Mobile measurements of black carbon: Comparison of normal traffic with reduced traffic conditions during COVID-19 lock-down

Abstract

A mobile monitoring campaign was conducted (by bicycle) to assess the black carbon (BC) concentrations in Cluj-Napoca city, Romania, in 2020, before, during and after COVID-19 lock-down. Over the entire study period, the BC concentrations ranged between 1.0 and 25.9 μg/m³ (averaged per street section and period characterized by different traffic conditions). Marked spatial and temporal differences were observed. Observed differences in BC concentrations between locations are attributed to traffic intensities, with average BC concentrations, under normal circumstances, of 6.6–14.3 μg/m³ at roads with high to intense traffic, compared to 2.8–3.1 μg/m³ at areas with reduced traffic, such as residential areas, parks and pedestrian streets. The COVID-19 measures impacted traffic volumes, and hence average BC concentrations decreased from 5.9 μg/m³ to 3.0 μg/m³ during lock-down and in a lower extent to 3.4 μg/m³ and 4.4 μg/m³ in post-lockdown periods with reduced and more normalized traffic. Two approaches to account for variations in background concentrations when comparing different situations in time are assessed. Subtracting background concentrations that are measured at background sites along the monitoring route is an appropriate method to assess spatio-temporal differences in concentrations. A reduction of about 1–2 μg/m³ was observed for the streets with low to medium traffic, and up to 6 μg/m³ at high traffic locations under lockdown. The approach presented in this study, using mobile measurements, is useful to understand the personal exposure to BC along the roads in different seasons and the influence of traffic reduction on BC pollution during prolonged restrictions. All these will support policymakers to reduce pollution and achieve EU directives targets and WHO recommendations.

Graphical abstract

1. Introduction

Traffic emissions are a major source of air pollution in cities. Negative health effects have been associated with short-term and long-term exposure to traffic-related air pollution (HEI, 2010) and there is toxicological evidence that acute exposure events have independent health effects compared to longer-term exposure (Zhou et al., 2017).

Black carbon (BC) is a component of fine particles, that is emitted from combustion sources (EPA, 2012; Petzold et al., 2013). Studies showed that BC is a better indicator of harmful particulate substances from combustion-related sources (especially traffic) and a better parameter than undifferentiated particulate matter (PM) to assess the impact of traffic measures (Keuken et al., 2012; Harrison et al., 2004; Lefebvre et al., 2011). Exposure to BC is associated with cardiovascular and cardiopulmonary mortality (Janssen et al., 2012).

In this study, a validated mobile monitoring approach for BC (www.airqmap.com), described in detail by Van den Bossche et al. (2015) was used. Previous research showed the potential of mobile measurements to assess personal exposure of cyclists (Berghmans et al., 2009) or other transport modes (Dons et al., 2012), and to assess the spatial variability of urban air quality of different pollutants (PM₁₀, ultrafine particles (UFP), BC) at (multi-)street level (Peters et al., 2013, 2014; Van Poppel et al., 2013). Also, other studies showed the potential of mobile measurements to assess personal exposure to traffic (Apparicio et al., 2016; Jereb et al., 2018; Hankey and Marshall, 2015; Hofman et al., 2018; Okonen et al., 2017; Targino et al., 2016, 2018), spatial air pollution mapping (Van den Bossche et al., 2015) and more specific the effect of bicycle infrastructure on the exposure levels of bicycle commuters (Betancourt et al., 2017; Lonati et al., 2017). However, the use of mobile monitoring to assess other traffic measures than bicycle infrastructure is not commonly used. The impact assessment of traffic measures, requires a data analysis approach to differentiate between variations in background concentrations and variations in local traffic contribution when comparing different situations in time. A common approach to assess the contribution of traffic to local concentrations is subtracting urban background concentrations (Harrison et al., 2004; Thorpe et al., 2007; Keuken et al., 2010, 2013; Dijkema et al., 2008) or use upwind concentrations as reference (Baldauf et al., 2013; Van Poppel et al., 2012).

