This work presents for the first time a classification of shrinkage events based on the aerosol processes that precede them. To this end, 3.5 years of continuous measurements (from 2009 to 2012) of aerosol size distributions, obtained with a Scanning Mobility Particle Sizer (SMPS) at an urban background site in Southern Europe, have been interpreted.
48 shrinkage events were identified and analysed, all occurring
during spring and summer when the atmospheric conditions are more
favourable for their development. In this study the shrinkage events
took place mostly towards the end of the day, and their occurrence
could be associated to atmospheric dilution conditions and
a reduction in photochemical activity. The shrinkage rate (SR)
varied between
Following the proposed methodology, three groups of events have been
identified: Group I (NPF
Although this analysis has confirmed that the triggering of shrinkage events is clearly linked to the atmospheric situation and the characteristics of the measurement area, this classification may contribute to a better understanding of the processes involved and the features that characterize shrinkage events.
Particle size is one of the most important properties of atmospheric aerosols (Hinds, 1999; Seinfeld and Pandis, 2006). This property determines the aerosol's physical and chemical characteristics such as its hygroscopicity (Sjogren et al., 2008) or optical properties, and consequently the processes wherein aerosols are involved, such as the ability to form Cloud Condensation Nuclei (CCN) (Dusek et al., 2006; Henning et al., 2002) or the processes of absorption/scattering of solar radiation in the atmosphere (Seinfeld and Pandis, 2006). Thus, aerosol size is a key property in the context of the impact of atmospheric particulates on human health (Englert, 2004), air quality (Charlson, 1969; Quan et al., 2014) and climate change (Lohmann and Feichter, 2005; IPCC, 2013).
Initially, the aerosol size depends on its genesis. However, the transformations that aerosols may undergo during their lifetime in the atmosphere will affect the particle size evolution (Dubovik et al., 2002).
Traditionally, the studies that analyze changes in aerosol size in the atmosphere have focused mainly on the mechanisms that condition and determine their growth. The environmental conditions, such as photochemical activity and relative humidity, which give rise to coagulation and condensation processes are responsible for the increase in the particle size during the course of different scenarios, such as new particle formation (NPF) (Zhang et al., 2012; Zhu et al., 2013; Guo et al., 2012) or combustion processes (Adler et al., 2011; Janhäll et al., 2010; Saarikoski et al., 2007). However, because of the many variables involved in the triggering of particle size reduction under real atmospheric conditions, the number of works which focus on this shrinkage phenomenon is scarce. In fact, the published works on this process are centered mainly on primary emissions of organic particles generated by traffic, due to the high content of semivolatile organic compounds in traffic emissions. These studies are more limited to experimental testing in laboratory settings, or predictive models supplemented by measurements under atmospheric conditions (Robinson et al., 2007; Shrivastava et al., 2006). Particle size reduction in traffic emissions has also been seen as a consequence of the atmospheric vertical dispersion processes (Dall'Osto et al., 2011).
In this last context, all authors agree on the fact that the processes of dilution, evaporation and chemical aging of semivolatile organics appear to be the mechanisms that contribute to the reduction of the particle size (Donahue et al., 2006; Shrivastava et al., 2006; Robinson et al., 2007; Hitchins et al., 2000). These mechanisms control the partitioning of semivolatile species between the particle phase and gas phase in order to maintain the balance between both phases.
Works found in the literature focusing on particle size reduction in the atmosphere have referred to this process as shrinkage (Young et al., 2013; Cusack et al., 2013; Skrabalova et al., 2015), aerosol growth reversals (Yao et al., 2010; Skrabalova et al., 2015) or shrunken particle size (Yao et al., 2010). All of these terms refer to a progressive reduction of the particle size as a result of the displacement of condensed semivolatile species from the particle phase to the gas phase, due to thermodynamic variations in the ambient conditions for a sufficiently long time that allows for its observation.
These processes have been identified in locations widely-separated geographically and therefore in measurement areas with rather different climates. On the Asian continent, Yao et al. (2010) documented these processes at a coastal suburban site and Young et al. (2013) observed shrinkage processes at a coastal, urban and downwind site. Both authors identified the shrinkage processes in subtropical climates. On the American continent, Backman et al. (2012) detected these processes at an urban background site also in a subtropical climate. In Europe, Cusack et al. (2013) studied these processes at a regional background site with a Mediterranean climate and Skrabalova et al. (2015) identified shrinkage processes at the urban background station in a continental climate. In some of these works, these processes have been identified mainly during NPF, particularly in the growth phase of newly nucleated particles (Yao et al., 2010; Young et al., 2013; Cusack et al., 2013; Skrabalova et al., 2015), but some authors have also documented them in the absence of NPF (Cusack et al., 2013; Backman et al., 2012). In these studies, all cases were identified during measurement campaigns or periods of continuous measurements ranging from seven to 24 months.
