Over the last two decades, new particle formation (NPF), i.e., the
formation of new particle clusters from gas-phase compounds followed by their
growth to the 10–50 nm size range, has been extensively observed in the
atmosphere at a given location, but their spatial extent has rarely been assessed. In
this work, we use aerosol size distribution measurements performed
simultaneously at Ersa (Corsica) and Finokalia (Crete) over a 1-year period
to analyze the occurrence of NPF events in the Mediterranean area. The
geographical location of these two sites, as well as the extended sampling
period, allows us to assess the spatial and temporal variability in atmospheric
nucleation at a regional scale. Finokalia and Ersa show similar seasonalities
in the monthly average nucleation frequencies, growth rates, and nucleation
rates, although the two stations are located more than 1000
New particle formation (NPF) events have been widely observed in the atmosphere in different environments (Kulmala et al., 2004) from remote areas at high altitude or latitude to polluted environments in different climates (Pey et al., 2008; Manninen et al., 2010; Yli-Juuti et al., 2011; Cusack et al., 2013). However, the exact mechanism and chemical species involved in the NPF process are not fully identified, especially regarding the diversity of environments to consider. Thus, most global climate models still do not represent this process well, and they use parameterizations which are based upon a limited number of mechanisms and gaseous precursors, even though they predict that it may contribute to a significant fraction of condensation nuclei (CN) and cloud condensation nuclei (CCN) concentration at the global scale (Spracklen et al., 2008; Merikanto et al., 2009; Makkonen et al., 2012).
The different features of NPF events (frequency, intensity, duration) may be influenced not only by meteorological variables
(temperature, relative humidity, and solar radiation) (Birmili et al., 2003; Jeong et al., 2004; Sihto et al., 2006; Young
et al., 2007) but also by the availability of gaseous precursors, regarding both their nature and their amount. It is
thus necessary to describe the occurrence and characteristics of NPF over a large variety of environments and assess to
what spatial extent these features can be applied. Although the characteristics of the NPF events have often been
documented in the literature (Hirsikko et al., 2007; Manninen et al., 2010; Yli-Juuti et al., 2009, 2011), analyses
dedicated to their spatial extent are rarer. This might be explained by the fact that such studies require airborne
measurements (Crumeyrolle et al., 2010; Rose et al., 2015a) or multi-site datasets. Such datasets have been analyzed by Vana
et al. (2004) and Hussein et al. (2009), who reported that NPF could take place in the form of regional events over up to
a thousand kilometers in Scandinavia, and at least 500 km over the western coast of South Korea (Kim
et al., 2016). Likewise, Dall'Osto et al. (2013) observed regional NPF events occurring in the northeast of Spain. Using
a similar methodology, Crippa and Pryor (2013) observed horizontal extents of
a hundred kilometers for the NPF process in USA and Canada. They also pointed
out a significant variability in the NPF characteristics (formation and
growth rates) within these large-scale events, suggesting that local influences could add to favorable synoptic conditions. In order to
allow for the analysis of the horizontal extent of NPF on a single-station
dataset, different methods based on air mass back-trajectory analysis and
particle growth rates have also recently been proposed (Kristensson et al., 2014;
Rose et al., 2015b). The Nanomap tool developed by Kristensson et al. (2014)
was reported to allow the identification of nucleation areas up to
500
These studies dedicated to the analysis of the horizontal extent of NPF were mainly conducted above continental regions. Similar analyses in marine environments are crucially missing although they are of high interest, as it has previously been shown that, in such pristine environments, cloud properties could be significantly impacted by changes in the aerosol loading (Tao et al., 2012; Koren et al., 2014; Rosenfeld et al., 2014). Although the Mediterranean area is particularly sensitive to the future evolution of atmospheric pollutants and climate change, only a few studies related to NPF in this area have been reported so far. Intensive campaigns were conducted on the eastern Spanish coast, in Barcelona and at the Montseny site (Pey et al., 2008; Cusack et al., 2013), while long-term measurements are performed at the Finokalia (Crete) station (Kalivitis et al., 2008, 2012, 2015; Manninen et al., 2010; Pikridas et al., 2012), where NPF event days are close to 30 %. The Mediterranean Basin is at the cross section of many different influences: there is a strong anthropogenic influence from densely populated coastal zones, in addition to marine and dust sources, as well as with emissions from Mediterranean forests and shrublands that emit both terpenes and isoprene. This geographical area is particularly exposed to high solar radiation compared to the rest of Europe; thus, we expect a strong contribution from photochemical processes.