Fixed air quality monitoring (AQM) stations are mostly not installed at representative locations to assess the impact of traffic on air quality. Moreover, traffic measures affecting a larger part of the city (e.g. Low Emission Zones (LEZ), improved mobility plans) need to be assessed at a city-wide scale. Kelly et al. (2011) studied the effect of Congestion Charging Schemes (CCS) in London and concluded that data from one roadside monitoring site is insufficient to assess the overall impact of the CCS.

In this paper, we used mobile measurements to assess the impact of traffic restrictions due to COVID-19 on air quality. The study was built on the previously developed methodology and applied in the second largest city of Romania, Cluj-Napoca, where no BC data is available. The impact of traffic (among other sources) on air quality may be of concern in this city since a previous study (Levei et al., 2020) showed that for last years (2017–2019) the PM₁₀ annual mean was 22.2 μg/m³, 24.4, 20.3 respectively, above the safety benchmark established by WHO in 2005 (20 μg/m³) and new WHO guidelines set in 2021 (World Health Organization, 2021) (15 μg/m³), but below the annual limit value set by the EU legislation (40 μg/m³).

2. Materials and methods

2.1. Measurement location and route

Cluj-Napoca, Romania, is located in the northwestern part of the country (Figure S1). The city covers an area of 179.5 km² and has 328,268 inhabitants. The fleet registered in 2020 in Cluj country is composed of 2100 buses, 282,000 passenger cars, 44,800 freight vehicles, 15,800 mopeds and motorcycles (INS, 2021). Residential heating and transport represent the main sources of black carbon emissions in Romania according to the national emission inventory, with estimated emission of respectively 9 Gg and 1 Gg in 2016 (Matthews and Paunu, 2019). However, at local level, to the best knowledge of the authors, there is no inventory data for BC emissions. The city has an average altitude of 340 m and continental climate with warm, dry summers and cold winters. The mobile monitoring campaign was performed in the northwestern part of the city in and around the city center. The measurement route had a total distance of 9.0 km and is characterized by mixed housing (family houses and flats in 2–10 level buildings) and different official buildings (city hall, bank headquarters, universities). It also contains two parks (Central Park and Sport Park) and the Somes river bank. Biking lane exists only in some parts of the measurement route. The measurement route was split into 18 sections that alternate green and highly urbanized areas with different traffic conditions (Fig. 1 ). The different sections are described in more detail in Table S1. In the proximity of the measurement route, there is a suburban type air quality monitoring station (CJ 3) and a traffic monitoring sensor (T1).

Fig. 1.

Map of the measurement route with different sections.

2.2. Measurement methodology and measurement platform

Measurements were performed using airQmap (www.airQmap.com), a user-friendly monitoring tool developed by VITO to map BC at street level via repeated measurement runs by bicycle (Van den Bossche et al., 2015). It comprises a measurement unit formed by a microaethalometer (microAeth®, AE51, AethLabs) and a GPS and an automated data processing software to obtain average BC concentration maps over user-defined aggregation periods. The flow rate of the microaethalometer was set at 150 mL min⁻¹, and measurements were made at a temporal resolution of 1s. To reduce the noise in BC measurements, the ONA (Optimized Noise-reduction Averaging, Hagler et al., 2011) algorithm was used with an attenuation threshold of 0.05. The geo-tagged measurements were aggregated (trimmed mean) and attributed to fixed points 20 m apart from each other along the cycling route. The resulting dataset was overlaid and joined with a street map. For further analysis, concentrations of 20 m road segments are further averaged over different sections (Section 1–18) characterised by green and highly urbanized areas with different traffic conditions (see Fig. 1).

2.3. Measurement period

Measurements were performed in 2020 on working days (Monday to Friday) during three time slots per day (morning, noon and afternoon) to represent an average effect of traffic on air quality during daytime. The impact of traffic can be larger when only considering the peak hours. The original aim of the study was to assess the impact of traffic on local air quality by repeating the measurements on different days over six weeks. While running the study, the exceptional traffic situations that occurred due to COVID-19 measures were taken as an opportunity to evaluate a methodology using spatio-temporal data collected by mobile measurements, to assess the impact of traffic measures. Therefore, the measurement period was extended to about 12 weeks and the resulting dataset was split in four measurements periods, characterized by different traffic conditions related to COVID-19 measures that impacted traffic (Table 1 , further details in Table S2). Most measurements were done before lockdown, but overall measurement runs were quite comparable. The measurements were evenly distributed over the time of the day in each of the measurement periods.

