursor vapor production, after which the particles are allowed to again grow a time t, corresponding to the growth time. In this case the resulting particle sizes are larger than during any of the observed events. The mode diameter goes off scale and seems to grow very rapidly, but the growth slows down in this case too, with the particle size after 100 s being around 25 nm, and after 1000 s around 30 nm. If the time between the first and second coast line passes is decreased, the resulting particle concentrations also decrease. [20] The majority of events used for the experimental fit correspond to a production time between 1 and 2 s. Typical wind speeds of 5 10 m s 1 translate to active tidal regions 5 20 m wide. In many cases the exposed regions are even wider, and air masses passing over these regions, together with air masses passing several tidal regions, can easily account for the largest particles we observed. Thus, with a realistic variability in the parameters, we were able to envelope all the experimental data points between the modeled values Ions and Charging State [21] As mentioned in the previous section, the concentrations of ions in the size range of 3 20 nm tended to be Table 1. Particle Sizes and Growth Rates at Selected Times After Formation, Based on a Logarithmic Fit to the Experimental Data a Time 10 s 30 s 1 min 10 min 30 min 1 h GR (nm h 1 ) Size (nm) a See Figure 4. 5of11

6 Figure 5. Charging state as a function of (a) particle concentration in the 3 20 nm range, (b) growth time, and (c) 3 20 nm ion concentration. The symbol colors indicate the mean particle size. The charging state is best correlated with the total particle concentration. higher when the mean size of the nucleation mode was bigger. This feature is consistent with the general increase of the particle charging probability with increasing particle size [Hoppel and Frick, 1986, Wiedensohler, 1988]. In other words, in charge equilibrium, the fraction of particles of a certain size that are charged, increases as the particle size increases. Therefore, the correlation between ion number and size should not be interpreted as a correlation between formation rate and growth rate. [22] When investigating the role of ions in the nucleation process, a useful quantity is the so called charging state [Laakso et al., 2007; Gagné et al., 2008]. The charging state is defined as the ratio of charged to neutral particles at a given size, to the ratio expected from the charging probability. A charging state greater than unity indicates some contribution by ion induced nucleation. In contrast, a charging state smaller than one suggests minor or no contribution by ion induced nucleation. The dynamic nature of the growing nuclei population complicates the interpretation of the measured charging state: When nuclei grow in size, their charging state approaches unity regardless of the initial nucleation mechanisms [Kerminen et al., 2007]. Influencing factors in this regard are the nuclei growth rate, cluster ion concentration, and in case the nucleation event is very strong, also the concentration of nucleated particles. [23] Here, we estimated the particle charging state first by taking the ratio between the ion and total particle concentration in the size range 3 20 nm and then dividing this quantity by the particle charging probability at the size where the ions had their concentration maximum. The charging state showed a large variability between the different events (Figure 5). The magnitude of the charging state seemed to be influenced most by the total particle concentration, whereas the size of the particles and total ion concentration did not show any consistent relation with it. Because of the limited amount of charges available in the air, it may take several minutes for the particles to reach charge equilibrium if the concentration of new particles is very high. Thus, as seen in Figure 5, the charging state is usually lower when particle number is high. Observed charging states were practically always below unity when the nuclei concentration exceeded 10 4 cm 3 or when the growth time was longer than 1 min. These results, while indicative of some contribution by ion induced nucleation during the weakest events, support the general dominance of neutral nucleation pathways in the coastal new particle formation events Observations on New Particle Formation Onboard the Celtic Explorer [24] To enlarge the geographical coverage of our observations and findings on the coastal new particle formation, we present