In the framework of the projects CHARMEX-ADRIMED (Mallet et al., 2016) and CHARMEX-SafMed, a large coordinated effort has been recently conducted to better characterize the physicochemical properties of the Mediterranean atmosphere. Measurements were conducted at ground stations on Mediterranean islands, such as Crete (Finokalia) and Corsica (Ersa) for an extended period of the years 2013–2014 and Mallorca (Cap d'Es Pinar) for several weeks during 2013. Forty research flights were also performed during the summers of 2013 and 2014. This vast dataset gave us a unique opportunity to characterize the spatial extent of the NPF process in the Mediterranean Basin. In this paper, we first report the long-term analysis of NPF event characteristics observed at Ersa (from May 2012 to August 2013) and Finokalia (from January to December 2013) using size distribution measurements in order to assess the large-scale space and time variability in NPF. We then focus our study on the special operation period (SOP) that took place during summer 2013. During this SOP additional measurements were performed in Mallorca (from 3 July to 12 August 2013) and aerosol particle size distributions and concentrations were measured onboard the ATR-42, which allowed for a deeper analysis of the horizontal and vertical development of the NPF process at daily scale.
Locations of the stations: Ersa (Corsica), Finokalia (Crete), and Cap d'Es Pinar (Mallorca). Aircraft flight paths from 30 July and 1 August are also shown.
Ground-based aerosol measurements reported in this work were performed at the Finokalia station (Crete) from January to December 2013, at the Ersa station (Corsica) from May 2012 to August 2013, and at the Cap d'Es Pinar station (Mallorca) from 3 July to 12 August 2013 (Fig. 1). Within these measurements periods, some gaps occurred in the Finokalia dataset (from 5 September to 15 October 2013) due to participation of the instrument in the ACTRIS (Aerosol Clouds and Trace gases Research Infrastructure) network mobility particle size spectrometer workshop, and in the Ersa dataset (from 1 September to 31 October 2012) because of instrumental failures.
The Finokalia station (35.24
The Ersa station is located on the northern tip of the island of Corsica, at Cape Corsica (43.00
The Cap d'es Pinar station is located on the northeastern side of the island of Mallorca (39.88
Airborne measurements were carried out onboard the ATR-42 French research aircraft operated by SAFIRE (Service des Avions
Français Instrumentés pour la Recherche en Environnement). Figure 1 shows the aircraft trajectory during the
flights performed on 30 July and 1 August, which are investigated in the next sections of the present work. The aerosol
size distribution in the 20–485
From ground-based observations, measurement days were classified according to Dal Maso et al. (2005) into four categories:
events days, including classes I and II; undefined; and non-events days. Class I events are characterized by a strong
increase in sub-25
Particle formation and growth rates are key entities to assess the strength of events belonging to Class I and II. While
formation rates (
Growth rates (GRs) were calculated from the SMPS nucleation-mode concentrations (16–20
From this growth rate, we derived the total particle formation rate at 16
The goal of this first section is to provide an overview of the seasonal variability in NPF in the Mediterranean area, and some insights into the spatial homogeneity of the NPF occurrence over the basin.
Classification of measurement days in Ersa and Finokalia (after filtering bad data).
Monthly mean NPF frequencies at Finokalia and Ersa.
