Canadian Wildfire Smoke Episode over Europe in October 2023: Lidar, Sun-Photometer, and Model Characterization of Smoke Layers Observed Above Sofia, Bulgaria

Abstract

Massive wildfires release enormous amounts of biomass-burning (BB) aerosols into the atmosphere, which might have a major impact on its thermal and radiative budget, as well as the environment and human health. This work presents the results of a study and characterization of a long-range transport episode of smoke aerosols from Canadian forest fires towards the entirety of Europe, as observed over Sofia, Bulgaria, in early October 2023. This study makes use of data from combined lidar, ceilometer, and sun-photometer measurements, supported by model and forecast data, meteorological radiosonde profiling, and (re)analyses, together with tracking and mapping of the aerosol air transport. A distinctive feature of the considered episode over Europe is the downward movement of the air masses, entraining smoke aerosols from the continental mid-troposphere down to the near-surface layers. The driving mechanism of the long-range transport of BB aerosols and their spread over Europe is revealed. Optical parameters of the registered aerosols are determined and vertically profiled with a high range resolution by lidar data analysis. A wide set of columnar optical and microphysical aerosol characteristics is also provided by sun-photometer measurements. The results show a dominance of relatively fine modes of dry smoke particles in the submicron size range, with a predominantly low degree of non-sphericity, indicating minimal up-size aging during the BB aerosol transport from Canada to the Sofia region. The average daily aerosol radiative forcing is determined by sun-photometer measurements and briefly discussed.

1. Introduction

Atmospheric aerosols have direct and strong impacts on the Earth’s radiative balance, the cloud formation, and the atmospheric chemical and physical processes, thus causing changes in the climate and influencing the air quality and human health [1,2,3,4]. This is why determining the aerosol optical, microphysical, and chemical properties and identifying the aerosol origins and types have been the subject of precise multiple-instrument observations and exhaustive investigations on a global and local scale and carried out within different research programs, networks, and infrastructures [5,6,7,8,9,10]. In order to categorize the aerosol types, various aerosol classification methods have been developed based on different combinations of aerosol characteristics [11,12,13,14,15]. Among the most investigated aerosol types are urban, continental, biomass-burning, desert dust, volcanic ash, marine, and mixed aerosols [11,14,16].

The biomass-burning (BB) aerosols are emitted into the atmosphere as a result of forest, grassland, or bush fires, smoldering combustion, or wood- and coal-based domestic heating. The biomass burning produces various particulate and gaseous pollutants, including greenhouse ones [17,18,19,20,21]. It is the major source of carbonaceous aerosols in the atmosphere, consisting mainly of black carbon and organic carbon [17,19,22]; the former strongly absorbs optical radiation, while the latter is rather a light-scattering substance with noticeable absorption mainly due to its brown-carbon component in the spectral range below 450–550 nm [22,23,24,25,26]. Most of the smoke components may have a significant impact on the global atmospheric thermal and radiative budget, the local/regional and global climate, the air quality, and human health [27,28,29,30,31,32,33].

The particulate matter (PM) released during the biomass burning belongs mainly to the PM_2.5 fraction (particles with sizes smaller than 2.5 µm) [34]. The initially released smoke aerosols may evolve during their propagation through the atmosphere, thus undergoing the so-called “aging” [35,36]. Those of them that are subject to long-range transport have a longer evolution time, resulting in more noticeable changes in their characteristics. In particular, a significant indicator of BB aerosols’ aging is the increase in the particle size due to processes as coagulation and aggregation of the particulate matter, photochemical oxidation, and particle formation [37,38,39]. The final aerosol particle size distribution observed over the measurement site is also influenced by a variety of factors, both local and related to the aerosol transport history.

The thorough worldwide and local monitoring and research of BB aerosols are based on the use of various active (lidar) and passive (photometric or radiometric) remote-sensing approaches, contact (in situ) measurement methods, and the corresponding ground-based, ship-borne, air-borne, and space-borne instruments [40,41,42,43,44,45,46,47,48,49,50,51]. Modern lidars are widely used to study biomass-burning aerosols, allowing for sensitive detection up to altitudes of tens of kilometers, with fast response, high accuracy, and range resolution [44,45,49,50,51], taking advantage of high-power multi-wavelength laser sources and various elastic-scattering, depolarization, and Raman-receiving channels [52,53,54,55]. Standardized lidar inversion algorithms provide reliable retrievals of the vertical (height) profiles of a variety of aerosols’ optical and microphysical parameters—backscattering, extinction, lidar ratio, Ångström exponents, volume/particle linear depolarization ratio, etc. [54,56,57,58,59,60,61]. The above-mentioned capabilities of the lidars also allow for obtaining, visualizing, and monitoring (including in real time) the dynamics of the BB aerosol density height distribution, accounting for the evolution of individual aerosol layers [62,63]. Sun/sky/lunar photometers are also employed for observations of BB aerosols [40,41,48,57]. They measure continuously the sun/moon irradiances and sky radiance at discrete wavelengths and provide automatically, in near-real time via specially developed inverse algorithms, a large set of column-integrated aerosol optical and microphysical parameters such as aerosol optical depths (AODs), the AODs of the fine and coarse particle fractions, Ångström exponents (AEs), particle volume size distribution (VSD), single-scattering albedo (SSA), real and imaginary parts of the particle complex refractive index, particle linear depolarization ratio (PLDR), particle sphericity factor (SF, the percentage of spherical particles in the observed aerosol), radiative forcing at the top and bottom of the atmosphere, etc. [64,65,66,67,68,69,70]. Contact probing and analyses also provide useful information on the microphysical, chemical, and optical properties and the type of the BB and other particulate matter of interest (e.g., [35,36,71,72,73,74,75]).

To facilitate the determination of the origins and types of detected aerosols, different modeling and forecasting web resources have been developed and used, such as air-mass transport and dispersion models [76], desert-dust spread forecasting models [77,78,79], and satellite-provided maps of the fire outbreaks worldwide [80].

Advanced inverse algorithms/codes have also been developed that combine in a synergistic way the processing of lidar and photometer data for retrieving range-resolved height profiles of mass/volume/number aerosol concentration, distinguishing between different particle shapes and size fractions. Such codes are, e.g., Lidar–Radiometer Inversion Code (LIRIC), Generalized Aerosol Retrieval from Radiometer and Lidar Combined data algorithm (GARRLiC), Polarization Lidar Photometer Networking (POLIFON), and Generalized Retrieval of Aerosol and Surface Properties (GRASP) [81,82,83,84].

