1. Introduction
As aerosols now represent one of the main causes of uncertainty in the Earth radiative budget [
1], measurements of their optical properties in various conditions and locations are needed to better quantify the aerosol radiative forcing and improve climate forecasts. Although there are large natural sources of aerosol such as deserts or forest fires, anthropogenic emissions are a major contributor to the budget of some aerosol families such as black carbon [
1,
2]. This is particularly true around cities where traffic, heating and industrial activities are concentrated.
The need for aerosol observations led to the development of large permanent observation networks, such as the Aerosol Robotic Network (AERONET) [
3], the Micropulse Lidar Network (MPLNET) [
4] and the Aerosol, Clouds and Trace gases Research Infrastructure Network (ACTRIS, formerly EARLINET) [
5]. These networks can provide the long time series needed for climatological studies, but Europe and North America are over-represented while Russia, Asia and Africa are still widely under-sampled. Moreover, in order for the observations to be representative of a larger area, these networks’ stations are often located away from direct aerosol sources, i.e., away from the cities. On the same continental scale, several airborne field experiments were organized in the framework of projects dedicated to long range transport of, among other things, pollution plumes. One can cite for instance the Aerosol Characterization Experiments (ACE-1, ACE-2, and ACE-Asia) [
6,
7,
8,
9]; the Indian Ocean Experiment (INDOEX) [
10,
11]; the Polar study using Aircraft, Remote sensing; surface measurements; and models of Climate chemistry, Aerosols and Transport project (POLARCAT) [
12] and the European Aerosol, Cloud, Climate and Air Quality Interactions project (EUCAARI) [
13].
To get a better insight on pollution aerosols, several field campaigns were conducted on a smaller, regional scale, near large pollution hotspots such as the megalopolises of Mexico City, with the Megacity Initiative: Local And Global Research Observations project (MILAGRO) [
14], and London, with the Emissions around the M25 motorway (EM25) project [
15]. Paris megalopolis also hosted several campaigns, such as the Air Pollution Over the Paris Region project (ESQUIF) [
16,
17], the Lidar pour la Surveillance de l’Air (LISAIR) [
18], and the Megacities: Emissions, urban, regional and Global Atmospheric Pollution and climate effects, and Integrated tools for assessment and mitigation project (MEGAPOLI [
19,
20]. Most of these urban field campaigns were performed in North America and Europe, even though most of the megalopolises are now located in Asia and Africa. A few observations exist in these cities, either thanks to the initiative of some research groups who investigated pollution hotspots such as Beijing or the Pearl River Delta [
21,
22,
23], or through programs such as EUCAARI (see above). The observations performed in the framework of this program, however, rather targeted the regional background, such as the large observations networks (e.g., near New Delhi [
24] or in the North China Plain, near Beijing [
25]). The Asian haze was also sampled during the already mentioned projects INDOEX [
10,
26] and ACE-Asia [
9].
Russia hosts only five stable AERONET station while covering 11.5% of the world dry lands. Moreover, except for a few projects involving international collaboration such as the joint Soviet–American campaign for the study of Asian dust [
27], observations have long been published only in Russian language, although the situation is changing, as proven by the recent increase in the number of papers, sometimes bearing on rather old data. For instance, Panchenko et al. [
28] published in 2012 about profiles of particle concentration and extinction collected in the Tomsk region, Southern Siberia, during airborne campaigns that started in 1986. Another airborne campaign took place more recently (2013) in the same Tomsk region [
29], while an itinerant campaign was conducted in Northern Siberia in the framework of the Airborne Extensive Regional Observations in Siberia project (YAK-AEROSIB) [
30]. These observations, however, were performed in remote locations and focused on desert dust or forest fire aerosols rather than on pollution particles, such as the measurement from the Zotino Tall Tower Observatory (ZOTTO) [
31,
32], which is located in the taiga, 600 km northwest of Krasnoyarsk.
Urban observations in Russia are scarcer—one can cite for instance Chubarova et al. [
33], who presented a variability and trend study from the Moscow AERONET station—and we were not able to find any lidar profiles of aerosol optical properties above Russian cities. Lidar observations have long been conducted on a regular basis in Tomsk, but are dedicated to stratospheric aerosols (e.g., [
34,
35]). Another lidar station exists in the nearby country of Kyrgyzstan, but observations there are focused on desert dust events [
36] Similarly, a one-year campaign took place in Tajikistan in the framework of the Central Asian Dust EXperiment (CADEX) [
37].
