The COVID-19 lockdown measures gradually implemented in Lombardy (northern
Italy) from 23 February 2020 led to a downturn in several economic sectors
with possible impacts on air quality. Several communications claimed in the
first weeks of March 2020 that the mitigation in air pollution observed at
that time was actually related to these lockdown measures without
considering that seasonal variations in emissions and meteorology also
influence air quality. To determine the specific impact of lockdown measures
on air quality in northern Italy, we compared observations from the European
Commission Atmospheric Observatory of Ispra (regional background) and from
the regional environmental protection agency (ARPA) air monitoring stations
in the Milan conurbation (urban background) with expected values for these
observations using two different approaches. On the one hand, intensive
aerosol variables determined from specific aerosol characterisation
observations performed in Ispra were compared to their 3-year averages. On
the other hand, ground-level measured concentrations of atmospheric
pollutants (NO2, PM10, O3, NO, SO2) were compared to
expected concentrations derived from the Copernicus Atmosphere Monitoring
Service Regional (CAMS) ensemble model forecasts, which did not account for
lockdown measures. From these comparisons, we show that NO2
concentrations decreased as a consequence of the lockdown by -30 % and
-40 % on average at the urban and regional background sites, respectively.
Unlike NO2, PM10 concentrations were not significantly affected by
lockdown measures. This could be due to any decreases in PM10 (and
PM10 precursors) emissions from traffic being compensated for by
increases in emissions from domestic heating and/or from changes in the
secondary aerosol formation regime resulting from the lockdown measures. The
implementation of the lockdown measures also led to an increase in the
highest O3 concentrations at both the urban and regional background
sites resulting from reduced titration of O3 by NO. The relaxation of
the lockdown measures beginning in May resulted in close-to-expected
NO2 concentrations in the urban background and to significant
increases in PM10 in comparison to expected concentrations at both
regional and urban background sites.
Introduction
The COVID-19 pandemic is an epidemic of the coronavirus disease 2019 (COVID-19),
of which the outbreak was first identified in Wuhan, China, in late December 2019. The World Health Organization declared COVID-19 a pandemic on 11 March 2020. The first case of COVID-19 in northern Italy was detected on 20 February 2020 in Codogno, about 60 km south-east of Milan
(Fig. 1). To reduce the virus spreading, the Italian government quickly adopted a series of measures, such as the quarantine for 10 municipalities, the cancellation of all main public events, and the closure of schools and universities in northern Italy (DL, 23 February 2020a). The lockdown started in all of Italy on 9 March 2020 (DPCM, 8 March 2020a). All commercial and retail activities were closed on 11 March, except for grocery shops and pharmacies (DPMCM, 11 March 2020b), and it was forbidden to move outside the place of residence, except for health issues or work. Further lockdown measures were decreed on 22 March 2020 (DPCM, 22 March 2020c), including the suspension of all non-essential industrial production activities. The lockdown lasted until 4 May 2020 (DPCM, 26 April 2020d), when a gradual relaxation of the measures was decided by the government. The reopening of manufacturing industries and construction sites was allowed, but schools and universities as well as some commercial activities such as restaurants remained closed. Movements from a region to another were still forbidden, but moving short distances to work and to visit relatives was possible. From 18 May 2020, most commercial businesses could reopen, and free movement was granted within regional borders (DL, 16 May 2020b). This lockdown provided a unique opportunity to determine how such dramatic measures can eventually influence air quality. This is the focus of this paper.
Lombardy, Piedmont and Emilia-Romagna in northern Italy produce roughly
50 % of the national gross domestic product (GDP), with Lombardy alone
producing 22 % of the national GDP (data from ISTAT, 2020). This economic
dynamism (mainly linked to industrial production and service-related
activities) is associated with significant pollutant emissions, which
together with unfavourable conditions for pollution dispersion (due to low
wind speeds and particular orography) cause high pollution levels, leading to
exceedances of the EU standards for nitrogen dioxide (NO2), particulate
matter (PM10) and ozone (O3) in northern Italy (European Environment Agency, 2019). In
this area, the impact of the lockdown on economic activities was quite
important, as illustrated by data relative to the production of electricity
and energy for heating and to transport-related activities (ARPA Lombardia,
2020a). Compared to 2019, the Italian thermal electricity production
(Fig. 2) fell in March (-18 %), April (-24 %)
and May 2020 (-16 %). The consumption of natural gas by the industrial
sector as reported by the Italian natural gas provider (https://www.snam.it, last access: 9 October 2020) also
fell by roughly -30 % at the end of March in comparison to the beginning
of March 2020.
Variations in activities resulting from lockdown measures (2020).
Percentages are calculated in comparison with 2019 data for thermal energy
production (source: https://www.terna.it, last access: 9 October 2020) and in comparison with data from the third
week of February 2020 for mobility data (source:
https://www.apple.com/covid19/mobility, last access: 9 October 2020).
Regarding transport, the Monitoring of Polluting Vehicles project (MOVE-IN)
managed by the Lombardy region provided data on the traffic changes derived
from its monitoring of “vehicle km” (the sum of kilometres travelled by all vehicles in the area) driven by light-duty vehicles and
passenger cars (for a small number of vehicles compared to the full fleet
circulating in the region though). MOVE-IN data show that the number of
vehicle km driven by light-duty vehicles remained quite constant till 9 March 2020, then dropped by -75 % to reach a minimum between 16 March and
13 April 2020 before returning to “usual” (i.e. as before the lockdown
period) values after 4 May 2020 (ARPA Lombardia, 2020a). For private cars,
the number of vehicle km driven also decreased by roughly -70 % between
the beginning and the end of March and started increasing again after 4 May but with a slower recovery than for light-duty vehicles. The
number of requests for driving directions (https://www.apple.com/covid19/mobility, last access: 9 October 2020)
showed similar variations (Fig. 2).
Numerous early communications based on preliminary measurement data analyses
associated observed improvements in air quality with the lockdown measures
taken to contain the spread of the COVID-19 epidemic. In Brazil, the
lockdown in São Paulo was followed by drastic reductions in NO (up to
-77 %) and NO2 (up to -54 %) and by an increase in O3
(approximately +30 %) compared to the previous 5-year means for the
same period (Nakada et al., 2020). In the Yangtze River Delta region
(China), Li et al. (2020) showed that concentrations of PM2.5, NO2
and SO2 decreased by 32 %, 45 % and 20 % during the first lockdown
phase and by 33 %, 27 % and 7 % during the second lockdown phase,
compared with the 2017–2019 average for the same period. O3 also
increased in that region. Across Europe, Grange et al. (2021) estimated that
NO2 and O3 concentrations at urban background sites were 32 %
lower and 21 % higher than expected, respectively, when maximum mobility
restrictions were in place. A clear decrease in NO2 concentrations in
Barcelona and Madrid (Spain) during the lockdown was also described by
Baldasano (2020). In France, the analysis by INERIS (Institut national de
l'environnement industriel et des risques) compared air pollution forecast
data (calculated without incorporating changes in emissions due to lockdown
measures) with adjusted simulations performed a posteriori by assimilating observation
data influenced by the lockdown measures. They estimated that NO2
concentrations were on average approximately 50 % lower than expected in
France's largest cities (INERIS, 2020).