Remote sensing observations and characterization of BB aerosols are being performed in cases of detecting both fresh smoke from fires occurring near the measurement sites [57,85,86] and aged smoke resulting from long-range transport from distant fire outbreaks [48,49,87,88,89,90]. Massive seasonal fires in the territories of the USA and Canada, for instance, can raise huge amounts of BB aerosols into the atmosphere, which are often transported by the air circulation system of the Northern Hemisphere to Europe [91,92,93,94,95] and, in particular, to the Balkan Peninsula [96,97].

On the basis of different remote sensing and in situ measurements, characteristic ranges of values have been outlined for some BB aerosol parameters of interest. According to sun-photometer and contact-sampling data, the fresh (minutes-to-hours old) smoke is characterized by a dominating optical influence of the fine submicron aerosol fraction, a high aerosol optical depth AOD₄₄₀ > ~0.3 at the wavelength of λ = 440 nm, a high Ångström exponent AE_440/870 > 1.5–1.7 for the wavelength pair 440/870 nm, a high particle sphericity factor SF > ~80–90%, and a low particle linear depolarization ratio PLDR₄₄₀ ~ 0.002–0.005 at λ = 440 nm [16,98,99]; the real n_r and imaginary n_i parts of the particle refractive index for λ from 440 nm to 1020 nm are ~1.41–1.59 and 0.0005–0.0256, respectively [71,98,100]. In the cases of aged aerosols of this type, the AE_440/870 may fall down to values around unity [100]. The particle sphericity should then remain high with the lidar-determined PLDR not exceeding several percent in the visible range (see below and [45,51,95,101,102,103]). However, some results were recently reported [44,45,90,94,95,101,104], indicating an increase in the PLDRs with the height reaching, e.g., values of ~20–30% in cases of long-range transported Canadian and Australian aged BB aerosol layers at altitudes of 15–20 km. Such an effect has usually been attributed to the presence at high altitudes of uncoated or partially coated irregular soot aggregates. It is also not excluded that the effect is partly due to icing, as, e.g., in the case of aged BB aerosols in the stratosphere or of smoke or volcanic dust entering the lower or upper troposphere [104,105,106,107]. A similar decrease in the Ångström exponent and particle sphericity and an increase in PLDR could also be caused by admixtures of (desert or continental) mineral dust and biogenic and marine aerosols [99,102,104,105]. Let us further note that the literature data about the lidar-determinable values of the extinction- and backscatter-related Ångström exponents (EAE and BAE) and volume linear depolarization ratio (VLDR) for BB smoke (or mixtures with predominant BB aerosols) in the troposphere vary widely depending on different factors such as fire location, burning materials, combustion type, stage of aging of the smoke particles, etc. (see [51,92] and the references therein). Thus, some reported EAE values of biomass burning smoke vary from 0 to 2 at the wavelength pair of 355/532 nm [42,57,87,107,108]. The BAE values vary in the range from 0.5 to 2 at the wavelength pairs of 355/532 nm and 532/1064 nm [57,101,104,107,109,110,111,112]. Some lidar-sensing-based estimates of VLDR of BB aerosol layers in the stratosphere (above 12 km) show that it is below 0.049 at 355 nm and below 0.059 at 532 nm [113]. According to other estimates, at 532 nm, the values of VLDR should most frequently be between 0.1 and 0.2 at heights of 12 to 16 km [44]. The values of the VLDR of BB aerosol layers in the troposphere have been similarly estimated (by lidar measurements) to be well below 0.1 at the wavelengths of both 355 nm and 532 nm, due to the tropospheric smoke particles’ higher sphericity [44,104].

The 2023 wildfire season in Canada was the most devastating on record, resulting in enormous emissions into the atmosphere, affecting not only the local air quality but also that in most of the Northern Hemisphere, including Europe, due to long-range transport [106,114,115,116]. It was estimated [117] that Canada’s 2023 wildfires produced carbon emissions with a magnitude of 647  TgC, i.e., five times more than Canada’s total annual emissions. The total burned area was approximately 15 million hectares, about 4% of the forest area of the country and more than six times the annual national average [118]. The Canadian wildfire season 2023 was record-breaking not only for its intensity and extent, but also for its duration—from mid-April to late October [119,120]. For instance, the area burned on 22 September 2023 was the largest total area burned in a single day ever recorded in Canada [120].

The aim of the present work is to study in detail the presence and effects of long-range transported fire-smoke layers, predominantly of Canadian origin, on the aerosol content and troposphere structure over Sofia, Bulgaria, on 3 October 2023, by using data characterizing the aerosol stratification and the aerosol optical and microphysical properties. The interest in studying this aerosol event was due to the extreme fire activity worldwide during 2023, and especially in Canada [117,121,122], including the last decade of September and the first decade of October [119,120]. Furthermore, it was interesting to detect and identify the BB aerosol layers descending from the upper troposphere down to the planetary boundary layer (PBL) over Sofia and to observe their sedimentation and dispersion [102,104,106,123,124], having, in principle, the potential of significantly and directly affecting the local ecology and human health [29,32,33].

In the following Section 2, the measurement site, instruments, and data of importance to be analyzed are briefly described and characterized. In Section 3, the aerosol layers and their evolution, the aerosol optical and microphysical characteristics, and the aerosol types are described and analyzed, as determined by vertically-resolved lidar and ceilometer and columnar sun/sky/lunar photometer measurements. The main results obtained and conclusions drawn in the work are summarized in Section 4.

2. Measurement Site, Instruments, and Data

2.1. Sofia Aerosol Remote Sensing Station

The experimental measurements were performed at the Sofia Aerosol Remote Sensing (ARS) station (42.653733°N, 23.387372°E, 610 m ASL) using a high-power aerosol lidar, an automatic ceilometer, and a sun/sky/lunar photometer. The three devices are installed on the roof of the Institute of Electronics at the Bulgarian Academy of Sciences (IE–BAS). The Sofia ARS station is contributing to the European Aerosol Research Lidar Network (EARLINET) [7], the program E-PROFILE for surface-based profile observations by European Meteorological Services Network (EUMETNET) [125], the AERONET (Aerosol Robotic Network) program established by NASA and PHOTONS (Photométrie pour le Traitement Opérationnel de Normalisation Satellitaire, University of Lille 1, CNES, and CNRS-INSU) [9,126]. Furthermore, the station is an observational National Facility for aerosol remote sensing in the Pan-European research infrastructure ACTRIS (The Aerosol, Clouds and Trace Gases Research Infrastructure) [10,127].

The measurement site is situated in the south-eastern part of the city of Sofia (Figure 1); the latter is located in a heavily urbanized mountain valley characterized by a continental climate with an average annual temperature of 10 °C and mainly westerly and easterly winds with an average annual speed of 2.4 m s⁻¹ [128].