In these articles aerosol growth reversals have been attributed to
changes in environmental conditions. The following three factors
have been identified as causing these processes:
Wind speed: atmospheric dilution caused by an increase in the wind speed
triggers changes in the concentrations of the atmospheric gaseous chemical
compounds and consequently the partitioning of the semivolatile species from
the particle phase to the gas phase in order to maintain the balance between
both phases. Air temperature: an increase in the ambient temperature facilitates the
evaporation of water and/or condensed semivolatile species which can produce
a reduction in the particle size. Photochemical activity: the degree of photochemical oxidation modifies
the availability and concentration of chemical species in the atmosphere
and, consequently, the distribution of chemical species between particle and
gas phases.
This paper proposes the first known classification of shrinkage
processes and presents the results obtained for a 3.5
This study has been carried out at an experimental site located
in the CIEMAT facilities (40
The Madrid region is surrounded by the Sierra de Guadarrama
mountain range to the north and by Sierra de Ayllon to the
northeast, both belonging to the Central System and situated
approximately at a distance of 50–70
The normal climatological values recorded in these stations reflect a Continental-Mediterranean climate (Köppen classification) influenced by urban features.
SMPS measurements in this study were obtained between 2009 and
2012. This instrument consists of a Differential Mobility
Analyser (TSI-SMPS: DMA 3081) connected to a Condensation
Particle Counter (CPC; TSI Model 3775). The SMPS allows to
measure the submicron aerosol fraction into 107 channels
(14–700
The equipment was checked and maintenance activities were carried out frequently throughout the whole study period. Furthermore, the response of the equipment was verified during the intercomparison campaigns of the Spanish Network on Environmental DMAs (REDMAAS, in its Spanish acronym) that took place from 2010 to 2012 (Gómez-Moreno et al., 2010, 2015).
The total number of particles (
The monthly data coverage obtained for the 3.5
Data loss during the measurement period was due to regular calibration, normal maintenance activities, equipment breakdown or transfer of the equipment to another measurement site. It is necessary to point out that, from June to December 2012, technical problems in the first nine channels of the SMPS were detected, and therefore those data have not been taken into account in the data processing.
A permanent meteorological station 52
To characterize the air mass composition arriving at the sampling
point, a Differential Optical Absorption Spectrometer (DOAS)
provided data of gaseous pollutants, such as NO,
Furthermore, standard information on air quality in the
measurement area was provided by the suburban station of Casa de
Campo (3
The methodology proposed by Dal Maso et al. (2005) to identify NPF and subsequently described in detail by Kulmala et al. (2012) has been adapted in this work to identify and study the shrinkage processes.
This methodology allows for the observation of the evolution of aerosol size distributions in the atmosphere by surface plots representing the particle size distributions as a function of time. These types of graphs were made on a daily basis (24 h), from 00:00 UTC on the first day to 00:00 UTC on the next day for the whole measurement period.
Given the data coverage of all the measurements that were
collected, thus giving an extensive and representative database,
a classification of shrinkages could be carried out. Based on
a visual analysis of the daily surface plots, the aerosol growth
reversals have been categorized into three groups according to
the processes that precede them:
Group I, NPF Group II, aerosol growth process Group III, pure shrinkage events: shrinkages that took place in the absence of a specific previous process.
In this paper, the authors have considered an event as the sum of
the process that precedes the shrinkage (NPF or aerosol growth
process) and the shrinkage process itself.
In the identification of these events, the evolution of the aerosol size distribution and the meteorological and air mass changes have been taken into account in order to identify mixtures of air masses, which may be responsible for “apparent shrinkages”. Most of the identified events showed a uni-modal size distribution.
Changes in aerosol concentrations due to dilution processes have also been studied. In this paper, the authors have considered the presence of dilution when the ratio between the two event phases (NPF or aerosol growth process phase vs. shrinkage phase) was higher than 10 %, coinciding with the measurement uncertainty established in the ACTRIS SMPS standards (Wiedensohler et al., 2010).