supporting evidence obtained during the MAP campaign conducted onboard the R/V Celtic Explorer. A time series of aerosol number size distributions during the period from 29 June to 1 July 2006 is presented in Figure 6. During this period, the twin DMPS detected three new particle formation events on three consecutive days, all of which were linked to coastal or continental influence. An event on 13 June 2006 was observed in clean marine air Coastal and Continental Influence [25] Figure 6 shows particle size distributions measured with the twin DMPS on 29 June to 1 July 2006, with the newly formed particles encircled. There are other short lived plumes visible in the data, mostly due to sampling of ship exhaust. During these 3 days, the ship was positioned north of Ireland, close to the Outer Hebrides. The ship traveled south into the wind during the days and during nights went back north. During the latter periods, ship plumes are clearly visible in the data. The engine exhaust also differs from the events in the way that it consist of short plumes on time scales on the order of seconds, whereas the three identified events have steady size distributions over at least an hour. [26] During the three event days, the Hysplit back trajectories during the nights came from the SW but during the days turned more southerly (Figure 7). On 29 June, the 6of11

7 Figure 6. (a) Aerosol particle number size distribution measured on board the Celtic Explorer during 29 June to 1 July 2006 and (b) corresponding total number concentration. The black circles highlight the periods when newly formed particles were detected. trajectory crossed over the Irish west coast; on 30 June, it did not appear to make it all the way to the coast. On 1 July, the trajectory passed over all of Ireland, and this is clearly seen in the features of the aerosol number size distribution (Figure 6) as the background aerosol concentration increased by almost an order of magnitude and the number size distribution changed from a typical bimodal marine to a more continental distribution. Nevertheless, also during this day, a new particle formation event was observed. [27] The origin of these new particles could not be narrowed down only to a single possibility. The hypotheses are listed below. [28] 1. The air masses all passed over Ireland before reaching the ship and can thus be coastal events similar to what was discussed in section 3.1. The main difference is that the air masses passed the Irish coast line very early in the morning and closer to a time of a high tide than a low tide. However, because of the uncertainties involved in back trajectory calculations and the mixing of the air parcel during transport, this option cannot be ruled out, especially for the event on 1 July. [29] 2. The particles were formed from plankton related emissions from the ocean, as proposed by O Dowd et al. [2002]. The lower limit of the particle modes were at 10 nm, which is an indication that the ship itself was most likely not located inside the plankton bloom. If there had been substantial precursor vapor emissions close to the ship, we would have detected the particle mode closer to the detection limit (3 nm) of the DMPS. As the particles were at 10 nm, it requires an area of very strong plankton bloom somewhere between the ship and the coast. Satellite images of surface water chlorophyll during this period were scarce because of frequent cloud cover, but earlier data had showed that this area was one of the most active and was the reason why this region was explored during these days. Rinaldi et al. [2009] presented water chlorophyll a concentration estimates measured on board the Celtic Explorer and showed that the region furthest SW of the area sailed during the three event days showed the highest readings, indicating a strong bloom upwind of the ship. [30] 3. During 29 June and 1 July 2006, there was a short period of rainfall around or during the periods when we saw the new particles. This may have cleaned the air from background particles and consequently decreased the condensation sink, thus allowing new particle formation to take place. At least during 29 June, the accumulation mode decreases during the event. Another rain related possibility is that the evaporating rain drops could have produced Figure 7. Hysplit back trajectories for 29 June to 1 July 2006 arriving at the coordinates of the Celtic Explorer at 500 m height above sea level. Dots on the trajectories are spaced to 1 h apart. 7of11