Because of Sofia’s specific location at the center of the Balkan Peninsula, the different aerosol types reaching the city [96,99,129,130,131,132,133,134,135] originate from global zones comprising powerful aerosol sources. Such zones, for instance, are North Africa and Near East with Sahara and Arabian deserts emitting desert dust; Equatorial and South Africa and North America with wildfire-covered areas emitting BB aerosols to be aged on their long path to Bulgaria; burning areas in Sofia or its vicinity, in Bulgaria or neighboring countries releasing fresh BB aerosols; densely-populated urbanized European regions emitting urban/industrial aerosols; and European roads, fields and woods emitting continental aerosols. In particular, it has been recently shown that during the 2020–2022 period, the relative appearance of BB aerosol events has been about 9% [99].

2.2. High-Power Aerosol Lidar

A model LR332-D3008 eight-channel depolarization and Raman lidar, developed by Raymetrics S.A. (Athens, Greece), in compliance with the optimum ACTRΙS requirements for Aerosol High Power LIDARs [136], is utilized for measuring and height-profiling the atmospheric aerosols’ optical and microphysical parameters with a height limit of 60 km and a range resolution of 3.75 m. The lidar transmitter is a high-power Nd:YAG laser (Quantel CFR 400 Series) with a pulse repetition rate of 20 Hz. It emits single-beam output radiation at three wavelengths (1064/532/355 nm) in short pulses of 11.5/5.80/6.20 ns duration and output pulse energies of 120.9/81.2/91.1 mJ, respectively. A beam expander (3×) reduces the laser beam divergence at the above three laser wavelengths to values of 0.86/0.86/0.78 mrad. The lidar contains three aerosol channels for recording elastically scattered signals (at 1064, 532, and 355 nm), two depolarization channels for parallel and cross polarization (at 355 and 532 nm), and three Raman channels (at 387, 408, and 607 nm). The latter are used for nighttime measurements only. The 1064 nm lidar channel operates only in analog mode, while all other channels operate in both analog and photon counting modes. The optical receiver of the lidar is a Cassegrain telescope with primary/secondary mirror diameters of 300/64 mm, focal length of 1500 mm, default field of view (FOV) of 2.3 mrad (adjustable from 0.5 to 3 mrad), and laser beam—telescope FOV full overlapping range of about 300 m. The lidar’s wavelength separation unit is equipped with narrow-passband interference filters providing high transmission for the operating wavelengths at each channel and optical density in the background blocking range greater than six. The 1064 nm channel employs an avalanche photodiode as a photo-detector, whereas the other channels make use of Hamamatsu type R9880U-113 photomultipliers. The acquisition system of the lidar is based on Licel transient recorders TP-40-16 bit equipped with an analogue-to-digital converter for the analog mode. It samples and digitizes the lidar signals at a sampling rate of 40 MHz. In photon counting mode, a 250 MHz fast discriminator provides detection of single photon events above the selected threshold voltage. The measurement data quality is ensured through the EARLINET/ACTRIS quality assurance/quality control procedures [137]. The lidar signals are analyzed to retrieve the aerosol backscatter and/or extinction coefficients height profiles by the Klett–Fernald inversion algorithm [59,60] for daytime measurements and the Raman inversion method [56] for the nighttime ones using the unified EARLINET data processing system Single Calculus Chain [138,139]. The calibrating molecular profiles are retrieved using radiosonde data obtained at a site near Sofia ARS station in parallel with the lidar measurements [140].

2.3. Automatic Ceilometer

The CHM15k “NIMBUS” ceilometer (Lufft, an OTT HydroMet brand, Fellbach, Germany) is a low-power lidar designed for unattended continuous automatic operation [141]. The device emits vertically into the atmosphere a series of short light pulses generated by a diode-pumped Nd:YAG solid-state laser at the wavelength of 1064 nm (pulse duration of 1 ns; repetition rate of 5–7 kHz; mean output power of 59.5 mW, with measuring range of up to 15 km; controllable vertical resolution of 5 to 30 m). The backscattered-light detection system operates in a photon-counting mode. Ceilometers were originally designed to provide the cloud base height. However, the sensitivity of the current devices is sufficient to provide useful information on the aerosol vertical distribution. The aerosol backscattering coefficient can also be retrieved from the ceilometer measurements by using the basic Klett–Fernald inversion algorithm and careful calibration [142,143].

2.4. Sun/Sky/Lunar Photometer

The sun/sky/lunar photometer CE318—TS9 (Cimel Electronique, Paris, France) [64,144] has eight optical channels for aerosol measurements at the central wavelengths λ = 340, 380, 440, 500, 675, 870, 1020, and 1640 nm and one channel for water vapor measurements (λ = 937 nm). The device field of view is 1.3°. Direct measurements of the sun (at all the available wavelengths) and moon (at λ ≥ 440 nm) irradiances are carried out every 3 min. Sky radiances are measured, in general, every hour at several wavelengths in almucantar and hybrid scenarios. A sufficient amount of accurate raw data for reliable and precise retrieval of the aerosol characteristics of interest is achieved by a suitable predefined scan configuration and measurement schedule. These characteristics are obtained automatically by using direct measurements and specially-developed AERONET data-processing algorithms, such as a spectral deconvolution algorithm (SDA) and an inversion algorithm [65,66,67,68,69]. The AOD uncertainty is estimated at less than 0.01 for wavelengths > 440 nm and less than 0.02 for those <440 nm. The relative uncertainty in measuring sky radiance is assumed to be about 5% [126,145,146]. The uncertainties in the inversion parameters are discussed in [67,146].

The lidar and ceilometer data considered in this work include colormaps of the total vertical attenuated backscatter (at λ = 1064 nm) and height profiles of the aerosol backscatter coefficient (BSC; at λ = 355, 532 and 1064 nm), the backscatter-related Ångström exponents (at wavelength pairs 355/532 nm and 532/1064 nm), and the volume linear depolarization ratio (at 355 and 532 nm) retrieved from the lidar data.

The aerosol characteristics provided by the photometer and considered here are AOD₄₄₀; AE_440/870; AODs of the fine and coarse aerosol fractions at λ = 500 nm (AOD_f500 and AOD_c500, respectively); the particle SSA, PLDR₄₄₀, n_r, and n_i; the particle volume size distribution (VSD), the corresponding inflection radius R_i, effective radii (total, R_effT, of the fine fraction, R_effF, and of the coarse fraction, R_effC), and SF.