Monthly classification of the measurement days into event (I and II), undefined, and non-event categories in Finokalia and Ersa.
The yearly average NPF frequencies, calculated as the number of event days over the total number of measurement days, are
very similar at Finokalia and Ersa, being 36 % (109 events) and 35 % (96 events), respectively
(Table 1). A comparable value was reported by Pikridas et al. (2012) at Finokalia, with a yearly average frequency of
The classification of the event days into the different categories (Fig. 3 and Table 1) shows that the occurrence of type I events in Finokalia follows the same seasonal variation as the total NPF frequency, being maximum during the spring season (up to 26 % of all days). This indicates that spring is favorable to both formation of new particles and their growth to larger sizes. Type II events are annually the most frequent, representing between 13 and 31 % of all measurement days with no clear seasonal variation. In contrast, undefined days are not frequently observed in Finokalia, around 9 % on average. Very similar features are observed in Ersa: type I events show the highest frequency of occurrence during spring and summer (up to 32 % of all days in August), while they represent less than 10 % of the measurement days during winter. The frequency of occurrence of type II events is on average 19 %, with no clear seasonal variation.
Annual median formation rates, growth rates, and annual CS in Ersa and Finokalia. Percentiles are also reported as additional information.
Particle formation and growth rates were calculated for type I events in
order to characterize the strength of the events observed at the two
stations. The yearly median particle growth rates in the range
16–20
Annual variation of particle growth rate calculated for the
range 16–20
The yearly median particle formation rates (
Annual variation of the 16 nm particle formation at Ersa and Finokalia for type I events. Small dots represent all values, while large dots stand for median values.
It is worth noting that in Ersa, even though NPF frequencies are lower in autumn compared to spring, particle formation
rates are comparable. This last observation suggests that, despite being less frequent, favorable conditions for NPF can
be found during autumn and lead to events with the same intensity as in spring, when radiation and biogenic emissions are
on average higher compared to the rest of the year (Manninen et al., 2010). The seasonal variation of nucleation
frequency, nucleation rates, and growth rates is most likely related to the availability of condensable gases. The amount of
such precursors results from the balance between a combination of emissions and radiation, which favor their production,
and their loss onto preexisting particles. In order to assess the influence of the preexisting aerosol population on NPF,
we calculated the condensational sink (CS) according to Pirjola et al. (1999). The CS was first derived from SMPS
measurements for the whole measurement period at both stations and was finally averaged over the 2 h period prior to
the onset of NPF events. On non-event days, the CS was averaged over the 2 h time period prior to the time at which
NPF is triggered on event days, i.e.,
Median values of condensation sink (CS) reported separately for event and non-event days in Finokalia and Ersa.
The CS has a strong seasonal cycle with a clear maximum during summer at both stations. This observation may explain the
lower NPF frequencies, formation rates, and growth rates that are on average observed during this season, which otherwise
shows high radiation (Fig. S1), and most probably high biogenic emissions. In addition, the CS is on average higher during
non-event days at both stations. This confirms that the CS is likely a limiting factor for the occurrence of NPF at these
stations. This has already been pointed out by Kulmala et al. (2005), Hamed et al. (2010), and Manninen et al. (2010) for
several boundary layer stations in Europe, including both industrialized locations and more pristine areas, such as boreal
forest. One should, however, note that during spring months (especially March and April), median CS is similar on event and
non-event days. This observation suggests that, during this period, the strength of precursors emissions together with
radiation might be driving the occurrence NPF to a major extent. Also, the CS is on average higher in Finokalia,
especially during spring and summer, with monthly CS twice as high compared to Ersa. It is worth noting that large
particles up to 848
SMPS particle number size distribution in
Based on the previous observations, Finokalia and Ersa show similar seasonality in the average nucleation frequency,
growth rates and nucleation rates although the two stations are more than 1000
In this section, we focus on the special observation period (SOP) that took place from 3 June to 12 August in the frame of the CHARMEX project. During this period, number size distribution measurements were additionally conducted at the Mallorca station (Cap d'Es Pinar).