Regarding Italy, maps of NO2 surface concentrations estimated from
satellite data (e.g. Sentinel-5p) were published by several websites and
media showing large reductions in NO2 concentrations over northern
Italy in March 2020 as compared to the previous months and to March 2019
(e.g. Copernicus Atmosphere Monitoring Service headlines published on 17 and
26 of March 2020; CAMS, 2020a, b). Observations and models were also combined in the
analysis from the German Aerospace Center (DLR) which estimated a decrease
of about 40 % in the total column-integrated NO2 tropospheric
concentrations over norther Italy due to the lockdown measures using
Sentinel-5p data. They also estimated reductions in ground level NO2
concentrations of about -20µg m-3 (-45 %) by comparing ground-based observations from 25 stations in Lombardy to a model simulation with
pre-lockdown emission levels (German Aerospace Center, 5 May 2020). In situ observations also
showed reduced ground level NO2 concentrations as lockdown measures
were implemented. The environmental protection agency ARPA Lombardia showed
that March 2020 NO2 concentrations were below the standard deviation
calculated from previous years, indicating a possible signal of reduced
emissions from traffic and economic sectors (ARPA Lombardia, 2020a). The
European Environment Agency (EEA) developed a viewer that tracks NO2
and particulate matter (PM10 and PM2.5) weekly average
concentrations
(https://www.eea.europa.eu/themes/air/air-quality-and-covid19, last access: 9 October 2020). It shows
that NO2 concentrations in Milan were at least 24 % lower after the
lockdown implementation than during previous weeks and 21 % lower
compared to the same period in 2019. Similar trends were found in other
cities of northern Italy and European countries where strong measures were
taken to contain the epidemic. In contrast, no consistent effect of the
lockdown measures on particulate matter (PM2.5 and PM10) could be
observed in the main European cities (European Environment Agency, 2020).
Air pollution did decline in northern Italy from February to May in 2020 as
it does every year, mainly due to seasonal variations in emissions and
weather conditions. The strength of certain sources does indeed change
during the course of the year, like, for example, domestic heating, while weather
conditions influence pollution concentration in diverse ways: advection and
dispersion of pollutants resulting from horizontal winds; dilution of
pollutants throughout the mixed boundary layer resulting from convection;
and pollutant lifetimes resulting from photochemical reactions (sun
radiation), wet removal (clouds and rain), etc. It is therefore not
straightforward to disentangle the effects of changing emissions due to
lockdown measure implementation from those of seasonal changes in emissions
and variability in meteorological conditions between different seasons and
different years. In the present study we determine how much of the changes
in air pollution observed during the lockdown period in northern Italy were
actually due to lockdown measures, independently from expected variations in
pollutants' emissions, lifetime and dispersion. Our results are based on
comparisons between air pollution observation data from Ispra (regional
background site) and the Milan conurbation (urban background sites) with
CAMS regional ensemble (hereafter ENSEMBLE) model forecast data for the same sites. To help understand the
effect of the lockdown measures in the regional background area, we also use 4
years of specific aerosol measurements from Ispra.
Material and methods
Model and observation air pollution data from four sites located in Lombardy
covering the time periods 17 February–24 May 2019 and 2020 were collected
and analysed. We selected three sites located in the Milan conurbation as
representative of the Milan urban background and the site of Ispra as
representative of the regional background of the upper Po Valley
(Fig. 1). Ground level concentrations of NO,
NO2, SO2, O3 and PM10 as measured in situ at the
monitoring stations and as calculated by the ENSEMBLE model forecast
were considered. Particle number size distribution and aerosol light
absorption Ångström exponent data from Ispra for the 2017–2020
period were also utilised.
Site description
The European Commission Atmospheric Observatory (ECAtmO) has been operated
in Ispra (45.815∘ N, 8.636∘ E; 209 m a.s.l.) since November 1985. It has
contributed to the CLRTAP-EMEP (co-operative programme for monitoring and
evaluation of the long-range transmission of air pollutants in Europe under
the Convention on Long-range Transboundary Air Pollution) and WMO-GAW (World
Meteorological Organization – Global Atmosphere Watch) air pollution
measurement programmes for several decades and to the European Research
Infrastructures ICOS (Integrated Carbon Observation System) and ACTRIS
(Research Infrastructure for the observation of Aerosol, Clouds and Trace
Gases) for several years. ECAtmO is located on the north-western edge of the
Po Valley, 20–60 km away from major pollution point sources but still in
a densely populated area (ca. 500 km-2) with significant
economic activity (GDP per capita = EUR 29 000; EUROSTAT, 2017).
Wood burning for domestic heating is also an important source of particulate
matter during the cold period of the year (Gilardoni et al., 2011). Past
measurements of HCHO / NO2 ratios compared to the threshold values
proposed by Tonnensen and Dennis (2000) suggest that the photochemical
production of O3 is limited by the availability of volatile organic
compounds (VOCs) in February–May in Ispra.
The Milan metropolitan area is the second most densely populated area in
Italy (ca. 2300 km-2), with a GDP per capita of about EUR 54 000
(EUROSTAT, 2017) and about 4100 circulating
vehicles km-2 (ISTAT, 2020). Three stations in the Milan
conurbation were selected as representative of the urban background in Milan
city, namely “Milan via Pascal” (45.478∘ N, 9.236∘ E; 122 m a.s.l.), “Cormano”
(45.548∘ N, 9.167∘ E; 155 m a.s.l.) and “Limito di Pioltello” (45.483∘ N,
9.328∘ E; 123 m a.s.l.). All three stations are operated by ARPA Lombardia.
We selected only urban background stations because pollutant concentrations
at traffic sites are hardly reproducible by regional air quality models with
a horizontal resolution of about 10 km. The station in Milan via Pascal is
located near the university, and it is considered to be the urban background
station of the city, while the other two stations are located in the
hinterland, near (<500 m) two main roadways used by commuters at
the northern (Cormano, with about 75 000 vehicles per day) and eastern (Limito,
20 000 vehicles per day) entrances of the city. Average population densities are
7500, 4500 and 2800 km-2 in Milan, Cormano and Limito di Pioltello,
respectively (ISTAT, 2020).