2.5. Model, Satellite, and Meteorological Data

Complementary data were also used in order to assist us in determining the origins and identifying the types of aerosol inclusions in the atmosphere above Sofia. Thus, data on fire outbreaks worldwide are provided by the Fire Information for Resource Management System (FIRMS, 2.0) maps. FIRMS is a part of NASA’s Land Atmosphere Near-real-time Capability for Earth observation (LANCE) [80] and uses satellite observations/data from the MODIS and VIIRS instruments. The backward trajectories of the air masses arriving above the measurement site at given height and time are determined by the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT, v5.3.0) transport and dispersion model developed in the Air Resource Laboratory of the National Oceanic and Atmospheric Administration (NOAA) and run interactively on the Real-time Environmental Applications and Display system (READY) website [76,147]. The information used in this work about the desert-dust spread is provided by the SKIRON Dust modeling system [77,148] of the Atmospheric Modeling and Weather Forecasting Group of the Department of Physics of the National and Kapodistrian University of Athens (NKUA) and by the Multiscale Online Nonhydrostatic AtmospheRe Chemistry model (MONARCH, v2.7.2) [78,79,149], developed at the Barcelona Supercomputing Center and accessible at the World Meteorological Organization (WMO) Barcelona Dust Regional Center website. The radiosonde data are provided by the WMO Station 15614 of the Bulgarian National Institute of Meteorology and Hydrology, located about 300 m away from the Sofia ARS station. The data are freely available at the University of Wyoming Atmospheric Science Radiosonde Archive [140]. Synoptic analyses also utilize data from the NCEP/NCAR (National Centers for Environmental Prediction/National Center for Atmospheric Research) Reanalysis Project [150,151], provided by the NOAA Physical Sciences Laboratory [152].

3. Results

In this section, the experimental results are presented of lidar, ceilometer, and sun photometer observations of long-range transported Canadian smoke layers above Sofia on 3 October 2023, along with modeling and forecasting data. Hereinafter, the cited values of height are as measured above ground level (AGL), whereas the time and time intervals are quoted in Universal Time Coordinate (UTC).

3.1. Aerosol Vertical Distribution and Dynamics

The lidar and ceilometer colormaps describing the evolution of the vertical (height) distribution of the total attenuated backscatter from 07:15 to 15:30 on 3 October and from 00:00 on 3 October to 00:00 on 4 October 2023 are shown in Figure 2a,b, respectively. The time resolution of the lidar data is 1 min, and that of the ceilometer data is 15 s. The colored diagrams reveal, in general, a multilayered aerosol stratification over Sofia with aerosol layers observed in the troposphere at altitudes reaching up to 5 km. Throughout the day, there is a tendency for the upper aerosol zone boundary to descend to 3.5–4 km by the end of lidar measurements and further to nearly 2 km by the end of the day. At the beginning of the lidar measurements, three aerosol layers can be distinguished above the PBL top (located at about 1.75 km), respectively numbered in Figure 2a. The clearly discernible aerosol layers 1 and 3 are positioned at heights of 4.7–4.9 km and 2.2–2.7 km. The less perceptible lidar-detected layer 2, consisting of a veil-like set of multiple fine sub-layers, can also be seen at heights about 3.6–4 km. In the period between 08:30 and 11:30, a thin transient aerosol layer (layer 4) is also observed at a height of about 1.8 km, afterward mixing with the rising PBL.

The lidar aerosol BSC vertical profiles obtained at wavelengths 355, 532, and 1064 nm averaged over one hour for the time intervals 07:15–08:15, 09:20–10:20, 10:20–11:20, and 13:30–14:30 are represented, respectively, in Figure 3a–d. The profiles are retrieved using a constant lidar ratio of 50 sr, with calibration heights falling in the range of 7–12 km. They reveal more clearly the time-averaged distribution of the aerosol layers in the troposphere above Sofia.

3.2. Aerosol Transport Tracking and Mapping

The NOAA HYSPLIT air-mass transport model [76,147] is used to determine the origin of the registered aerosol layers. Figure 4a shows representative long-range (240 h) air-mass backward trajectories starting on 24 September 2023 and ending above Sofia at the height range of the detected aerosol layers above the PBL top at 08:00 on 3 October 2023. They begin mainly from Canada and pass over regions with intensive fire outbreaks. As mentioned in Section 1, the year 2023 is characterized by unprecedented record-breaking fires in Canada from April to October (with a peak of the total burned area on 22 September) that could have seriously affected the atmospheric aerosol situations and air quality in many countries of the Northern Hemisphere, including Bulgaria. As an illustration, Figure 4c shows the map of the active fires in a selected part of the Northern Hemisphere for the period 21–27 September 2023, provided by FIRMS-NASA and based on detections from the MODIS and VIIRS sensors [80].

Thus, the HYSPLIT backward trajectories presented in Figure 4a indicate that the aerosol layers registered by lidar at heights of about 2 to 5 km in the troposphere above Sofia on 3 October 2023 are mainly smoke layers originating from Canadian forest fires. The trajectories in Figure 4a have also long, high, and distant sections over the Atlantic Ocean, with heights ~4–7 km, and shorter, lower, and closer sections, with heights ~2.5–5.5 km, over Central Europe. Therefore, one may expect small admixtures of urban/industrial and continental aerosols rather than of marine aerosols in the incoming aged Canadian BB aerosols. At the same time, the back trajectories ending at heights of 1 km and 1.5 km begin and pass over Western, Central, and Eastern Europe low above the Earth’s surface, almost crawling along it (Figure 4b); the one ending at 0.5 km height begins and passes (3–6 km) high above the Atlantic Ocean, crosses the European atmosphere from west to east and is again crawling along the earth’s surface near Bulgaria and Sofia (Figure 4b). Thus, at heights up to 1.5–2 km above the Sofia ARS Station, the presence of admixtures of urban/industrial and continental aerosols, and aged and fresh BB aerosols from the fires north of the Black Sea (Figure 4b,c) is possible.

3.3. Evolution of the Registered Biomass-Burning Layers

As seen in Figure 2a and Figure 3a,b, after the start of the lidar measurement at 07:15, the highest layer 1 is descending, further stratified and widened with time likely due to diffusion and/or different settling rates of the particle size fractions; further, its upper part gradually fades and vanishes, whereas its lower part closely approaches the layer underneath. Layer 2, at heights about 3.6–4 km, does not seem to be descending until about 11:15, but is rather subject to intense diffusion leading to its widening. Around 11:15, it begins descending in parallel with the bottom of the dispersed and widened highest layer (Figure 2a). As a result of the mixing of layers 1 and 2 (Figure 2a and Figure 3d), several relatively high-contrast layers arise at heights between 3 km and 3.5 km. Further, as shown in Figure 2b, during the last third of the day, these layers settle at the same rate as the most contrasting layer 3. The fourth detected layer is practically glued to the PBL and evolves along with it. Let us also note that the BSC values of the lowest aerosol layers 3 and 4 and the PBL increase with time, likely as a result of particulate matter settling from the space above them, of such rising from the earth’s surface, and of horizontal inflow of aerosols (Figure 3a–d).