Only space-borne instruments have the capability to cover other Russian regions, such as the industrial cities of southern Siberia. One can cite, for instance, the Moderate Resolution Imaging Spectrometer (MODIS) (e.g., [
38,
39]), the Polarization and Directionality of the Earth Reflectance/Polarization and Anisotropy of Reflectances for Atmospheric Sciences coupled with Observations from a Lidar (POLDER/PARASOL) (e.g., [
40]), and the Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) aboard the Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation (CALIPSO) (e.g., [
41,
42]). However, space-borne observations are limited by cloud-coverage and by the satellite overpass time so that ground-based observations are still useful to better characterize the aerosol optical properties over Russia, particularly using a Raman lidar that allows a proper determination of the intensive property that is the extinction-to-backscatter ratio (so-called lidar ratio (LR)).
In June 2013, we performed the first road transect from Paris to Ulan-Ude, near Lake Baikal, profiling aerosols all the way from Europe to Siberia with a N
2-Raman lidar embedded on a van. The general analysis of the data recorded during this journey and the case studies corresponding to forest fire and desert dust outbreaks have already been presented [
43]. The general analysis of the data showed a great variability of the aerosol optical properties all along the journey, but this feature had to be confirmed by a more detailed analysis of the urban observations that could not fit in this first paper. Hence, we now present in detail the pollution aerosol optical properties derived from the mobile N
2-Raman lidar over various urban areas. Due to the mentioned scarcity of the data over Russia, this offers a new scientific insight despite the shortness of the sample over time.
By comparing the observations recorded in 11 cities visited along the journey, we will try to answer the question: do pollution particles in Russia have optical properties similar to what can be observed in Europe or are these properties closer to what exists in Asian countries? We use two relevant optical parameters to identify the major types of aerosols (dust-like, pollution/fires, and sea salt) in the atmosphere: the lidar ratio and the linear Particle Depolarization Ratio (PDR). They obviously do not give direct access to the granulometry or the chemical composition of the aerosols. Nevertheless, the LR is inversely proportional to the single scattering albedo and the backscatter phase function, making this parameter sensitive to the size and absorbing properties of aerosols, which are directly relevant to their radiative impact. The PDR is sensitive to the particles’ shapes, so it is used to separate the dust-like particles from the other types of aerosol.
This paper is therefore organized as follows: in
Section 2, the campaign itinerary and the N
2-Raman lidar instrument are exposed while the data processing methods used to retrieve the LR and PDR are presented in
Section 3. Then, five case studies are detailed in
Section 4 (Paris, Berlin, Moscow, Ufa and Irkutsk) and the other six cases are more briefly described. In
Section 5, the similarities and differences between the 11 cities are discussed in terms of LR and PDR, and a comparison is performed with CALIOP observations available during the end of spring and summer 2013 above Russia. Finally,
Section 6 presents a brief summary of the paper and its conclusions.
4. Case Studies
The results detailed below involve five of the eleven cities included in this study: the megalopolis of Paris (France,
Section 4.1), the metropolis of Berlin (Germany,
Section 4.2), the megalopolis of Moscow (Russia,
Section 4.3), and two large industrial cities of central or eastern Russia (Ufa,
Section 4.4, and Irkutsk,
Section 4.5). The other cases are briefly described in
Section 4.6. Results in terms of LR and PDR values at 355 nm are summarized in
Table 2. The different contributions to the global uncertainties on the LR and PDR values presented in
Table 2 are detailed in
Appendix A (
Table A1 and
Table A2).