Additionally, the growth rate (GR) during these events was
calculated as outlined by Kulmala et al. (2012). The particle
shrinkage rate (SR) was estimated using the same equation as for
GR, with the resulting value being negative. The calculation was
made from the mode/s of the aerosol size distributions averaged
every 15 min (
Finally, the calculation of the condensation sink (CS) and an
estimation of
The aerosol condensation sink (CS) determines how rapidly molecules will condense onto pre-existing aerosols and depends strongly on the shape of the size distribution (Pirjola et al., 1999; Lehtinen et al., 2003). NPF takes place during periods in which the CS is low because it indicates that the concentration of condensable gases present in the atmospheric is high.
The CS has been calculated according to Kulmala et al. (2001) and
is defined by the expression (1):
Sulfuric acid is formed from
Many authors suggest that gas-phase sulphuric acid is the main initiating agent of nucleation processes (Fiedler et al., 2005; Kusaka et al., 1998; Kulmala and Laaksonen, 1990; Kulmala et al., 2000), which makes its study of utmost importance during NPF.
In this work, sulfuric acid in the gas phase
The estimation of
A total of 48 shrinkage events were observed during the study
period. The number of cases identified was irregularly distributed
over the four years. In 2010 and 2012, the number of shrinkages
identified was 21 and 16 respectively, whereas only seven and four
cases were observed in 2009 and 2011 respectively (Fig. 3). The
difference in the number of cases identified between both groups
of years did not seem to be motivated by bias of the database as
a consequence of the data loss. The study of the evolution of the
monthly and daily averages of the meteorological parameters
recorded during the four years of study showed no significant
interannual variations, nor for the annual averages (temperature
of
The occurrence of shrinkage has been detected only during spring and summer seasons, in the months of May, June, July and August, when the necessary conditions occur. On the one hand, the higher solar irradiance during spring and summer enhances photochemical activity. On the other hand, higher photosynthetic activity results in higher emissions of biogenic volatile organic compounds (BVOCs) from vegetation. Thus, during these periods there was increased production and availability of gaseous chemical compounds (precursors) in the atmosphere which facilitate NPF and also the growth of preexisting particles. This process has been described previously in some studies (Dal Maso et al., 2009; Eerdekens et al., 2009). Moreover, the atmospheric dynamics in the study area during these months, mostly local air flows and recirculation processes, gave rise to an increased residence time of aerosols in the atmosphere, therefore making the aerosols prone to suffer more changes during their prolonged atmospheric lifetime. This was evidenced by the significant number of NPF and growth processes observed during the period studied, which often took place on the same days as the shrinkage events.
The greater number of shrinkages observed in June and July
compared to May and August (Fig. 3) was probably due to the
predominance of the optimal environmental conditions during June
and July (average meteorological values for June and July compared
to May and August during the 3.5 years of study: temperature
The highest number of cases observed corresponded to pure
shrinkage events whereas the less frequent were those of aerosol
growth process
A summary of the five shrinkage studies found in the literature, and the present work, is given in Table 1. All shrinkages have been identified under temperate climates, with the exception of Skrabalova et al. (2015). The climatic characteristics of each region determined the temporal variability observed in these processes. However, the occurrence of particle shrinkage was mainly observed on the warm seasons.
Except for those of Cusack et al. (2013), all cases have occurred in measurement areas with a clear influence of anthropogenic emissions. Furthermore, the shrinkage events took place in the middle of the day (around 12:00 UTC) (Fig. 4a), contrasting with the cases identified in this study where the shrinkage phases were triggered around 18:00 UTC (Fig. 4b).
The studies related to aerosol shrinkages concluded that atmospheric dilution and the high ambient temperature were found to be the main causes of all these processes (Table 1), and exceptionally a decrease in photochemical activity as Yao et al. (2010) and Skrabalova et al. (2015) pointed out in their papers.
In the present work, the atmospheric dilution was the leading
cause of the reduction in particle size, the 80 % of the
shrinkage processes occurred under a wind speed that exceeded
4
As a consequence of the particle size reduction, a displacement of particle concentrations towards smaller size modes was observed. Furthermore, in some case studies, the shrinkage was accompanied by a reduction in the concentration of total number of particles, which exceeded 25 % for some events.
In the following sections, the formation patterns of each type of shrinkage event outlined previously will be discussed based on a selected case study chosen as an example.
