1 s2.0 S0304389422003739 Main

Journal of Hazardous Materials 432 (2022) 128585
Contents lists available at ScienceDirect
Journal of Hazardous Materials

journal homepage: www.elsevier.com/locate/jhazmat
Research Paper
Sources and fate of atmospheric microplastics revealed from inverse and

dispersion modelling: From global emissions to deposition
Nikolaos Evangeliou a, *, Ondřej Tichý b, Sabine Eckhardt a, Christine Groot Zwaaftink a,
Janice Brahney c
a
Norwegian Institute for Air Research (NILU), Instituttveien 18, 2007 Kjeller, Norway
b
The Czech Academy of Sciences, Institute of Information Theory and Automation, Prague, Czech Republic
c
Department of Watershed Sciences and Ecology Center, Utah State University, Logan, UT 84322, USA
H I G H L I G H T S G R A P H I C A L A B S T R A C T
• About 9.6 and 6.5 Tg y-1 of micro

plastics and microfibers 1 are globally
released.
• Global average monthly surface MPs
(MFs) were 47 ng m-3 (33 ng m-3) at
maximum.
• 1.8% of the emitted microplastics from
ocean to land, 1.4% from land to ocean.
• Validation suggests that removal of
microplastics in global models needs
update.
• Results can be used as a proxy of the
expected global levels for
experimentalists.
A R T I C L E I N F O A B S T R A C T
Editor: Zaher Hashisho We combine observations from Western USA and inverse modelling to constrain global atmospheric emissions of
microplastics (MPs) and microfibers (MFs). The latter are used further to model their global atmospheric dy
Keywords: namics. Global annual MP emissions were calculated as 9.6 ± 3.6 Tg and MF emissions as 6.5 ± 2.9 Tg. Global
Microplastics average monthly MP concentrations were 47 ng m-3 and 33 ng m-3 for MFs, at maximum. The largest deposition
Microfibers
of agricultural MPs occurred close to the world’s largest agricultural regions. Road MPs mostly deposited in the
Transport
East Coast of USA, Central Europe, and Southeastern Asia; MPs resuspended with mineral dust near Sahara and
Dispersion
Airborne contaminants Middle East. Only 1.8% of the emitted mass of oceanic MPs was transferred to land, and 1.4% of land MPs to
ocean; the rest were deposited in the same environment. Previous studies reported that 0.74–1.9 Tg y-1 of land-
based atmospheric MPs/MFs (< 5 mm) are transported to the ocean, while riverine transport is between 3.3 and
14 Tg y-1. We calculate that 0.418 ± 0.201 Tg y-1 MPs/MFs (size up to 250 and 2500 µm) were transported from
the land to ocean (large particles were ignored). Model validation against observations showed that particle
removal must be urgently updated in global models.
* Corresponding author.
E-mail address: Nikolaos.Evangeliou@nilu.no (N. Evangeliou).
https://doi.org/10.1016/j.jhazmat.2022.128585
Received 29 October 2021; Received in revised form 23 February 2022; Accepted 24 February 2022
Available online 26 February 2022
0304-3894/© 2022 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
N. Evangeliou et al. Journal of Hazardous Materials 432 (2022) 128585
1. Introduction influence the global climate indirectly, as enhanced plastic production

needs larger consumption of fossil fuels and, in turn, larger emissions
Since the first reports on the presence of plastic debris in the marine (Höök and Tang, 2013; Royer et al., 2018). They also have a direct
environment in the early 70 s (Carpenter et al., 1972; Carpenter and impact, as MPs are usually colourful and can absorb incoming solar
Smith, 1972; Colton et al., 1974), there has been an increased awareness radiation in the atmosphere and where deposited. On snow or ice sur
of plastic accumulation in the environment. The global production of faces, they might decrease surface albedo enhancing melting, similar to
plastics in 2019 reached 368 Mt (from 225 Mt in 2004) (PlasticsEurope, conventional pollutants (Hegg et al., 2009). Revell et al. (2021) recently
2019) without PET-fibers (polyethylene terephthalate), PA-fibers (aro calculated a weak radiative impact, though highlighting that without
matic polyamide) and polyacrylic-fibers included. An estimated 10% of serious attempts to overhaul plastic production, the abundance and
total production is believed to end up in the sea every year by riverine of effective radiative forcing of airborne MPs will continue to increase.
washout transport (Barnes et al., 2009; Mattsson et al., 2015), although Although MPs’ role in freshwater, marine and terrestrial ecosystems
Weiss et al. (2021) recently argued that this might be overestimated. As and biota has been discussed extensively, very little is known on the
a result of widespread waste mismanagement, plastic pollution has been exact primary and secondary sources and budgets of airborne MPs, due
confirmed in many freshwater (Blettler et al., 2018), and terrestrial the lack of consistent measurements. Recently, a set of fallout samples
(Chae and An, 2018) ecosystems. Although the majority of plastics exists from remote and protected areas of the Western USA was collected and
in the form of macroplastics (>5 mm) (Lebreton et al., 2019), they may analysed for MPs/MFs in both wet and dry atmospheric deposition
fragment into microplastics (MPs, 1 µm to 5 mm) (Peeken et al., 2018) (Brahney et al., 2020). We make use of these measurements (i) to build a
and nanoplastics (NPs, <1 µm) (Wagner and Reemtsma, 2019) via robust methodology that combines atmospheric transport and Bayesian
photodegradation, physical abrasion, hydrolysis and biodegradation inverse modelling for source quantification; (ii) to calculate emissions of
(Gewert et al., 2015). MPs and MFs and determine their main source locations in Western USA.
MPs have been found in various shapes and sizes in the environment (iii) We calculate global emissions of MPs/MFs by extrapolating regional
such as 1-D fibers, 2-D fragments (flat particles) and 3-D spherules (Dris emissions, and (iv) report on the global dispersion of MPs/MFs calcu
et al., 2015). Their origin can be primary, when manufactured in smaller lating the respective budgets from their emissions to the regions they are
sizes for scientific and medical applications, paint, (Gregory, 1996) or deposited. Finally, (v) we create a product that comprises the respective
cosmetic products (Fendall and Sewell, 2009) or when originate from levels of surface concentrations and deposition rates in high spatio
abrasion of large plastic objects during manufacturing, use or mainte temporal resolution (0.5◦ ×0.5◦ , daily), as a tool for scientists conducting
nance (e.g., road dust) (An et al., 2020; Boucher and Friot, 2017; Coyle MP/MF measurements to forecast their expected levels.
et al., 2020; Goßmann et al., 2021; Grigoratos and Martini, 2015; Habib
et al., 2020; Halle et al., 2020; Hartmann et al., 2019; Jan Kole et al., 2. Materials and methods
2017a; Patil et al., 2021; Sharma and Chatterjee, 2017; Szymańska and
Obolewski, 2020; Wang et al., 2018; Yukioka et al., 2020; Zhang et al., 2.1. Fallout measurements of dry and wet deposition
2020). Secondary microplastics are produced by decomposition
(O’Brine and Thompson, 2010). The largest portion of the secondary The detailed methodologies for the determination of MPs and MFs in
MPs is synthetic microfibers (MFs) produced after washing synthetic fallout samples is described in Brahney et al. (2020). Briefly, fallout
clothes (Browne et al., 2011). Athey et al. (2020) reported that a single samples were collected at 11 National Park and Wilderness sites be
pair of jeans discharges 56,000 fibers per wash into the wastewater. MFs tween 2017 and 2019 using Aerochem Metrics model 31 wet/dry col
comprise a range of different shapes, and a widely acceptable nomen lectors (ACMs), which include precipitation sensors that opens the wet
clature is missing. Here, we consider MFs as synthetic fibers having a bucket, and closes the dry bucket, while precipitation and vice versa.
base diameter of less than 10 µm with a height to base diameter ratio of Wet samples were filtered through 0.45 µm polyethersulfone (excluded
up to 103 (J. Liu et al., 2019). from the study) filters every week, whereas dry ones were collected at
Once MPs are released into the environment, they are subject to monthly or bi-monthly intervals using custom-built dry sampling units.
physical (e.g. mechanical), radiative, chemical, and biological degra In total, 236 wet and 103 dry samples were weighed and counted at
dation, which changes their size, shape, surface, composition, and 100x magnification using a BX50 Olympus Microscope and cellSens
environmental mobility. They become easily airborne following turbu Imaging Software and were separated into the following size classes: <
lent processes at the surface, similar to dust, not only when they are 10 µm, 10–25 µm, 25–50 µm, 50–100 µm, 100–250 µm, 250–500 µm,
deposited in continental regions (Qian and Ferro, 2008), but also from 500–1000 µm, 1000–1500 µm, 1500–2000 µm, 2000–2500 µm, and
the surface of the ocean (Allen et al., 2020), and undergo long-range 2500–3000 µm. Respective densities were assumed between 0.92 and
transport. Airborne MPs will eventually deposit on land or ocean, but 2.2 g cm-3 with a mean of 1.22 g cm-3 based on literature values and
may be resuspended again as a result of grasshopping processes (Gouin, detected polymers. For justification, mass deposition rates were also
2021). The global atmospheric transport of MPs is generally more effi estimated using FTIR (Fourier Transform Infra-Red) mapping data, with
cient than the oceanic or the terrestrial one, as it occurs in much shorter the limitation that particles < 20 µm cannot be determined. The com
time-scales (weeks compared to years for hemispheric distances, parison of the results obtained with both techniques (count-based
respectively) (Evangeliou et al., 2020; Mountford and Morales Maqueda, deposition and FTIR-based deposition showed a strong correlation (R =
2021). Lately, MPs have been determined in remote regions, from the 0.89, p < 0.001) revealing high-quality measurements. As regards to
Alps (Bergmann et al., 2019), the Pyrenees (Allen et al., 2019), and US MFs, in subsamples collected, almost all brightly colored particles that
national parks (Brahney et al., 2020), as far as to Antarctica fell within the counting criteria described in Brahney et al. (2020) were
(González-Pleiter et al., 2020; Kelly et al., 2020) and the high Arctic identified as synthetic using FTIR spectroscopy mapping.
(Bergmann et al., 2019), whereas modelling reveals atmospheric
transport almost everywhere on earth (Evangeliou et al., 2020; Brahney 2.2. Atmospheric transport modelling
et al., 2021). Notice that presence of MPs in remote regions, distant from
major waterways, can result only via the atmosphere. The source receptor matrices (SRMs) for each fallout sample were
MPs have been found to affect marine (Wilcox et al., 2018), terres calculated with the Lagrangian particle dispersion model FLEXPART
trial animals (Harne, 2019), and potentially human health (Lehner et al., version 10.4 (Pisso et al., 2019). The model was driven with 3-hourly
2019; Wright and Kelly, 2017), as MPs have been detected in human operational meteorological wind fields retrieved from the European
stool (Schwabl et al., 2019) and all placental portions (Ragusa et al., Centre for Medium-Range Weather Forecasts (ECMWF) consisting of
2021). Beyond organismal and ecosystem effects, MPs may also 137 vertical levels and a horizontal resolution of 1◦ × 1◦ . The SRMs were
2
calculated in backwards time (retroplume) mode, using a new feature of domain resulting into the estimated source term which is, in essence,
FLEXPART that reconstructs wet and dry deposition with backward averaged emission from each spatial element. Second, we use the
simulations (Eckhardt et al., 2017). This new feature is an extension of emission estimated in the first step as the prior emission for the second
the traditional backward simulation for atmospheric concentrations step, where the inversion problem (Eq. 1) is solved for each spatial
(Seibert and Frank, 2004). To our knowledge, FLEXPART is the only element of the assumed spatial domain. Using this concept, the solution
model with the capability to calculate SRMs of the deposited mass. of the second step is more stable and less depends on regularization
More specifically, for the reconstruction of wet deposition of MPs parameters. Also, to stabilize the inversion, we do not consider mea
and MFs, computational particles were released at altitudes 0–20 km at surements that have zero computed concentration by the FLEXPART
the locations of the samples (receptors), as scavenging can occur at any model for the whole period for a given spatial element, i.e. have zero
height of the atmosphere, depending on the location of clouds and associated row in the SRM, M.
precipitation. For dry deposition, particles were released at 0–30 m at While the inversion formulation (Eq. 1) appears simple, its solution is
the respective receptors, as this shallow layer is equal to the height of the non-trivial. Least-squares solution fails due to typically ill-conditioned
layer in which, in forward mode, particles are subject to dry deposition. matrix M in atmospheric problems, due to the associated uncertainties
All released particles represent a unity deposition amount, which was and the nature of the problem (Ganesan et al., 2014; Liu et al., 2017).
converted immediately (i.e. upon release of a particle) to atmospheric Therefore, the problem needs to be regularized, i.e. additional infor
concentrations using the deposition intensity as characterized by either mation on model parameters or variables needs to be assumed. One such
dry deposition velocity or wet scavenging rate (in-cloud and below- regularization could be the addition of the term λ‖x‖22 , λ > 0, to the
cloud scavenging) (Eckhardt et al., 2017; Grythe et al., 2017). The minimization problem known as the Tikhonov regularization (Golub
concentrations were then treated as in normal “concentration mode” et al., 1999). When further regularizations are needed, more parameters
backward tracking (Seibert and Frank, 2004) to establish SRMs between such as λ are introduced and, notably, need to be set manually or using
emissions and deposition amounts (30 d backward tracking). The model heuristics (Hansen and O’Leary, 1993). This is, however, also the case of
output consists of a spatially gridded sensitivity of MPs and MFs depo the present inversion problem. In order to reduce manual tuning, we
sition at the receptor points to the respective emissions, equivalent to the follow the Bayesian formalism, in which the least-squares minimization
backwards time mode output for concentrations (Seibert and Frank, arising from Eq. 1 can be equivalently formulated as the maximization of
2004). Deposition rates of MPs and MFs (particles m-2 d-1) can be the logarithm of Gaussian distribution:
computed by multiplying the SRMs (in m) divided with the lowest model ( )
layer (100 m) with gridded emissions (particles m-2 d-1).
( ( )) − 1
argx min‖y − Mx‖22 ⟺argx max lnN Mx, Ip = argx max ‖y − Mx‖22
Except for dry and wet deposition (Grythe et al., 2017), FLEXPART 2
accounts for turbulence (Cassiani et al., 2014), unresolved mesoscale (2)
motions (Stohl et al., 2005) and convection (Forster et al., 2007). A point ( )
Therefore, the Gaussian distribution N Mx, Ip on the right side of the
that adds uncertainty in our calculations is the efficiency with which
Eq. 2 is chosen as the prior observation model. Similarly, all other reg
particles are scavenged by precipitation. Plastics are generally hydro
ularization terms can be included in the forms of prior distributions.
phobic and should therefore act as inefficient cloud condensation (CCN)
These prior distributions, appropriately chosen, can form hierarchical
or ice nuclei (IN). However, coatings formed during ageing of the
priors for all the unknown variables of the inverse model. The key
aerosols may make the particles more hydrophilic with time (Bond et al.,
advantage of the Bayesian formalism is that all variables and regulari
2013). A recent study by Ganguly and Ariya (2019a) reveals that NPs
zation terms are estimated within the method.
and MPs may become important for cloud formation and thus anthro
We follow the variational Bayesian methodology (Smidl and Quinn,
pogenic climate change. However, since their exact scavenging co
2006), in which the posterior distributions of the parameters remain in
efficients are currently unknown, we distinguish between three different
the same form as their prior distributions. Following Eq. 2, the residual
in-cloud scavenging properties (low, medium, and high CCN/IN effi
model of y is formulated as the Gaussian distribution with mean value of
ciency, Table S1) in each of the aforementioned particle sizes and
Mx and an unknown scalar precision of the noise parameter ω as below:
quantify the uncertainty that is associated with the scavenging effi
( )
ciency. We adopted the exact particle densities (1.22 g cm-3) and size p(y|ω, x) = N Mx, ω− 1 Ip (3)
distribution of MPs and MFs in the model as those reported by Brahney
( )
et al. (2020). − 1 −1
∝exp ω ‖y − Mx‖22 (4)
2
2.3. Linear inversion problem and Bayesian inversion where ∝ denotes equality up to the normalizing constant. For tractability
of the model, the prior distribution of the parameter ω is chosen con
The concept of the SRM is used here assumes that the relationships jugate to the observation model in Eq. 3 using Gamma distribution as
between the source and the receptor are linear such as that mij = ci /xj , follows:
where xj is a hypothetical release from the site in j-th time and ci is the
p(ω) = G(ϑ0 , ρ0 ) (5)
calculated response at the i-th receptor at the given time period. Then,
the measurement, yi , can be approximated as the sum of the emission, x, where ϑ0 and ρ0 are scalar constants, which are set to non-informative
weighted by the model predictions, mij . values 10-10 and serve for numerical stabilization in the resulting algo
The linear equation describing the measurements, y ∈ Rp , based on rithm to avoid division by zero, when necessary.
the modelled SRMs, M ∈ Rp×n , and the unknown release, x ∈ Rn , can be The prior model for the source term is based on principles of the LS-
formulated as: APC (least squares with adaptive prior covariance) model (Tichý et al.,
y = Mx (1) 2016), in which the emission is assumed to alter between sparse or
smooth character, modified using the assumption of the prior source
This formulation can be used for each spatial element as well as for term here. The prior emission term x0 could be a zero vector, as in the
the assumed spatial domain as a whole. We used both these formulations first step for overall emission from the whole domain where no prior
in a two-step procedure to reduce the ill-conditionality of the inversion information is available, or non-zero, as in the second step where the
problem arising when computing spatial distribution of the emission, estimated emission from the first step is used as the prior emission term
which is ill-posed problem due to sparse measuring network. First, we for each spatial element. The prior distribution of the emission is chosen
solve the inversion problem (Eq. 1) for the whole assumed spatial
3
as a Gaussian distribution truncated to the positive values with mean for land (cropland, dust, road dust, and population) and oceanic sources
vector x0 and precision (inverse covariance) matrix, Ξ, as: (sea salt). Also, we calculated the average posterior release of MPs in our
( ) inversion domain for every day. Then, we took their ratio, which we
p(x|Ξ) = tN x0 , Ξ− 1 , [0, + ∞] (6)
used to calculate global emissions of MPs based on our estimates for each
inventory and particle size class with daily temporal resolution. Since
where the precision matrix, Ξ, is in the specific form of Cholesky
the proportion of each MP source to the total emissions is unknown, we
decomposition as Ξ = LVLT . The matrix V = diag(v) is the diagonal
assumed different proportions of agricultural, dust, road dust and sea
matrix favouring sparse, i.e. zero, solution(Tipping, 2001), whereas the
salt, respectively, that constitute the MPs emissions (see Table S2).
matrix L is the lower bi-diagonal matrix with ones on diagonal and
As reported earlier, we assumed that MFs are linked with population
vector l ∈ Rn− 1 on sub-diagonal, such as: density owing its presence to clothing. Hence, in the same way as for
⎛ ⎞ MPs, we computed the average population in the studied domain and
1 0 ⋯ ⋯ 0
calculated the ratio with MF releases for every day. Using this ratio and
⎜ l1 1 0 ⋯ ⋮ ⎟ the global population density, we extrapolated our Western USA MF
⎜ ⎟
L=⎜ 0 l2 ⋱ ⋱ ⋮ ⎟ (7) emissions in a global grid for each particle size class with daily temporal
⎝ ⋮ 0 ⋱ 1 0
⎠ resolution.
0 0 0 ln− 1
3. Results
1
Following Tipping (2001), we select the prior distribution for the

