INTRODUCTION

Large-scale smoke hazes transform drastically the radiation regime in the atmosphere [1, 2]. Smoke aerosol formed during mass fires and tiny gas impurities, including carbon monoxide, negatively affect the environmental setting [3]. The optical and microphysical characteristics (OMCh) of smoke aerosol can be judged by the data of tropospheric aerosol monitoring on the AERONET global network of stations [4]. The AERONET website (aeronet.gsfc.nasa.gov), in addition to the measured data of aerosol optical depth (AOD) (extinction) τex and indicatrix of sky brightness, presents the results of reconstructing [5] (for wavelengths of 440, 675, 870, and 1020 nm) spectral dependences of the real (n) and imaginary (κ) parts of the refractive index of a substance, τex for fine and coarse modes of aerosol (τf and τс), absorption AOD (τab), the single-scattering albedo (ω), the asymmetry parameter of the scattering indicatrix and the function of aerosol particle size distribution in the form of \(dV(r){\text{/}}d\ln r\), where r is the particle radius and V(r) is the accumulated volume of particles.

We studied earlier [6–8] large-scale smoke hazes in Northern Eurasia in the twenty-first century and in particular the large-scale smoke haze in July 2016, when eastern air-mass transport led to the spread of Siberian smoke haze to the European part of Russia and further west [9, 10]. During the long-range transport of Siberian smoke haze, there was a decrease in the absorption coefficient of the smoke aerosol due to the condensation of volatile organic compounds nonabsorbing in the visible region on the smoke aerosol particles, which are formed under photochemical transformation of gas components in the smoke haze [10] during the long-range transport of smoky air masses.

In the first decade of the twenty-first century, the optical and microphysical characteristics of smoke aerosol during mass forest fires on the territory of Alaska were observed using the AERONET data [11]. Since during climate warming, many processes that are negative for humans have intensified, we analyzed the consequences of the mass forest fires on the territory of Alaska in July 2019.

In particular, the changes in the absorption coefficient of smoke aerosol (ACSA) were of interest. During the analysis of ACSA based on the AERONET data, as a rule, focus is given to the variations in the single-scattering albedo [12] \(\omega (\lambda ) = {{\tau }_{{{\text{sc}}}}}(\lambda ){\text{/}}{{\tau }_{{{\text{ex}}}}}(\lambda )\), where λ is the optical wavelength, τsc is the aerosol optical depth of scattering, and \({{\tau }_{{{\text{ex}}}}} = {{\tau }_{{{\text{sc}}}}} + {{\tau }_{{{\text{ab}}}}}\). The results of measuring absorption AOD τab are often analyzed. However, ω(λ) and τab(λ) depend on the aerosol particle size distribution. Therefore, in order to determine ACSA, it is expedient first of all to use the spectral dependences of the refractive index of the imaginary part κ(λ).

The purpose of this work is to study the variations in OMCh of smoke aerosol during the mass fires in the boreal forests of Alaska in July 2019 using the AERONET data (level L1.5, version 3) and primarily to determine the main types of spectral dependences of the imaginary part of the refractive index of the substance of the smoke aerosol. It is shown that, along with numerous cases of nonselective absorption in the spectral region of 440‒1020 nm, determined by the presence of black carbon (BC) in the aerosol particles [13] and by the comparatively frequently observed spectra κ(λ) with increased values of κ at the wavelength of 440 nm, there is the presence of organic compounds in the smoke aerosol particles (together with black carbon) that have an absorption band in the ultraviolet and in the short-wave part of the visible region (brown carbon or BrC) [14–16].

In July 2019, anomalous absorption of smoke aerosol was recorded in some cases, which was evidently due to smoke aerosol particles having substance with  an absorption band at least in the region of 440‒1020 nm. This work presents the results of measuring the spectral dependences of extinction AOD and the function of the particle size distribution, which illustrates the dominant role of the fine mode in the processes of scattering and absorption of electromagnetic radiation by smoke aerosol (for the region of 440‒1020 nm), including for the cases of nonselective absorption. The variations in absorption AOD and the single-scattering albedo during the mass forest fires on the territory of Alaska in July 2019 are discussed.

In this work, we used the results of monitoring the optical and microphysical characteristics of tropospheric aerosol during large-scale smoke haze over the territory of Alaska in July 2019 at AERONET stations: Bonanza_Creek (abbreviated BZC) with the coordinates 64°45′ N and 148°32′ W, Kluane_Lake or KNL (61°02′ N, 138°24′ W), and NEON_DEJU or NDU (63°32′ N, 145°45′ W).

For comparison, we presented some results of monitoring the smoke aerosol characteristics during the savanna fires at the AERONET Mongu_Inn station or MGI (15°16′ S, 23°08′ E) in August 2022.

IMAGINARY PART OF THE REFRACTIVE INDEX FOR SMOKE AEROSOL

Work [11] presents the results of studying the variations in OMCh of smoke aerosol during the mass fires in the boreal forests of Alaska in 2004 and 2005.

It was shown that in most cases the ACSA was determined by the black carbon (soot) contained in the aerosol articles, which is characterized by a weak dependence of the imaginary part of the refractive index κ of the aerosol substance on the optical wavelength λ [13].

The typical spectra κ(λ) in the wavelength region spanning from 440 nm to 1020 nm were also recorded during the large-scale smoke haze of the Alaska territory in July 2019 (1‒5 in Fig. 1). Spectra 1, 3, 4, and 5 were obtained from the monitoring data at Bonanza_Creek station, while spectrum 2 was found based on the data from Kluane_Lake station. For comparison we provided a similar spectrum (6 in Fig. 1) by using the monitoring data at Mongu_Inn station in August 2022 during the savannah fire. Table 1 provides information on OMCh of smoke aerosol in measuring the spectra under consideration.