Table 1.

Dataset split according to different traffic conditions.

Measurement perioda	Traffic condition	Period	No. of days measured	No. of runs
				Morning	Noon	Afternoon	Total
Pre-lockdown NT	Normal traffic	24th February −13th March	15	15	15	14	44
Lockdown LT	Low traffic	27th April – 13th May	11	10	10	8	28
Post-lockdown RT	Reduced traffic	18th May −29th May	9	9	9	9	27
Post-lockdown NT	Normal traffic	2nr June –1st July	16	11	13	10	34
Total			51	45	47	41	132

^aNT = normal traffic, LT = low traffic, RT = reduced traffic.

2.4. Reference air quality measurements, traffic data and meteorological conditions

One AQM stations at Cluj-Napoca (CJ3) is located near the measuring route, at about 7 m from the road. No BC measurements are available from this station. However, among other air quality parameters, NO₂, which is also a traffic-related pollutant, is measured. The average NO₂ concentration was higher during the pre-lockdown period, whereas the other periods showed similar concentrations (Table 2 ). Differences in NO₂ concentrations can be explained by different background and meteorological conditions and reductions in non-traffic sources. However, given the proximity of traffic at this AQM station, it is also probable that the reduced concentration during the second and third period (and maybe to a lesser extent during the fourth period when traffic was normal) are (partly) attributed to the reduced traffic in the city. Fig. 2 shows the average NO₂ concentrations at CJ3 station in 2017, 2018, 2019 and 2021 in similar periods as pandemic time frame, together with data measured during mobile campaigns in 2020. This shows that concentrations are typically higher during the Pre-lockdown period (in February-March) compared to other periods in warmer season (April–July).

Table 2.

Meteorological conditions, NO₂ concentrations at CJ3 air quality monitoring station and traffic intensity at T1-T4 traffic monitoring sensors during mobile measurements.

	Pre-lockdown NTa	Lockdown LT	Post-lockdown RT	Post-lockdown NT
Average temperature °C	6.02	12.83	13.08	19.02
Average relative humidity %	75.73	67.08	68.25	75.90
Wind direction NE (<0-90°) %	33.33	25.00	21.15	27.69
Wind direction SE (91-180°) %	11.90	19.23	1.92	16.92
Wind direction SW (181-270°) %	22.61	15.38	9.61	7.69
Wind direction NW (271-360°) %	32.14	40.38	67.30	47.69
Average wind speed NE m/s	2.093	2.2	2.36	1.82
Average wind speed SE m/s	2.16	2.05	1.02	2.8
Average wind speed SW m/s	2.97	3.21	2.67	3.7
Average wind speed NW m/s	3.77	3.57	3.34	2.78
T1 (traffic counts in vehicle/h)	NAd	584	NA	NA
T2 (traffic counts in vehicle/h)	1249b	650c	NA	NA
T3 (traffic counts in vehicle/h)	944b	436c	NA	NA
T4 (traffic counts in vehicle/h)	608b	383c	NA	NA

^aNT = normal traffic, LT = low traffic, RT = reduced traffic.

^bbased on traffic measurements in 2019 during similar season as lock-down in 2020.

^cbased on a few days of traffic measurements during lockdown.

^dNA = not available.

Fig. 2.

NO₂ concentrations at CJ3, measured in 2017–2021 in similar periods as pandemic time frame (Pre-lockdown NT, Lockdown LT, Post-lockdown RT, Post-lockdown NT).

Traffic counts (Table 2) were obtained using the Swarco magnetic sensors of the traffic management system from the Local Council of Cluj-Napoca, Traffic Safety Service. Traffic data from 4 locations were available for certain periods (mostly several days) in 2019 and 2020. On average, a traffic reduction of about 45% was observed during the lockdown period.