Measurements
At ECAtmO in Ispra, online in situ air pollution measurements are performed
from appropriate inlets located at 6.5 and 9 m above ground level for
gaseous and particulate pollutants, respectively. The inlet for reactive gas
is made of PTFE (inner diameter: 2.7 cm). The sample residence time in the inlet tube
is ca. 2 s. Each analyser samples from the main inlet through a Nafion
dryer. In 2019–2020, the measurement programme included CO, NO, NOx,
NO2, SO2, O3, non-methane hydrocarbons (until 6 March 2020)
and NH3 (since 28 January 2020) as gaseous pollutants. The NOx (i.e. NO
and NO2), SO2 and O3 data reported in this work were
obtained with trace level instruments based on infrared (IR) (1200 nm)
chemiluminescence and a Mo converter (Thermo Fisher 42iTL), ultraviolet (UV) (214 nm)
fluorescence (Thermo Fisher 43i TLE), and UV (254 nm) absorption
(Thermo Fisher 49C), respectively. These instruments are calibrated every 3
months using zero air and certified gas cylinders (NO and SO2) or a
primary standard ozone generator (O3). In 2019, annual average
concentrations of NO, NO2, SO2, O3 and PM10 were 4, 16,
0.4, 38 and 21 µg m-3, respectively. Particulate matter is
sampled through metal-made inlets characterised by negligible losses. Each
instrument samples isokinetically from the main aerosol inlet through Nafion
dryers. In 2019–2020, the aerosol online in situ measurement programme
included PM10 mass concentration, particle number concentration and
number size distribution, particle light extinction, absorption, scattering,
and backscattering at several wavelengths. The PM10 mass, particle
number concentration and light absorption data reported in this work were obtained with a
TEOM-FDMS (Thermo Fisher 1405-DF), a differential mobility particle sizer
(home-made Vienna-type differential mobility analyser + TSI 3772
condensation particle counter) covering the particle mobility diameter range
10–800 nm and a seven-wavelength aethalometer (Magee AE31), respectively.
The TEOM has been calibrated using a standard filter provided by the
manufacturer, while the differential mobility particle sizer (DMPS) and the aethalometer are operated, maintained
and controlled according to ACTRIS guidelines (https://www.actris.eu, last access: 9 October 2020).
They were both calibrated at the specific ACTRIS central facility
(https://www.actris-ecac.eu, last access: 9 October 2020) on 3–7 June 2019. Near-real-time data are
available from the JRC data catalogue at http://data.jrc.ec.europa.eu/collection/abcis (last access: 9 October 2020).
The three stations in the Milan conurbation are part of the ARPA Lombardia
air quality network, compliant with Directive 2008/50/EC requirements in
terms of measurement methods, macro and micro localisation, and data
coverage. Inlets are located at 2.5 m above ground level for all pollutants.
The measurement programmes comprise NO, NOx, NO2, SO2 and O3
at all three sites. Additional measurements include benzene, toluene,
xylenes, PM10, PM2.5, benzo[a]pyrene and NH3 in Milan and CO and
PM10 in Limito. Each gas analyser samples from the main inlet through a
Nafion dryer. The NOx data reported in this work for the Milan conurbation
were obtained with trace level instruments based on IR (1200 nm)
chemiluminescence and a Mo converter (Teledyne API 201E, Thermo Fisher 42i and Thermo Fisher 42c in Milan Pascal, Limito and Cormano, respectively), and O3 was measured by UV (254 nm) fluorescence (Thermo Fisher 49i at
all three sites). All measurements are performed according to a specific quality assurance and quality control programme. All gas monitors are calibrated every 3 months using zero air and
certified gas cylinders (NO) and every 6 months using a primary standard
ozone generator for O3. The PM10 mass concentrations in the Milan
conurbation reported in this work were measured using beta absorption
analysers (FAI SWAM DC and 5A models in Milan Pascal and Limito,
respectively). The PM analysers are checked for temperature, pressure, flow
rates, leaks and other operational parameters every 3 months. A periodical
comparison with gravimetric samples has been performed once yearly in Milan
Pascal and upon a specific audit programme in Limito. In 2019, annual
average concentrations of NO, NO2 and O3 were, respectively, 25, 37 and 46 µg m-3 in Milan Pascal; 29, 45 and 46 µg m-3 in
Cormano; and 26, 34 and 44 µg m-3 in Limito. For PM10, 2019
annual averages were 29 and 31 µg m-3 in
Milan Pascal and Limito, respectively. Data are available online at
https://www.arpalombardia.it (last access: 9 October 2020).
CAMS regional ensemble forecast description
The Copernicus Atmospheric Monitoring Service (CAMS) provides 4 d
ahead air quality forecasts daily for Europe from currently nine different
regional air quality models (CHIMERE, DEHM, EMEP, EURAD-IM, GEM-AQ,
LOTOS-EUROS, MATCH, MOCAGE, SILAM). Hourly pollutant concentrations are
calculated for altitudes ranging from the 40 m thick surface layer to 5 km.
The outputs of the different individual models are interpolated on a common
regular 0.1∘× 0.1∘ latitude × longitude grid (about 10 km × 10 km) over the European domain (30–72∘ N, 25∘ W–45∘ E). Forecasts are performed independently by
all the individual regional air quality systems: each air quality model is
based on different chemical (gas and aerosols) and physical
parameterisations but uses the same meteorological drivers as input (the
ECMWF Integrated Forecasting System, IFS) and the same anthropogenic
emissions data (Kuenen et al., 2014; Denier van der Gon et al., 2015) based
on 2011 emission inventories until June 2019 and on 2016 emission
inventories afterwards. An ensemble (named “ENSEMBLE forecast”) is
calculated from individual model outputs with a median approach (Marécal
et al., 2015). This method provides an optimal estimate (Riccio et al., 2007)
which is rather insensitive to outliers and generally yields better
estimates than the individual models (Galmarini et al., 2018). CAMS regional
air quality forecasts are routinely quality-controlled, and dedicated
evaluation reports are published every third month for both individual and
the ENSEMBLE models (see https://atmosphere.copernicus.eu/regional-services, last access: 9 October 2020). In this work, we used daily
averages of the ENSEMBLE surface concentration forecast each day for
the next 24 h (D0). For the period March–May 2019, the differences
between daily mean D0 forecasts and measurements performed at various
reference stations across northern Italy (expressed as median of the root
mean square errors, RMSEs) were 10.5, 10.6 and 24.5 µg m-3 for
NO2, PM10 and O3, respectively. Additional statistical scores
are available in quarterly CAMS reports (CAMS, 2019, 2020c). Note that
the actual ENSEMBLE RMSEs relative to the stations and time periods we
analyzed are part of our statistical analysis described in Sect. 2.4.1.
As the anthropogenic emissions used by the individual models did not change
to account for any lockdown measure, the ENSEMBLE model continued to
forecast pollutants' concentrations as if the COVID-19 epidemic had not
occurred in 2020.