8 Figure 8. The particle growth rates determined from the DMPS data during three new particle formation events as a function of time of day. The dots represent the modal diameter of the peaks during each DMPS scan. supersaturated conditions with respect to some precursor compounds that can form new particles. [31] 4. The particles can also have been formed at higher altitudes and are mixed downward to the ship. This option is weakened by the fact that during 1 July, the event seemed to be stronger and thus implied that the initial particle formation happened under the influence of the terrestrial emissions. [32] Option 2, where the particles were formed in the presence of phytoplankton emissions, seems the most likely option out of the ones listed above. In addition, assuming a coastal source, the travel time from the coast to the ship (i.e., the growth time) varied greatly between the days. However, the new particles were detected at the same time every day implying that photochemistry was a driving factor during the events. During 29 June, the observed mode did not grow, but during the following 2 days, we were able to calculate the growth rates from the DMPS data. The results are presented in Figure 8. The danger with this method is that the growth rates can be biased since the estimate is based on data measured on a moving platform. If the ship was moving toward the particle formation region, and into the wind, the determined growth rates would be larger than they really were. Additionally, if the new particle formation happened only in a limited area, the ship might have moved in or out of regions downwind of the production zone. This would cause abrupt changes in the observed number size distributions. On 1 July, the growth rate changes during the event around the same time the ship s heading changed by 20 westward. [33] Although these events differ in many ways, we believe that they all can be of marine origin, and thus, we calculated an average GR of all the five GR in Figure 8, to get a mean GR of 3.2 nm h 1. Inclusion of the event Figure 9. Aerosol number size distribution during 13 June 2006 on board the Celtic Explorer. New particle formation is detected in the evening. 8of11

9 Figure 10. Back trajectory arriving at Mace Head during a marine new particle formation event where the air mass passed over the Celtic Explorer approximately 8 h earlier. on 29 June with GR 0, the average is 2.7 nm h 1. Although many uncertainties are involved, a best estimate for average marine GR based on our data is 3 nm h Clean Marine Air [34] The size distribution for 13 June is plotted in Figure 9. A mode of new particles was observed possibly already after noon, but it became more pronounced around This event lasted longer than those discussed in section 3.1, and in this case, the back trajectories show no continental contact during at least the last 3 days, implying that this is a purely marine new particle formation event. This is the first time that new particle formation on the open ocean in this part of the Atlantic has been observed, although Dall Osto et al. [2010] observed nucleation mode particles at Mace Head that they attributed to open ocean nucleation. As a fortunate coincidence during the event observed during 13 June, the air mass, passing over the Celtic Explorer during the event, continued on to Mace Head several hours later, as can be seen from Figure 10. Figure 11 shows ion size distributions for June measured by the AIS in Mace Head. They show a mode of particles present starting from the evening of 13 June, to around noon 14 June. The shape and duration of the events are very similar, indicating that the new particles indeed were transported to Mace Head. As these particles crossed the coast line mostly during night, the influence of possible coastal emissions was negligible. If these are the same particles as observed on the ship, the long duration of the particle formation suggests a very large area over which the formation took place. The wind speed on board the ship was measured to be 5 8 m/s during the event and following night, which means that an air parcel travels around km during 10 h. This gives an estimate of the spatial scale over which the formation happened. [35] The growth rates of these particles were estimated in two ways. In Figure 8, the evolution of the mode as measured on the Celtic Explorer is presented. As discussed in the previous section, this method yields an uncertainty due to the nonstationary measurements but nevertheless gives the best estimate for the initial growth of these