introduced variables, vectors v and l as below: 3.1. Annual posterior emissions of microplastics and microfibers in the
( ) Western USA
p vj = G(α0 , β0 ), j = 1, …, n (8)
( ⃒ ) ( ) The annual posterior emissions of MPs and MFs can be seen in Fig. 1a
p lj ⃒ψ j = N − 1, ψ −j 1 , j = 1, …, n − 1 (9) and c for the inversion domain (124–91◦ W, 29–47◦ N). The calculated
daily posterior emissions can be of primary (direct emissions) or sec
( )
p ψ j = G(ζ0 , η0 ), j = 1, …, n − 1 (10) ondary origin (emissions from resuspension of previously deposited
material). In total, 22 ± 10 million MPs m-2 y-1 were estimated in five
where the prior constants α0 , β0 , ζ0 , and η0 are selected again to serve for different sizes (5–10 µm, 10–25 µm, 25–50 µm, 50–100 µm,
numerical stability. The prior model of the variable l in Eq. 9, favours a 100–250 µm) following the measured size distribution. The later
smooth solution using prior mean value of − 1 and an unknown variance together with the respective measured densities (average: 1.22 g cm-3)
vector ψ . and volume of each size bin gave a total annual emitted mass of MPs of
Variational Bayes method (Smidl and Quinn, 2006) seeks for 9.0 ± 3.8 kt y-1 in the inversion domain. MF number emissions were in
approximation of posteriors distribution in the form of conditional in the same order as MPs (24 ± 11 million MFs m-2 y-1), albeit they were
dependence distribution, so that: measured in much larger sizes of up to 2500 µm (Brahney et al., 2020).
An accurate conversion of number to mass emissions of MFs is practi
p(x, υ, l, ψ , ω|y) ≈ ̃
p(x|y)̃
p(v|y)̃ p(ψ |y)̃
p(l|y)̃ p(ω|y) (11)
cally impossible, due to the chaotic shape of fibers. The most realistic
The best possible solution minimizes the Kullback-Leibler divergence approach to resemble their capillary shapes would be to assume they are
(Kullback and Leibler, 1951) between the posterior and the hypothetical thin cylinders. Here, we distinguish between three different base di
true posterior as follows: ameters, all below 10 µm as defined in Liu et al. (2019) ( 1 µm for MF
[ ] size 10–25, 25–50, and 50–100 µm; 5 µm for MF size 100–250 µm,
p(θi |, y)∝exp Ẽp(θ− i |,y) ln(p(θ, y))
̃ (12) 250–500 µm, 500–1000 µm; 10 µm for MF size 1000–1500 µm,
1500–2000 µm, 2000–2500 µm, and 2500–3000 µm). The relevant
where θi denotes the i-th variable from the set {x, v, l, ψ , ω} and θ− i de equations can be found in Supplementary Information. Then, adopting
notes complement of θi in θ. More details on the method and its the measured sizes and their respective densities, we calculated a total
implementation can be found in Tichý et al. (2020). emitted mass of 244 ± 129 kt y-1. This number is much smaller than the
respective for MPs, in contrast to the particle number emissions, because
2.4. Extrapolation on a global domain fibers have a much smaller volume than particles. For both MPs and
MFs, the calculated mass emitted was higher for larger sizes, although
To extrapolate our estimates in a global domain, we used global greater number emissions were found at smaller sizes.
datasets of the main sources for MPs and only population density for Fig. 1b and d show the Taylor diagrams of the mismatches between
MFs. Specifically for MPs, global emission inventories of mineral dust, deposited number concentrations and reconstructed modelled concen
road dust, sea salt, and agriculture were used as the main sources of MPs trations of MPs and MFs for each size bin. An accurate validation of the
(Brahney et al., 2021; Chen et al., 2020; Evangeliou et al., 2020; Piehl posterior emissions requires observations that were not included in the
et al., 2018). Mineral dust emissions were calculated using the FLEX inversion (independent observations). However, the small number of
DUST model (Groot Zwaaftink et al., 2017). Road dust emissions were deposition measurements that were available to perform the inversion
adopted from ECLIPSEv6 emission inventory and are the same as those prevents us from using independent observations from Brahney et al.
used in (Evangeliou et al., 2020). Sea salt emissions were taken from (2020) here. Hence, the comparison shows only the posterior deposited
(Grythe et al., 2014) as the average emissions from 20 models. Agri concentration mismatches to the observations simulated with FLEX
cultural activities are represented by a global dataset of croplands and PART for any given size bin. For almost all size bins, the Pearson’s
pastures developed by combining agricultural inventory data and correlation coefficient was 0.4 – 0.6, while the normalised root mean
satellite-derived land cover data (Ramankutty et al., 2008). For MFs, we square error (RMSE) and standard deviation were kept low both for MPs
assumed that their main source is from clothing and should be therefore and MFs (Fig. 1).
linked with the distribution of the global population (Henry et al., 2019; The spatial distribution of the MP and MF emissions in the Western
O’Brien et al., 2020), which we adopted from NASA (Gao, 2017; Jones USA can be seen in Fig. 2 for all size classes and in for each size class. The
and O’Neill, 2016). lowest emissions were calculated close to the measurement stations
For each global emission inventory serving as source of MPs, we opposite to the respective footprint emission sensitivity (SRM), which
computed the average value in our spatial inversion domain, separately were the highest closer to the measurements (Supplementary Figure S3).
4
Fig. 1. (a) Posterior monthly emissions of MPs calculated using wet and dry deposition measurements in Western USA. Five different sizes were considered following
a measured size distribution, which resulted in total annual MPs emissions of 21.6 million particles m-2. (b) Modelled deposition of MPs against observations both for
dry and wet samples for each of the five sizes presented in a Taylor diagram. The latter shows the Pearson’s correlation coefficient (gauging similarity in pattern
between the modelled and observed deposition) that is related to the azimuthal angle (blue contours); the standard deviation of modelled deposition is proportional
to the radial distance from the origin (black contours) and the centered normalised RMSE of modelled deposition is proportional to the distance from the reference
standard deviation (green contours). (c) Posterior monthly emissions of MFs in Western USA for nine sizes resulting in an annual total of 23.6 million fibers m-2. (d)
Taylor diagram modelled versus observed deposition of MFs both for dry and wet samples for each of the nine sizes.
Fig. 2. Spatial distribution of MP and MF emissions in the Western USA calculated from deposition measurements and Bayesian inverse modelling. The largest cities
are shown in black circles and the measurement stations in white stars. Note that emissions are stronger away from the stations due to the remote location of the
measurement stations away from artificial sources.
This behavior can be expected considering that the measuring stations level), thus, far from any man-made activity that would emit MPs and
are located in US national parks and remote areas (45–300 km from MFs. On the other hand, our findings are necessarily biased by the
urban centers and at elevations ranging from 1240 to 3520 m above sea remoteness of measurement sites showing the need for sample collection
5
also close to populated areas. larger than those of Brahney et al. (2021) (8.7 Tg y-1). The calculated
global annual emissions per source, size and region can be seen in
Table 1. The ocean dominates atmospheric emissions with 8.9 ± 3.5 Tg
3.2. Global emissions of microplastics and microfibers y-1, as the insoluble plastic debris accumulates at the surface of the ocean
and can be resuspended by bubble bursts similar to other sea spray
The exact composition of atmospheric MPs from individual sources is aerosols (Allen et al., 2020). It is slightly higher than that of (Brahney
currently under uncertain. What is known with some certainty is that et al., 2021) (8.6 Tg y-1), albeit it shows a completely different distri
primary MPs are estimated to represent between 15% and 31% of MPs in bution, because of the 20 member ensemble used for its calculation.
the oceans (Boucher and Friot, 2017). Accordingly, Brahney et al. However, it should be noted that the selected inversion domain suffi
(2020) determined that 10% of the counted microplastics were primary cient to accurately constrain MP emissions covers a very small oceanic
microbeads. Secondary MPs account for 69–81% of oceanic MPs mostly surface and any interpolation might be uncertain. Furthermore, the
originating from degradation of larger plastic objects, such as plastic deposition measurements used in the inversion (Brahney et al., 2020)
bags, bottles or agricultural or fishing nets (World Economic Forum, are far from the ocean and the footprint emission sensitivities very weak
2016). In the present study, MPs were assumed to be produced by road () inducing an additional uncertainty. A good example for the latter is
dust (primary source), mineral dust (secondary source), agriculture the part of the Atlantic Ocean and the Gulf of Mexico, where footprint
(secondary source) and sea salt (secondary source), although other emission sensitivities are very low. For more accurate emission calcu
sources might be also important (e.g., direct emissions from industrial lations, oceanic stations measuring MPs is a necessity. Agricultural ac
regions). We give 23% (average of 15 – 31%) to primary sources (in our tivities resuspend around 0.31 ± 0.13 Tg y-1 of plastics previously
case, road dust) and we split the rest to all the other secondary sources deposited in the soil (Nizzetto et al., 2016) or from the use of agricultural
(mineral dust, agriculture, sea salt) forming 30 scenarios as indicated in mulch (Fakour et al., 2021). Road dust contributed another 0.28 ± 0.12
Table S2. Their average was used as the basis for the calculation of the Tg y-1, almost three times higher than in Brahney et al. (2021) (0.096 Tg
MP emissions. MFs have been calculated separately assuming that they y-1), and similar to Evangeliou et al. (2020) (0.43 Tg y-1, range: 0.20–1.1
only originate from the average population. Tg y-1) for tire wear (TWPs) and brake wear particles (BWPs) of size
The global annual posterior emissions of MPs and MFs can be seen in < 10 µm. TWPs are produced by shear forces between the tread and the
Fig. 3. For the MPs, emissions were estimated to be 9.6 ± 3.6 Tg y-1, 10%
Fig. 3. (a) Annual emissions of MPs revealed from inverse modelling in the Western USA and then extrapolated globally. (b) Sensitivity of the MP emissions to
different emission sources (road dust, mineral dust, agricultural activity, and sea salt). The sensitivity is calculated here as the standard deviation of the 30 different
scenarios that consider different proportions of emission sources with respect to total MP emissions (Table S2). (c) Global annual MP emissions calculated with the
source proportions described in (Brahney et al., 2021) (Table S1). (d) Global annual MP emissions from (Brahney et al., 2021). (e) Annual emissions of MFs revealed
from inverse modelling in the Western USA and then extrapolated globally.
6
Table 1
Global annual emissions of MPs and MFs (in Tg y-1) per different source, size and region (according to Supplementary Figure S4).
*described as dustpop in Brahney et al. (2021).
road pavement (Rogge et al., 1993) or by volatilization (Wagner et al., consumer stage. The total atmospheric emissions calculated in the pre
2018), and the whole wearing process depends on the type of tire, the sent study are at least an order of magnitude higher.
road surface, different vehicle characteristics, the state of operation and The continental distribution of the calculated MP and MF emissions
overall condition of the vehicle (Grigoratos and Martini, 2014). BWPs can be seen in Table 1 calculated using continental masks as defined in.
are produced via mechanical abrasion and corrosion (Penkała et al., Due to the lack of emission data, the only comparison can be performed
2018; Sommer et al., 2018). Jan Kole et al. (2017b) reported global against the emissions calculated by Brahney et al. (2021). The annual
emissions of TWPs to be about 6.1 Tg y-1 (BWP emissions add another total MP emissions in Asia were estimated to be 0.25 ± 0.12 Tg y-1 (MF:
0.5 Tg y-1) not specifying any size range. However, they reported that 3.7 ± 1.3 Tg y-1), in contrast to 0.089 Tg y-1 in Brahney et al. (2021).
3–7% of the PM2.5 (1.2 – 2.8 Tg y-1), is estimated to consist of tire wear North America contributes another 0.13 ± 0.055 Tg y-1 (0.024 Tg y-1 in
and tear. The emissions reported here are one order of magnitude lower. Brahney et al., 2021) (MF: 0.28 ± 0.12 Tg y-1), and Africa 0.11 ± 0.055
MPs resuspended with mineral dust are the least important (0.10 Tg y-1 (0.093 Tg y-1 in Brahney et al., 2021) on MP emissions (MF: 0.97
± 0.052 Tg y-1) in close agreement with Brahney et al. (2021) (0.068 Tg ± 0.45 Tg y-1). Europe (MPs/MFs: 0.088 ± 0.042/0.46 ± 0.14 Tg y-1)
y-1). If the proportion of different sources to total are to be used from emits twice as much MPs as Russia (MPs/MFs: 0.044 ± 0.020/0.44
Brahney et al. (2021), as reported for deposition (see Table S2), the total ± 0.25 Tg y-1) or South America (MPs/MFs: 0.045 ± 0.022/0.52 ± 0.25
emitted MPs are one quarter (~2.4 Tg y-1) of those presented here (9.6 Tg y-1), while the rest of the continents have smaller shares in the annual
Tg y-1) (Fig. 3c). emissions of MPs and MFs. Brahney et al. (2021) calculated MP emis
As regards to MF emissions (Table 1), the global annual mass emitted sions in Europe to be 0.048 Tg y-1, 0.0023 Tg y-1only in Russia and
was 6.5 ± 2.9 Tg y-1 assuming that the only source was fibers from 0.0071 Tg y-1 in South America.
human clothes. Emissions increased with fiber size with a peak at
1000–1500 µm (2.9 ± 1.1 Tg). Gavigan et al. (2020) estimated that 5.6
3.3. Atmospheric transport and deposition
Tg of synthetic MFs were released from apparel washing between 1950
and 2016, half of it during the last decade, though not specifying what
The global atmospheric transport of MPs emitted from agricultural
fraction might become airborne. Godfrey (2021) reported that about
sources, with mineral and road dust and with sea spray, as well as MFs
0.12 Tg y-1 of synthetic MFs are released into the environment annually
from the global population can be seen in Video 1 and 2. In the latter,
at the pre-consumer stage, or one shirt for every 500 manufactured. This
transport is shown in daily temporal resolution for the year of the
means that this number rises after accounting for the MF loss at the
inversion (2018). The large size of the particles modelled (up to 250 µm
7
for MPs and 2500 µm for MFs), following the observations used in the activities, road and mineral dust, and sea spray) is illustrated in Fig. 5
inversion, do not present long-range transport characteristics. The latter together with the respective one for MFs. As expected, all the emitted
is reinforced by the assumption that the emissions took place at the mass has been deposited by the end of the simulation year, while
surface (within the first 10 m), thus they are not lofted significantly high maximum deposition occurred near the largest sources indicating
enough to be transported over longer distances, but they are rather limited transport, because of the large particle/fiber sizes. The largest
removed fast. Their lifetimes depend not only on the size, but also from MPs deposition originating from agricultural sources (annual deposi
how fast they are scavenged by rain droplets and, in turn, removed from tion: 310 ± 131 kt y-1) was seen in Central USA, in the Indo-Gangetic
the atmosphere (Evangeliou et al., 2020). Here, we have assumed a Plain and the North China Plain (Fig. 5a), which are well-known re
modelled CCN/IN efficiency to be moderate (Table S1). Hence, for the gions of agricultural emissions and have been quantified as the most
smallest particles (<10 µm), the lifetime is 8.3 ± 1.0 days and drops important sources of agricultural ammonia (Evangeliou et al., 2021).
with increasing size until 2.5 ± 1.1 days for the largest size Road MP annual deposition (mainly TWPs and BWPs) was equal to 279
(100–250 µm). The same characteristics are also seen for the MFs (Video ± 125 kt y-1, with the largest deposition occurring in the East Coast of
2), which were modelled as particles. Although this is inaccurate, as the USA, Central Europe, and Southeastern Asia (Fig. 5b). Secondary
their shape is rather capillary causing different aerodynamic properties, MPs were assumed to be resuspended with mineral dust deposited
it only gives an indication of how far from the main land sources they mainly near Sahara and Middle East (Fig. 5c) with a total annual
should be expected. Their sizes were somewhat higher than those of MPs deposition to be 100 ± 52.2 kt y-1. Similar patterns were calculated for
(15–5000 µm). Nevertheless, particles larger than 2500 µm were not MPs remobilized with sea spray, which deposited mostly in the Ocean
determined, which means that they are removed from the atmosphere so (Fig. 5d). The observed MF deposition complies with the general pop
fast that cannot travel at all. This is in agreement with the calculated ulation density, with maxima in Beijing (China) and the Indo-Gangetic
modelled lifetimes for MFs that were found to be < 2 days for sizes Plain (Fig. 5f).
above 500 µm. The observed deposition of MPs and MFs in different continental and
Global mean concentrations of MPs and MFs at the surface of the oceanic regions (according to Fig. S 4) can be found in Table 2. The
atmosphere can be seen in Fig. 4. Average monthly mass concentrations deposition in different continents mainly originates from the land-based
of MPs ranged between 6 and 47 ng m-3 or between 1 and 8 particles m-3 sources (agriculture, transportation, mineral dust), while those in
d-1, if density and different volume for each size bin are to be considered. oceanic regions mainly from sea spray. The largest continental annual
MF surface monthly mass concentrations were significantly lower MP deposition occurred in Asia (267 ± 121 kt y-1), followed by North
(2.4–33 ng m-3) and daily number concentrations a few fibers m-3. Both America (160 ± 71.1 kt y-1), Africa (114 kt y-1) and Europe (102
for MPs and MFs, one can immediately notice the very rapid removal ± 45.4 kt y-1). The largest continental deposition of MFs occurred in
from the atmosphere, because of the extremely large sizes that were Asia (3792 ± 1933 kt y-1), Africa (801 ± 372 kt y-1) and Europe (598
considered in the present. Although larger particles are rarer (Fig. 1), ± 253 kt y-1), whereas another 846 ± 209 kt y-1 were deposited over
they are much heavier and subsequently removed faster having lifetimes the American continent. As regards to deposition over the ocean, about
< 1 day. This is shown by almost constant day-to-day variation of daily 1718 ± 899.5 kt y-1 were deposited in the Atlantic Ocean, 2751
concentrations in each month, which rather fail to accumulate with time ± 1225 kt y-1 in the Pacific, 1435 ± 765.5 kt y-1 in the Indian and 2289
in the atmosphere. Considering that MPs/MFs emissions occur near the ± 1185 kt y-1 in the Southern Ocean, while deposition in the Mediter
surface and are characterized by large sizes, concentrations decline ranean Sea was one order of magnitude less (102 ± 55.6 kt y-1). The
substantially at higher altitudes of the planetary boundary layer (PBL) or oceanic deposition of MFs was tiny for two reasons, (a) MF emissions
in the free troposphere. We calculate that global average daily concen were calculated to be negligible in oceanic regions of the inversion
trations of MPs in the PBL to be 1.7–13 times lower (for particle di domain (see Fig. 2b), hence no emissions from sea spray could be
ameters of 5–10 µm and 100–250 µm) than at surface, while in the free assumed and, subsequently, no direct deposition to the ocean was
troposphere 8–400 times lower (for particle diameters of 5–10 µm and simulated; (b) the sizes of MFs was very large (see Section 2.1) and no
100–250 µm) than at surface. For MFs, PBL concentrations are 2.5–16 significant transport from the land could be expected.
times smaller (for fiber heights between 10 and 25 µm and
2000–2500 µm) than at surface, and free tropospheric concentrations
between 9 and 1000 times (for fiber heights between 10 and 25 µm and
2000–2500 µm) smaller than at surface.
The global annual deposition for each MP source (agricultural
Fig. 4. Timeseries of global mean daily number (blue) and monthly mass (red) concentrations of MPs and MFs at the surface of the atmosphere (0–100 m). The latter
can be used as a proxy for the expected surface atmospheric levels by researchers conducting MP measurements.
8
Fig. 5. Annual global deposition of MPs emitted from agriculture, with road and mineral dust and with sea spray. The annual global deposition of MFs from the
global population is also given in the lower right panel.
Table 2
Global annual deposition of MPs and MFs (in kt y-1) in different continental, mountainous and oceanic regions (as defined in Supplementary Figure S4).
9
4. Discussion measurements do not contribute to the reconstruction of the posterior