Data analysisPollutant concentrations
To determine the specific impact of lockdown measures on concentrations of
air pollutants, we compared daily observations (Obs2020) with daily
expected concentrations (Exp2020) for the period 17 February–24 May 2020, which comprises the 8 lockdown weeks (D= 9 March–3 May 2020), the 3 weeks before the beginning (A= 17 February–8 March 2020)
and the 3 weeks after the end of the lockdown period (P= 4–24 May 2020). NO2, PM10, NO, O3 and SO2 observed and expected
concentrations are shown in Fig. 3. Expected
concentrations were derived from 2020 ENSEMBLE forecasts, which
account for variations in meteorological conditions and seasonal changes in
emission source strengths in a “business as usual” world, i.e. without
lockdown measures. However, since data from 2019 show that the agreement
between ENSEMBLE forecasts (CAMS2019) and observations
(Obs2019) improves from February to May (see Figs. S1–S3 in
Supplement), CAMS2020 was corrected for this seasonality. Thus, 2020
daily expected pollutant concentrations (Exp) were calculated as follows:
Exp2020=CAMS2020CAMS2019Obs2019.
The comparison of observations with these expected concentrations for 2020
has the great advantage of being insensitive to the fact that the emissions
inventories used to calculate ENSEMBLE forecast data for 2019 and 2020
were different. The disadvantage of this approach is that Obs and Exp
cannot be compared to each other on a daily basis since Exp values are
affected by random variations in the Obs2019/ CAMS2019
ratio. Therefore Obs2020 and Exp2020 data were compared
statistically for the three periods A, D and P. Since changes in pollutant
emission rates are expected to result in changes in pollutant concentrations
in terms of percentages or ratios, statistical analyses were performed on
Obs2020/ Exp2020 daily ratios. We calculated
occurrence frequency distributions of the Obs2020/ Exp2020 ratio using eight class bins ranging from <0.25 to
>2, all equally wide on a logarithmic scale (except the last one
when specifically indicated). Cumulative frequencies of occurrence were also
plotted to facilitate comparisons (Fig. 4). To
detect possible specific impacts of lockdown measures on the highest
concentrations, specific occurrence frequency distributions were also
calculated by selecting the 28 d on which ENSEMBLE forecast data
were greater than the median during the lockdown period. These days are
different for each pollutant and each site. The statistical significance of
the differences in Obs2020/ Exp2020 ratios
during the lockdown period in comparison with before and after the lockdown
period (i.e. between A and D or P and D) was assessed by applying a t test assuming
unequal variances to the means A‾, P‾ and D‾, defined
as follows:
D‾=meanlogObs/CAMSduring
lockdownObs/CAMS10 Mars–25 May 2019,A‾=meanlogObs/CAMSbefore lockdownObs/CAMS17 Feb–9 Mars 2019,P‾=meanlogObs/CAMSafter
lockdownObs/CAMS5–25 May 2019.
The null hypotheses (D‾=A‾ and D‾=P‾)
were tested at the 95 % confidence level, and results were used to
determine if differences between D‾ and A‾ and D‾ and
P‾ were statistically significant.
Observed (dots) and expected (lines) 2020 concentrations (µg m-3) of NO2, PM10, NO, O3 and SO2 in Ispra (left-hand side) and the Milan conurbation (right-hand side). Vertical lines indicate
the beginning and end of the lockdown period.
Occurrence frequency distributions of 2020 observed/expected
concentration ratios (Obs2020/ Exp2020) for
NO2, PM10, NO, O3 and SO2 during the lockdown period
and during the 3 weeks before and after the lockdown period in Ispra (left)
and the Milan conurbation (right). Lines show cumulative frequencies of
occurrence. Dashed lines show the cumulative frequency of occurrence of
Obs2020/ Exp2020 ratios for the 28 d
corresponding to the highest CAMS forecast values. Note: the last bin for NO
in Ispra contains all values >2.
Intensive aerosol variables
To complement our analyses based on pollutant concentrations, we also looked
at two characteristics of the atmospheric aerosol measured at ECAtmO in
Ispra. The first one is the percentage of number of tiny particles with
mobility diameters (Dp) between 15 and 70 nm as compared with the
“total” number of particles with mobility diameters between 15 and 800 nm.
This percentage was calculated from full particle number size distributions
(10<Dp<800 nm). The smallest particles (10<Dp<15 nm) were not considered because their measurement is
affected by larger uncertainties (Wiedensohler et al., 2018) and by
nucleation particle bursts. The range 15<Dp<70 nm
was selected as representative of particles emitted by primary sources
(Giechaskiel et al., 2019; Giechaskiel, 2020; Ozgen et al., 2017; Tiwari et
al., 2014). The second variable is the aerosol light absorption
Ångström exponent (AÅE). It represents the wavelength dependence
of light absorption by aerosol particles. AÅE values vary with particle
sources and have commonly been used to apportion pollution particles between, for example, traffic and wood burning (Sandradewi et al., 2008). Traffic-emitted
particles (mainly from diesel engines) have an AÅE close to 1, while
particles from wood combustion have more variable AÅEs around 2
(Sandradewi et al., 2008). The mixture of pollution particles with primary or
secondary aerosol of biogenic origin can also lead to AÅE values much
greater than 1. Since both variables are insensitive to air pollution
dispersion, they are much less variable than the extensive variables (i.e.
atmospheric concentrations) they are derived from (e.g. Putaud et al.,
2014). The values expected for these so-called intensive variables were
calculated as the arithmetic averages observed during the 2017–2019 period.
Results and discussion
The observation and ENSEMBLE forecast data used to estimate the values
expected for the air pollution variables discussed in this section are
described in Sects. 1 and 2 of the Supplement to this article. The high
PM10 concentrations observed at all sites on 28 and 29 March 2020 were
related to desert dust advection from the east (see maps from the World Meteorological Organisation Sand and Dust Storm Warning Advisory and Assessment System, 2021). The data from these two dates were not excluded from our
statistical analysis since they did not affect its results.
Regional background (Ispra)
The trend in AÅE observed in Ispra in 2017–2019 (and also in 2020) is
consistent with a decreasing contribution of wood burning to particulate
pollution from winter to summer. The AÅE values measured in 2020 can of
course not be compared point to point to the 2017–2019 average in
Fig. 5 because the use of wood fuel for domestic
heating also depends on weekend and cold-evening occurrences. However, the
clear increase in the AÅE average between 9 March and 4 May 2020
compared to the 3 weeks before, the 3 weeks after and the corresponding
period in 2017–2019 undoubtedly shows a change in particle sources related
to lockdown measures (Table 1). A specific analysis focused on the first 4 weeks of the lockdown period (before significant numbers of biogenic
aerosols are expected) suggests a -45 % reduction in aerosol from traffic
(and a concomitant +45 % increase in aerosol from wood combustion)
during that period.
Aerosol light absorption Ångström exponent (AÅE) in
2020 (dots) compared to its 2017–2019 average (lines). The shaded area
represents ±1 standard deviation of the average. Vertical lines
indicate the beginning and the end of the lockdown period.