Figure 7 shows the SMPS particle size distributions recorded at the three ground-based stations during the SOP. From this synoptic overview, we clearly observe similar trends in the evolution of the particle size distributions in Ersa and Cap d'Es Pinar, with three distinct NPF periods during which NPF events occurred daily over several days (first period from 4 to 9 July, second period from 28 July to 3 August, and third period from 9 to 12 August) (see Table S1 in the Supplement). This observation would confirm the spatial extent of NPF events at a large scale. However, these periods of intense NPF activity were not observed in Finokalia, where both the occurrence and strength of NPF events seem to be more homogeneous over the SOP. These contrasting observations might be explained by an environmental contrast between the eastern and western part of the Mediterranean Basin.
As reported in Table S1, during this 41-day period, NPF was observed to occur at one station (at least) on 23 days. Among these 23 event days, 8 events were observed on the same day on two stations at least. This frequency of simultaneous NPF events occurrence is very similar to the one observed at Korean coastal sites (5 out of 21 observation days, Kim et al., 2016). NPF was detected at all sites on 9 August, and three events were reported on the same day for each of the station pairs Ersa–Finokalia and Ersa–Mallorca, and one event for the pair Finokalia–Mallorca. In order to further investigate the link that might exist between the events observed at the three stations, we first chose to focus our analysis on 3 days that belong to the three different NPF periods identified: 5 July, 29 July, and 9 August are presented as case studies. Type one events were observed in Ersa and Cap d'Es Pinar on those specific days, thus allowing for particle formation and growth rate calculations, and further direct comparison of event intensity at these two sites.
Average growth rates and formation rates computed for the three case studies at Ersa and Cap d'es Pinar.
We calculated the total formation rate of 20
Temporal evolution of the particle concentrations in the size range
11–15
Back trajectories of air masses sampled in Ersa and Cap d'Es Pinar on
9 August at
On 5 July, although NPF occurs both at Ersa and Cap d'Es Pinar, the time evolution of particle concentrations are very different from one site to the other. Particles of the smallest size range are detected in the morning at Ersa, but only later in the afternoon at Cap d'Es Pinar, and at larger sizes and lower concentrations (Fig. S2). The 24 h air mass back-trajectory analysis (HYSPLIT transport and dispersion model; Draxler et al., 2003) shows that air masses arriving at both stations are of northerly origin (Fig. S3). Hence, it is unlikely that particles formed during the NPF event detected at Ersa in the morning have been transported west and detected later in the afternoon at Cap d'es Pinar.
In order to further evaluate the spatial extent of nucleation, we estimated for each site the distance between the station
the place where nucleation was initially triggered upstream of the station. The method we used is based on the time evolution
of the aerosol size distribution and was previously described by Rose et al. (2015b). We assumed that 20
On 5 July, previous calculations led to distances of at least 9
In Finokalia, both for 5 and 29 July, significant
On 9 August, newly formed particles are detected in air masses originating from the near southern area in Ersa and from
northwestern sector in Cap d'Es Pinar (see Fig. 9). The concentration of particles measured in the first SMPS size channels
in Ersa (11–15
As shown in Fig. 9 for Cap d'Es Pinar, the place where nucleation initially occurred is at least 49
The three case studies showed that NPF events could be detected, with some time offset, on two remote stations separated by several hundred kilometers in the Mediterranean area. In particular for the case of 9 August, the fact that these events can be detected in air masses from different origins suggest that the NPF is, for both sites, initiated above the sea, either in the marine boundary layer or higher in the free troposphere. In any case, the NPF process is likely not subject to the availability of precursors that would be specific to the air mass type reaching the sites. It could rather depend on synoptic meteorological conditions at the European scale, including low condensational sinks following precipitations periods. Indeed, the analysis of the meteorological conditions along back-trajectories shows that precipitation did occur prior to their arrival at both stations on 29 July (during the passage of low-pressure systems), but not on the two other case studies. The minimum areas that we determined for nucleation onset at both sites did not overlap. However, the estimates we obtained are some lower limits of the actual values, and there are no elements which could justify that the NPF was interrupted between both sites. Airborne measurements will be used in the next section to further investigate this aspect. In addition, these flights will allow an analysis regarding the origin of the clusters and their precursors, from the marine boundary layer or from the upper levels of the atmosphere, as previously shown by Rose et al. (2015a).