Shrinkage processes were identified during the growth phase of the
newly nucleated particles in 17 cases. The shrinkages were mainly
observed during NPF of type Ia, and a minority in NPF of type Ib,
with 13 and 4 cases, respectively. The duration of these events
ranged between 7 and 14.75
These types of events typically started between 09:00 and 12:00 UTC (Fig. 4a), when the conditions were suitable for NPF development i.e. low concentrations of preexisting particles in the atmosphere (low CS); high solar activity; and low relative humidity (Hamed et al., 2011). In addition, NPF was typically observed in the measurement area when the CIEMAT site was downwind of big green areas, that is, under wind directions from W to N sector (El Pardo forest area) and from S to W sector (Casa de Campo Park). This effect has been already observed in the measurement area and described by Gómez-Moreno et al. (2011). The origin of these air masses suggests the presence of high concentrations of BVOCs, which favor the growth of the newly nucleated particles.
At the beginning of the NPF phase, the average
During these events, the wind circulation in the study area played an important role. During the months that shrinkages were identified, as a result of the high solar radiation, local air flows were thermally driven and the wind direction in the area followed a well-defined pattern (Salvador, 2004; Artíñano et al., 1994; Plaza and Artíñano, 1994; Artíñano et al., 2003; Pujadas et al., 2000). At night (21:00–08:00 UTC) flows had a dominant NE–ENE origin, whereas during the daytime (09:00–20:00 UTC) the dominant origin sector was SW–WSW. The wind direction after noon maintained a directional component from the SW, when the daily wind speed reached the maximum value. In these cases, an early and atypical change of the wind direction towards the NE sector that occurred around 18:00 UTC, accompanied by an increase in wind speed, was responsible for a significant number of shrinkages. The factors that determine this change in the wind pattern could not be identified, and could probably attributed to a transition regime associated to a change of the pressure field at synoptice scale.
An example of a shrinkage process associated with NPF of type Ia
was observed on 1 July 2012 (Fig. 5). This event lasted
8.5
The event took place under clean air mass conditions, as
demonstrated by the low and invariable concentrations of pollutant
gases NO and
NPF phase began at 11:15 UTC and concluded at 15:45 UTC. The
wind speed and direction remained constant throughout the NPF
event. The average wind speed was low (
As shown in Fig. 5, high solar radiation (943
Furthermore, the evolution of the concentration of particles corresponding to each of the three modes was characteristic of this type of process (Guo et al., 2012; Du et al., 2012; Shen et al., 2011). From 11:15 until 13:15 UTC, the nucleation mode was the main contributor to the total particle concentration (above 50 %), while after 13:15 UTC and due to growth of freshly nucleated particles, the main contributor to the total concentration was the Aitken mode, with more than 50 % of the total. This trend continued until the end of the NPF event.
The beginning of the shrinkage phase occurred at 15:45 UTC, when
A type of shrinkage process not previously documented in the literature is the shrinkage associated with a previous aerosol growth event in the absence of nucleation.
Nine aerosol growth process
Three events corresponded to the first subgroup (Group IIa). The
growth phase occurred at noon, between 12:00 and 14:00 UTC
(Fig. 4), under stagnant conditions (wind speed around
2
The growth phase of these events lasted between 4.5 and
6.5
Six events fit in the second subgroup (Group IIb). These events
occurred during the late afternoon, around 18:00 UTC (Fig. 4),
and were produced under polluted air masses with traffic
emissions. An increase in concentrations of NO and NO
Aerosol growth process
A clear example of this group was the shrinkage event observed on 31 May and 1 June 2010 (Fig. 6).
This event commenced at 13:00 UTC on 31 May and continued until
02:30 UTC on 1 June, lasting 13.5
On 31 May, significant traffic emissions affected the measurement
site between 05:30 and 07:00 UTC. The average particle
concentration was
Under this situation, the growth phase emerged at 13:00 UTC and
ended at 19:30 UTC.
Between 19:30 and 02:30 UTC the shrinkage phase
emerged.