The well-pronounced downward movement of smoke-containing aerosol layers over Europe from the upper part of the troposphere (heights 6–7 km) down to the PBL is a significant, distinctive feature of the BB aerosol transport from Canadian fires in the period 30 September–3 October 2023. Above the measurement site, this is clearly visible in Figure 2a,b and is confirmed by the aerosol stratification pictures based on E-profile data from a number of ceilometer stations across Europe [125]. Another important feature to note here is the downward trend during the above-mentioned period (Figure 4a) in the backward trajectories of air masses traveling from Canada to Bulgaria and Sofia. As is clear from the aerosol dynamics colormap (Figure 2a), as well as from the BSC profiles (Figure 3), as a result of the sedimentation of smoke aerosols, the latter accumulate in the dense aerosol layer 3 confined within the height interval 2.2–2.7 km and centered at a height of about 2.3 km, as well as in layer 4 and PBL. The sedimentation process of smoke aerosols limited from below leads to a sharp contrast between the very high density of the bottom layer 3 and that of the overlying looser and transitional smoke layers. This is also reflected in the measured BSC values for these layers. The maximum values of the BSC for the dense smoke layer 3 are in the range of 0.6–1.4 Mm⁻¹sr⁻¹ for 1064 nm, 2.2–3.4 Mm⁻¹sr⁻¹ for 532 nm, and 5–8 Mm⁻¹sr⁻¹ for 355 nm. These values are 8 to 15 times greater than the maximum values of the BSC at the three wavelengths for the weaker smoke layers. They are comparable to the maximum values of the BSC in the PBL for 1064 nm and to some extent for 532 nm, while for 355 nm, they are two to three times greater than their respective values for the PBL. To explain such a spectral behavior of BSC, we may suppose that layer 3 consists mainly of relatively fine submicron particles, while the PBL content is dominated by coarse (supra-micron-sized) particles. At relatively large wavelengths (1064 nm), exceeding the submicron sizes, but smaller than the over-micron sizes, the (back)scattering mechanism of the fine particles (nearer the Rayleigh one [153]) will be much weaker than that of the coarse particles, according to the Mie theory [153]. Then, despite the much higher particle concentration supposed in layer 3, the lidar return from it may not exceed that from the PBL. At relatively small wavelengths (355 nm), comparable with the submicron sizes and much less than the micron sizes, the (back)scattering mechanisms from both the above-mentioned layers would be similar [153]. Then, the several times stronger lidar return from layer 3 actually reveals its higher saturation with particulate matter.

3.4. Meteorological Situation

Figure 5a presents meteorological radiosonde profiles of the temperature (T), dew point (DP), and relative humidity (RH) measured in Sofia at 12:00 UTC on 3 October 2023. Two significant features can be distinguished in the radiosonde profiles in the height interval occupied by the densest smoke layer (at about 2–3 km) and slightly above it. First, a temperature quasi-inversion is present in the height range 1.2–3 km of the temperature profile (marked in Figure 5a by black open circles). Temperature inversions are also present near the local PBL top, according to radiosonde data from the previous period, which ran from 30 September to 2 October 2023 [140]. Second, the relative humidity drops sharply as one moves up the quasi-inversion, from roughly 80% at the lower end (at 1.5 km) to about 5% at and above the upper end (at 3–3.8 km). Respectively, within these intervals, the DP profile moves away from the T profile by a significant 40 °C.

We associate the accumulation of Canadian smoke observed in the densest layer above the top of the PBL and slightly in the layers underneath with the barrier effect of the temperature quasi-inversion, which prevents the further downward motion of precipitating aerosols. On the other hand, the aforementioned significant drop in the relative humidity suggests that the smoke aerosols in the dense layer and those directly above it are extremely dry.

Figure 5b presents NCEP/NCAR reanalysis diagrams at the 700 mb level of the relative humidity composite mean for the Northern Hemisphere in the period 25–27 September 2023—the days during which, according to the backward trajectories in Figure 4a, air masses passing over Canada could have mainly captured smoke aerosols from the forest fires raging there. As can be seen from the map, during the period in question, most of the Canadian territory, in particular its central and eastern parts, was occupied by a zone of relatively low humidity, which in the southeastern region was around and below 20%. This area of low humidity is fed by a vast synoptic-scale dry zone located at subtropical latitudes over the Pacific Ocean (visible whale-shaped in the lower left quadrant of the map). Over the western parts of the United States, the dry zone rises to temperate latitudes, passing over central and Eastern Canada, extending further over the Atlantic towards the Mediterranean and Europe. Thus, the backward trajectories depicted above pass through primarily moistureless atmospheric domains practically along their entire path. Therefore, smoke aerosols, being naturally warm and dry over the fire zones in Canada, were likely to remain dry throughout the long-range transport to Europe, and in particular to Bulgaria, as confirmed by the RH profile in Figure 5a.

Figure 5c displays NCEP/NCAR reanalysis diagrams at the 700 mb level of the air-temperature composite anomaly for the region of Europe, the Mediterranean, and North Africa on 3 October 2023. The core of the positive temperature anomaly moves eastward in the period 30 September–3 October. During this shift, Bulgaria, being in the core of the negative T-anomaly on 30 September, passes into the zone of the positive one by 3 October, as can be seen on the diagram in Figure 5c. This is a non-equilibrium transient process of temperature changes, occurring at different speeds at different heights, which, in our opinion, creates the prerequisites for the temperature inversions observed over the considered period in the troposphere above Sofia, as shown in Figure 5a.

3.5. Driving Mechanism of the Large-Scale Spread of the Biomass-Burning Aerosol

An essential question concerning the clarification of the nature and features of the transatlantic transport of smoke aerosols from Canada to Europe considered here is identifying the meteorological driving mechanism of this transport. To this end, Figure 6 presents NCEP/NCAR reanalysis composite diagrams at the 700 mb level of meteorological parameters relevant to the transport, such as winds, geopotential height, and air temperature, for the geographic regions and time periods of interest.

Figure 6a shows the NCEP/NCAR reanalysis diagram at the 700 mb level of the vector wind composite mean for a part of the Northern Hemisphere (20–75°N, 80°W–40°E) in the period 1–3 October 2023. From Eastern Canada to the right across the diagram to Northern Europe (60°W–40°E), in the temperate range of latitudes 45–55°N, a laminar zonal air flow with vector speeds in the range of 10–20 m s⁻¹ is clearly visible, with a high probability of carrying fire smoke aerosols trapped over the territory of Canada. The NCEP/NCAR reanalysis diagrams of the air temperature composite mean indicate that the zonal flow carries cold air with temperatures in the range of 270–275 K. Over the British Isles and Northern Europe, the zonal flow merges with a southwest–northeastward extended closed elliptical circulation configuration (20–55°N, 30°W–30°E) with negative (clockwise) vorticity. The latter is caused by the synoptic blocking pattern (stationary ridge) shown on the diagram of the geopotential height composite mean in Figure 6b. It is centered over the Western Mediterranean and covers parts of the Eastern Atlantic, Northwest Africa, and practically all of Europe. In the area of confluence of the zonal flow and the circular circulation, the aerosol contents of the two systems mix, which most likely leads to the capture of Canadian smoke aerosols from the zonal flow and their inclusion in the circular one. We believe that this is the mechanism by which smoke aerosols spread over Europe, including Bulgaria.