Among the 11 flights performed during the SOP, particles in the lowest size range (
The first event to be investigated was observed on 30 July. Regarding aircraft measurements, the analysis was focused on
the flight legs performed at constant altitude and during which
SMPS size distributions measured at Ersa (left panel) and onboard
the ATR-42 at high altitude (
In order to explore the link that may exist between the events detected simultaneously from the aircraft and from the
ground, we first investigated the origin of the air masses. Figure 10b shows the 72 h back-trajectories of the air masses
sampled by the ATR-42 every 10
In addition, Fig. 11 shows the evolution of the particle size distributions measured onboard the ATR-42 and at Ersa. The
spectra are color-coded according to the position of the aircraft indicated in the insert included in the middle panel of
Fig. 11. At Ersa, the shape of the particle size distribution remains similar during the whole measurement period, with
a nucleation mode around 20–25
Ground-based (left panel) and airborne (right panel) SMPS size distributions measured on 1 August. The color coding of the spectra corresponds to the location of the aircraft, as shown on the insert of the left panel.
These last observations, together with the air mass back-trajectory analysis shown in Fig. 10b, suggest that for this first event, new particles were initially formed at low altitude over the continent and further transported above the sea to be finally detected over a large area, and more especially in Ersa. Decreasing particle concentrations observed while moving further off the continent make the hypothesis of new small particle formation from an additional marine source less probable; instead, they show the effect of dispersion process that may have taken place during particle transport.
The second event included in this analysis was observed on 1 August. Compared to the previous case study, the flight was
performed over a larger area (
Ratio of nucleation-mode diameters measured onboard the ATR-42 over that calculated in Ersa as a function of the distance between the aircraft and Ersa on 1 August. The color coding of this scatter plot matches with the location of the aircraft showed in the insert of the left panel of Fig. 13.
The evolution of the particle size distributions together with the location of the aircraft is shown in Fig. 13. Unlike
during the flight performed on 30 July, the shape of the distributions measured onboard the ATR-42 remains similar during
the whole measurement period despite the changing origin of air masses. In contrast, the shape of the particle size
distributions measured at Ersa shows significant variability. In particular, the nucleation mode displays increasing
diameters from 20 to 30
In order to further investigate the origin of the nucleation-mode particles and the connection that may exist between ground-based and airborne measurements, we compared the diameters of the corresponding nucleation modes. For that purpose, Fig. 14 shows the ratio of the nucleation-mode diameter obtained onboard the ATR-42 over that from Ersa as a function of the distance between the aircraft and the station. This ratio is in the range 0.6–1.2, with on average decreasing values while increasing the distance between the two measurement points. Nucleation-mode diameter getting smaller along the air mass back-trajectory above the sea could be the result of intense inputs of nucleated particles initially below the SMPS size detection limit and feeding the nucleation mode as they grow, as confirmed by the detection of 3–10 nm particles from the ATR-42. In this particular case, particles detected in the nucleation mode observed onboard the ATR-42 would be the result of an event occurring above the sea from marine precursors, which combines with a preexisting particle mode.
We investigated the occurrence of NPF in the Mediterranean area using particle size distributions measured at three ground-based stations (Ersa, Cap d'Es Pinar, and Finokalia) as well as airborne measurements performed in 2013 in the frame of the CHARMEX-ADRIMED and CHARMEX-SafMed projects.