Observed/expected concentration ratios for pollutant
concentrations and aerosol characteristics before, during and after the
lockdown measures in Ispra (regional background) and Milan (urban
background).
Particle number size distribution measurements in Ispra typically show modes
at 25–50 nm during morning rush hours as well as in the evening in
winter. Particle primary sources include fuel combustion by thermal engines
and liquid (oil) or solid fuel (e.g. wood) combustion for domestic heating.
The ultrafine mode diameters of primary particle emissions range from 50 to
100 nm for domestic heating (e.g. Tiwari et al., 2014; Ozgen et al., 2017)
and range from 10 to 90 nm for engines (e.g. Giechaskiel et al., 2019;
Giechaskiel, 2020). Measurements also show that peaks in the number of 15–70 nm particles can result from the growth of nucleation particles in the
afternoon. The percentage of 15–70 nm particles generally increased from
mid-February till end of May in 2017–2019 (Fig. 6). Considering that (1) wood burning combustion for domestic heating did
not decrease during the lockdown period, (2) nucleation and growth of
secondary aerosol particles were observed on sunny days during the lockdown
period from 6 April 2020, and (3) that mostly morning peaks in particle
number diminished during the lockdown period especially from 11 March to 13 April 2020, the relative “disappearance” of 15–70 nm particles during
the lockdown period (Fig. 6) can be attributed to
a decrease in traffic related to lockdown measures. Since the atmospheric
lifetime of 15–70 nm particles is 3–12 h, local to regional traffic
was concerned. Although it significantly increased after 4 May 2020, the
percentage of 15–70 nm particles did not get back to the level observed
before the lockdown as lockdown measures were relaxed.
Percentage of sub-70 nm particles in 2020 (dots) compared to 2017–2019 average (line). The shaded area represents ±1 standard deviation of the average. Vertical lines indicate the beginning and the end of the lockdown period.
The decrease in emissions from local traffic indicated by the drop in the
percentage of the smallest particles (Fig. 6) is
the most probable cause for the decrease in NO related to the lockdown
measures in Ispra (Fig. 4). NO daily mean
concentrations are indeed dominated by their morning peak corresponding to
traffic rush hours (which disappears during weekends). During daytime, NO is
rapidly converted to NO2, and NO concentrations reach very low steady-state values. Decreased NO emissions should therefore result in decreases in
NO2. Such a decrease in NO2 (-40 % on average) actually occurred
in Ispra as a result of the lockdown measures from 9 March 2020
(Fig. 4). In contrast, NO2 did not
“recover” from the lockdown measures, unlike NO, of which concentrations
increased again in comparison to expected concentrations as lockdown
measures were relaxed on 4 May 2020 (Table 1). Due to its lifetime of about
1–2 d (Seinfeld and Pandis, 2016), NO2 can travel rather long
distances. Nitrogen oxides are also emitted by large combustion sources like
thermal power plants, which also emit SO2. However, our analysis of
SO2 data also reveals that sources of SO2 that affect
concentrations in Ispra decreased due to lockdown measures (Figs. 3 and 4). The fact that NO2 observed/expected concentration ratios remained
as low after as during the lockdown period could be explained by a slower
increase in traffic on the regional scale as compared to the local scale.
Regarding secondary pollutants, the highest O3 concentrations
significantly increased compared to expected concentrations during the
lockdown period in comparison with the 3 weeks before
(Figs. 7 and 4). This suggests that O3 peaks
are usually diminished by NO titration during this period of the year in
Ispra and that the abatement in NOx emissions revealed by NO and
NO2 data analyses led to a reduction in this effect. The relaxation of
lockdown measures led to a further increase in O3. Since O3
production is generally VOC-limited in May in Ispra, this increase in
O3 is probably due to an increase in anthropogenic emissions of VOCs
from, for example, local traffic. In the case of PM10, which is mainly composed
of secondary particulate species in Ispra (Larsen et al., 2012), no
significant decrease compared to expected concentrations could be
identified as lockdown measures were implemented (Figs. 3 and 4). This is
because the decrease in PM10 related to traffic was compensated by the
increase from wood burning for domestic heating, at least during the first
half of the lockdown period. In contrast, PM10 significantly but
marginally increased as lockdown measures were relaxed on 4 May 2020, at a
time of the year (from May onwards) when wood burning combustion for
domestic heating is largely reduced.
Changes in observed and expected concentrations for the 3 weeks
before and after the lockdown in comparison with the 28 d of lockdown
corresponding to the CAMS daily forecasts above the median for each pollutant.
Filled bars represent statistically significant differences. Empty bars
represent differences that are not significantly different from zero.
Urban background (Milan conurbation)
To represent the Milan urban background, we used data from three urban background
sites located in Milan's hinterland and in Milan's city centre (Fig. 1).
NO2 significantly decreased (-30 % on average) compared to expected
concentrations as lockdown measures were implemented (Figs. 3 and 4).
NO2 significantly but not totally “recovered” when lockdown measures
were relaxed (Table 1), which suggests that not all
sources determining NO2 concentrations in Milan were fully reopened.
However, the increase in NO after the end of the lockdown period suggests
that local traffic largely resumed. Perhaps NOx emissions on a broader scale
did not yet reach their usual intensity during the first 3 weeks after the
end of the lockdown period, as already suggested by NO2 data from
Ispra. Regarding NO, it should be noted that a significant decrease in
comparison to the weeks before the lockdown period could only be detected at
the city centre station (Fig. S8). Both sites in the hinterland are much
closer to highways and might reflect more NO emissions from heavy-duty
vehicles, whose traffic did not decrease that much, at least during the first
weeks of the lockdown period.
As in Ispra, the implementation of lockdown measures on 9 March 2020 led to
increases in O3 in the Milan conurbation compared to expected
concentrations (Figs. 3 and 4). This can be explained by the decrease in
O3 titration by NO in a pollution regime where photochemical O3
production is limited by the availability of volatile organic compounds. The
relaxation of lockdown measures did not lead to the expected decrease in
O3 (Fig. 4), perhaps because NOx emissions did not fully recover
during the 3 weeks following the end of the lockdown period.
Again, as in Ispra, no significant change in PM10 could be detected
when lockdown measures were implemented. This is very likely due to the fact
that decreased emissions of PM10 (and PM10 precursors) were
compensated by increases from other sources like domestic heating and
enhanced formation of secondary PM. In particular, Huang et al. (2021)
reported that increased oxidative capacities of the atmosphere (e.g. higher
O3 concentrations) resulted from the drastic reductions in NOx
emissions following from the lockdown measures in China, which in turn lead
to an increase in the formation rate of nitric acid (HNO3). Such a
phenomenon in northern Italy, together with sustained emissions of ammonia
from agriculture, which was not affected by the lockdown (ARPA Lombardia, 2020a),
could have resulted in increased formation of particulate ammonium nitrate
(NH4NO3) and therefore an increase in PM10 concentrations
beyond expected concentrations in the Milan conurbation. For the 3 weeks
following 4 May 2020, the relaxation of lockdown measures led to a further
increase in PM10 in comparison to expected concentrations in Milan.