The daily traffic pattern shows a large increase during morning rush hours. However, there was no clear bimodal pattern with morning and evening peaks, because traffic volumes stayed high during the day.

During the measurement campaign, the lowest temperatures were measured during pre-lockdown NT and the highest during post-lockdown NT (Table 2). During Pre-lockdown, wind was coming mainly from NE and NW, whereas during the other periods, the predominant wind direction was NW (see wind rose in Figure S2).

3. Results and discussion

3.1. Spatio-temporal variability in relation to traffic and street characteristics

The maps of the BC concentration under different traffic conditions: normal conditions (Fig. 3 (A)), lockdown with limited traffic (Fig. 3(B)), post-lockdown with reduced traffic (Fig. 3(C)), and post-lockdown with normal traffic (Fig. 3(D)) are used to investigate the difference in BC concentration at different locations and also to evaluate the impact of different traffic situations on the BC concentration. Concentrations are shown for every 20 m road segment. The average BC concentrations during the normal traffic situation (Fig. 3(A)) shows large spatial differences between streets and locations that can be related to traffic density. Relative low concentrations (2.25–3.75 μg/m³) are measured in the parks and motor vehicle free areas (sections 7, 15, see Fig. 1), as well as in residential areas with limited traffic and characterized by high and medium height building areas (four levels blocks) and green areas in the front of the buildings (sections 4, 5, see Fig. 1). Very high concentrations are measured at locations with very busy multiple lane two-way roads with traffic lights (e.g. 9 and 12, see Fig. 1, Fig. 3) where, especially during rush hours, traffic congestion occurs. These two sections (9 and 12, see Fig. 1) are part of one of the most important roads in Cluj-Napoca, linking the west side with the east side of the city. The BC concentration was lower in section 9 (with concentrations on 20 m segments up to 36.5 μg/m³) than in section 12 (with concentrations on 20 m segments up to 89.0 μg/m³) because the bicycle lane was alongside the road and on the road, respectively (see Fig. 3 (A)). Residential areas with medium traffic (like sections 17 and 18, see Fig. 1, Fig. 3) show intermediate BC concentrations (typically between 3.75 and 5.25 μg/m³).

Fig. 3.

BC concentrations during pre-lockdown with normal traffic (A), lockdown with limited traffic (B), post-lockdown with reduced traffic (C), post-lockdown with normal traffic (D).

High differences in BC concentrations are observed at adjacent street segments within a short distance from each other; such as sections 1 (low traffic), 2 (medium to high traffic), 3 (high traffic) with concentrations on 20 m segments of respectively 2.25–3.75 μg/m³, 3.00–8.00 μg/m³ and >12.00 μg/m³ (up to 60 μg/m³), (see Fig. 1, Fig. 3); the observed differences can be explained by the impact of traffic volume. Sometimes a high spatial variability of concentrations is observed within one street (section 6, see Fig. 1, Fig. 3). In this case, this can be explained by road topology and traffic flows. The part of the road with smooth traffic has concentrations between 3.75 and 5.25 μg/m³ until a roundabout where the road narrows, and the traffic intensifies until a second roundabout with typical concentrations on 20 m segments between 6.0 and 10.0 μg/m³.

The presence and the position of the bicycle lane also influences the BC concentrations: e.g. whereas sections 2, 6 and 16 are characterized by similar ‘medium to high’ traffic, in section 16 where the bicycle route was located on the sidewalk the BC concentration was lower compared to section 6 where bicycle lane is next to the road and to section 2 where there was no bicycle lane and the bicycling was made on the road (see Fig. 1, Fig. 3). Residential areas with medium traffic (like sections 17 and 18, see Fig. 1, Fig. 3) show intermediate BC concentrations (typically between 3.75 and 5.25 μg/m³).