4.1. Paris
On Tuesday 4 June 2013, the van departed shortly before 07:00 UTC (09:00 LT) from Palaiseau, a town located 19 km south-southwest from Paris; then the van headed to the southeast to exit the agglomeration. The day was sunny and a light wind was blowing from the north-northeast, bringing Paris morning pollution plume over the lidar. All data recorded in the densely-urbanized area surrounding Paris (up to 30 km from Paris center) were gathered into a 1-h average profile covering the end of the morning traffic peak (07:00–08:00 UTC). For this case, only a single-layer constrained Klett inversion was possible due to: (i) the limited range of the Raman channel (these are daytime observations, recorded ~3 h after sunrise); and (ii) the large variations of the aerosol optical properties above the Raman channel maximal range. Hence, the Raman partial AOT between the complete overlap range and the morning Planetary Boundary Layer (PBL) top was used to constrain the lidar ratio, resulting in a LR value of 83 ± 6 sr. As a comparison, using the AOT measured by the Palaiseau AERONET sun-photometer (~0.35 at 355 nm) to constrain the LR gave an almost identical value of 83 ± 18 sr, despite the 2 h-time gap between the lidar profile and the sun-photometer first observation (at 09:50 UTC), a delay which resulted from the presence of high level clouds.
In a second step, the LR value retrieved from the lidar/AERONET synergy was used to invert the 5-min average profiles and plot the AOT time series and the time–height cross-sections of both the ABC and the PDR (
Figure 4). The ABC was the highest into the morning PBL where pollution particles from the morning traffic peak were trapped below 0.7 km above mean sea level (a.m.s.l.). Pollution from the previous day or/and contribution of remote pollution plumes was present above, in the residual layer (0.7 km–1.4 km a.m.s.l.). The ABC had a secondary maximum near the residual layer top, commonly attributed to the water-coating of particles. In both layers, the ABC gradually decreased as the van moved away from Paris, which resulted in a decrease of the AOT from ~0.35 near Paris to ~0.20 into the countryside. The PDR was very low in Paris pollution plume, with an average value of 0.6 ± 0.7% in the boundary and residual layers (average between 07:00 and 08:00 UTC and 0.4 and 1.4 km a.m.s.l.). As the van moved away from Paris, the aerosol mix incorporated a higher fraction of terrigenous particles and the PDR slightly increased (~1.5%). The elevated aerosol layers (>2 km a.m.s.l.) are not discussed in this paper.
4.2. Berlin
On Thursday 6 June 2013, the van travelled through Eastern Germany. It passed near Leipzig around 08:30 UTC and entered Berlin agglomeration from the southwest at 10:30 UTC. The van left Berlin area from the southeast at 15:10 UTC and headed towards the Polish border. The LR of Berlin aerosols is determined using a 30-min average profile gathering the data recorded when crossing southwest Berlin (10:30–11:00 UTC). For the same reasons as in Paris, only a single-layer constrained Klett inversion was possible to retrieve the lidar ratio over Berlin. In this case (observation recorded close to solar noon), the Raman channel range was actually so limited that the columnar LR had to be constrained using the AOT derived from MODIS Terra (0.26 at 550 nm, satellite overpass at 10:25 UTC). In addition, the Ångström exponent used to retrieve the cumulative AOT on this peculiar day was 1.8, following the daily mean value retrieved by the AERONET sun-photometer in Leipzig. Indeed, back-trajectories from the Hybrid Single-Particle Lagrangian Integrated Trajectory model (HYSPLIT) show that both cities were overflown by the same air mass on this day, which came from northwestern Poland and the Baltic Sea (
Figure S1). Data recorded over southeast Berlin were not used because boundary layer cumulus clouds had formed at this time, making the aerosol profile too heterogeneous for the constrained Klett procedure to work properly. The LR value retrieved using the morning data was 68 ± 21 sr.
Figure 5 presents the AOT, ABC and PDR from the 5-min average profiles, inverted using the LR value retrieved from the lidar/MODIS synergy. Several aerosol layers were present in the free troposphere coming from Eastern Europe, but only the local aerosols trapped in the PBL are considered here. The profile recorded nearest from Leipzig stands outs from the rural background in terms of ABC and AOT. Both variables increased when approaching Berlin and a pronounced ABC maximum appeared near the PBL top as humidity increased (profiles with cumulus clouds were filtered, hence the gaps in the data). The PDR profile in Berlin showed a maximum in the lowest visible layer and decreased regularly with altitude throughout the PBL. The particularly large amount of pollens reported this day might explain this profile: pollens are very coarse particles that are too heavy to reach the upper levels of the PBL, but have a high depolarizing power [
55,
56]. This combined with the effect of relative humidity, as aerosols reaching the upper levels of the PBL became less depolarizing due to water-coating. Due to pollens, the PDR in Berlin lower PBL was therefore slightly higher than in Paris, with a value of 2.3 ± 1.2% (average between 10:30 and 15:00 UTC and below 0.8 km a.m.s.l.).