This might be attributed to the upturn in traffic and particularly to the
re-suspension of dust from roads that had been little used for several
weeks.
Conclusions
Northern Italy has been an air pollution hotspot for decades. Northern
Italy also hosted the very first clusters of the COVID-19 epidemic in Europe, and
from February 2020, containment measures were gradually implemented,
culminating in strict lockdown measures in force between 9 March and 4 May 2020. We isolated specific impacts of the lockdown measures on air pollution
by comparing observed with expected data at one regional background site
(Ispra) and three urban background sites (in the Milan conurbation) across the
period 17 February–24 May 2020. All four stations were in the COVID-19 “red
zone”. Expected pollutant concentrations were derived from ENSEMBLE
forecasts, which are based on actual meteorological conditions and
historical emissions estimates that ignored the COVID-19 epidemic and
related lockdown measures. Changes in observed versus computed expected
concentrations for the lockdown period and the 3 weeks before and after the
lockdown period should therefore directly reflect the impact of lockdown
measures on air pollution.
We showed that lockdown measures had statistically significant impacts on
concentrations of most gaseous pollutants (Table 1).
However, we were not able to highlight systematic significant effects on
PM10 concentrations.
Focusing on those days for which the ENSEMBLE model forecast
concentrations were above the median for the lockdown period
(Fig. 7), the following can be said:
NO2 concentrations decreased by about -30 % and -50 % at the urban
and regional background sites, respectively, as a result of the lockdown
implementation on 9 March 2020. The relaxation of lockdown measures on 4 May led to a partial recovery in NO2 concentrations in Milan (urban
background) but not in Ispra (regional background).
Unlike NO2, PM10 concentrations were not significantly affected by
the lockdown measures. We showed that the decrease in traffic-related
PM10 was compensated by an increase in PM10 associated with wood
burning for domestic heating in Ispra. PM10 concentrations in Milan are
to a great extent influenced by PM10 “non-urban” and “non-traffic”
sources (Thunis et al., 2018), including the formation of secondary aerosol.
Sustained regional background PM10 concentrations and a modified
HNO3 production regime associated with continuing NH3 emissions
from agriculture could explain the lack of decrease in PM10 resulting
from the lockdown measures in Milan too. In contrast, the relaxation of
lockdown measures led to an increase in PM10 concentrations at both
urban and regional background sites (+30 % and +20 %, respectively)
in May, when domestic heating is much reduced.
The lockdown measures led to an increase in the highest O3
concentrations at both the urban and regional background sites.
The sad experience of the COVID-19 epidemic and subsequent lockdown measures
shows that drastic changes in mobility and economic activity can lead to
0 % (insignificant) to -30 % reductions in air pollution in urban
background areas. These figures suggest that the abatement of air pollution
down to levels that do not have adverse effects on human health in northern
Italy may require structural changes in other sectors, including energy
production, domestic heating and agriculture in addition to transport.
Data availability
Observation data from Ispra are available at https://data.jrc.ec.europa.eu/collection/abcis (European Commission, 2021) and https://actris.nilu.no/ (NILU, Norsk Institutt for Luftforskning, 2021). Observation data from Milan are available at
https://www.arpalombardia.it/Pages/Aria/Richiesta-Dati.aspx (ARPA Lombardia, 2020b).
Model forecast data for all sites are available at https://ads.atmosphere.copernicus.eu/cdsapp#!/dataset/cams-europe-air-quality-forecasts?tab=form (CAMS, 2020d).
The supplement related to this article is available online at: https://doi.org/10.5194/acp-21-7597-2021-supplement.
Author contributions
JPP, EP and LP contributed to conception and design. SMDS, FL and UDS contributed to acquisition of data. JPP, EP, LP, GL and AC contributed to analysis and interpretation of data. JPP, EP, LP, GL and AC drafted the article.
Competing interests
The authors declare that they have no conflict of
interest.
Acknowledgements
The authors thank the Copernicus Atmosphere Monitoring Service Information, in particular the Regional Production
Service. Jean-Philippe Putaud, Luca Pozzoli, Enrico Pisoni, Sebastiao Martins Dos Santos and Friedrich Lagler thank their colleagues from JRC for
fruitful tele-discussions during the whole lockdown period and for helpful
comments on the manuscript.
Financial support
This research has been supported by the European Commission, H2020 Research Infrastructures (ACTRIS-2 (grant no. 654109) and ACTRIS IMP (grant no. 871115)).
Review statement
This paper was edited by Rolf Müller and reviewed by S. Fadnavis and one anonymous referee.