particles. The sudden decrease in GR happened consecutively with a change in heading of about 50 eastward. The second option for calculating the GR is based on comparing size distributions at Mace Head and the Celtic Explorer. The air mass travels to Mace Head during the evening and night in roughly 10 h, and the observed mode at the ship is at 22 nm and in Mace Head 26 nm, giving an estimated growth rate of 0.4 nm h 1 even during night. However, the uncertainties in this calculation are high. 4. Conclusions [36] The coastline is a strong source for new aerosol particles, which also grow to climate relevant sizes [O Dowd et al., 2002; Whitehead et al., 2009]. In this study we explored coastal formation of nanoparticles at Mace Figure 11. The ambient ion number size distribution during June 2006, at Mace Head, showing a mode of new particles during the night, of a similar shape and duration as seen on board the Celtic Explorer the previous day (Figure 9). 9of11

10 Head in western Ireland and applied a concept called growth time that describes the elapsed time between the initial formation of the particles at the coastline and their subsequent sampling at a fixed measurement location. The location of Mace Head on a peninsula enabled us to acquire data with variable growth times. [37] To summarize our findings, the apple type nucleation events frequently observed in Mace Hade are fully consistent with the exposure of the measured air to a strong, very restricted source of nucleating vapors, such as the coast line during a low tide. For a single such source region, the mean size of the nucleated particles appears to increase with growth time, whereas the growth rate decreases. The mean size of the nucleated particles is also sensitive to the strength (or size) of the restricted source area. The presence of multiple restricted source areas of different strength around the coast of Mace Head could easily explain the mixedtype nucleation events that are most frequently observed in Mace Head [Vana et al., 2008]. These experimental observations were successfully reproduced in a box model. [38] As part of the MAP project a field study onboard the R/V Celtic Explorer was conducted. We observed new particle formation during four different days in the period 11 June to 5 July Three of these events took place during consecutive days and were linked to a variable coastal and continental influence. Several explanations were proposed, with the most likely one being that the particles were mainly of marine origin, with the precursor vapors having been emitted from biologically active areas with strong plankton bloom north/northwest of Ireland. The first two were most likely relatively free of anthropogenic influence, whereas the last day revealed particle number size distribution properties and higher background number concentrations related to terrestrial and anthropogenic influences. This influence was also seen in the newly formed particles, with clearly higher concentrations during the polluted event. One new particle formation event observed onboard the Celtic Explorer seemed to occur in a clean marine atmosphere over an area spanning several hundred kilometers west of Ireland. The average growth rate for all four events was estimated to be 3 nm h 1. [39] Acknowledgments. Financial support from the European Commission via project Marine Aerosol Production, MAP, FP6 STREP, GOCE is gratefully acknowledged, as well as EPA Ireland for the research support at Mace Head. NOAA Air Resources Laboratory (ARL) is acknowledged for the provision of the HYSPLIT transport and dispersion model and READY Web site ( used in this publication. References Aalto, P., et al. (2001), Physical characterization of aerosol particles during nucleation events, Tellus, Ser. B, 53, Berresheim, H., T. Elste, H. G. Tremmel, A. G. Allen, H. C. Hansson, K. Rosman, M. Dal Maso, J. M. Mäkelä, M. Kulmala, and C. D. O Dowd (2002), Gas aerosol relationships of H 2 SO 4, MSA, and OH: Observations in the coastal marine boundary layer at Mace Head, Ireland, J. Geophys. Res., 107(D19), 8100, doi: /2000jd Covert, D. S., V. N. Kapustin, T. S. Bates, and P. K. Quinn (1996), Physical properties of marine boundary layer aerosol particles of the mid Pacific in relation to sources and meteorological transport, J. Geophys. Res., 101, Dall Osto, M., et al. (2010), Aerosol properties associated with air masses arriving into the North East Atlantic during the 2008 Mace Head EUCAARI intensive observing period: An overview, Atmos. Chem. Phys., 10, Dal