4.3. Moscow
The van passed near Moscow on the early morning of Monday 17 June 2013. It entered the agglomeration from the south at 06:15 LT (02:15 UTC), bypassed the inner city using the third ring road (~14 km from the city center) and exited the agglomeration towards the east at 07:45 LT (03:45 UTC). The wind was very light on this morning, blowing from the north-northwest direction so that the van’s ride was downwind of Moscow city center. To determine the LR, all data recorded in Moscow agglomeration were gathered into a single 90-min average profile. The sun was lower than when the van crossed Paris or Berlin so the multi-layer constrained Klett inversion could be used up to 1.5 km a.m.s.l. The resulting average profiles of ABC, LR and PDR are presented on
Figure 6a. According to HYSPLIT back-trajectories (
Figure S2), the upper layer of low LR (46 ± 8 sr) and relatively high PDR (up to 9%) was part of the dust event described in Dieudonné et al. [
43] that brought desert dust from the Caspian-Aral region all over southern Russia. Moscow pollution particles were present in the residual layer (0.5–0.9 km a.m.s.l.) and in the shallow early morning boundary layer (below 0.5 km a.m.s.l.). The LR of 82 ± 13 sr retrieved in the lowest bin of the inversion procedure (0.3–0.5 km a.m.s.l.) is therefore the most representative value for Moscow fresh pollution.
The LR profile on
Figure 6a was extended (blue line) and used to invert the 5-min average profiles and plot the AOT time series and the time–height cross-sections of the ABC and PDR (
Figure 7). The ABC showed a maximum near the residual layer top, probably due to a humidity effect, similar to in Berlin. The residual layer depth increased notably near the city center and this dome-like shape possibly resulted from the effect of the urban heat island on the previous day PBL. Contrary to the LR, the PDR was slightly higher in Moscow than in Paris, with 2.3 ± 1.0% (average below 0.9 km a.m.s.l. for the 90 min of data), a value comparable with the one retrieved in Berlin.
4.4. Ufa
Located ~1200 km east of Moscow, Ufa (56°41′N, 55°58′E) is the last city before the Ural Mountains, which mark the western limit of Siberia. With 1.07 million inhabitants, Ufa is a large industrial center with a specialty in oil refining and petrochemistry, in relation to the numerous oil fields that are exploited southwest from the city, in the triangle between Samara, Orenburg and Ufa cities [
47]. The van was stationed in Ufa during the night from Wednesday 19 to Thursday 20 June 2013. Fixed observations were recorded from 00:40 to 10:40 LT (19:40–05:40 UTC), which correspond to the second part of the night and morning, as sunrise occurred at 04:00 LT (23:00 UTC). The measurement site was located about 5 km northeast from the city center, on the shore of the Belaya River which is ~100 m below the city ground level. The wind was blowing from the southeast, both near the ground and in the free troposphere, so that the lidar was downwind from both the city center and the oil fields. On the contrary, Ufa major industrial sites were in the north-northeast direction and did not a priori impact the lidar.
The aerosol optical properties were first determined from a 3.3-h average profile grouping the night data (00:40–04:00 LT, i.e., 19:40–23:00 UTC). The time-averaged ABC, LR and PDR profiles are presented on
Figure 6b. Due to the very low aerosol load, the ABC was noisy, the LR profile exhibited fluctuations and the inversion gave unrealistically high LR values below 0.8 km a.m.s.l. The lowest layer, between 0.8 and 1.5 km a.m.s.l., was the residual layer with an average LR of 96 ± 9 sr and an average PDR of 4.7 ± 1.4%. The uppermost layer (2.3–4.2 km a.m.s.l.) had a much lower LR (35 ± 9 sr) and a similarly low PDR (3.5 ± 1.4%), which is usually associated with biomass burning particles or polluted dust, i.e., a dust-smoke or dust-pollution mix as classified by CALIOP operational algorithm (e.g., [
57,
58,
59]). As this elevated layer could not be attributed with certainty to pollution, for instance from the regional oil fields, it was not included in
Table 2.