ReferencesARPA Lombardia: Stima preliminare delle emissioni in Lombardia durante l'emergenza COVID-19, available at: https://www.arpalombardia.it/sites/DocumentCenter/Documents/ARIA/report-stima-emissiva-COVID19-lombardia.pdf, last access: 9 October 2020a.ARPA Lombardia: Richiesta Dati, available at: https://www.arpalombardia.it/Pages/Aria/Richiesta-Dati.aspx, last access: 9 October 2020b.Baldasano, J. M.: COVID-19 lockdown effects on air quality by
NO2 in the cities of Barcelona and Madrid (Spain), Sci. Total
Environ., 741, 140353, 10.1016/j.scitotenv.2020.140353, 2020.Copernicus Atmosphere Monitoring Service (CAMS): Quarterly report on ENSEMBLE NRT productions (daily analyses and
forecasts) and their verification, at the surface and above surface March–April–May 2019, available at: https://atmosphere.copernicus.eu/sites/default/files/2020-02/CAMS50_2018SC1_D5.2-3.1.ENSEMBLE-2019MAM_201908_NRTProduction_Report_v1_APPROVED_Ver1.pdf (last access: 15 January
2021), 2019.Copernicus Atmosphere Monitoring Service (CAMS): Air quality information confirms reduced activity levels due to
lockdown in Italy, 17 March 2020, available at: https://atmosphere.copernicus.eu/air-quality, last access: 9 October 2020a.Copernicus Atmosphere Monitoring Service (CAMS): Flawed estimates of the effects of lockdown measures on air quality
derived from satellite observations, 26 March 2020, available at:
https://atmosphere.copernicus.eu/flawed-estimates-effects-lockdown-measures-air-quality-derived-satellite-observations, last access: 9 October 2020b.Copernicus Atmosphere Monitoring Service (CAMS): Quarterly report on ENSEMBLE NRT productions (daily analyses and
forecasts) and their verification, at the surface and above surface,
available at: https://atmosphere.copernicus.eu/sites/default/files/2020-09/CAMS50_2018SC2_D5.2-3.1.ENSEMBLE-2020MAM_202008_NRTProduction_Report_v1.pdf (last access: 15 January 2021), 2020c.Copernicus Atmosphere Monitoring Service (CAMS): CAMS European air quality forecasts, available at: https://ads.atmosphere.copernicus.eu/cdsapp#!/dataset/cams-europe-air-quality-forecasts?tab=form (last access: 15 January 2021), 2020d.Decreto del Presidente del Consiglio dei Ministri (DPCM): Ulteriori disposizioni attuative del decreto-legge
23 febbraio 2020, n. 6, recante misure urgenti in materia di contenimento e
gestione dell'emergenza epidemiologica da COVID-19, 8th of March, 2020, (20A01522) (GU Serie
Generale n. 59 del 08-03-2020), 2020a.Decreto del Presidente del Consiglio dei Ministri (DPCM): Ulteriori disposizioni attuative del decreto-legge
23 febbraio 2020, n. 6, recante misure urgenti in materia di contenimento e
gestione dell'emergenza epidemiologica da COVID-19, applicabili sull'intero
territorio nazionale, 11th of March 2020, (20A01605) (GU Serie Generale n. 64 del 11-03-2020), 2020b.Decreto del Presidente del Consiglio dei Ministri (DPCM): Ulteriori disposizioni attuative del decreto-legge
23 febbraio 2020, n. 6, recante misure urgenti in materia di contenimento e
gestione dell'emergenza epidemiologica da COVID-19, applicabili sull'intero
territorio nazionale, 22nd of March 2020, (20A01807) (GU Serie Generale n. 76 del 22-03-2020), 2020c.Decreto del Presidente del Consiglio dei Ministri (DPCM): Ulteriori disposizioni attuative deldecreto-legge
23 febbraio 2020, n. 6, recante misure urgenti in materia di contenimento e
gestione dell'emergenza epidemiologica da COVID-19,applicabili sull'intero
territorio nazionale, 26th of April 2020, Pubblicato nella Gazz. Uff. n. 108, 27 aprile 2020, 2020d.Decreto Legge (DL): Misure urgenti in materia di contenimento e
gestione dell'emergenza epidemiologica da COVID-19, 23rd of February 2020, n. 6, (20G00020) (GU Serie
Generale n. 45 del 23-02-2020), 2020a.Decreto Legge (DL): Ulteriori misure urgenti per fronteggiare
l'emergenza epidemiologica da COVID-19, 16th of May 2020, n. 33, (20G00051) (GU Serie Generale n. 125
del 16-05-2020), 2020b.Denier van der Gon, H. A. C., Bergström, R., Fountoukis, C., Johansson, C., Pandis, S. N., Simpson, D., and Visschedijk, A. J. H.: Particulate emissions from residential wood combustion in Europe – revised estimates and an evaluation, Atmos. Chem. Phys., 15, 6503–6519, 10.5194/acp-15-6503-2015, 2015.European Commission: Joint Research Centre Data Catalogue, available at: https://data.jrc.ec.europa.eu/collection/abcis (last access: 9 October 2020), 2021.European Environment Agency (EEA): Air quality in Europe – 2019 report, European Environment Agency Report
No 10/2019, available at:
https://www.eea.europa.eu/publications/air-quality-in-europe-2019 (last access:
25 March 2020), 10.2800/822355, 2019.European Environment Agency (EEA): Air quality and COVID-19, available at:
https://www.eea.europa.eu/themes/air/air-quality-and-covid19/air-quality-and-covid19, last access: 25 March 2020.EUROSTAT: Data Browser, available at: https://ec.europa.eu/eurostat/, last access: 9 October 2020.Galmarini, S., Kioutsioukis, I., Solazzo, E., Alyuz, U., Balzarini, A., Bellasio, R., Benedictow, A. M. K., Bianconi, R., Bieser, J., Brandt, J., Christensen, J. H., Colette, A., Curci, G., Davila, Y., Dong, X., Flemming, J., Francis, X., Fraser, A., Fu, J., Henze, D. K., Hogrefe, C., Im, U., Garcia Vivanco, M., Jiménez-Guerrero, P., Jonson, J. E., Kitwiroon, N., Manders, A., Mathur, R., Palacios-Peña, L., Pirovano, G., Pozzoli, L., Prank, M., Schultz, M., Sokhi, R. S., Sudo, K., Tuccella, P., Takemura, T., Sekiya, T., and Unal, A.: Two-scale multi-model ensemble: is a hybrid ensemble of opportunity telling us more?, Atmos. Chem. Phys., 18, 8727–8744, 10.5194/acp-18-8727-2018, 2018.German Aerospace Center (DLR): Despite the influence of weather patterns, the effect of the
Coronavirus on air quality is now visible, available at:
https://www.dlr.de/content/en/articles/news/2020/02/20200505_effect-of-the-coronavirus-on-air-quality-is-now-visible.html, last access: 9
October 2020.Giechaskiel, B.: Particle Number Emissions of a Diesel Vehicle during and
between Regeneration Events. Catalysts, 10, 587, 10.3390/catal10050587, 2020.Giechaskiel, B., Joshi, A., Ntziachristos, L., and Dilara, P.: European
Regulatory Framework and Particulate Matter Emissions of Gasoline Light-Duty
Vehicles: A Review, Catalysts, 9, 586, 10.3390/catal9070586, 2019.Gilardoni, S., Vignati, E., Cavalli, F., Putaud, J. P., Larsen, B. R., Karl, M., Stenström, K., Genberg, J., Henne, S., and Dentener, F.: Better constraints on sources of carbonaceous aerosols using a combined 14C – macro tracer analysis in a European rural background site, Atmos. Chem. Phys., 11, 5685–5700, 10.5194/acp-11-5685-2011, 2011.Grange, S. K., Lee, J. D., Drysdale, W. S., Lewis, A. C., Hueglin, C., Emmenegger, L., and Carslaw, D. C.: COVID-19 lockdowns highlight a risk of increasing ozone pollution in European urban areas, Atmos. Chem. Phys., 21, 4169–4185, 10.5194/acp-21-4169-2021, 2021.Huang, X., Ding, A., Gao, J., Zheng, B., Zhou, D., Qi, X., Tang, R., Wang, J.,
Ren, C., Nie, W., Chi, X., Xu, Z., Chen, L., Li, Y., Che, F., Pang, N., Wang, H., Tong, D., Qin, W., Cheng, W., Liu, W., Fu, Q., Liu, B., Chai, F., Davis, S. J., Zhang, Q., and He, K.: Enhanced secondary pollution offset reduction of primary
emissions during COVID-19 lockdown in China, Natl. Sci. Rev., 8, nwaa137,
10.1093/nsr/nwaa137, 2021.INERIS (Institut national de l’environnement industriel et
des risques): The lockdown and air quality: nitrogen-dioxide pollution falls in France’s largest cities, available at: https://www.ineris.fr/en/ineris/news/lockdown-and-air-quality-nitrogen-dioxide-pollution-falls-france-largest-cities, last access: 9 October 2020.Istituto Nazionale di Statistica (ISTAT): Principali fattori di pressione sull'ambiente nelle città italiane, available at: https://www.istat.it/, last access: 9 October 2020.Kuenen, J. J. P., Visschedijk, A. J. H., Jozwicka, M., and Denier van der Gon, H. A. C.: TNO-MACC_II emission inventory; a multi-year (2003–2009) consistent high-resolution European emission inventory for air quality modelling, Atmos. Chem. Phys., 14, 10963–10976, 10.5194/acp-14-10963-2014, 2014.Larsen, B. R., Gilardoni, S., Stenström, K., Niedzialek, J., Jimenez, J., and Belis C. A.: Sources for PM air pollution in the Po Plain, Italy: II.