The interaction of low-level (i.e., at altitudes below 550 mb) cold air advection and negative vorticity advection systems creates favorable conditions for the occurrence of downward motion of the air masses [154]. The theoretical description of a vertical motion in the atmosphere used in numerical weather models is based on the so-called omega equation [154,155]. Omega is the vertical velocity of air masses in pressure coordinates (ω = dp/dt) and is expressed in units of pressure per time (Pa s⁻¹). Since the atmospheric pressure decreases with height, upward convection is described by negative omega values, while downward motion is described by positive ones.

Figure 6c displays NCEP/NCAR reanalysis diagrams at the 700 mb level of the composite mean of the pressure vertical velocity omega over Europe on 3 October 2023. As can be seen, on the day of lidar measurements, Bulgaria, including Sofia, falls into the core of the most intense descending motion of the air masses (red areas), with record daily omega values for the depicted geographical window of about 0.15 Pa s⁻¹.

In the period 30 September–3 October 2023, the core of the stationary ridge slowly moved eastward. Along with it, the accompanying meteorological processes, including the zones of descending air masses over the Mediterranean and Europe, moved eastward. Thus, in the range of latitudes 30–50°N, these zones, being over Western Europe (longitudes 5°W–5°E) on 30 September, over the next two days crossed the central regions of the continent and the Mediterranean and by 3 October reached the pattern depicted in Figure 6c.

Thus, the observed downward-motion processes of the smoke-containing air masses taking place over all of Europe in late September and early October 2023 find their natural explanation arising from the analysis of the synoptic-scale meteorological processes described above.

3.6. Lidar Profiling and Analysis of the Biomass-Burning Aerosol Optical Properties

Figure 7a–d presents height profiles of the backscatter-related Ångström exponent for the two wavelength pairs 355/532 nm (BAE_355/532) and 532/1064 nm (BAE_532/1064), based on the BSC profiles shown in Figure 3a–d. As is known, BAE is a microphysical aerosol characteristic giving a qualitative assessment of the dominant aerosol size fractions in inverse proportion [156]. A 10-point smoothing of the profiles was applied in order to compensate for electronic noise oscillations and better visualize the essential profile structure. In areas around and below 2 km, the BAE_532/1064 profiles have discontinuities due to the low signal-to-noise ratio for the BSC profiles at both wavelengths, as can be seen in Figure 3a–c. As a general feature, the BAE profiles in Figure 7 show a two-step structure with lower flat sections in the PBL zone below 1.5–2 km and higher ones above 2 km, where the smoke aerosol layers are located. The BAE values for the former are in the range 0.5–1, indicating larger aerosol fractions around 1 μm, while for the latter, including the dense smoke layer at 2–3 km, they are in the range 1.5–2, indicating finer aerosol modes in the submicron size range. Such BAE values for the smoke layers are in good qualitative agreement with data for smoke aerosols from Canada from other groups of EARLINET stations in Europe [51,97]. Based on the relatively high BAE values, it can be assumed that the smoke aerosols from Canada have not aggregated to a significant extent during their transport. This is probably also influenced by the fact that Canadian smoke aerosols are predominantly dry. This is evidenced by the radiosonde profile of the relative humidity in Figure 5a, which in the area of the dense smoke layer at 2–3 km shows a strong drop in the relative humidity to values of about 10%, and even to 5% immediately above. In the area of upper smoke layers in the range of 4–5 km, a drop in the relative humidity is also observed, down to 10% in the middle of the range.

While BAE gives an idea of the predominant size fractions of aerosols, the shape of the particles can be estimated from their depolarization properties. Figure 8a,b presents the height profiles of the volume linear depolarization ratio at the two wavelengths, λ = 355 and 532 nm, as obtained by the lidar depolarization channels during the hours of interest on 3 October 2023. The profiles at the two wavelengths show similar structure and dynamics in clear correspondence with the presence and evolution of the aerosol/smoke layers delineated in Figure 2a and Figure 3a–d. The VLDR values obtained generally fall between 0.02 and 0.1, which can be linked to particles that have a small to moderate non-sphericity. In the boundary layer up to 1.5 km, the VLDR profiles at 532 nm show nearly twice as high values and a steeper decrease with height, compared to those at 355 nm. Given the high values of the atmospheric relative humidity in this height interval, reaching 80% at 1.5 km at a temperature quite near the dew point, this decrease can be attributed to the contribution of water droplets or hydrated aerosol particles that have a higher degree of sphericity. At heights above 1.5 km, a local increase in VLDR values is observed for all distinct smoke layers 1–4, proportional to the density of the layers and following their topological evolution. Thus, in the case of the particle-accumulating layer 3 occupying the heights between 2 and 3 km, which has a high, weakly changing density over time, the VLDR at 355 nm and 532 nm reach values of 0.06–0.08, remaining practically unchanged during the measurements.

The highest layer 1, centered in the initial period of the measurements at a height of about 4.8 km, is clearly present in Figure 8a,b (the blue profile for 07:15–08:20) with a well-pronounced VLDR peak with relatively high values of 0.042 at 355 nm and 0.073 at 532 nm. The sections of the VLDR profiles corresponding to the following measurement intervals presented (09:20–10:20, 10:20–11:20 and 13:30–14:30) reflect the downward movement of the smoke aerosols and the formation of transitional layers at heights from 4.5 km to 3 km with fluidly changing shapes and values about 0.035–0.04 at 355 nm and 0.04–0.05 at 532 nm. Layer 4 (at a height of about 1.8 km immediately below the dense layer 3) shows VLDR values close to those of layers 1 and 2. This is also an indication of the possibility that the layer comprises a smoke aerosol of Canadian origin along with the corresponding backward trajectory shown in Figure 3a.

We associate the proportionality between the density of the smoke layers and the corresponding VLDR values with the increase in the relative contribution of the aerosol component compared to the molecular one in denser layers, and with that of the relative fraction of particles with a lower degree of sphericity.

The lidar and model data and results presented above allow one to infer that the Canadian smoke layers recorded on 3 October 2023 in the 2–5 km altitude range were mainly made up of dry, partially aggregated smoke particles that belong predominantly to submicron size fractions and have a rather low degree of non-sphericity.