emissions; however, they remain in the loss function (Eq. 2), which re
4.1. Uncertainty of the posterior emissions in the Western USA sults in a less stable solution with larger emissions. For each of the three
inversion algorithms, three different MP and MF species were assumed,
The calculation of the uncertainty of the posterior emissions was each with a different CCN/IN efficiency (Table S1), which gives a total of
performed in two different ways to show the robustness of the meth nine ensemble members for each size of MPs and MFs.
odology used in the present study. The first method is based on a The second method for calculating uncertainty of the posterior
sensitivity study with an ensemble of inversion algorithms employing emissions in the Western USA is based on the relation between mea
different scavenging characteristics for MPs and MFs; the uncertainty surement and reconstruction (Eq. 1). In Eq. (1) many types of un
was calculated as the standard deviation of the different posterior certainties affect the results. For instance, measurements are affected by
emissions. The different members of the ensemble were built from three the spatial and temporal quality of the monitoring network (in addition
different inversion algorithms. The first is the one already described in to the uncertainty of the measurement methodology), i.e., sparse
Section 2.3; the second algorithm is a modification of the first one with network can significantly bias the results (De Meutter et al., 2020). On
the choice of the mean value of prior emissions to be equal to zero (x0 = the other hand, SRMs accumulate all the biases of the respective atmo
0) in Eq. (7). This choice leads to lower estimates since zeros are spheric transport model parametrizations and the uncertainties of the
assumed when no information on releases can be calculated from the meteorological data used (Sørensen et al., 2020), as well as the param
observations, which may cause biased estimates. The third algorithm is etrization of the source term prior model (Tichý et al., 2020). To
another version of the first one, where the prior mean value in Eq. (6) is determine the overall posterior emission uncertainty, variants of
assumed again to be zero (x0 = 0). Moreover, we remove the assump log-normal models have been used recently (Dumont Le Brazidec et al.,
tion of not considering measurements that have zero computed SRM 2021; Liu et al., 2017). Here, we use a Gaussian model, due to its trac
sensitivities (calculated with FLEXPART) for the whole studied period in tability and interpretability of posterior estimates.
each spatial element. This is a very demanding case, since these Specifically, the uncertainty quantification of the estimated posterior
Fig. 6. Calculation of the posterior emission uncertainty of MPs and MFs (a, b) based on the sensitivity of the emissions to different scavenging coefficients, (c, d)
using a Gaussian model (see Section 4.1), and (e, f) the propagated (combined) uncertainty.
10
emissions is based on the form of posterior distribution of the estimated emissions. Thus, any change is this source fraction () is crucial for the
emissions that is Gaussian with estimated mean value μ and covariance global emissions. A good example is Fig. 3c, which is a result of our
matrix Σ x . Therefore, the uncertainty quantification of the estimated calculated emission, but after applying source fractions from Brahney
release rate is a direct output of the inversion method. Hence, we can et al. (2021), as reported for deposition (see Table S2); the latter gives
calculate the uncertainty of each release element as: total emitted MPs to be four times lower (~2.4 Tg y-1). Another problem
√̅̅̅̅̅̅̅̅ with the oceanic emissions of MPs reported here is the restricted
xi = μx,i ± Σ x,ii (13) network of measurements (see also Section 3.2). Although an assessment
of the introduced uncertainty over the inversion domain took place in
where μx,i is ith element of the estimated mean value and Σ x,ii is ith di
the previous section, we admit that the deposition measurements used
agonal element of the estimated covariance matrix, whose square root for inverse modelling in the present study are not unique to assess
represents the standard deviation of the posterior emissions. To identify oceanic emissions. The reason is that they are located far from the US
the total release uncertainty, we used properties of Gaussian distribution coast covering only a very small domain in the center of the inversion
and calculated the uncertainty bounds of the posterior releases as fol domain. On the other hand, oceanic surface area in the selected inver
lows: sion domain lies in the west (Pacific Ocean) and in the south (Gulf of
U=
∑ √̅̅̅̅̅̅̅̅̅̅̅̅
μx,i ± Σ i Σ x,ii (14) Mexico) covering only 10%. As expected, SRMs that are used in the
i inversion algorithm to define the connection between sources and ob
servations are extremely low in oceanic regions (). A more careful
The calculated posterior relative uncertainties for MPs and MFs are
assessment of the oceanic sources of MPs would require several obser
depicted in Fig. 6 for the first (ensemble) and the second (Gaussian) case,
vations to be taken in different oceanic locations, e.g., during ship
while the combined relative uncertainty is a propagation of the latter
campaigns and/or remote islands, which lacks in the current study.
two. The uncertainty using the inversion algorithm ensemble of nine
Global uncertainty estimated for the land-based sources was much
members shows that uncertainties grow up to 50% as we move to the
lesser, in the order of about 30% (Fig. 3b), as a result of the perturbation
eastern part of the inversion domain. In this part of the domain the SRMs
in the source fractions for agriculture, road and mineral dust. It follows
were near zero (). The uncertainty quantification using Gaussian prop
the same pattern with emissions with maxima close to the largest
erty of posterior distribution (Fig. 6c and d) depends on the estimated
sources.
covariance matrix within the LS-APC model (Section 2.3) and is, in its
essence, uncertainty of the posterior model caused by the sparsity of
4.3. Land – Ocean interactions
measurements rather than uncertainty of estimated emissions. The
principle of the model is to tighten the values of the covariance matrix in
MPs and MFs emitted in the atmosphere, especially at smaller sizes,
spatial elements, where only few non-zero SRM are available (see on the
undergo long-range transport. Today, they have been already detected
eastern part). That subsequently tightens the estimated emissions to the
almost everywhere on earth (Allen et al., 2019, 2020; Bergmann et al.,
prior value x0 (see Eq. 6). However, this leaves low uncertainty in the
2019; Dris et al., 2015, 2016; González-Pleiter et al., 2021b; Kelly et al.,
posterior. On the other hand, in spatial elements with strong SRM, the
2020; Qian and Ferro, 2008 and many others). However, it has been
estimated values of covariance matrix are so large that the estimated
shown that oceanic emissions of MPs can be transported (Allen et al.,
emissions do not depend on the prior value x0 , but rather on the data
2020), similar to marine particles, when breaking waves cause bubbles
term. This causes larger variability in the covariance matrix resulting in
of trapped air to rise to the surface and burst (Erinin et al., 2019). How
larger uncertainty in Eq. (12).
much of the land-based emissions end into the ocean and vice versa
Since the principles in uncertainty quantification in the case of
remains unknown and is a frequent question by researchers conducting
ensemble approach and in the case of Gaussian approach are not
measurements often trying to interpret whether their measurements
consistent, we calculate a propagation of these two in Fig. 6e and f. We
refer to primary or secondary sources.
end up with uncertainties reaching 50% near the measurement stations
To identify this, the reported modelled annual deposition of the
and in the easternmost parts of the inversion domain.
emitted MPs caused by sea spray was masked towards land, whereas
annual deposition from the land-based sources of agriculture, road and
4.2. Sensitivity to different source fractions mineral dust was masked towards ocean, and the resulting budgets were
calculated (Fig. 7). We report that 13 ± 6.5 kt y-1 of MPs were trans
The extrapolation of the posterior emissions in global scale is based ferred from land to ocean or about 1.8% of the land emitted mass, due to
on the assumption of specific sources for MPs and MFs. While for MFs the already mentioned limited transport due to the large particle sizes.
the main source is mostly clothing, this is not the case for MPS, as they The deposition occurred close to the shoreline, while a more consistent
are known to originate mainly from transportation (primary source), transport occurred at regions surrounded by land, such as the Mediter
mineral dust (secondary source), agricultural activity (secondary ranean, North and Black Sea. Accordingly, about 122 ± 66.1 kt y-1 were
source) and sea spray (secondary source), but perhaps from other transferred from ocean to land or about 1.4% of the oceanic emitted
sources not yet defined. The main question is what the exact fractions of mass and mostly accumulated in mid-latitudes of the northern hemi
these specific sources constitute total MPs. To tackle this lack of sphere (30–60◦ N), where the largest oceanic emissions were calculated
knowledge, we accepted that primary MPs (here road dust) contribute (Fig. 3). The same latitudinal band of the southern hemisphere also gives
15–31% (average 23%), as seen in the ocean (Boucher and Friot, 2017; high oceanic emissions, but land largely lacks there, except for the
Goßmann et al., 2021; Zhao et al., 2019) and we perturbed the rest of the southernmost parts of South America and Australia. As regards to MFs,
sources building 30 ensemble members, each with different source we estimate that 405 ± 201 kt are transported and deposited to the
contribution scenarios that can be seen in Table S2. We calculate the ocean annually.
global emission uncertainty as the standard deviation of the global re Boucher and Friot (2017) reported that around 15% of marine
leases that resulted from this 30-member ensemble. plastics is a result of atmospheric transport and deposition to the global
The absolute uncertainty is depicted in Fig. 3b, side-by-side with the ocean. Between 5.3 and 14 Tg of plastics enter the global ocean annually
global annual releases of MFs. Note that such an assessment is not (Eriksen et al., 2014; Jambeck et al., 2015; Jang et al., 2016). Although a
possible for MFs, as we only assumed that they originated from clothing rough estimation, combining these two numbers, it is found that
of the global population. Uncertainty reaches 80% where the largest 0.80–2.1 Tg y-1 is the number of plastics that are transported by air.
oceanic emissions were calculated. This is a direct consequence of the Around 92% of these oceanic plastics are MPs (Auta et al., 2017; Eriksen
fact that oceanic emissions calculated here are the vast majority of total et al., 2014), or 0.74–1.9 Tg y-1 (< 5 mm). Here, we report that 0.405
11
Fig. 7. (a) MPs emitted from the land that were deposited over the ocean annually and vice versa (ocean to land). (b) MFs emitted from global population that were
deposited in the ocean.
range, albeit we have not considered larger sizes than 2500 µm for MFs
± 0.201 and 0.013 ± 0.0065 Tg y-1 of synthetic MFs (<2500 µm) and
and 250 µm for MPs that are heavier. Furthermore, atmospheric trans
MPs (<250 µm), respectively, are transported from the land to ocean.
port of MPs that end to the ocean are an order of magnitude lower than
This is comparable with the reported values although in the lowest
Table 3
Observations of MPs and MFs in deposition and surface air samples from the global literature for various years. Considering that the modelled lifetime of MPs and MFs
is very short, we assume that transport is less important; therefore, we use the global simulation for the year 2018 to assess how the modelled MPs and MFs compare
with observations.
Sampling method Measured size Location Concentration Year This study
-2 -1
(Roblin et al., 2020) Deposition 50 µm Ireland 12 fibers m d 2017–2018 32 fibers m-2 d-1
(Dris et al., 2016, 2015) Deposition 50 µm Paris 53–118 items m-2 d-1 2014 200 items m-2 d-1
(Trainic et al., 2020) Outdoor active 5 µm North Atlantic 0.01 items m-3 2016 0.6 items m-3
(Allen et al., 2019) Deposition 5 µm French Pyrenees 365 items m-2 d-1 2017–2018 150 items m-2 d-1
(Allen et al., 2020) Outdoor active 5 µm French Atlantic 2.9–9.6 items m-3 2018 1.5 items m-3
(Peñalver et al., 2021) Outdoor active 10 µm South Spain 36 ng m-3 2017 48 ng m-3
(Wright et al., 2020) Deposition 5 µm London 575–1008 items m-2 d-1 2018 500 items m-2 d-1
(Bergmann et al., 2019) Snow deposition 11 µm Arctic/Swiss Alps/Germany 1.4–66 items m-2 y-1 2015–2017 180 items m-2 d-1
(Materić et al., 2021) Snow deposition < 1 µm Alps 42 kg km-2 y-1 2017 NP not considered
(Abbasi et al., 2019) Outdoor active 2 µm Iran 0.3–1.1 items m-3 2017 0.8 items m-3
(Abbasi and Turner, 2021) Deposition < 100 µm Iran 7–120 items m-2 d-1 2019–2020 450 items m-2 d-1
(Ding et al., 2021) Outdoor active < 200 µm South China Sea 0.035 items m-3 2019 0.5 items m-3
(Klein and Fischer, 2019) Deposition 5–13 µm Hamburg 136–512 items m-2 d-1 2017–2018 410 items m-2 d-1
(Szewc et al., 2021) Deposition 5 µm Baltic Sea 136–512 items m-2 d-1 2017–2018 115 items m-2 d-1
(K.Liu et al., 2019a) Outdoor active < 1 mm Shanghai, China 1.42 items m-3 2018 1.5 items m-3
(Cai et al., 2017) Deposition 200–700 µm Dongguan, China 175–313 items m-2 d-1 2016 586 items m-2 d-1
(Zhou et al., 2017) Deposition < 0.5 mm Yantai, China 475 items m-2 d-1 2016 489 items m-2 d-1
(K.Liu et al., 2019b) Outdoor active 20 µm West Pacific coast 0.0–1.4 items m-3 2019 0.7 items m-3
(Knobloch et al., 2021) Deposition 20 µm New Zealand 1018 items m-2 d-1 2020 549 items m-2 d-1
(Huang et al., 2021) Deposition < 50 µm Guangzhou, China 51–178 items m-2 d-1 2018–2019 286 items m-2 d-1
(Wang et al., 2020) Outdoor active 60 µm S. China Sea / E. Indian Ocean 0.04–0.08 items m-3 2019 0.12–0.21 items m-3
(Wang et al., 2021) Outdoor active 20 µm China Sea 0.0039 items m-3 2020 0.45 items m-3
(Liao et al., 2021) Outdoor active 5 µm – 5 mm Wenzhou, China 189 items m-3 2019 0.74 items m-3
(Tunahan Kaya et al., 2018) Outdoor active 50 µm – 5 mm Turkey 116–3424 items m-3 2016–2017 0.19 items m-3
(Asrin and Dipareza, 2019) Outdoor active 500 µm – 5 mm Indonesia 131–174 items m-3 2017 1.5 items m-3
(Li et al., 2020) Outdoor active 5 µm – 2 mm Beijing, China 5600–5700 items m-3 2019 17 items m-3
(Gaston et al., 2020) Outdoor active 20–3000 µm Cal State Univ., USA 13–22 items m-3 2019 22 items m-3
(Syafei et al., 2019) Outdoor active 500–5000 µm Indonesia 56–175 items m-3 2018 1.8 items m-3
(Akhbarizadeh et al., 2021) Outdoor active < 2.5 Iran 0.0–14 items m-3 2016–2017 2.5 items m-3
(González-Pleiter et al., 2021a) Outdoor active 30–70 µm Madrid, Spain 1.5–14 items m-3 2020 2.9 items m-3
(González-Pleiter et al., 2021b) Deposition 2.3–12.6 mm Collins Glacier, Antarctica 0.08–0.17 items m-2 d-1 2020 1.5 items m-2 d-1
(Truong et al., 2021) Deposition 50–5000 µm Ho Chi Minh, Vietnam 71–917 items m-2 d-1 2018–2019 211 items m-2 d-1
(Stanton et al., 2019) Deposition 38 µm – 5 mm Nottingham, UK 0–31 items m-2 d-1 2018 188 items m-2 d-1
(Hamilton et al., 2021) Deposition 80–5000 µm Nunavut, Canada 500–6000 items m-2 d-1 2018 15 items m-2 d-1
(Yukioka et al., 2020) Deposition 75 µm – 5 mm Kusatsu, Japan 0.4 items m-2 d-1 2017 12 items m-2 d-1
(Yukioka et al., 2020) Deposition 75 µm – 5 mm Da Nang, Vietnam 4.0 items m-2 d-1 2017 8.6 items m-2 d-1
(Yukioka et al., 2020) Deposition 75 µm – 5 mm Kathmandu, Nepal 12.5 items m-2 d-1 2017 101 items m-2 d-1
(Zhu et al., 2021) Outdoor active 5–5000 µm Beijing, China 393 items m-3 2019 23 items m-3
(Zhu et al., 2021) Outdoor active 5–5000 µm Tianjin, China 324 items m-3 2019 14 items m-3
(Zhu et al., 2021) Outdoor active 5–5000 µm Nanjing, China 177 items m-3 2019 12 items m-3
(Zhu et al., 2021) Outdoor active 5–5000 µm Sanghai, China 267 items m-3 2019 20 items m-3
(Zhu et al., 2021) Outdoor active 5–5000 µm Hangzhou, China 246 items m-3 2019 13 items m-3
(Allen et al., 2021) Outdoor active < 50 µm Pic du Midi, France 0.09–0.66 items m-3 2017 0.05 items m-3
12
the riverine transported MPs to the ocean (3.3–14 Tg y-1).
4.4. Independent validation and forecast of expected levels
As it was already mentioned in Section 3.1, the observations from