The different road sections (shown in Fig. 1 and described in more detail in Table S1) were grouped based on traffic intensity, to assess the impact on traffic on BC concentrations. As no traffic data are available for the different streets, selected road sections were grouped based on perceived traffic intensity categories “in normal times” (no traffic restrictions due to COVID-19): reduced traffic (residential, very reduced, pedestrian), medium, medium to high, high, intense, very intense and extreme intense traffic taken into account the traffic level in relation to the street infrastructure. For example, the label ‘high’ is used for larger roads, whereas ‘intense’ is used for smaller roads. Fig. 4 (A-D) shows the BC concentrations, grouped per traffic intensity category for the different monitoring periods. For the pre-lockdown period (Fig. 4(A)), characterized by normal traffic conditions, we observe a positive relationship between BC concentration and traffic intensity: intense traffic results in high concentrations, whereas locations with reduced traffic show lower concentrations. This indicates that BC concentrations in Cluj-Napoca are mainly impacted by traffic. BC concentration (median) at extreme traffic sites is approx. five times higher than at remote sites. In addition, sites with higher traffic intensities show higher variability in concentrations reflected by the box-width. This indicates that in these streets there are moments/locations with lower/higher BC concentrations (higher temporal or within street variability). The same trends were observed for the four periods (Fig. 4) and for each of the four periods significant differences were found in BC concentration between the traffic intensity classes (Kruskall-Wallis test, p < 0,01). A pairwise comparison between traffic intensity classes revealed significant (p < 0.01) differences between most of the classes, except for some of the higher intensity classes (excluding extremely intense which was a traffic class with significantly higher BC concentrations in all four periods) and lower intensity classes (reduced). During lockdown, the higher intensity classes (very intense, intense and high) did not differ significantly in BC-concentrations (at confidence level of 0.01), whereas in the other three periods significant differences were observed between two or more of these classes (p < 0.01). So, the reduction in traffic during lockdown resulted in more comparable and lower BC-concentrations at normally busy roads.

Fig. 4.

BC concentrations during pre-lockdown with normal traffic (A), lockdown with limited traffic (B), post-lockdown with reduced traffic (C), post-lockdown with normal traffic (D), grouped per perceived traffic intensity.

3.2. Comparison with other European cities

By comparing these data with other studies of mobile BC measurements using AE51 (Magee Scientific or Aethlabs), we observe that the levels are similar. For example, in Antwerp, a Belgian city (of 200 km² and about 529,000 inhabitants), the lowest BC concentrations were measured in a city park (2.8 μg/m³), and the highest concentrations were found in street canyons and traffic intensive streets (6.0–9.5 μg/m³). Large open areas with busy cross-roads and bus stops had moderate BC concentrations (around 5.0 μg/m³) significantly lower than at street canyons with much less traffic (Peters et al., 2014). These measurements were performed in winter only (February–March), resulting in slightly higher average background concentrations.

In Mechelen, a Belgian city (28 km² and 72,000 inhabitants), extrapolated yearly averages – based on mobile measurements in different seasons – ranged from 1.5 to 1.8 μg/m³ at background locations, pedestrian zone and low-traffic streets; followed by 2.3 μg/m³ at residential areas, 3.0–4.2 μg/m³ at access roads and circulation streets. Individual street segments showed higher concentrations, up to 10 μg/m³ and more during some seasons. The concentrations were strongly impacted by the season, showing city-averaged concentrations between 1.2 μg/m³ (in summer) and 3.2 μg/m³ (in autumn) (Van Poppel et al., Unpublished result).

In a smaller municipality in Belgium, the median values of mobile BC concentrations were 1.5 μg/m³ in background zones and 6.0 μg/m³ at higher traffic-impacted areas (Van Poppel et al., 2013). A mobile study done in Braunswheich, Germany (Ruths et al., 2014) showed that BC at an urban park site was only 1.4 μg/m³ whereas BC was 4.1 μg/m³ at a curbside location. In California, Westerdahl et al. (2005) performed mobile measurements using an Aethalometer (Magee, model 42) and reported BC concentrations at residential areas from 0.75 to 1.5 μg/m³ and 12 μg/m³ at the freeway. In developing megacities outside Europe, much higher BC concentrations are measured due to less stringent emission legislation, like in Manila (Alas et al., 2018), where average BC concentrations at different street segments ranged from 3.0 to 80 μg/m³.