During the aerosol growth phase, the Aitken mode is the dominant
mode, followed by the nucleation and accumulation modes. However,
as in the NPF
The wind speed increase was associated not only with a significant
reduction in the particle size but also in a reduction in the
particle number concentration. A 24 % decrease of the total
particle concentration was observed during the shrinkage
phase. During this phase,
Shrinkage processes have also been observed in the absence of
a preceding NPF event or particle growth process. Twenty two pure
shrinkage events were identified during the 3.5
Most of these cases were identified during a period of high wind
speeds, normally higher than the previous two hours i.e. average
wind speeds of
The shrinkage in this group had a longer duration compared with
the shrinkage phases of the other types of events analyzed, with
an average of 4
As it was similarly hypothesized for the Group II: aerosol growth
process
An example of a pure shrinkage event was identified on 29 July 2010
(Fig. 7), which lasted 4
The aerosol growth reversal occurred under cleaned air masses which
transported a significant concentration of biogenic secondary
organic aerosol. The cleaned air plumes began to arrive at the
study site from the green areas near the CIEMAT at 14:00 UTC, when
the wind direction changed from a fluctuating direction to a clear
dominance of the W–N sector. This change was also accompanied by
an increase in the wind speed, from
The NO
The shrinkage began at 16:15 UTC and ended at 20:15 UTC, during
which the
As occurred in the rest of the events, changes in concentrations corresponding to each mode were also observed. Between 16:15 and 18:00 UTC, the Aitken mode was the main contributor to the total particle concentration, accounting for over 80 % of the total particles measured. From 18:00 until 20:15 UTC, when the shrinkage ended, a gradual increase in the nucleation mode occurred simultaneously with a gradual decrease in the Aitken mode. At the end of the event, the nucleation mode was the main contributor (59 %) to the concentration of total particles. The accumulation mode did not undergo significant variations and the ambient particle number concentration remained elevated throughout the event, indicating there was no significant dilution of the submicrometer atmospheric aerosol measured.
In this case study, the particle size reduction, and consequently the variations in the particle concentrations corresponding to each mode, seemed to be a result of the loss of BVOCs from biogenic aerosol during a period of high atmospheric dilution. In addition, the decrease in irradiance would produce a decrease in the photochemical activity, and therefore, a reduction in the photochemical formation of semivolatile gases.
This type of shrinkage have also been observed by Cusack et al. (2013) and Backman et al. (2012). However, while both authors related these processes to the evaporation of semi-volatile gases from the surface of particles during the hours of maximum solar radiation, in this work they have been identified at the end of the day, mainly as a result of atmospheric dilution caused by a significant increase of wind speed.
This paper provides the first study of aerosol shrinkage processes
based on a long time series of data covering 3.5
All shrinkages occurred during the months of May, June, July and August, when the atmospheric dynamics allowed the aerosol to have a longer residence time in the atmosphere and therefore an extended exposure time to incur physical and chemical changes. In addition, the environmental conditions, mainly intense photochemical activity in the presence of elevated concentrations of BVOCs, facilitated new particle formation and subsequent growth of atmospheric particles.
The shrinkage events identified in this study all occurred during
the final part of the day, and were caused mostly by atmospheric
dilution by increased wind speeds, 80 % of the shrinkage
processes occurred under a wind speed that exceeded
4
The shrinkage phases usually had a shorter duration than NPF and
aerosol growth phases. The SR values ranged between
As a result of the shrinkage, changes in the total particle number concentration and predominantly in the nucleation and Aitken modes were observed. Generally, an increase of particle concentration in the nucleation mode coincided with a reduction in the Aitken mode as particles reduced in size. During some events these changes were not so evident because the particle size reduction was maintained within the Aitken mode. Furthermore, the shrinkages were in some cases accompanied by a reduction in the aerosol concentration.
Based on the identified cases, three types of shrinkage events
were proposed: shrinkages proceeded by a new particle formation
and growth process (Group I: NPF
Pure shrinkage events were the most frequent group followed by
NPF
This work demonstrates that the proposed methodology is suitable for accurately and reliably classifying these processes. In addition, as it has been evidenced in this paper, environmental conditions and typical characteristics of the measurement area (climate, atmospheric dynamics, land use…) are crucial for their development. However, it is necessary to further deepen our understanding of these processes and their dynamics and patterns.
This work has been supported by the Spanish National Research Plan through funding of the projects, PHAESIAN (CGL2010-1777), REDMAAS (CGL2011-15008-E), MICROSOL (CGL2011-27020), PROACLIM (CGL2014-52877-R) and by the Madrid Regional Research Plan through TECNAIRE (P2013/MAE-2972). E. Alonso-Blanco acknowledges the FPI grant to carry out the doctoral thesis/PhD at the Research Center for Energy, Environment and Technology (CIEMAT). Thank are also given to José Luis Mosquera for his help in the processing of data which are the core of this work and Iván Alonso for his help in preparing some of the figures included in this paper.
Summary of studies related to shrinkage processes, including the present study.
Location of Madrid within Spain (inset) and the measurement site at CIEMAT facilities (red marker).The red line represents the Madrid municipality and the white lines the main traffic thoroughfares.
2009–2012 data coverage of the submicron aerosol fraction.
Interannual variation in the number and group of shrinkage events
during the 3.5
NPF
Aerosol growth process
Pure shrinkage case: evolution of the aerosol size distributions,
total particle number concentration (