Brahney et al. (2020) used in the inversion algorithm should not serve as
measurements to validate the posterior emissions of MPs and MFs. The
reason for this is because the inversion algorithm has been designed to
reduce the model–observation mismatches. This means that the reduc
tion of the posterior concentration mismatches with the observations is
determined by the weighting that is given to the observations and,
hence, such a comparison depends on this weighting (dependent
observation). Therefore, the ideal comparison would be against mea
surements that were not included in the inversion algorithm. There are
several literature records on MP and MF measurements in the environ
ment using various techniques, both for surface concentrations and
deposition rates (Table 3). They refer to the recent years between 2014
and 2020, whereas measured particles sizes are extremely large (up to
5000 µm) in most cases due to specific limitations of the available
analytical techniques. In the present study, we report optimised MPs and
MFs emissions for sizes up to 250 µm and 5000 µm, respectively. Our Fig. 8. Independent validation of modelled concentration and deposition of
results suggest limited transport due to the large particle size consid MPs and MFs against observations from the relevant literature (Table 3).
ered. Therefore, we use MP and MF concentrations and deposition rates Scatterplots of modelled results against observations were plotted using the
from Table 3 as a proxy to assess how modelled results from our global Kernel density estimation, which is a way to estimate the probability density
simulation for the year 2018 compare with observations assuming function (PDF) of a random variable in a non-parametric way. The Mean
year-by-year meteorology has limited effect on transport of large Fractional Bias (MFB) is also computed separately for concentrations and
particles. deposition rates.
We use the Gaussian kernel density estimation (KDE), which is a non-
parametric way to estimate the probability density function (PDF) of a we only assumed that MPs and MFs are scavenged moderately in and
random variable (Parzen, 1962): below clouds (Table S1), as they are synthetic polymers (macromole
cules) and should have hydrophobic behavior. However, without spe
1 ∑N (x − xi )
f (x) = K (15) cific measurements of the scavenging coefficients for MPs and MFs, it is
Nh i=1 h
impossible to know how they behave in the atmosphere, in order to
where K is the kernel, xi the univariate independent and identically accurately reproduce their transport and removal in global models.
distributed point of the relationship between modelled and measured These properties are crucial for accurate representation of MPs, both in
ammonia and h is a smoothing parameter called the bandwidth. KDE is a forward modelling (atmospheric dispersion of a known source), as well
fundamental data smoothing tool that attempts to infer characteristics of as in inverse algorithms (source quantification). Nevertheless, the sim
a population, based on a finite dataset. It weighs the distance of all ulations presented in this study can be used by researchers who plan to
points in each specific location along the distribution. If there are more perform sampling and analysis of MPs and MFs in order to forecast the
points grouped locally, the estimation is higher as the probability of expected levels at any place on earth in high temporal and spatial res
seeing a point at that location increases. The kernel function is the olutions (0.5◦ ×0.5◦ , daily).
specific mechanism used to weigh the points across the data set and it
uses the bandwidth to limit the scope of the function. The latter is
4.5. Robustness of the inverse modelling methodology
computed using the Scott’s factor (Scott, 2015). We also provide the
mean fractional bias (MFB) for modelled and measured separately for
Brahney et al. (2021) have successfully managed to calculate total
concentrations and deposition rates as follows:
emissions of MPs, though presenting somewhat large model-observation
∑N
1 (C − C ) mismatches. The core of their methodology was to minimize the cost
MFB = (i=1 m) o (16)
N ∑N Cm +Co function based on the goodness of fit between modelled values and
× 100%
i=1 2 measurements weighted by the model-observation error. Moreover, a
scalar regularization term was added to suppress negative values of
where Cm and Co are the modelled and measured quantities and N is the estimated emissions. Their estimated emissions are based on an opti
total number of observations. MFB is a symmetric performance indicator mized combination of five known annual sources, road dust, ocean,
that gives equal weights to under- or overestimated concentrations agricultural dust, population dust, and population. The combination is
(minimum to maximum values range from − 200% to 200%). calculated using a global search method for these five sources with 30
The comparison of modelled surface concentrations and deposition possible strengths for each source (from zero to the value explaining the
rates with the observations is shown in Fig. 8. As seen both in Fig. 8 and whole measurements). Since the estimated emissions rely on the usage
Table 3 the modelled concentrations are in the same order with obser of annual sources, the calculated emissions lack any temporal variation.
vations except for some outliers (e.g., in Nunavut, Canada from Ham This simplification imposes the shape of spatial distribution of emissions
ilton et al., 2021). The calculated MFBs suggest that the model tends to and might be the key reason for the poor fit of modelled values with
underestimate concentrations (MFBcon = − 57%) and overestimate measurements. In contrast with the aforementioned methodology, here
deposition rates (MFBdep = + 39%). The latter shows that more infor we do not make use of any pre-computed source, but rather use a
mation is required to understand how efficient CCN or IN (Ganguly and completely data-driven approach (Tichý et al., 2016) for the current
Ariya, 2019b) MPs and MFs are, how dry deposition affects removal spatial inversion domain (124–91◦ W, 29–47◦ N). This method calculates
from the atmosphere and, in turn, how they should be modelled in MP and MF emissions with large spatiotemporal resolution (0.5◦ ×0.5◦ ,
global models. As regards to scavenging that is a more uncertain process, daily).
13
5. Conclusions - The largest continental MP deposition occurred in Asia (267