3.3. Impact of lockdown due to COVID-19: comparison of different periods

A summary of the BC concentrations (averaged per street section, see Fig. 1) for the different periods characterized by various COVID-19 measures and related traffic conditions is given in Table 3 .

Table 3.

Summary statistics of BC (μg/m³) concentrations during the four periods.

Measurement perioda	Minimum	1st Quartile	Mean	3rd Quartile	Maximum
pre-lockdown NTa	2.4	3.4	5.9	6.8	24.8
lockdown LT	1.3	1.8	3.0	3.3	15.9
post-lockdown RT	1.0	1.6	3.4	3.6	25.9
post-lockdown NT	1.2	1.9	4.4	5.2	24.4

^aNT = normal traffic, LT = low traffic, RT = reduced traffic.

Fig. 3(A-D) shows the maps of averaged BC concentration for each period. Each individual map represents the average daytime BC concentration for that respective period; so, the average is taken of the different runs, taken at a different time of the day for each of the respective 2 to 4-weeks periods. Only the third period with reduced traffic was shorter because measures changed; however, it was considered a separate period because it represents a situation between lock-down and normal traffic.

The results showed that BC concentrations are different for the different measurement periods. The highest concentrations are measured during the pre-lockdown (first period), and the lowest concentrations are measured during the lockdown (second period).

Similar spatial patterns are observed for the different periods. This is also reflected in relation with traffic intensity categories (Fig. 4(A-D)). A similar increasing trend in BC with increasing traffic intensity is observed in all periods. However, BC concentration (median) at extreme traffic sites is approx. five times higher than at remote sites in pre-lockdown period (Fig. 4(A)) and about three times higher during lockdown (Fig. 4(B)). The difference between extreme traffic and reduced traffic is less than 5 μg/m³ during lockdown compared to 10 μg/m³ in normal traffic conditions. In addition, differences observed between sites with different traffic categories are less apparent compared to the normal traffic situation. The post-lockdown reduced traffic situation (Fig. 4(C)) shows again a larger variation between locations, whereas not to the same extent as the normal traffic situations. This can be due to reduced traffic even when COVID-19 traffic restrictions were abandoned. A study by Skiriene et al. (2021) also demonstrated the impact of reduced economic activities due to COVID-19 on air pollution. Results of the first period also showed higher concentrations at locations with no or limited traffic compared to the other three periods. In Cluj, BC concentrations (without correction for background concentrations) were significantly lower during Lockdown compared to the normal Pre-corona situation for all the traffic intensity categories including the zones with reduced traffic (p < 0,01).

It is known that local air quality in cities is affected by background concentrations, dispersion conditions, and local emission sources. The measured concentration along the routes depends on local conditions such as local traffic density, stop and go traffic, the contribution of heavy-duty traffic, proximity of biking lane, traffic, features of the build environment (e.g. street canyon) and background concentrations. If background concentrations are higher, the concentrations of the entire route are increased, resulting in a more orange or red-colored maps (Fig. 3(A-D)). The NO₂ concentration measured at CJ 3 in similar periods of 2017–2021 indicated that, in each year, the highest values are recorded at the end of winter and the beginning of spring (i.e. pre-lockdown NT). In the spring and summer (i.e., Lockdown, Post-lockdown RT, Post-lockdown NT), the NO₂ concentrations are comparable. However, the lowest NO₂ concentrations are recorded in 2020 during the Lockdown period when traffic is restricted. The analysis of NO₂ data over different years illustrate that, apart from the expected lower emissions from traffic, reduced concentrations during Lockdown LT are also due to the different seasonal conditions, affecting the dispersion of atmospheric pollutants. To further assess the impact COVID-19 traffic restrictions, we have to consider the changes in background concentrations between the different periods.

3.4. Absolute and relative correction to account for differences in background concentrations

A common approach to assess impact of local sources to air pollution is by subtracting ‘background concentrations’ measured at a nearby station (absolute correction, e.g. Lenschow et al., 2001). But also, other rescaling techniques using both multiplicative and additive methods are commonly used to compare air quality data that is not collected simultaneously, to account for changing background concentrations (e.g. to build Land Use Regression models, Dons et al., 2014).