A representative LR profile (blue line on
Figure 6b) is built from the raw profile produced by the standard Raman inversion; this nocturnal LR profile was used to process the 5-min average profiles recorded during the following morning. The resulting AOT time series and time–height cross-sections of the ABC and PDR are presented on
Figure 8 (only morning data from 00:00 UTC are plotted because all night and dawn profiles recorded before 01:00 UTC were very similar). The left part of
Figure 8 shows a low ABC background extending up to 3–4 km a.m.s.l. that already appeared on the nocturnal average profile (
Figure 6b). Between 01:20 and 02:15 UTC, a highly depolarizing layer passed over the lidar, with an average PDR of 20 ± 7% and a maximal value reaching 24%. To ensure these high PDR values did not result from an inappropriate LR, the 55-min average profile grouping data recorded during the layer’s overpass were inverted using the single-layer constrained Klett inversion method, using the Raman partial AOT between 0.4 and 1.3 km a.m.s.l. as constraint. The resulting LR value (96 ± 7 sr) was almost identical to the 96 ± 9 sr observed in the residual layer during the night and used in the representative profile, confirming the PDR values presented for this layer on
Figure 8 were trustworthy, although we lack information about the local wind field and particle sources to identify the nature of these aerosols.
Several other aerosol layers are visible on the right side of
Figure 8 (after 08:00 LT/03:00 UTC). The free tropospheric layer (above 1.5 km a.m.s.l.) was very likely part of the dust event already described in Dieudonné et al. [
43]: it took place between the two case studies detailed in this paper (Kazan on 18 June and Omsk on 22 June) but was not included in the dust case studies because the LR of the layer could not be determined (the Raman channel range was too low at this time). The depolarization being around 13–17%, it was likely a mixture of desert dust with biomass burning or pollution particles. The nature of the layers appearing below 1.5 km a.m.s.l. remains undetermined.
4.5. Irkutsk
Irkutsk (52°17′N, 104°17′E) is a town of 586,000 inhabitants located 4200 km east from Moscow and 50 km west from Lake Baikal. Although it is half as populous as cities such as Ufa, Irkutsk experiences high levels of pollution due to the numerous industries settled all along the valley of River Angara. Going down the valley, in the northwest direction from Irkutsk, there are four other industrial towns: Angarsk (40 km from Irkutsk) which has petrochemical and oil refining facilities, Usolie-Sibirskoe (60 km) with open-air salt mining, Cheremkhovo (120 km) with open-air coal mining, and Bratsk (490 km) where stands one of the world largest aluminum smelting factory and a paper mill. When the van was stationed in Irkutsk, on Thursday 27 June 2013 and during the following night, the wind was blowing from the northwest, putting the Angara Valley and all its industrial facilities directly upwind of the lidar.
The LR in the afternoon urban PBL was determined from a 4-h average profile (15:35–19:40 LT/07:35–11:40 UTC). The Klett procedure was constrained by the Raman partial AOT over a single layer extending from the complete overlap to the PBL top (1.6 km a.g.l.). The resulting LR value was 50 ± 4 sr, which is compatible with the 58 ± 12 sr reported for Central European pollution by Mattis et al. [
60] though it is lower than observations in Western Europe or Moscow at 355 nm (this paper, [
18,
48]). A second average-profile was computed from nighttime data (23:10–03:15 LT/15:10–19:15 UTC) and inverted using the multi-layer Raman constrained Klett procedure. The resulting ABC, LR and PDR profiles are presented on
Figure 6c. The lowest layer (0.6–0.8 km a.m.s.l.) corresponds to the upper limit of the shallow nocturnal inversion layer and exhibits a LR of 80 ± 6 sr. Such a LR is higher than the one retrieved in the afternoon PBL and close to the ones derived from observations in Western Europe or Moscow (see
Section 4.1,
Section 4.2 and
Section 4.3). The scattering structure, between 0.8 and 1.2 km a.m.s.l., encompasses a thin layer evolving inside the residual layer (
Figure 9), which was very likely a pollution plume from one of the first cities upwind. In the free troposphere (>1.6 km a.m.s.l.), the aerosol layer had a higher LR of 83 ± 20 sr suggesting either a pollution plume advected from a more remote location such as Bratsk, or a forest fire plume. This hypothesis could not be checked as MODIS was blinded by a dense cloud cover over the whole region from 23 to 28 June.