Maso, M., M. Kulmala, K. E. J. Lehtinen, J. M. Mäkelä, P. Aalto, and C. D. O Dowd (2002), Condensation and coagulation sinks and formation of nucleation mode particles in coastal and boreal forest boundary layers, J. Geophys. Res., 107(D19), 8097, doi: /2001jd Draxler, R., and G. Rolph (2003), HYSPLIT (HYsplit Single Particle Lagrangian Intregrated Trajectory) model, Web site ( noaa.gov/ready/hysplit4.html), NOAA Air Resources Laboratory, Silver Spring, Md, USA. Gagné, S., L. Laakso, T. Petäjä, V. M. Kerminen, and M. Kulmala (2008), Analysis of one year of Ion DMPS data from the SMEAR II station, Finland, Tellus, Ser. B, 60, Heintzenberg, J., W. Birmili, A. Wiedensohler, A. Nowak, and T. Tuch (2004), Structure, variability and persistence of the submicrometer marine aerosol, Tellus, Ser. B., 56, Hoppel, W. A. (1978), Determination of the aerosol size distribution from the mobility distribution of the charged fraction of aerosols, J. Aerosol Sci., 9, Hoppel, W. A., and G. M. Frick (1986), Ion aerosol attachment coefficients and the steady state charge distribution on aerosols in a bipolar ion environment, Aerosol Sci. Technol., 5, Iida, K., M. R. Stolzenburg, P. H. McMurry, and J. N. Smith (2008), Estimating nanoparticle growth rates from size dependent charged fractions: Analysis of new particle formation events in Mexico City, J. Geophys. Res., 113, D05207, doi: /2007jd Kerminen, V. M., K. E. J. Lehtinen, T. Anttila, and M. Kulmala (2004), Dynamics of atmospheric nucleation mode particles: A time scale analysis, Tellus, Ser. B., 56, Kerminen, V. M., T. Anttila, T. Petäjä, L. Laakso, S. Gagné, K. E. J. Lehtinen, and M. Kulmala (2007), Charging state of the atmospheric nucleation mode: Implications for separating neutral and ion induced nucleation, J. Geophys. Res., 112, D21205, doi: /2007jd Koponen, I., A. Virkkula, R. Hillamo, V. M. Kerminen, and M. Kulmala (2002), Number size distributions and concentrations of marine aerosols: Observations during a cruise between the English Channel and the coast of Antarctica, J. Geophys. Res., 107(D24), 4753, doi: / 2002JD Korhonen, H., K. E. J. Lehtinen, and M. Kulmala (2004), Multicomponent aerosol dynamics model UHMA: Model development and validation, Atmos. Chem. Phys., 4, Kuang, C., P. H. McMurry, and A. V. McCormick. (2009), Determination of cloud condensation nuclei production from measured new particle formation events, Geophys. Res. Lett., 36, L09822, doi: / 2009GL Kuang, C., I. Riipinen, S. L. Sihto, M. Kulmala, A. V. McCormick, and P. H. McMurry (2010), An improved criterion for new particle formation in diverse atmospheric environments, Atmos. Chem. Phys., 10, Kulmala, M., T. Petäjä, P. Mönkkönen, I. K. Koponen, M. Dal Maso, P. Aalto, K. Lehtinen, and V. M. Kerminen (2005), On the growth of nucleation mode particles: Source rates of condensable vapor in polluted and clean environments, Atmos. Chem. Phys., 5, Kulmala, M., H. Vehkamäki, T. Petäjä, M. Dal Maso, A. Lauri, V. M. Kerminen, W. Birmili, and P. H. McMurry (2004), Formation and growth rates of ultrafine atmospheric particles: A review of observations, J. Aerosol Sci., 35, Kulmala, M., and V. M. Kerminen (2008), On the growth of atmospheric nanoparticles, Atmos. Res., 90, Laakso, L., S. Gagné, T. Petäjä, A. Hirsikko, P. P. Aalto, M. Kulmala, and V. M. Kerminen (2007), Detecting charging state of ultrafine particles: Instrumental development and ambient measurements, Atmos. Chem. Phys., 7, Mäkelä, J. M., T. Hoffmann, C. Holzke, M. Väkevä, T. Suni, T. Mattila, P. P. Aalto, U. Tapper, E. I. Kauppinen, and C. D. O Dowd (2002), Biogenic iodine emissions and identification of end products in coastal ultrafine particles during nucleation bursts, J. Geophys. Res., 107(D19), 8110, doi: /2001jd McFiggans, G., et al. (2004), Direct evidence for coastal iodine particles from Laminaria macroalgae Linkage to emissions of molecular iodine, Atmos. Chem. Phys., 4, Merikanto, J., D. V. Spracklen, G. W. Mann, S. J. Pickering, and K. S. Carslaw (2009), Impact of nucleation on global CCN, Atmos. Chem. Phys., 9, Mirme, A., E. Tamm, G. Mordas, M. Vana, J. Uin, S. Mirme, T. Bernotas, L. Laakso, A. Hirsikko, and M. Kulmala (2007), A wide range multichannel air ion spectrometer, Boreal Environ. Res., 12, Modini, R. L., Z. D. Ristovski, G. R. Johnson, C. He, N. Surawski, L. Morawska, T. Suni, and M. Kulmala (2009), New particle formation 10 of 11