± 121 kt y-1), North America (160 ± 71 kt y-1), Africa (114
We have used a robust Bayesian inverse modelling algorithm com ± 59 kt y-1) and Europe (102 ± 45 kt y-1) similar to MFs (Asia: 3792
bined with a Lagrangian particle dispersion model suitable to track ± 1933 kt y-1, America: 846 ± 209 kt y-1, Africa: 801 ± 372 kt y-1,
deposited particles in backward mode to calculate high spatiotemporal Europe: 598 ± 253 kt y-1).
emissions of MPs and MFs from observations in the Western USA. We - The largest oceanic MP deposition occurred in the Atlantic (1718
further extrapolate these calculations to retrieve global estimates of ± 899.5 kt y-1), Pacific (2751 ± 1225 kt y-1), Indian (1435
atmospheric MP and MF fluxes from different sectorial emissions. ± 766 kt y-1) and Southern Ocean (2289 ± 1185 kt y-1).
Finally, we feed the calculated emissions into a dispersion model to track
global atmospheric dynamics and budgets of MPs and MFs. Our results Since the particles considered in the present study are large and their
are openly accessible and can be used as a proxy from research groups atmospheric lifetimes short, we validated the present results with global
conducting surface concentration and deposition measurements to know measurement taken during the last decade assuming that meteorology
the expected levels of MPs and MFs at any given place on earth. Further does not have a major effect on particles that are removed from the
conclusions are summarised below: atmosphere very fast. We report that the current model set-up un
derestimates surface concentrations and overestimates deposition rates.
- Around 9.0 ± 3.8 kt y-1 of atmospheric MPs of size up to 250 µm and This means that the coefficients for in-cloud and below-cloud scav
244 ± 129 kt y-1 of MFs of size up to 2500 µm were released in the enging and dry deposition processes that are considered in global
inversion domain that includes central and west USA. Lower emis models need to be updated.
sions closer to the observations are due to remoteness of the stations
(far from any man-made activity). CRediT authorship contribution statement
- Global MP emissions were estimated to be 9.6 ± 3.6 Tg y-1, whereas
MF emissions equal to 6.5 ± 2.9 Tg y-1. Nikolaos Evangeliou: Conceptualization, Lagrangian modelling,
- Ocean dominates MP emissions with 8.9 ± 3.5 Tg y-1, as insoluble Investigation, Writing – review & editing, Supervision. Ondřej Tichý :
plastics accumulate at the surface of the ocean with time and are Methodology, Inverse modelling algorithm development & optimisa
resuspended similar to sea spray aerosols. We note that this calcu tion, Writing – review & editing. Sabine Eckhardt: Lagrangian
lation is highly uncertain because (i) the inversion domain covers a modelling, Writing – review & editing. Christine Groot Zwaaftink:
small oceanic surface that can be used in interpolation, (ii) the Calculation of global dust emissions using FLEXDUST model, Writing –
deposition measurements used in the inversion are far from the review & editing. Janice Brahney: Sample collection & analysis, Re
ocean, thus inappropriate to constrain oceanic emissions, and (iii) it view & editing.
is assumed that a constant number of oceanic microplastics is
released globally following sea spray.
Declaration of Competing Interest
- Agricultural activities resuspend around 0.31 ± 0.13 Tg y-1, road
dust contributes another 0.28 ± 0.12 Tg y-1, and mineral dust 0.10
The authors declare that they have no competing financial interests
± 0.052 Tg y-1.
or personal relationships that could have influenced the work reported
- The largest emissions are calculated for Asia (MPs: 250 ± 120 kt y-1
in this paper.
– MFs: 3700 ± 1300 kt y-1), North America (MPs: 130 ± 55.4 kt y-1 –
MFs: 280 ± 120 kt y-1), Africa (MPs: 110 ± 55.1 kt y-1 – MFs: 970
Acknowledgements
± 452 kt y-1) and Europe (MPs: 88 ± 42 kt y-1 – MFs: 460
± 140 kt y-1), while the rest of the continents have smaller shares in
The work was funded by NILU and COMBAT (Quantification of
annual emissions.
Global Ammonia Sources constrained by a Bayesian Inversion Tech
- Global average monthly mass concentrations were at maximum
nique) project funded by ROMFORSK – Program for romforskning of the
47 ng m-3 for MPs and 33 ng m-3 for MFs at the surface, while both
Research Council of Norway (Project ID: 275407), website: https://
are rapidly removed from the atmosphere, due to the small lifetimes
prosjektbanken.forskningsradet.no/project/FORISS/275407?Kil
affected by their large particle sizes.
de=FORISS&distribution=Ar&chart=bar&calcType=funding&Sprak=
- The largest deposition from agricultural sources (annual total: 310
no&sortBy=date&sortOrder=desc&resultCount= 30&offset= 0&Prog
± 155 kt y-1) occurred in Central USA, in the Indo-Gangetic Plain
Akt.3 =ROMFORSK-Program+for+romforskning. Dr. Ondřej Tichý was
and the North China Plain, all regions of great agricultural activity.
supported by the Czech Science Foundation, grant no. GA20-27939S.
Road MPs (annual total: 279 ± 145 kt y-1) were mostly deposited in
the US East Coast, Central Europe, and Southeastern Asia, whereas
mineral dust MPs deposited near Sahara and Middle East (annual Novelty
total: 100 ± 53 kt y-1). Oceanic MPs were deposited mostly in the
Ocean. We combine high-quality deposition measurements from the West
- Only 1.8% of the land MP mass emissions (13 ± 6.6 kt y-1) were ern USA with a dispersion model and an inverse modelling algorithm to
transferred to ocean, due to the limited transport of large particle constrain global atmospheric emissions of microplastics (MPs) and
considered. About 1.4% (122 ± 65 kt y-1) were transferred from microfibers (MFs) in high spatiotemporal resolution. The only global
ocean to land and accumulated in mid-latitudes of the north dispersion model that can track deposition backward in time is used for
hemisphere. the first time in an inverse modelling approach. The constrained emis
- It is reported that 0.74–1.9 Tg y-1 of MPs (< 5 mm) are globally sions are used further to model global atmospheric dynamics of MPs and
transported by air from land to ocean. We calculate that 0.418 MFs. We address the expected surface concentrations, and deposition
± 0.201 of synthetic MFs (<2500 µm) and MPs (<250 µm), are rates of atmospheric MPs and MFs in a gridded product that aims at
transported from the land to ocean. This close to the reported values assisting researchers conducting measurements at a global scale.
although in the lowest range, due to exclusion of larger sizes (>
2500 µm for MFs and >250 µm for MPs) from this study that are Appendix A. Supporting information
heavier. Atmospheric transport of MPs that end to the ocean are an
order of magnitude lower than the riverine transported MPs to the Supplementary data associated with this article can be found in the
ocean (3.3–14 Tg y-1) online version at doi:10.1016/j.jhazmat.2022.128585.
14
References Colton, J.B., Knapp, F.D., Bums, B.R., 1974. Plastic particles in surface waters of the
Northwestern Atlantic. Science 185, 491–497.
Coyle, R., Hardiman, G., Driscoll, K.O., 2020. Microplastics in the marine environment: a
Abbasi, S., Keshavarzi, B., Moore, F., Turner, A., Kelly, F.J., Dominguez, A.O.,
review of their sources, distribution processes, uptake and exchange in ecosystems.
Jaafarzadeh, N., 2019. Distribution and potential health impacts of microplastics
Case Stud. Chem. Environ. Eng. 2, 100010 https://doi.org/10.1016/j.
and microrubbers in air and street dusts from Asaluyeh County, Iran. Environ. Pollut.
cscee.2020.100010.
244, 153–164. https://doi.org/10.1016/j.envpol.2018.10.039.
De Meutter, P., Hoffman, I., Ungar, K., 2020. On the model uncertainties in Bayesian
Abbasi, S., Turner, A., 2021. Dry and wet deposition of microplastics in a semi-arid
source reconstruction using the emission inverse modelling system FREARtool v1.0
region (Shiraz, Iran). Sci. Total Environ. 786, 147358 https://doi.org/10.1016/j.
and the Lagrangian transport and dispersion model Flexpart v9.0.2. Geosci. Model
scitotenv.2021.147358.
Dev. Discuss. 1–23. https://doi.org/10.5194/gmd-2020-162.
Akhbarizadeh, R., Dobaradaran, S., Amouei Torkmahalleh, M., Saeedi, R., Aibaghi, R.,
Ding, Y., Zou, X., Wang, C., Feng, Z., Wang, Y., Fan, Q., Chen, H., 2021. The abundance
Faraji Ghasemi, F., 2021. Suspended fine particulate matter (PM2.5), microplastics
and characteristics of atmospheric microplastic deposition in the northwestern South
(MPs), and polycyclic aromatic hydrocarbons (PAHs) in air: Their possible
China Sea in the fall. Atmos. Environ. 253, 118389 https://doi.org/10.1016/j.
relationships and health implications. Environ. Res. 192, 110339 https://doi.org/
atmosenv.2021.118389.
10.1016/j.envres.2020.110339.
Dris, R., Gasperi, J., Rocher, V., Saad, M., Renault, N., Tassin, B., 2015. Microplastic
Allen, S., Allen, D., Baladima, F., Phoenix1, V.R., Thomas, J.L., Roux, G., Le, Sonke, J.E.,
contamination in an urban area: a case study in Greater Paris. Environ. Chem. 12,
2021. Evidence of free tropospheric and long-range Observatory. Nat. Commun. 12,
592–599. https://doi.org/10.1071/EN14167.
7242. https://doi.org/10.1038/s41467-021-27454-7.
Dris, R., Gasperi, J., Saad, M., Mirande, C., Tassin, B., 2016. Synthetic fibers in
Allen, S., Allen, D., Moss, K., Le Roux, G., Phoenix, V.R., Sonke, J.E., 2020. Examination
atmospheric fallout: a source of microplastics in the environment? Mar. Pollut. Bull.
of the ocean as a source for atmospheric microplastics. PLoS One 15, 1–14. https://
104, 290–293. https://doi.org/10.1016/j.marpolbul.2016.01.006.
doi.org/10.1371/journal.pone.0232746.
Dumont Le Brazidec, J., Bocquet, M., Saunier, O., Roustan, Y., 2021. Quantification of
Allen, S., Allen, D., Phoenix, V.R., Le Roux, G., Durántez Jiménez, P., Simonneau, A.,
the modelling uncertainties in atmospheric release source assessment and
Binet, S., Galop, D., 2019. Atmospheric transport and deposition of microplastics in a
application to the reconstruction of the autumn 2017 Ruthenium 106 source. Atmos.
remote mountain catchment. Nat. Geosci. 12, 339–344. https://doi.org/10.1038/
Chem. Phys. Discuss. 1–28. https://doi.org/10.5194/acp-2020-1129.
s41561-019-0335-5.
Eckhardt, S., Cassiani, M., Evangeliou, N., Sollum, E., Pisso, I., Stohl, A., 2017. Source-
An, L., Liu, Q., Deng, Y., Wu, W., Gao, Y., Ling, W., 2020. Sources of microplastic in the
receptor matrix calculation for deposited mass with the Lagrangian particle
environment. Handbook of Environmental Chemistry. Springer, pp. 143–159.
dispersion model FLEXPART v10.2 in backward mode. Geosci. Model Dev. 10
https://doi.org/10.1007/698_2020_449.
https://doi.org/10.5194/gmd-10-4605-2017.
Asrin, N., Dipareza, A., 2019. Microplastics in ambient air (case study: Urip Sumoharjo
Eriksen, M., Lebreton, L.C.M., Carson, H.S., Thiel, M., Moore, C.J., Borerro, J.C.,
Street and Mayjend Sungkono Street of Surabaya City, Indonesia). IAETSD J. Adv.
Galgani, F., Ryan, P.G., Reisser, J., 2014. Plastic pollution in the world’s oceans:
Res. Appl. Sci. 6, 54–57.
more than 5 trillion plastic pieces weighing over 250,000 tons afloat at sea. PLoS
Athey, S.N., Adams, J.K., Erdle, L.M., Jantunen, L.M., Helm, P.A., Finkelstein, S.A.,
One 9, 1–15. https://doi.org/10.1371/journal.pone.0111913.
Diamond, M.L., 2020. The widespread environmental footprint of indigo denim
Erinin, M.A., Wang, S.D., Liu, R., Towle, D., Liu, X., Duncan, J.H., 2019. Spray generation
microfibers from blue jeans. Environ. Sci. Technol. Lett. 7, 840–847. https://doi.org/
by a plunging breaker. Geophys. Res. Lett. 46, 8244–8251. https://doi.org/10.1029/
10.1021/acs.estlett.0c00498.
2019GL082831.
Auta, H.S., Emenike, C.U., Fauziah, S.H., 2017. Distribution and importance of
Evangeliou, N., Balkanski, Y., Eckhardt, S., Cozic, A., Van Damme, M., Coheur, P.-F.,
microplastics in the marine environmentA review of the sources, fate, effects, and
Clarisse, L., Shephard, M., Cady-Pereira, K., Hauglustaine, D., 2021. 10–year
potential solutions. Environ. Int. 102, 165–176. https://doi.org/10.1016/j.
satellite–constrained fluxes of ammonia improve performance of chemistry transport
envint.2017.02.013.
models. Atmos. Chem. Phys. 21, 4431–4451. https://doi.org/10.5194/acp-21-4431-
Barnes, D.K.A., Galgani, F., Thompson, R.C., Barlaz, M., 2009. Accumulation and
2021.
fragmentation of plastic debris in global environments. Philos. Trans. R. Soc. B Biol.
Evangeliou, N., Grythe, H., Klimont, Z., Heyes, C., Eckhardt, S., Lopez-Aparicio, S.,
Sci. 364, 1985–1998. https://doi.org/10.1098/rstb.2008.0205.
Stohl, A., 2020. Atmospheric transport is a major pathway of microplastics to remote
Bergmann, M., Mützel, S., Primpke, S., Tekman, M.B., Trachsel, J., Gerdts, G., 2019.
regions. Nat. Commun. 11. https://doi.org/10.1038/s41467-020-17201-9.
White and wonderful? Microplastics prevail in snow from the Alps to the Arctic. Sci.
Fakour, H., Lo, S.L., Yoashi, N.T., Massao, A.M., Lema, N.N., Mkhontfo, F.B.,
Adv. 5, 1–11. https://doi.org/10.1126/sciadv.aax1157.
Jomalema, P.C., Jumanne, N.S., Mbuya, B.H., Mtweve, J.T., Imani, M., 2021.
Blettler, M.C.M., Abrial, E., Khan, F.R., Sivri, N., Espinola, L.A., 2018. Freshwater plastic
Quantification and analysis of microplastics in farmland soils: characterization,
pollution: Recognizing research biases and identifying knowledge gaps. Water Res.
sources, and pathways. Agriculture 11. https://doi.org/10.3390/