The smoothed LR profile from
Figure 6c (blue line) was used to invert the 5-min average profiles recorded during the night. During daytime, the upper part of this profile was kept but the boundary layer LR value was used below 1.6 km a.m.s.l., and a smooth transition was applied around twilight (21:30–23:00 LT/13:30–15:00 UTC). The resulting AOT time series, ABC and PDR time–height cross-sections are presented in
Figure 9. The lidar-derived AOT is higher than the values provided by Irkutsk sun-photometer but this is not surprising as the AERONET station is located ~100 km southwest from the city and out of the Angara Valley, so it was not impacted during our measurements by urban or industrial pollution. The ABC time–height cross section shows the passage of afternoon turbulent updrafts (enriched in aerosols from fresh ground emissions) and downdrafts (poorer in aerosols), resulting in a large temporal variability of the lidar-derived AOT and to a lesser extent, of the PDR. A decrease of the PDR with altitude is also visible (such as in Berlin), as the coarse depolarizing particles did not reach the PBL upper levels and/or as they got coated by water near PBL top. These phenomena also resulted in a PDR gradient between updrafts and downdrafts (variations from one profile to the next were ±0.2% in average, and up to 0.6%).
In Irkutsk, we did not observe an outstanding amount of pollens such as in Berlin; however, similar to all Russian cities that we crossed, we noticed the large amounts of terrigenous particles lifted from the city ground where many bare surfaces remain (wastelands, traffic islands, etc.). The average PDR value remained almost constant between the afternoon PBL (3.2 ± 1.2%) and the shallow nocturnal inversion layer, that becomes clearly visible from 17:00 UTC (3.8 ± 1.1%). The elevated layers were of different nature, as shown by their different PDR value: 5.0 ± 1.6% for the pollution plume evolving around 1 km a.m.s.l. (up to 6.4%) and 2.2 ± 1.0% for the upper, more diluted layer between 2 and 3 km a.m.s.l.
4.6. Other Cities
Riga is the capital of Latvia with 638,000 inhabitants. The van was stationed ~20 km northeast from the city during the night from Sunday 9 Monday 10 June 2013. The wind was from the southwest, bringing the city’s plume toward the lidar. An almost 3-h average profile recorded during the first part of the night and a standard Raman inversion were used to retrieve the LR profile. Two sub-layers containing aerosols with different optical properties were visible inside the residual layer and presented separately in
Table 2.
Pskov is the first city after the Estonia–Russia border. It is located ~200 km south of Saint Petersburg and has 203,000 inhabitants. The van was stationed in the city center during the night of Thursday 14 to Friday 15 June. The LR in the residual layer was determined from a 70-min average profile recorded during the middle of the night, processed using a standard Raman inversion.
Yartsevo is a town of 56,000 inhabitants in Western Russia, located ~50 km northeast of Smolensk and ~320 km west of Moscow. The van was stationed ~3.5 km west from the city during the night of Friday 15 to Saturday 16 June. A 2.5-h average profile recorded during the early morning and the multi-layer Raman constrained Klett procedure were used to retrieve the lidar ratio in the residual layer (below 600 m a.g.l.). The high LR and very low PDR values identify pollution aerosols as MODIS did not show any fire in the region during the previous days. However, the origin of these aerosols is unclear: during the previous day, the wind was blowing from the southeast (from the town and its foundry) but, at the time of the lidar observations, it had turned to the northwest, a region of forests and possibly dried lakes according to satellite images. The very low PDR might have been related to a high humidity level (~90% at the local weather station) and to a small shower that had occurred at the beginning of the night.
Chelyabinsk is the first city encountered in Siberia. It is located 1400 km east of Moscow and 270 km south of Yekaterinburg. With 1.15 million inhabitants, it is a major industrial center with notable activity in metallurgy, metal-working and zinc smelting. The van was stationed in the southern part of the city during the night of Thursday 20 to Friday 21 June. The wind was very light (~1–2 m/s) and turned from northeast to southeast, then southwest during the night, putting alternatively several large industrial facilities upwind of the lidar (several metallurgical sites, an open-air mine, located from 4 to 16 km). The lidar ratio in the residual layer was determined from a 2.5-h average profile recorded during the middle of the night and from a standard Raman inversion.