Probabilistic uncertainty characterization and sensitivity analysis of
secondary and primary sources, Atmos. Environ., 50, 203–213,
10.1016/j.atmosenv.2011.12.038, 2012.Li, L., Li, Q., Huang, L., Wang, Q., Zhu, A., Xu, J., Liu, Z., Li, H., Shi,
L., Li R., Azari, M., Wang, Y., Zhang, X., Liu, Z., Zhu Y., Zhang K., Xue
S., Chel, M., Ooi, G., Zhang, D., and Chan, A.: Air quality changes during the
COVID-19 lockdown over the Yangtze river delta region: An insight into the
impact of human activity pattern changes on air pollution variation, Sci.
Total Environ., 732, 139282, 10.1016/j.scitotenv.2020.139282, 2020.Marécal, V., Peuch, V.-H., Andersson, C., Andersson, S., Arteta, J., Beekmann, M., Benedictow, A., Bergström, R., Bessagnet, B., Cansado, A., Chéroux, F., Colette, A., Coman, A., Curier, R. L., Denier van der Gon, H. A. C., Drouin, A., Elbern, H., Emili, E., Engelen, R. J., Eskes, H. J., Foret, G., Friese, E., Gauss, M., Giannaros, C., Guth, J., Joly, M., Jaumouillé, E., Josse, B., Kadygrov, N., Kaiser, J. W., Krajsek, K., Kuenen, J., Kumar, U., Liora, N., Lopez, E., Malherbe, L., Martinez, I., Melas, D., Meleux, F., Menut, L., Moinat, P., Morales, T., Parmentier, J., Piacentini, A., Plu, M., Poupkou, A., Queguiner, S., Robertson, L., Rouïl, L., Schaap, M., Segers, A., Sofiev, M., Tarasson, L., Thomas, M., Timmer
mans, R., Valdebenito, Á., van Velthoven, P., van Versendaal, R., Vira, J., and Ung, A.: A regional air quality forecasting system over Europe: the MACC-II daily ensemble production, Geosci. Model Dev., 8, 2777–2813, 10.5194/gmd-8-2777-2015, 2015.Nakada, L. Y. K. and Urban, R. C.: COVID-19 pandemic: Impacts on the air
quality during the partial lockdown in São Paulo state, Brazil, Sci. Total Environ., 730, 139087, 10.1016/j.scitotenv.2020.139087, 2020.Norsk Institutt for Luftforskning (NILU): ACTRIS Data Centre, available at: https://actris.nilu.no/ (last access: 9 October 2020), 2021.
Ozgen, S., Becagli, S., Bernardoni, V., Caserini, S., Caruso, D., Corbella, L.,
Dell'Acqua, M., Fermo, P., Gonzalez, R., Lonati, G., Signorini, S., Tardivo, R.,
Tosi, E., Valli, G., Vecchi, R., and Marinovich, M.: Analysis of the chemical
composition of ultrafine particles from two domestic solid biomass fired
room heaters under simulated real-world use, Atmos. Environ., 150, 87–97,
doi.org/10.1016/j.atmosenv.2016.11.048, 2017.Putaud, J. P., Cavalli, F., Martins dos Santos, S., and Dell'Acqua, A.: Long-term trends in aerosol optical characteristics in the Po Valley, Italy, Atmos. Chem. Phys., 14, 9129–9136, 10.5194/acp-14-9129-2014, 2014.Riccio, A., Giunta, G., and Galmarini, S.: Seeking for the rational basis of the Median Model: the optimal combination of multi-model ensemble results, Atmos. Chem. Phys., 7, 6085–6098, 10.5194/acp-7-6085-2007, 2007.Sandradewi, J., Prevot, A. S. H., Szidat, S., Perron, N., Alfarra, M. R.,
Lanz, V. A., Weingartner, E., and Baltensperger, U.: Using aerosol light
absorption measurements for the quantitative de-termination of wood burning
and traffic emission contributions to particulate matter, Environ. Sci.
Technol., 42, 3316–3323, 10.1021/Es702253m, 2008.Seinfeld, J. H. and Pandis, S. N., Atmospheric Chemistry and Physics: From
Air Pollution to Climate Change, John Wiley & Sons, Inc., New York, USA, 2016.Thunis, P., Degraeuwe, B., Pisoni, E., Trombetti, M., Peduzzi, E., Belis, C. A.,
Wilson, J., Clappier, A., and Vignati, E.: PM2.5 source allocation in European
cities: A SHERPA modelling study, Atmos. Environ., 187, 93–106,
doi.org/10.1016/j.atmosenv.2018.05.062, 2018.Tiwari, M., Sahu, S. K., Bhangare, R. C., Yousaf, A., and Pandit, G. G.: Particle size
distributions of ultrafine combustion aerosols generated from household
fuels, Atmos. Pollut. Res., 5, 145–150,
10.5094/APR.2014.018, 2014.Tonnensen, G. S. and Dennis, R. L.: Analysis of radical propagation
efficiency to assess ozone sensitivity to hydrocarbons and NOx: 2. Long-lived
species as indicators of ozone concentration sensitivity, J. Geophys. Res.,
105, 9227–9241, 10.1029/1999JD900372, 2000.Wiedensohler, A., Wiesner, A., Weinhold, K., Birmili, W., Hermann, M.,
Merkel, M., Müller, T., Pfeifer, S., Schmidt, A. Tuch, T., Velarde, F.,
Quincey, P., Seeger, S., and Nowak, A.: Mobility particle size spectrometers:
Calibration procedures and measurement uncertainties. Aerosol Sci. Technol.
52, 146–164, 10.1080/02786826.2017.1387229, 2018.World Meteorological Organisation Sand and Dust Storm Warning Advisory and Assessment System: Dust Forecasts, available at: https://sds-was.aemet.es/forecast-products/dust-forecasts/ensemble-forecast, last access: 15 January 2021.