Since there are no fixed (background) stations for BC available in or near Cluj-Napoca, we propose a methodology that used the background data collected during the mobile monitoring to correct for changes in background concentrations. Table 4 shows two sections considered to be not affected by local traffic: these include Central Park (section 7, see Fig. 1) and Sport Park (section 15, see Fig. 1). BC concentrations are 70–80% higher during pre-lockdown compared to the lockdown period. The high background concentrations in the pre-lockdown period can be due to (a combination of)

• •higher (regional) background concentrations outside the city
• •higher traffic intensity in the city affecting neighboring streets and resulting in a higher overall urban background
• •meteorological conditions resulting in lower dilutions and as a result, high concentrations for similar city-wide emissions
• •additional sources of BC in the pre-lockdown period that were not there (or to a lesser extent) in the lock-down period or periods after lockdown (measurements during lockdown started in April whereas pre-lockdown was in February–March), changes in regional background sources in Cluj-Napoca.

Table 4.

BC concentrations (μg/m³) at locations that are not directly (or only limited) impacted by traffic.

Perioda	Central Park (section 7)	Sport Park (section 15)
pre-lockdown NTa	3.07	2.74
lockdown LT	1.78	1.52
post-lockdown RT	1.61	1.32
post-lockdown NT	2.05	1.42

^aNT = normal traffic, LT = low traffic, RT = reduced traffic.

For further analysis, we use the concentration at Central Park (section 7-S7, see Fig. 1) as a reference for the background concentration. Whereas the concentrations were slightly lower for the Sport Park, this location was selected because of its pedestrian zone, whereas the route through Sport Park was along a road that had almost no vehicles, but the impact of an occasional vehicle passage cannot be excluded. To make abstraction of the background concentrations, BC values are expressed as function of background values. Two approaches are compared: relative and absolute background correction by respectively applying a ratio and absolute difference to correct for changing background concentrations (Fig. 5 ). The ratio (Fig. 5(A)) shows the BC concentration of location Sx (BC_Sx) over S7 (BC_S7) and relative rescaled BC concentration is calculated as BC_{rescaled, rel} = BC_Sx/BC_S7. This means that if the number is higher than 1, the location Sx has higher concentration compared to the selected background S7. The absolute difference (Fig. 5 (B)) shows the BC concentration of location Sx minus the BC concentration of S7, so calculating the absolute rescaled BC concentration as BC_{rescaled, abs} = BC_Sx - BC_S7. This means that if the number is higher than zero, the BC concentration at location Sx is higher compared to the selected background S7. Some locations might also show lower concentrations compared to the background reference, resulting in values < 1 or <0 in case of the relative or absolute rescaling, respectively. This can be explained by the variability in the overall ‘background concentration’ in a city, resulting in slightly lower concentrations at other background concentrations than the one selected.

Fig. 5.

BC concentrations compared to background site as the ratio (A) or as the difference (B).

When looking at the relative comparison using the ratio (Fig. 5(A)) we see that for the lockdown period, the ratio is lower for most traffic locations (e.g. sections 9, 10, 12 with very or extreme intense traffic, and section 8 with intense traffic), whereas the ratio for the other remote location is similar (close to 1, e.g. sections 1, 4, 5, 7, 15 with very reduced or reduced traffic). For the period of post-lockdown with reduced traffic, most traffic locations show similar ratios or even higher (e.g. Motilor street in sections 9 and 12 and Donath street in section 3). The post-lockdown period with normal traffic showed slightly higher ratios for most busy traffic locations than during pre-lockdown. There is no indication of increased traffic in this period that could result in higher concentrations compared to the background site. This suggests that the approach using the relative method (with ratio) could result in overestimation of traffic-related pollutants at traffic locations. Also, the background sites show a slightly lower factor (more below 1) compared to the other periods.