3.7. AERONET Sun-Photometer-Derived Columnar Aerosol Characteristics

The AERONET data (level 1.5, version 3.0 [145,146]) are presented in

Table 1. AERONET data based on sun/sky/lunar photometer measurement of the optical and microphysical characteristics of the aerosol ensembles over the city of Sofia on 3 October 2023: daily mean values of AOD₄₄₀, AOD_f500, AOD_c500, and AE_440/870, prevalent ranges of variation during the day of SF and PLDR₄₄₀, and ranges of variation during the day of n_r and n_i, R_i, R_effT, R_effF, and R_effC.

Parameter	Value	Parameter	Value
AOD440 ± SD	0.28 ± 0.03	nr (440 nm ≤ λ ≤ 1020 nm)	1.4–1.6
AODf500 ± SD	0.23 ± 0.02	ni (440 nm ≤ λ ≤ 1020 nm)	0.004–0.02
AODc500 ± SD	0.02 ± 0.01	Ri [μm]	0.76–0.99
AE440/870 ± SD	1.22 ± 0.04	ReffT [μm]	0.28–0.56
SF [%]	91.9–99.0 *	ReffF [μm]	0.14–0.24
PLDR440	0.002–0.009 *	ReffC [μm]	2.52–3.34

* Refer to the text and Figure 9d for more detail.

. They outline a relatively high daily-mean AOD₄₄₀ = 0.28 ± 0.03; a moderate daily-mean AE_440/870 = 1.22 ± 0.04; a daily-mean AOD_f500 = 0.23 ± 0.02 and AOD_c500 = 0.02 ± 0.01 (Figure 9a); comparable fine and coarse fraction modes of the particle VSDs throughout the day (Figure 9b) with R_effT, R_effF, and R_effC of 0.28–0.56 µm, 0.14–0.24 µm and 2.52–3.34 µm, respectively, and R_i of 0.76–0.99 µm; and a complicated increasing–decreasing behavior of the SSA with λ indicating a rather noticeable aerosol absorption of radiation (Figure 9c). In 73% of the daily measurements, the PLDR₄₄₀ ~ 0.009–0.002 corresponds to the high SF ~92–99% (see also Figure 9d), and n_r and n_i for 440 nm ≤ λ ≤ 1020 nm vary between 1.4 and 1.6 and 0.004 and 0.02, respectively (Figure 9e). As seen in Table 1 and Figure 8, the AERONET-provided PLDR₄₄₀ are about one order of magnitude smaller than the VLDRs obtained by lidar at λ = 355 nm and 532 nm. At the same time, the results regarding PLDR/VLDR of BB aerosols obtained here are similar to the results reported by other authors, both in the case of AERONET data [157] and in the case of lidar data [44,104]. The difference existing between the lidar and AERONET results may be due to various reasons. Among them are the following: the lidar is measuring VLDR, while the AERONET algorithms retrieve PLDR; the operational wavelengths are different for the two devices; and the sun photometer provides columnar (height-integrated over the whole atmosphere) data, while the range-resolved lidar data cover a limited height range (lower half of the troposphere).

The optical and microphysical aerosol characteristics listed above concerning 3 October 2023 outline, in general, a mixed aerosol situation over Sofia [16], where, at a moderate value of AE_440/870 and strongly prevailing optical impact of the fine particle fraction (AOD_f500 >> AOD_c500), the coarse-particle mode peaks are a little higher, on average, than the accumulation-particle mode peaks (Figure 9b, [100]). Such a conclusion is in agreement with the lidar-measurement-based results for BAE_355/532 and BAE_532/1064 shown in Figure 7, where they are seen to vary from 1, mainly at heights up to 1.5 km, to 1.5–2.0 at higher altitudes. That is, the coarse particles occupy mainly the heights up to 1.5 km, while the fine particles prevail at heights above 1.5–2.0 km. Correspondingly, the SSA vs. λ exhibits a complicated increasing–decreasing behavior (Figure 9c). Let us further discuss the increased values of R_effT, R_effF, and R_i and the generally high values of SF and low values of PLDR₄₄₀ (Figure 9d). Bearing in mind the high, as a whole, sphericity of the aerosol ensembles during that day, one may conclude that the aerosol situation was characterized by prevailing aged BB smoke particles accompanied by urban/industrial aerosols [16,98,99,100] and some amounts of fresh BB aerosols from close local fires. This is confirmed by the backward trajectories of the air masses arriving over Sofia from Canada, at heights above 2 km, and from closer and distant fire outbreaks north of the Black Sea, at heights below 2 km (Figure 4a–c). According to the analysis performed above of the weather conditions illustrated in Figure 5 and Figure 6, the aging process of the smoke particles arriving from Canada should not have been so intensive because they have traveled through a mostly dry environment. At the same time, the backward trajectories from distant fire regions north of the Black Sea have not passed through extremely dry atmospheric zones and could have undergone more intensive aging. This is likely one of the reasons why the values of BAE at heights below 1.5 km are, in general, lower than the ones at heights above 1.5–2.0 km. The process of aging leads to enlargement of the fine particles and, correspondingly, to higher R_effT, R_effF, and R_i and lower BAE and AE_440/870 [100]_. The columnar photometric value of AE_440/870 = 1.22 provided by AERONET is in good agreement with the prevailing lidar-measurement-based BAE values of ~1–1.5. As it seems, the carbonaceous particles in the case considered have retained the usually intrinsic to them relatively high degree of sphericity during their transport to Sofia. This feature is confirmed by both the AERONET (Table 1, Figure 9d) and the lidar (Figure 8) data. Let us note, in addition, that the nearest end sections of the back trajectories ending over Sofia, especially those from Eastern Europe, pass entirely over the continent and low above the Earth’s surface (Figure 4a,b). So, in addition to BB aerosols, they could carry some amounts of continental aerosols. These aerosols, however, would further increase the coarse particle content and reduce AE and BAE, the degree of sphericity (lower SF and higher PLDR and VLDR), and the light absorption ability of the aerosol ensembles [16,98]. Such are also the effects of Saharan dust, some traces of which could be present over Sofia according to the forecasts of the models SKIRON and MONARCH.

Based on daily photometric observations, one can recognize the cases when coarse, non-spherical, and weakly absorbing aerosol particles pass over the Sofia ARS Station. Indeed, as seen in Figure 9d, for instance, a short-term passage occurs around 10:26 of non-spherical particles with a relatively low radiation absorbance; the graphs of the corresponding SSA and n_i vs. λ in Figure 9c,e are the highest and the lowest, respectively. According to the MONARCH forecast, the Saharan dust load at that time was ~10 mg m⁻². A somewhat decreased particle sphericity was observed as well at 08:26 and 09:26 (Figure 9d), which may be due to passages of continental aerosols with moderate radiation absorbance (see Figure 9c,e and also [16,98]). Compared to the cases of fresh BB aerosols and urban aerosols, the reduction in AE_440/870 down to 1.22, the increase of the VSD effective and inflection radii, and the complicated behavior of SSA with λ should be due to the presence of aged smoke and continental aerosol particles and, possibly, small amounts of desert dust [99,100].