11 and growth at a remote, sub tropical coastal location, Atmos. Chem. Phys., 9, Mönkkönen, P., I. K. Koponen, K. Lehtinen, K. Hämeri, R. Uma, and M. Kulmala (2005), Measurements in a highly polluted Asian mega city: Observations of aerosol number size distributions, modal parameters and nucleation events, Atmos. Chem. Phys., 5, O Dowd, C. D. (2001), Biogenic coastal aerosol production and its influence on aerosol radiative properties, J. Geophys. Res., 106, O Dowd, C. D., and G. de Leeuw (2007), Marine aerosol production: A review of the current knowledge, Phil. Trans. R. Soc. A, 365, O Dowd, C. D., et al. (2002), Coastal new particle formation: Environmental conditions and aerosol physicochemical characteristics during nucleation bursts, J. Geophys. Res., 107(D19), 8107, doi: / 2000JD O Dowd, C. D., M. C. Facchini, F. Cavalli, D. Ceburnis, M. Mircea, S. Dececari, S. Fuzzi, Y. J. Yoon, and J. P. Putaud (2004), Biogenically driven organic contribution to marine aerosol, Nature, 431, O Dowd, C. D., J. L. Jimenez, R. Bahreini, R. C. Flagan, J. H. Seinfeld, K.Hämeri,L.Pirjola,M.Kulmala,S.G.Jennings,andT.Hoffmann (2002), Marine aerosol formation from biogenic iodine emissions, Nature, 417, O Dowd, C. D., Y. J. Yoon, W. Junkermann, P. Aalto, M. Kulmala, H. Lihavainen, and Y. Viisanen (2007), Airborne measurements of nucleation mode particles I: Coastal nucleation and growth rates, Atmos. Chem. Phys., 7, Pierce, J. R., and P. J. Adams (2006), Efficiency of cloud condensation nuclei formation from ultrafine particles, Atmos. Chem. Phys., 7, Pierce, J. R., and P. J. Adams (2009), Uncertainty in global CCN concentrations from uncertain nucleation and primary emission rates, Atmos. Chem. Phys., 9, Rolph, G. (2003), Real time Environmental Applications and Display system (READY), NOAA Air Resources Laboratory, Silver Spring, Md, USA. ( Sellegri, K., Y. J. Yoon, S. G. Jennings, C. D. O Dowd, L. Pirjola, S. Cautenet, C. Hongwei, and T. Hoffmann (2005), Quantification of coastal new ultra fine particles formation from in situ and chamber measurements during the BIOFLUX campaign, Environ. Chem., 2, Spracklen, D. V., et al. (2008), Contribution of particle formation to global cloud condensation nuclei concentrations, Geophys. Res. Lett., 35, L06808, doi: /2007gl Stolzenburg, M. R., P. H. McMurry, H. Sakurai, J. N. Smith, R. L. Mauldin III, F. L. Eisele, and C. F. Clement (2005), Growth rates of freshly nucleated atmospheric particles in Atlanta, J. Geophys. Res., 110, D22S05, doi: /2005jd Vaattovaara, P., P. E. Huttunen, Y. J. Yoon, J. Joutsensaari, K. E. J. Lehtinen, C. D. O Dowd, and A. Laaksonen (2006), The composition of nucleation and Aitken modes particles during coastal nucleation events: Evidence for marine secondary organic contribution, Atmos. Chem. Phys., 6, Vana, M., M. Ehn, T. Petäjä, H. Vuollekoski, P. Aalto, G. de Leeuw, D. Ceburnis, C. D. O Dowd, and M. Kulmala (2008), Characteristic features of air ions at Mace Head on the west coast of Ireland, Atmos. Res., 90, Vuollekoski, H., V. M. Kerminen, T. Anttila, S. L. Sihto, M. Vana, M. Ehn, H. Korhonen, G. McFiggans, C. D. O Dowd, and M. Kulmala (2009), Iodine dioxide nucleation simulations in coastal and remote marine environments, J. Geophys. Res., 114, D02206, doi: / 2008JD Wen, J., Y. Zhao, and A. S. Wexler (2006), Marine particle nucleation: Observations at Bodega Bay, California, J. Geophys. Res., 111, D08207, doi: /2005jd Whitehead, J. D., G. B. McFiggans, M. W. Gallagher, and M. J. Flynn (2009), Direct linkage between tidally driven coastal ozone deposition fluxes, particle emission fluxes, and subsequent CCN formation, Geophys. Res. Lett., 36, L04806, doi: /2008gl Wiedensohler, A. (1988), An approximation of the bipolar charge distribution for particles in the submicron size range, J. Aerosol Sci., 19, Yoon, Y. J., C. D. O Dowd, S. G. Jennings, and S. H. Lee (2006), Statistical characteristics and predictability of particle formation at Mace Head, J. Geophys. Res., 111, D13204, doi: /2005jd P. Aalto, M. Ehn, M. Kulmala, T. Petäjä, and H. Vuollekoski, Department of Physics, P O Box 64, FI 00014, University of Helsinki, Finland. (mikael.ehn@helsinki.fi) D. Ceburnis, R. Dupuy, and C. D. O Dowd, School of Physics and Centre for Climate and Air Pollution Studies, Environmental Change Institute, National University of Ireland Galway, Galway, Ireland. V. M. Kerminen, Finnish Meteorological Institute, Climate Change Unit, P O Box 503, FI Helsinki, Finland. G. de Leeuw, Netherlands Organization for Applied Scientific Research, The Hague, Built Environmental and Geosciences, PO Box 80015, 3508 TA Utrecht, Netherlands. M. Vana, Institute of Physics, University of Tartu, Ülikooli 18, Tartu, Estonia. 11 of 11