Omsk, with 1.16 million inhabitants, is the second Siberian city by population after Novosibirsk (which the van also crossed by but where the observations were not exploitable due to clouds). Omsk is located in Central Siberia, ~2200 km east of Moscow. Like in Ufa, the region is specialized in oil extraction, oil refining and petrochemistry. The van was stationed in the city center during the night from Saturday 22 to Sunday 23 June. The LR in the residual layer was determined at twilight and in the middle of the night using two 2.5-h average profiles and standard Raman inversions [
43].
Ulan-Ude, with 421,000 inhabitants, was the last step of the campaign. It is the next city after Irkutsk, located on the southeastern side of Baikal Lake, ~50 km from the shore. Several abandoned open-air mines (coal, tungsten, molybdenum) exist in the area, where the terrain have been left without restoration [
61]. The van was stationed in the city center during the night of Monday 1 to Tuesday 2 July. A 6-h average profile grouping all the night data and a standard Raman inversion were used to retrieve the lidar ratio in the residual layer. Like in Riga, two sublayers with aerosols of different optical properties were visible inside the residual layer and presented separately in
Table 2.
6. Conclusions
In June 2013, a van equipped with a 355 nm N2-Raman and depolarization lidar travelled over 10,000 km from Paris to Ulan-Ude (near Lake Baikal), providing a unique picture of aerosol optical properties over Europe, Western Russia and Siberia. Even if this campaign represents only a snapshot, ground-based observations over Russia are scarce (or not accessible), making the results of this campaign precious. The observations recorded in 11 cities have been exploited, including the megalopolises of Paris and Moscow, the metropolis of Berlin, four large industrial Russian cities (Ufa, Chelyabinsk, Omsk, and Irkutsk), three medium size cities (Riga, Pskov, and Ulan-Ude) and one small Russian town (Yartsevo).
The LR values show no trend in longitude, with most observations (14 out of 17 in the 11 studied cities) within the 67–96 sr range, corresponding to what has previously been observed in Western or Central Europe at the same 355 nm wavelength. However, a few lower values (three observations from 38 to 50 sr) have been retrieved in medium-size Siberian cities under certain conditions (isolated plume and convective afternoon PBL). These LR values are closer to what has been retrieved in Asian countries such as China or India. The PDR is almost always lower than 5%, corresponding to what has been observed for urban haze in Europe and Asia. Only a slight increase of PDR is visible from Western cities to Russian cities, for instance with a +1.7% between Paris and Moscow, two megalopolises of comparable size. This PDR increase is attributed to a higher fraction of coarse terrigenous particles, a fact that is supported by an updraft/downdraft contrast, and by a decrease with altitude and with time around twilight.
These results suggest that pollution particles in large Russian cities are broadly similar to what exists in European cities, i.e., with high LR values and low PDR values, reflecting an aerosol mix dominated by carbonaceous sources (traffic, heating, power plants, etc.), only with a slightly higher fraction of coarse terrigenous particles in the lower PBL. In Russia, the exact proportions of the mix appear to depend on the size of the city: the larger it is, the more carbonaceous sources are likely to be important, explaining why we retrieved lower values of LR above medium size Siberian cities such as Irkutsk and Ulan-Ude (~0.5 million inhabitants) compared to the largest Siberian cities that are Ufa, Chelyabinsk and Omsk (~1.1 million inhabitants). The industrial activities will of course also play a role: cities hosting large oil refining and petrochemical facilities (Ufa and Omsk for instance) will have higher emission of carbonaceous particles, whereas cities hosting metallurgical and open-air mining activities (Chelyabinsk and Ulan-Ude for instance) will have higher emissions of mineral particles. One must keep in mind, however, that the weather conditions will overlay on the source effects (aerosol washing by previous showers such as in Yartsevo, humidity, cloud-capped boundary layer, etc.).
Finally, this study highlighted one regional specificity, i.e., a source producing aerosols with very different optical properties from what has previously been reported in the literature. These particles had both a very high LR (96 ± 9 sr) and very high PDR (20 ± 7%) even though these two features are normally incompatible as they correspond to different types of particles. Unfortunately, this layer remains unidentified due to the lack of ancillary data in the surroundings. This confirms the interest of developing observations in under-sampled regions such as Russia to highlight the differences or similarities with the already broadly sampled aerosols of industrial countries, and to accurately identify all the aerosol sources and model them correctly at a global scale.