143, 416–424. https://doi.org/10.1016/j.watres.2018.06.015.
agriculture11040330.
Bond, T.C., Doherty, S.J., Fahey, D.W., Forster, P.M., Berntsen, T., Deangelo, B.J.,
Fendall, L.S., Sewell, M.A., 2009. Contributing to marine pollution by washing your face:
Flanner, M.G., Ghan, S., Kärcher, B., Koch, D., Kinne, S., Kondo, Y., Quinn, P.K.,
Microplastics in facial cleansers. Mar. Pollut. Bull. 58, 1225–1228. https://doi.org/
Sarofim, M.C., Schultz, M.G., Schulz, M., Venkataraman, C., Zhang, H., Zhang, S.,
10.1016/j.marpolbul.2009.04.025.
Bellouin, N., Guttikunda, S.K., Hopke, P.K., Jacobson, M.Z., Kaiser, J.W., Klimont, Z.,
Forster, C., Stohl, A., Seibert, P., 2007. Parameterization of convective transport in a
Lohmann, U., Schwarz, J.P., Shindell, D., Storelvmo, T., Warren, S.G., Zender, C.S.,
Lagrangian particle dispersion model and its evaluation. J. Appl. Meteorol. Clim. 46,
2013. Bounding the role of black carbon in the climate system: A scientific
403–422. https://doi.org/10.1175/JAM2470.1.
assessment. J. Geophys. Res. Atmos. 118, 5380–5552. https://doi.org/10.1002/
Ganesan, A.L., Rigby, M., Zammit-Mangion, A., Manning, A.J., Prinn, R.G., Fraser, P.J.,
jgrd.50171.
Harth, C.M., Kim, K.R., Krummel, P.B., Li, S., Mühle, J., O’Doherty, S.J., Park, S.,
Boucher, J., Friot, D., 2017. Primary Microplastics in the Oceans: a Global Evaluation of
Salameh, P.K., Steele, L.P., Weiss, R.F., 2014. Characterization of uncertainties in
Sources, Primary microplastics in the oceans: A global evaluation of sources. IUCN
atmospheric trace gas inversions using hierarchical Bayesian methods. Atmos. Chem.
Gland, Switzerland. https://doi.org/10.2305/iucn.ch.2017.01.en.
Phys. 14, 3855–3864. https://doi.org/10.5194/acp-14-3855-2014.
Brahney, J., Hallerud, M., Heim, E., Hahnenberger, M., Sukumaran, S., 2020. Plastic rain
Ganguly, M., Ariya, P.A., 2019a. Ice nucleation of model nanoplastics and microplastics:
in protected areas of the United States. Science 368, 1257–1260. https://doi.org/
a novel synthetic protocol and the influence of particle capping at diverse
10.1126/science.aaz5819.
atmospheric environments. ACS Earth Space Chem. 3, 1729–1739. https://doi.org/
Brahney, J., Mahowald, N., Prank, M., Cornwell, G., Klimont, Z., Matsui, H., Prather, K.
10.1021/acsearthspacechem.9b00132.
A., 2021. Constraining the atmospheric limb of the plastic cycle. Proc. Natl. Acad.
Ganguly, M., Ariya, P.A., 2019b. Ice nucleation of model nanoplastics and microplastics:
Sci. U.S.A. 118, e2020719118 https://doi.org/10.1073/pnas.2020719118.
a novel synthetic protocol and the influence of particle capping at diverse
Browne, M.A., Crump, P., Niven, S.J., Teuten, E., Tonkin, A., Galloway, T., Thompson, R.,
atmospheric environments. ACS Earth Space Chem. 3, 1729–1739. https://doi.org/
2011. Accumulation of microplastic on shorelines woldwide: sources and sinks.
10.1021/acsearthspacechem.9b00132.
Environ. Sci. Technol. 45, 9175–9179. https://doi.org/10.1021/es201811s.
Gao, J., 2017. Downscaling Global Spatial Population Projections from 1/8-degree to 1-
Cai, L., Wang, J., Peng, J., Tan, Z., Zhan, Z., Tan, X., Chen, Q., 2017. Characteristic of
km Grid Cells. NCAR Tech. Note 9.
microplastics in the atmospheric fallout from Dongguan city, China: preliminary
Gaston, E., Woo, M., Steele, C., Sukumaran, S., Anderson, S., 2020. Microplastics differ
research and first evidence. Environ. Sci. Pollut. Res. 24, 24928–24935. https://doi.
between indoor and outdoor air masses: insights from multiple microscopy
org/10.1007/s11356-017-0116-x.
methodologies. Appl. Spectrosc. https://doi.org/10.1177/0003702820920652.
Carpenter, E.J., Anderson, S.J., Harvey, G.R., Miklas, H.P., Peck, B.B., 1972. Polystyrene
Gavigan, J., Kefela, T., Macadam-Somer, I., Suh, S., Geyer, R., 2020. Synthetic microfiber
spherules in coastal waters. Science 178, 749–750. https://doi.org/10.1126/
emissions to land rival those to waterbodies and are growing. PLoS One 15, 1–13.
science.178.4062.749.
https://doi.org/10.1371/journal.pone.0237839.
Carpenter, E.J., Smith, K.L., 1972. Plastics on the Sargasso sea surface. Science 175,
Gewert, B., Plassmann, M.M., Macleod, M., 2015. Pathways for degradation of plastic
1240–1241. https://doi.org/10.1126/science.175.4027.1240.
polymers floating in the marine environment. Environ. Sci. Process. Impacts 17,
Cassiani, M., Stohl, A., Brioude, J., 2014. Lagrangian stochastic modelling of dispersion
1513–1521. https://doi.org/10.1039/c5em00207a.
in the convective boundary layer with skewed turbulence conditions and a vertical
Godfrey, M., 2021. Toward eliminating pre-consumer emissions of microplastics from the
density gradient: formulation and implementation in the FLEXPART model. Bound.
textile industry. San Francisco.
-Layer. Meteor. 154, 367–390. https://doi.org/10.1007/s10546-014-9976-5.
Golub, G.H., Hansen, P.C., O’Leary, D.P., 1999. Tikhonov regularization and total least
Chae, Y., An, Y.J., 2018. Current research trends on plastic pollution and ecological
squares. SIAM J. Matrix Anal. Appl. 21, 185–194. https://doi.org/10.1137/
impacts on the soil ecosystem: a review. Environ. Pollut. 240, 387–395. https://doi.
S0895479897326432.
org/10.1016/j.envpol.2018.05.008.
González-Pleiter, M., Edo, C., Aguilera, Á., Viúdez-Moreiras, D., Pulido-Reyes, G.,
Chen, G., Feng, Q., Wang, J., 2020. Mini-review of microplastics in the atmosphere and
González-Toril, E., Osuna, S., de Diego-Castilla, G., Leganés, F., Fernández-Piñas, F.,
their risks to humans. Sci. Total Environ. 703, 135504 https://doi.org/10.1016/j.
Rosal, R., 2021a. Occurrence and transport of microplastics sampled within and
scitotenv.2019.135504.
15
above the planetary boundary layer. Sci. Total Environ. 761, 143213 https://doi. Kelly, A., Lannuzel, D., Rodemann, T., Meiners, K.M., Auman, H.J., 2020. Microplastic
org/10.1016/j.scitotenv.2020.143213. contamination in east Antarctic sea ice. Mar. Pollut. Bull. 154, 111130 https://doi.
González-Pleiter, M., Edo, C., Velázquez, D., Casero-Chamorro, M.C., Leganés, F., org/10.1016/j.marpolbul.2020.111130.
Quesada, A., Fernández-Piñas, F., Rosal, R., 2020. First detection of microplastics in Klein, M., Fischer, E.K., 2019. Microplastic abundance in atmospheric deposition within
the freshwater of an Antarctic specially protected area. Mar. Pollut. Bull. 161, 1–6. the Metropolitan area of Hamburg, Germany. Sci. Total Environ. 685, 96–103.
https://doi.org/10.1016/j.marpolbul.2020.111811. https://doi.org/10.1016/j.scitotenv.2019.05.405.
González-Pleiter, M., Lacerot, G., Edo, C., Pablo Lozoya, J., Legane´s, F., Ferna´ndez- Knobloch, E., Ruffell, H., Aves, A., Pantos, O., Gaw, S., Revell, L.E., 2021. Comparison of
Pina&tild;s, F., Rosal, R., Teixeira-De-Mello, F., 2021b. A pilot study about Deposition Sampling Methods to Collect Airborne Microplastics in Christchurch,
microplastics and mesoplastics in an Antarctic glacier. Cryosphere 15, 2531–2539. New Zealand. Water Air Soil Pollut. 232. https://doi.org/10.1007/s11270-021-
https://doi.org/10.5194/tc-15-2531-2021. 05080-9.
Goßmann, I., Halbach, M., Scholz-Böttcher, B.M., 2021. Car and truck tire wear particles Kullback, S., Leibler, R.A., 1951. On information and sufficiency. Ann. Math. Stat. 22,
in complex environmental samples – a quantitative comparison with “traditional” 79–86. https://doi.org/10.1214/aoms/1177729694.
microplastic polymer mass loads. Sci. Total Environ. 773, 145667 https://doi.org/ Lebreton, L., Egger, M., Slat, B., 2019. A global mass budget for positively buoyant
10.1016/j.scitotenv.2021.145667. macroplastic debris in the ocean. Sci. Rep. 9, 1–10. https://doi.org/10.1038/
Gouin, T., 2021. Addressing the importance of microplastic particles as vectors for long- s41598-019-49413-5.
range transport of chemical contaminants: perspective in relation to prioritizing Lehner, R., Weder, C., Petri-Fink, A., Rothen-Rutishauser, B., 2019. Emergence of
research and regulatory actions. Micro Nanoplastics 1, 1–19. https://doi.org/ nanoplastic in the environment and possible impact on human health. Environ. Sci.
10.1186/s43591-021-00016-w. Technol. 53, 1748–1765. https://doi.org/10.1021/acs.est.8b05512.
Gregory, M.R., 1996. Plastic scrubbers in hand cleansers: a further (and minor) source for Li, Y., Shao, L., Wang, W., Zhang, M., Feng, X., Li, W., Zhang, D., 2020. Airborne fiber
marine pollution identified. Mar. Pollut. Bull. 32, 867–871. https://doi.org/ particles: types, size and concentration observed in Beijing. Sci. Total Environ. 705,
10.1016/S0025-326X(96)00047-1. 135967 https://doi.org/10.1016/j.scitotenv.2019.135967.
Grigoratos, T., Martini, G., 2015. Brake wear particle emissions: a review. Environ. Sci. Liao, Z., Ji, X., Ma, Y., Lv, B., Huang, W., Zhu, X., Fang, M., Wang, Q., Wang, X.,
Pollut. Res. 22, 2491–2504. https://doi.org/10.1007/s11356-014-3696-8. Dahlgren, R., Shang, X., 2021. Airborne microplastics in indoor and outdoor
Grigoratos, T., Martini, G., 2014. Non-exhaust traffic related emissions. Brake and tyre environments of a coastal city in Eastern China. J. Hazard. Mater. 417, 126007
wear PM. Publications Office of the European Union. Luxembourg. https://doi.org/ https://doi.org/10.1016/j.jhazmat.2021.126007.
10.2790/21481. Liu, J., Yang, Y., Ding, J., Zhu, B., Gao, W., 2019. Microfibers: a preliminary discussion
Groot Zwaaftink, C.D., Arnalds, Ó., Dagsson-Waldhauserova, P., Eckhardt, S., on their definition and sources. Environ. Sci. Pollut. Res. 26, 29497–29501. https://
Prospero, J.M., Stohl, A., 2017. Temporal and spatial variability of Icelandic dust doi.org/10.1007/s11356-019-06265-w.
emissions and atmospheric transport. Atmos. Chem. Phys. 17, 10865–10878. Liu, K., Wang, X., Fang, T., Xu, P., Zhu, L., Li, D., 2019a. Source and potential risk
https://doi.org/10.5194/acp-17-10865-2017. assessment of suspended atmospheric microplastics in Shanghai. Sci. Total Environ.
Grythe, H., Kristiansen, N.I., Groot Zwaaftink, C.D., Eckhardt, S., Ström, J., Tunved, P., 675, 462–471. https://doi.org/10.1016/j.scitotenv.2019.04.110.
Krejci, R., Stohl, A., 2017. A new aerosol wet removal scheme for the Lagrangian Liu, K., Wu, T., Wang, X., Song, Z., Zong, C., Wei, N., Li, D., 2019b. Consistent transport
particle model FLEXPARTv10. Geosci. Model Dev. 10, 1447–1466. https://doi.org/ of terrestrial microplastics to the ocean through atmosphere. Environ. Sci. Technol.
10.5194/gmd-10-1447-2017. 53, 10612–10619. https://doi.org/10.1021/acs.est.9b03427.
Grythe, H., Ström, J., Krejci, R., Quinn, P., Stohl, A., 2014. A review of sea-spray aerosol Liu, Y., Haussaire, J.M., Bocquet, M., Roustan, Y., Saunier, O., Mathieu, A., 2017.
source functions using a large global set of sea salt aerosol concentration Uncertainty quantification of pollutant source retrieval: comparison of Bayesian
measurements. Atmos. Chem. Phys. 14, 1277–1297. https://doi.org/10.5194/acp- methods with application to the Chernobyl and Fukushima Daiichi accidental
14-1277-2014. releases of radionuclides. Q. J. R. Meteorol. Soc. 143, 2886–2901. https://doi.org/
Habib, R.Z., Thiemann, T., Al Kendi, R., 2020. Microplastics and wastewater treatment 10.1002/qj.3138.
plants—a review. J. Water Resour. Prot. 12, 1–35. https://doi.org/10.4236/ Materić, D., Ludewig, E., Brunner, D., Röckmann, T., Holzinger, R., 2021. Nanoplastics
jwarp.2020.121001. transport to the remote, high-altitude Alps. Environ. Pollut. 288. https://doi.org/
Halle, L.L., Palmqvist, A., Kampmann, K., Khan, F.R., 2020. Ecotoxicology of micronized 10.1016/j.envpol.2021.117697.
tire rubber: past, present and future considerations. Sci. Total Environ. 706, 135694 Mattsson, K., Hansson, L.A., Cedervall, T., 2015. Nano-plastics in the aquatic
https://doi.org/10.1016/j.scitotenv.2019.135694. environment. Environ. Sci. Process. Impacts 17, 1712–1721. https://doi.org/
Hamilton, B.M., Bourdages, M.P.T., Geoffroy, C., Vermaire, J.C., Mallory, M.L., 10.1039/c5em00227c.
Rochman, C.M., Provencher, J.F., 2021. Microplastics around an Arctic seabird Mountford, A.S., Morales Maqueda, M.A., 2021. Modeling the accumulation and
colony: particle community composition varies across environmental matrices. Sci. transport of microplastics by sea ice. J. Geophys. Res. Ocean 126, 1–19. https://doi.
Total Environ. 773, 145536 https://doi.org/10.1016/j.scitotenv.2021.145536. org/10.1029/2020JC016826.
Hansen, P.C., O’Leary, D.P., 1993. The use of the L-curve in the regularization of discrete Nizzetto, L., Futter, M., Langaas, S., 2016. Are agricultural soils dumps for microplastics
ill-posed problems. Siam J. Sci. Comput. 1487–1503. of urban origin? Environ. Sci. Technol. 50, 10777–10779. https://doi.org/10.1021/
Harne, R., 2019. Studies on plastic bezoar ingestion in free range axis deer in summer. acs.est.6b04140.
J. Anim. Res. 383–386. https://doi.org/10.30954/2277-940x.02.2019.25. O’Brien, S., Okoffo, E.D., O’Brien, J.W., Ribeiro, F., Wang, X., Wright, S.L.,
Hartmann, N.B., Hüffer, T., Thompson, R.C., Hassellöv, M., Verschoor, A., Daugaard, A. Samanipour, S., Rauert, C., Toapanta, T.Y.A., Albarracin, R., Thomas, K.V., 2020.
E., Rist, S., Karlsson, T., Brennholt, N., Cole, M., Herrling, M.P., Hess, M.C., Ivleva, N. Airborne emissions of microplastic fibres from domestic laundry dryers. Sci. Total
P., Lusher, A.L., Wagner, M., 2019. Are we speaking the same language? Environ. 747, 141175 https://doi.org/10.1016/j.scitotenv.2020.141175.
Recommendations for a definition and categorization framework for plastic debris. O’Brine, T., Thompson, R.C., 2010. Degradation of plastic carrier bags in the marine