When looking at the absolute difference with background (Fig. 5(B)) a significant change between pre-lockdown and lockdown was noticed. We observed a large reduction in rescaled BC concentration, especially at road with higher traffic volumes. In road sections 12 and 9, for example, rescaled BC concentrations drop from 11.2 to 4.9 μg/m³, and from 6.8 to 2.4 μg/m³, respectively. At the same time, we observe a similar difference between the more remote areas when comparing both periods which could be expected because the contribution of traffic emissions to BC concentrations is low at these sites. During the post-lockdown with normal traffic, the concentration difference is similar to the values during the pre-lockdown, except for one of the traffic locations (Emil Isac Street in section 8).

The analysis shows that pre-lockdown and post-lockdown periods show comparable results when using the absolute background correction method; in addition, other remote locations show more similar results in different periods. Therefore, we conclude that it is better to use the absolute method to assess the impact of traffic and to take into account differences in background conditions. For background sites, it is straightforward that absolute correction is right. However, for traffic emissions we can expect that the impact of concentrations is larger when meteorological conditions result in lower dilution and therefore relative correction could be better. On the other hand, using the relative approach will result in overcompensation of traffic increment (when background concentrations are high during normal traffic scenario) and might result in underestimation of traffic impact on concentrations. Both methods can be considered as complementary approaches to look at the data. The relative method might work better when a traffic location is used as reference location instead of a background location.

4. Conclusions

A mobile monitoring campaign was conducted to assess the BC concentrations in Cluj-Napoca, Romania, before, during and after COVID-19 lock-down. Results showed a clear spatial and temporal variability, with BC concentrations in a range (averaged per street section and period) between 1.0 and 25.9 μg/m³ depending on location (street section) and time of measurement. Spatial gradients could be attributed to traffic intensities with average BC concentrations (under normal traffic conditions) of 18 μg/m³ at roads with high traffic, compared to 1.3–4.1 μg/m³ at streets and areas with reduced traffic, such as residential areas, parks and pedestrian streets. The temporal variability was due to background concentrations and changes in traffic volumes due to COVID-19 measures. To assess the impact of COVID-19 measures, two approaches were applied to correct for variations in background concentrations between the periods: correction by absolute and relative methods. Concentrations measured at a low-traffic zone along the measurement route were used to take into account background variability. Average concentrations at these locations during each measurement period were compared to the overall measurement period. The study showed that absolute correction (by subtraction of background values) was better to assess the impact of traffic measures because similar results are obtained when applying this method to periods with similar traffic conditions and different background concentrations, for both traffic and remote areas. In addition, both methods can be considered as complementary ways of looking at data collected in different periods, to take into account variation in background concentrations. After applying the correction, a reduction of about 1–2 μg/m³ for streets with low to medium traffic and up to 6 μg/m³ at high traffic locations under lockdown situation was observed. This study showed that spatio-temporal data collected by mobile measurements can be used to assess the impact of traffic measures and that background areas in the route can be used to take into account (slightly) different background concentrations between the different measurement periods (scenario's).

Funding

This work was supported by the European Commission Directorate-General Environment under the project “Analyzing the effect of residential solid waste burning on ambient air quality in central and eastern Europe and potential mitigation measures” [grant number 07.027737/2018/788206/SER/ENV.C.3], the European Union’s Horizon 2020 research and innovation programme under the project "Research Infrastructures Services Reinforcing Air Quality Monitoring Capacities in European Urban & Industrial AreaS (RI-URBANS)" [grant number 101036245] and by the Ministry of Research, Innovation and Digitization under the Core Program [grant number18N/2019].

CRediT authorship contribution statement

Martine Van Poppel: Conceptualization, Methodology, Supervision. Jan Peters: Formal analysis, Methodology, Writing – original draft, Visualization. Erika Andrea Levei: Investigation, Writing – original draft, Visualization. Luminița Mărmureanu: Writing – original draft, Visualization. Ana Moldovan: Investigation. Maria-Alexandra Hoaghia: Investigation. Cerasel Varaticeanu: Investigation. Jo Van Laer: Experimental preparation, Data processing.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article.

Data availability

Data will be made available on request.