The AERONET data on the aerosol radiative forcing over Sofia on 3 October 2023 are illustrated in Figure 9f, where we present the daily evolution of the radiative forcing at the Top Of the Atmosphere (TOA), ΔF_TOA, at the Bottom Of the Atmosphere (BOA), ΔF_BOA, and the atmospheric radiative forcing ΔF_Atm = ΔF_TOA − ΔF_BOA (e.g., [158,159]). As is seen, throughout the day, the radiative forcing at TOA and BOA are negative, but the overall atmospheric radiative forcing is positive, which implies warming of the atmosphere. The average daily radiative forcing is −16.11 ± 2.07 Wm⁻² for the TOA, and −37.84 ± 5.89 Wm⁻² for the BOA, with a resulting atmospheric forcing of 21.72 ± 4.81 Wm⁻². Close values for biomass burning aerosols have been reported for other sites across the world: ΔF_BOA ~ −40 Wm⁻² and ΔF_TOA ~ −20 Wm⁻² [159], and ΔF_Atm ~ 21.6 and 25.7 Wm⁻² [160].

4. Conclusions

The primary conclusions derived from this study and the results presented in this work are summarized below.

According to lidar and ceilometer data, it is found that the registered aerosol layers distributed within the altitude range 2–5 km above Sofia on the day of measurement (3 October 2023) represent smoke aerosols originating from Canadian wildfires (convincingly proven by HYSPLIT backward trajectories and confirmed by the obtained lidar, photometric, and model data). The conducted detailed analysis of aerosol stratification and its dynamics reveals a specific picture of the distribution and evolution of the smoke aerosol density during the observation period. The main characteristic of the latter is the downward movement (sedimentation) of smoke aerosols from the middle of the troposphere (about 5 km) down to the local PBL top (1.5–2 km), where they accumulate in a dense aerosol layer. This turns out to be a characteristic feature of the considered smoke transport over the whole of Europe in the period 30 September–3 October, according to data from a number of European ceilometer stations. In the sedimentation height range 2.5–5 km, the smoke aerosols are distributed in a pattern of diffusing and veil-like fields and local filaments of low density, emphasizing the transient nature of the aerosol precipitation process.

Based on the presented radiosonde profiles of meteorological parameters for the day of measurement, it can be concluded that the accumulation of smoke aerosols in the densest layer above the top of the local PBL is caused by the barrier effect of a temperature quasi-inversion present in the altitude range 1.2–3 km, which is characteristic of the period under consideration. In addition, the smoke aerosols are markedly dry, with record low relative humidity values of about 5% measured in the altitude range 3–3.8 km.

The driving mechanism of the long-range transport of smoke aerosols and their spread over Europe, in particular, over Sofia, is the merger and interaction at heights in the lower troposphere of a zonal flow at moderate latitudes (a cold air advection system) carrying Canadian smoke and a gradually migrating eastward, blocking synoptic pattern over the Mediterranean and Europe with a circular air flow (a negative vorticity advection system), which captures and spreads the smoke aerosols over the continent with specific downward motion (as supported by NCEP/NCAR reanalysis composite diagrams of relevant meteorological parameters, such as geopotential height, air temperature anomaly and pressure vertical velocity omega).

The smoke aerosols have probably undergone a low degree of upsize aging (e.g., through particle aggregation) during their transport from Canada to the Sofia region, as shown by the relatively high values of the backscatter-related Ångström exponent for the two lidar wavelength pairs (355/532 nm and 532/1064 nm), which for the height range of the smoke layers are in the range 1.5–2, indicating aerosol modes in the submicron size range.

The registered smoke particles have a predominantly low degree of non-sphericity, which can be concluded from the low values of the lidar-determined depolarization ratio (0.02–0.07 at 355 nm and 0.03–0.08 at 532 nm).

The AERONET-provided columnar optical and microphysical properties of the aerosol ensembles over Sofia on 3 October 2023 outline a mixed aerosol situation with a rather high AOD₄₄₀ and AE_440/870 well below that of fine particles, as BB or urban/industrial aerosols, and well above that of coarse particles, as Saharan dust and continental or marine aerosols. Moreover, typically, the aerosol optical depth of the fine particle fraction at the wavelength of 500 nm is one order of magnitude larger than the one of the coarse particle fraction, while both fractions have comparable mode peaks in the particle volume size distribution; the particle single-scattering albedo has a complicated increasing–decreasing behavior depending on the wavelength and indicates rather high light absorption ability, and the fine particle fraction has become, in fact, an accumulation mode fraction with effective radius exceeding several times 100 nm. The above-listed peculiarities, along with the generally high particle sphericity established by lidar and AERONET sun photometer measurements, suggest that the aerosol field over Sofia on that day is dominated by transatlantic smoke from Canadian wildfires (above the PBL top) and aged and fresh European BB aerosols mixed with urban/industrial ones (in the PBL).

The HYSPLIT-provided backward trajectories of the air masses arriving over Sofia, the FIRMS’ maps of the fire outbreaks worldwide, the analysis of the weather situation over Europe and Mediterranean, as well as the lidar and radiosonde profiling of the atmosphere above the city, all show that the dry BB aerosol masses arriving from Canada are layered mainly above the PBL; in contrast, those originating from Europe follow trajectories slightly above the ground into the PBL space. Coarse-particle continental aerosols could also have accompanied the European BB smoke masses. According to SKIRON and MONARCH model forecasts, one may also expect the presence of traces of Saharan dust. Indeed, passages of coarse, non-spherical and weakly absorbing particles have been observed by tracking the daily evolution of their sphericity factor and linear depolarization ratio and the spectra of their single-scattering albedo and imaginary refractive index.

We can conclude, in summary, that smoke aerosols from the devastating forest fires raging in Canada in the early fall of 2023, along with the large-scale circulation patterns that transported and spread them, have had a significant impact on the general state and dynamics of the troposphere over the Mediterranean and all of Europe, in particular, over Bulgaria. These effects relate to the aerosol composition and stratification, meteorological conditions and parameters, air-mass motion, as well as the atmospheric thermal and radiative budget, with direct and indirect projections on local and regional climate. Furthermore, regional advection transport of aerosol mixtures observed in the near-ground atmospheric layer, such as moderately aged fire smoke, general continental aerosols, traces of mineral dust, etc., have had the potential of directly affecting the local environment and human health.