Environ. Sci. Technol. 53, 1039–1047. https://doi.org/10.1021/acs.est.8b05297. environment. Mar. Pollut. Bull. 60, 2279–2283. https://doi.org/10.1016/j.
Hegg, D.A., Warren, S.G., Grenfell, T.C., Doherty, S.J., Larson, T.V., Clarke, A.D., 2009. marpolbul.2010.08.005.
Source attribution of black carbon in arctic snow. Environ. Sci. Technol. 43, Parzen, E., 1962. On the estimation of probability density functions and mode. Ann.
4016–4021. https://doi.org/10.1021/es803623f. Math. Stat. 33, 1065–1076.
Henry, B., Laitala, K., Klepp, I.G., 2019. Microfibres from apparel and home textiles: Patil, S., Bafana, A., Naoghare, P.K., Krishnamurthi, K., Sivanesan, S., 2021.
prospects for including microplastics in environmental sustainability assessment. Sci. Environmental prevalence, fate, impacts, and mitigation of microplastics—a critical
Total Environ. 652, 483–494. https://doi.org/10.1016/j.scitotenv.2018.10.166. review on present understanding and future research scope. Environ. Sci. Pollut. Res.
Höök, M., Tang, X., 2013. Depletion of fossil fuels and anthropogenic climate change-a 28, 4951–4974. https://doi.org/10.1007/s11356-020-11700-4.
review. Energy Policy 52, 797–809. https://doi.org/10.1016/j.enpol.2012.10.046. Peeken, I., Primpke, S., Beyer, B., Gütermann, J., Katlein, C., Krumpen, T., Bergmann, M.,
Huang, Y., He, T., Yan, M., Yang, L., Gong, H., Wang, W., Qing, X., Wang, J., 2021. Hehemann, L., Gerdts, G., 2018. Arctic sea ice is an important temporal sink and
Atmospheric transport and deposition of microplastics in a subtropical urban means of transport for microplastic. Nat. Commun. 9. https://doi.org/10.1038/
environment. J. Hazard. Mater. 416, 126168 https://doi.org/10.1016/j. s41467-018-03825-5.
jhazmat.2021.126168. Peñalver, R., Costa-Gómez, I., Arroyo-Manzanares, N., Moreno, J.M., López-García, I.,
Jambeck, J., Geyer, R., Wilcox, C., Siegler, T.R., Perryman, M., Andrady, A., Narayan, R., Moreno-Grau, S., Córdoba, M.H., 2021. Assessing the level of airborne polystyrene
Law, K.L., 2015. Plastic waste inputs from land into the ocean 347. https://doi.org/1 microplastics using thermogravimetry-mass spectrometry: results for an agricultural
0.1126/science.1260352. area. Sci. Total Environ. 787, 1–8. https://doi.org/10.1016/j.
Jan Kole, P., Löhr, A.J., Van Belleghem, F.G.A.J., Ragas, A.M.J., 2017a. Wear and tear of scitotenv.2021.147656.
tyres: a stealthy source of microplastics in the environment. Int. J. Environ. Res. Penkała, M., Ogrodnik, P., Rogula-Kozłowska, W., 2018. Particulate matter from the road
Public Health 14. https://doi.org/10.3390/ijerph14101265. surface abrasion as a problem of non-exhaust emission control. Environments 5, 9.
Jan Kole, P., Löhr, A.J., Van Belleghem, F.G.A.J., Ragas, A.M.J., 2017b. Wear and tear of https://doi.org/10.3390/environments5010009.
tyres: a stealthy source of microplastics in the environment. Int. J. Environ. Res. Piehl, S., Leibner, A., Löder, M.G.J., Dris, R., Bogner, C., Laforsch, C., 2018. Identification
Public Health 14, 1–4. https://doi.org/10.3390/ijerph14101265. and quantification of macro- and microplastics on an agricultural farmland. Sci. Rep.
Jang, M., Shim, W.J., Han, G.M., Rani, M., Song, Y.K., Hong, S.H., 2016. Styrofoam 8, 1–9. https://doi.org/10.1038/s41598-018-36172-y.
Debris as a Source of Hazardous Additives for Marine Organisms. Environ. Sci. Pisso, I., Sollum, E., Grythe, H., Kristiansen, N., Cassiani, M., Eckhardt, S., Arnold, D.,
Technol. 50, 4951–4960. https://doi.org/10.1021/acs.est.5b05485. Morton, D., Thompson, R.L., Groot Zwaaftink, C.D., Evangeliou, N., Sodemann, H.,
Jones, B., O’Neill, B.C., 2016. Spatially explicit global population scenarios consistent Haimberger, L., Henne, S., Brunner, D., Burkhart, J.F., Fouilloux, A., Brioude, J.,
with the shared socioeconomic pathways. Environ. Res. Lett. 11. https://doi.org/ Philipp, A., Seibert, P., Stohl, A., 2019. The Lagrangian particle dispersion model
10.1088/1748-9326/11/8/084003. FLEXPART version 10.4. Geosci. Model Dev. 12, 4955–4997. https://doi.org/
10.5194/gmd-12-4955-2019.
16
PlasticsEurope, 2019. Plastics - the Facts 2019 [WWW Document]. URL 〈https://www. Tichý, O., Šmídl, V., Hofman, R., Stohl, A., 2016. LS-APC v1.0: A tuning-free method for
plasticseurope.org/en/resources/market-data〉 (accessed 5.12.20). the linear inverse problem and its application to source-term determination. Geosci.
Qian, J., Ferro, A.R., 2008. Resuspension of dust particles in a chamber and associated Model Dev. 9, 4297–4311. https://doi.org/10.5194/gmd-9-4297-2016.
environmental factors. Aerosol Sci. Technol. 42, 566–578. https://doi.org/10.1080/ Tichý, O., Ulrych, L., Šmídl, V., Evangeliou, N., Stohl, A., 2020. On the tuning of
02786820802220274. atmospheric inverse methods: comparisons with the European Tracer Experiment
Ragusa, A., Svelato, A., Santacroce, C., Catalano, P., Notarstefano, V., Carnevali, O., (ETEX) and Chernobyl datasets using the atmospheric transport model FLEXPART.
Papa, F., Rongioletti, M.C.A., Baiocco, F., Draghi, S., D’Amore, E., Rinaldo, D., Geosci. Model Dev. 13, 5917–5934. https://doi.org/10.5194/gmd-13-5917-2020.
Matta, M., Giorgini, E., 2021. Plasticenta: First evidence of microplastics in human Tipping, M.E., 2001. Sparse Bayesian learning and the relevance vector machine.
placenta. Environ. Int. 146, 106274 https://doi.org/10.1016/j.envint.2020.106274. J. Mach. Learn. Res. 1, 211–244. https://doi.org/10.1162/15324430152748236.
Ramankutty, N., Evan, A.T., Monfreda, C., Foley, J.A., 2008. Farming the planet: 1. Trainic, M., Flores, J.M., Pinkas, I., Pedrotti, M.L., Lombard, F., Bourdin, G., Gorsky, G.,
Geographic distribution of global agricultural lands in the year 2000. Global Boss, E., Rudich, Y., Vardi, A., Koren, I., 2020. Airborne microplastic particles
Biogeochem. Glob. BioCycles 22, 1–19. https://doi.org/10.1029/2007GB002952. detected in the remote marine atmosphere. Commun. Earth Environ. 1, 1–9. https://
Revell, L.E., Kuma, P., Ru, E.C., Somerville, W.R.C., Gaw, S., 2021. Direct radiative doi.org/10.1038/s43247-020-00061-y.
effects of airborne microplastics. Nature 598. https://doi.org/10.1038/s41586-021- Truong, T.N.S., Strady, E., Kieu-Le, T.C., Tran, Q.V., Le, T.M.T., Thuong, Q.T., 2021.
03864-x. Microplastic in atmospheric fallouts of a developing Southeast Asian megacity under
Roblin, B., Ryan, M., Vreugdenhil, A., Aherne, J., 2020. Ambient atmospheric deposition tropical climate. Chemosphere 272, 129874. https://doi.org/10.1016/j.
of anthropogenic microfibers and microplastics on the western periphery of Europe chemosphere.2021.129874.
(Ireland). Environ. Sci. Technol. 54, 11100–11108. https://doi.org/10.1021/acs. Tunahan Kaya, A., Yurtsever, M., Çiftçi Bayraktar, S., 2018. Ubiquitous exposure to
est.0c04000. microfiber pollution in the air. Eur. Phys. J. 133. https://doi.org/10.1140/epjp/
Rogge, W.F., Hildemann, L.M., Mazurek, M.A., Cass, G.R., Simoneit, B.R.T., 1993. i2018-12372-7.
Sources of fine organic aerosol. 3. road dust, tire debris, and organometallic brake Wagner, S., Hüffer, T., Klöckner, P., Wehrhahn, M., Hofmann, T., Reemtsma, T., 2018.
lining dust: roads as sources and sinks. Environ. Sci. Technol. 27, 1892–1904. Tire wear particles in the aquatic environment - a review on generation, analysis,
https://doi.org/10.1021/es00046a019. occurrence, fate and effects. Water Res. 139, 83–100. https://doi.org/10.1016/j.
Royer, S.-J., Ferron, S., Wilson, S.T., Karl, D.M., 2018. Production of methane and watres.2018.03.051.
ethylene from plastic in the environment. PLoS One 13, e0200574. https://doi.org/ Wagner, S., Reemtsma, T., 2019. Things we know and don’t know about nanoplastic in
10.1371/journal.pone.0200574. the environment. Nat. Nanotechnol. 14, 300–301. https://doi.org/10.1038/s41565-
Schwabl, P., Koppel, S., Konigshofer, P., Bucsics, T., Trauner, M., Reiberger, T., 019-0424-z.
Liebmann, B., 2019. Detection of various microplastics in human stool: a prospective Wang, J., Zheng, L., Li, J., 2018. A critical review on the sources and instruments of
case series. Ann. Intern. Med. 171, 453–457. https://doi.org/10.7326/M19-0618. marine microplastics and prospects on the relevant management in China. Waste
Scott, D.W., 2015. Multivariate density estimation: Theory, practice, and visualization: Manag. Res. 36, 898–911. https://doi.org/10.1177/0734242X18793504.
Second edition, Multivariate Density Estimation: Theory, Practice, and Visualization: Wang, X., Li, C., Liu, K., Zhu, L., Song, Z., Li, D., 2020. Atmospheric microplastic over the
Second Edition. https://doi.org/10.1002/9781118575574. South China Sea and East Indian Ocean: abundance, distribution and source.
Seibert, P., Frank, A., 2004. Source-receptor matrix calculation with a Lagrangian J. Hazard. Mater. 389, 121846 https://doi.org/10.1016/j.jhazmat.2019.121846.
particle dispersion model in backward mode. Atmos. Chem. Phys. 4, 51–63. https:// Wang, X., Liu, K., Zhu, L., Li, C., Song, Z., Li, D., 2021. Efficient transport of atmospheric
doi.org/10.5194/acp-4-51-2004. microplastics onto the continent via the East Asian summer monsoon. J. Hazard.
Sharma, S., Chatterjee, S., 2017. Microplastic pollution, a threat to marine ecosystem and Mater. 414, 125477 https://doi.org/10.1016/j.jhazmat.2021.125477.
human health: a short review. Environ. Sci. Pollut. Res. 24, 21530–21547. https:// Weiss, L., Ludwig, W., Heussner, S., Canals, M., Ghiglione, J.F., Estournel, C.,
doi.org/10.1007/s11356-017-9910-8. Constant, M., Kerhervé, P., 2021. The missing ocean plastic sink: gone with the
Smidl, V., Quinn, A., 2006. The Variational Bayes Method in Signal Processing. Springer, rivers. Science 373, 107–111. https://doi.org/10.1126/science.abe0290.
Berlin Heidelberg New York. Wilcox, C., Puckridge, M., Schuyler, Q.A., Townsend, K., Hardesty, B.D., 2018.
Sommer, F., Dietze, V., Baum, A., Sauer, J., Gilge, S., Maschowski, C., Gieré, R., 2018. A quantitative analysis linking sea turtle mortality and plastic debris ingestion. Sci.
Tire abrasion as a major source of microplastics in the environment. Aerosol Air Rep. 8, 1–11. https://doi.org/10.1038/s41598-018-30038-z.
Qual. Res. 18, 2014–2028. https://doi.org/10.4209/aaqr.2018.03.0099. World Economic Forum, 2016. The New Plastics Economy: Rethinking the future of
Sørensen, J.H., Bartnicki, J., Blixt Buhr, A.M., Feddersen, H., Hoe, S.C., Israelson, C., plastics. Ellen MacArthur Foundation.
Klein, H., Lauritzen, B., Lindgren, J., Schönfeldt, F., Sigg, R., 2020. Uncertainties in Wright, S.L., Kelly, F.J., 2017. Plastic and human health: a micro issue? Environ. Sci.
atmospheric dispersion modelling during nuclear accidents. J. Environ. Radioact. Technol. 51, 6634–6647. https://doi.org/10.1021/acs.est.7b00423.
222. https://doi.org/10.1016/j.jenvrad.2020.106356. Wright, S.L., Ulke, J., Font, A., Chan, K.L.A., Kelly, F.J., 2020. Atmospheric microplastic
Stanton, T., Johnson, M., Nathanail, P., MacNaughtan, W., Gomes, R.L., 2019. deposition in an urban environment and an evaluation of transport. Environ. Int.
Freshwater and airborne textile fibre populations are dominated by ‘natural’, not 136, 105411 https://doi.org/10.1016/j.envint.2019.105411.
microplastic, fibres. Sci. Total Environ. 666, 377–389. https://doi.org/10.1016/j. Yukioka, S., Tanaka, S., Nabetani, Y., Suzuki, Y., Ushijima, T., Fujii, S., Takada, H., Van
scitotenv.2019.02.278. Tran, Q., Singh, S., 2020. Occurrence and characteristics of microplastics in surface
Stohl, A., Forster, C., Frank, A., Seibert, P., Wotawa, G., 2005. Technical note: the road dust in Kusatsu (Japan), Da Nang (Vietnam), and Kathmandu (Nepal). Environ.
Lagrangian particle dispersion model FLEXPART version 6.2. Atmos. Chem. Phys. 5, Pollut. 256, 113447 https://doi.org/10.1016/j.envpol.2019.113447.
2461–2474. https://doi.org/10.5194/acp-5-2461-2005. Zhang, Q., Xu, E.G., Li, J., Chen, Q., Ma, L., Zeng, E.Y., Shi, H., 2020. A review of
Syafei, A.D., Nurasrin, N.R., Assomadi, A.F., Boedisantoso, R., 2019. Microplastic microplastics in table salt, drinking water, and air: direct human exposure. Environ.
pollution in the ambient air of Surabaya, Indonesia. Curr. World Environ. 14, Sci. Technol. 54, 3740–3751. https://doi.org/10.1021/acs.est.9b04535.
290–298. https://doi.org/10.12944/cwe.14.2.13. Zhao, Y.B., Gao, P.P., Ni, H.G., 2019. A chemical time bomb: future risks of microplastics.
Szewc, K., Graca, B., Dołęga, A., 2021. Atmospheric deposition of microplastics in the Water Air Soil Pollut. 230. https://doi.org/10.1007/s11270-019-4320-9.
coastal zone: characteristics and relationship with meteorological factors. Sci. Total Zhou, Q., Tian, C., Luo, Y., 2017. Various forms and deposition fluxes of microplastics
Environ. 761, 143272 https://doi.org/10.1016/j.scitotenv.2020.143272. identified in the coastal urban atmosphere. Chin. Sci. Bull. 62, 3902–3909. https://
Szymańska, M., Obolewski, K., 2020. Microplastics as contaminants in freshwater doi.org/10.1360/N972017-00956.
environments: a multidisciplinary review. Ecohydrol. Hydrobiol. 20, 333–345. Zhu, X., Huang, W., Fang, M., Liao, Z., Wang, Y., Xu, L., Mu, Q., Shi, C., Lu, C., Deng, H.,
https://doi.org/10.1016/j.ecohyd.2020.05.001. Dahlgren, R., Shang, X., 2021. Airborne microplastic concentrations in five
megacities of Northern and Southeast China. Environ. Sci. Technol. https://doi.org/
10.1021/acs.est.1c03618.
17