The analysis of long-term datasets from Ersa and Finokalia first revealed similar features, although the two stations are
more than 1000
This investigation, initially performed at a monthly resolution, was downscaled in a second step at the daily resolution
over a two months period, in order to further assess the simultaneity of NPF over a large part of the Mediterranean
Basin. Three simultaneous nucleation periods of several days appeared clearly for Ersa and Cap d'Es Pinar, and less clearly
at Finokalia. NPF formation was observed to occur simultaneously at least at two of the three stations on 8 days over the
41 days of observation, which confirms the frequent occurrence of regional-scale NPF events in the Mediterranean
area. Three case study events were selected within these three distinct NPF periods for a more detailed analysis. These
three case studies showed that NPF events could be detected, with some time offset, on two remote stations separated by
several hundred kilometers in the Mediterranean Basin, without the stations being directly linked to each other within
a single air mass trajectory. While featuring local characteristics, the occurrence of NPF events was likely not dependant
on the availability of precursors that would be specific to the air mass type reaching the sites, but rather on synoptic
meteorological conditions at the European scale. Komppula et al. (2006) also concluded from observation from two different sites 250
The case studies also showed that despite the fact that nucleation monthly frequencies, monthly nucleation rates, and growth rates had similar seasonal variations in Ersa and Finokalia, different behaviors were observed on a daily basis between the western and eastern Mediterranean basins. Again, the combination of favorable synoptic conditions and seasonal variations in general emission schemes may favor a seasonal behavior of the NPF frequency and characteristics, but local conditions are modulating the general behavior of regional NPF.
Airborne measurements were finally used to further investigate the horizontal and vertical extent of NPF, as well as to determine the origin of the clusters and their precursors. Two case studies were again selected within the NPF periods identified previously from ground-based observations, during which newly formed clusters were observed onboard the ATR-42 and from Ersa on the same day. Airborne measurements confirmed the regional spatial extent of NPF events, and further showed regional NPF events can have different sources. The selected events depicted contrasting situations where particles were initially probably formed above the continent for one of them, both in the boundary layer and in the free troposphere, and probably formed above the sea for the other.
This work, together with the previous study by Rose et al. (2015a), demonstrates the occurrence of NPF in the Mediterranean Basin, thus highlighting the possibility for the process to be triggered above open seas. These results are of great interest to improve the parameterizations of nucleation in models, which actually only consider a limited number of precursors, commonly including sulfuric acid and ammonia but excluding those more specifically emitted in the marine atmosphere. Model predictions would also benefit from the analysis of the vertical extent of the NPF process provided in these studies. Besides the identification of preferential altitudes for the occurrence of the process, these results aid in understanding the transport of the newly formed clusters and their precursors between the boundary layer and the free troposphere. Future studies should focus on understanding the chemical precursors that contribute to these new particle formation processes.
SMPS data from ERSA and Cap d'Es Pinar are available at
The authors declare that they have no conflict of interest.
This article is part of the special issue “CHemistry and AeRosols Mediterranean EXperiments (ChArMEx) (ACP/AMT inter-journal SI)”. It is not associated with a conference.
This study was performed with the financial support of the French National Research Agency (ANR) project ADRIMED (contract ANR-11-BS56-0006) the ANR project “SAf-Med” (grant number: SIMI-5-6 022 04) and is part of the ChArMEx project supported by ADEME, CEA, CNRS-INSU, and Météo-France through the multidisciplinary programme MISTRALS (Mediterranean Integrated Studies aT Regional And Local Scales). The financial support for the ACTRIS Research Infrastructure Project by the European Union's Horizon 2020 research and innovation program under grant agreement no. 654169 and previously under grant agreement no. 262254 in the 7th Framework Programme (FP7/2007–2013) is gratefully acknowledged. Edited by: Evangelos Gerasopoulos Reviewed by: two anonymous referees