1 s2.0 S0304389422003739 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S0304389422003739 Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0304389422003739 Main

Uploaded by

Copyright:

Available Formats

Journal of Hazardous Materials 432 (2022) 128585

Contents lists available at ScienceDirect

Journal of Hazardous Materials

Sources and fate of atmospheric microplastics revealed from inverse and

• About 9.6 and 6.5 Tg y-1 of micro

1. Introduction influence the global climate indirectly, as enhanced plastic production

Following Tipping (2001), we select the prior distribution for the

*described as dustpop in Brahney et al. (2021).

4. Discussion measurements do not contribute to the reconstruction of the posterior

the riverine transported MPs to the ocean (3.3–14 Tg y-1).

4.4. Independent validation and forecast of expected levels

As it was already mentioned in Section 3.1, the observations from

5. Conclusions - The largest continental MP deposition occurred in Asia (267

You might also like

1 s2.0 S0304389422003739 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S0304389422003739 Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0304389422003739 Main

Uploaded by

Copyright:

Available Formats

Journal of Hazardous Materials 432 (2022) 128585

Contents lists available at ScienceDirect

Journal of Hazardous Materials

Sources and fate of atmospheric microplastics revealed from inverse and

• About 9.6 and 6.5 Tg y-1 of micro­

1. Introduction influence the global climate indirectly, as enhanced plastic production

Following Tipping (2001), we select the prior distribution for the

*described as dustpop in Brahney et al. (2021).

4. Discussion measurements do not contribute to the reconstruction of the posterior

the riverine transported MPs to the ocean (3.3–14 Tg y-1).

4.4. Independent validation and forecast of expected levels

As it was already mentioned in Section 3.1, the observations from

5. Conclusions - The largest continental MP deposition occurred in Asia (267

You might also like

• About 9.6 and 6.5 Tg y-1 of micro