Fig. 1.
figure 1

Spectral dependences of the imaginary part of the refractive index of the smoke aerosol substance in the presence of black carbon (curves 1‒6) and brown and black carbon (curves 7‒10) in the particles and for the cases of anomalous absorption in the spectral region of 440‒1020 nm (curves 11‒15). The spectral dependence for black carbon is 16.

Full size image
Table 1. Optical and microphysical parameters of smoke aerosol
Full size table

Table 1 contains the values (for the wavelength of 440 nm) of extinction AOD, including separately for fine and coarse aerosol modes (\(\tau _{f}^{{440}}\) and \(\tau _{c}^{{440}}\)), the values of the real and imaginary parts of the refractive index of the smoke aerosol substance and the corresponding values of the single-scattering albedo.

In addition, Table 1 contains the \({{\beta }_{{{\text{ex}}}}}\) and \({{\beta }_{{{\text{ab}}}}}\) values of power approximations of spectral dependences \({{\tau }_{{{\text{ex}}}}}(\lambda )\) and \({{\tau }_{{{\text{ab}}}}}(\lambda )\) for the wavelength range of 440‒870 nm (Angstrom exponents) for extinction and absorption AOD \(\tau (\lambda ) = {{\tau }_{0}}{{({{\lambda }_{0}}{\text{/}}\lambda )}^{\beta }}\), where \({{\tau }_{0}},{{\lambda }_{0}}\), and \(\beta \) are the approximation parameters, as well as the maximum values \({v}_{m}^{f}\) of distribution \(dV(r){\text{/}}d\ln r\) for the fine aerosol mode and the corresponding radii of particles \(r_{m}^{f}\), at which the indicated maxima are reached.

It was noted in [11] that, during the large-scale smoke hazes over Alaska in 2004 and 2005, the absorption coefficient of smoke aerosol was affected by the presence of brown carbon in the particles. However, this influence was not quantified in [11].

Figure 1 illustrates the κ(λ) spectra obtained from the measurement data in the smoky atmosphere of Alaska in July 2019, indicating the noticeable influence of brown carbon at the wavelength of 440 nm (spectra 7‒9). Such spectra were also observed during the savanna fires in August 2022 (spectrum 10).

In analyzing the monitoring data (level L1.5) of smoke aerosol OMCh during the mass forest fires on the territory of Alaska in July 2019, we detected anomalous absorption in the wavelength range of 440‒1020 nm by κ(λ) spectra (spectra 11‒15 in Fig. 1). We note that, out of 74 OMCh measurements at Bonanza_Creek Station in July 2019, we assigned 15 cases to situations with noticeable manifestations of anomalous absorption of the smoke aerosol.

The κ(λ) spectra 11 and 15 (Table 1 and Fig. 1) recorded at AERONET NEON_DEJU Station and at Bonanza_Creek Station on July 10 at 3:45 p.m. and on July 23 at 5:53 p.m., respectively, are of primary interest to us. When the wavelength changed from 440 to 1020 nm, the imaginary part of the refractive index at NEON_DEJU Station (spectrum 11) increased from 0.134 to 0.315. We propose a power approximation of anomalous spectra \(\kappa (\lambda ) = A{{\lambda }^{\alpha }}\), where A and α are the approximation parameters. For spectra 11‒14, the indicated index of power varies in the range from 0.7 to 1.05. The maximum rate of growth in κ reached (α = 2.3) as the wavelength was increasing in the case of spectrum 15. The variability of the anomalous spectra of κ(λ) is likely to be determined by the variations in the relative contributions of anomalous and nonselective absorption and the possible changes in the shape of the anomalous absorption band.

Thus, a broad band of smoke aerosol absorption was discovered in the visible and the near infrared region during mass fires in the burial forests of Alaska.

ABSORPTION AEROSOL OPTICAL DEPTH

The absorption aerosol optical depth is an important characteristic of the absorption coefficient of tropospheric aerosol. Figure 2 presents the spectral dependencies of the absorption AOD. In cases 1‒6, when the ACSA is determined by black carbon, the spectral dependencies of the absorption AOD are approximated with satisfactory accuracy by power functions with Angstrom exponents \({{\beta }_{{{\text{ab}}}}}\) calculated for the wavelength range of 440‒870 nm, spanning from 1.08 to 1.46 (Table 1). It is known that for small particles at a weak dependence of κ on the wavelength, absorption AOT is proportional to λ–1 [17] \(({{\beta }_{{{\text{ab}}}}} = 1)\). For the large particles, the Angstrom exponent is close to 1.3 [17]. According to the data from Table 1, the values of \({{\beta }_{{{\text{ab}}}}}\) in July 2019 at Bonanza_Creek station (the spectra 1, 3‒5) were found in the range of 1.23‒1.28, which conforms to the data published earlier.

Fig. 2.
figure 2

Spectral dependences of the absorption aerosol optical depth of smoke aerosol in the presence of black carbon (curves 1‒6) and brown and black carbon (curves 7‒10) in the aerosol particles and for the cases of anomalous absorption in the spectral region of 440‒1020 nm (curves 11‒15).

Full size image

Noticeable deviations were observed for the spectra of \({{\tau }_{{{\text{ab}}}}}(\lambda )\), recorded at Kluane_Lake station (\({{\beta }_{{{\text{ab}}}}}\) = 1.46) and Mongu_Inn station (\({{\beta }_{{{\text{ab}}}}}\) = 1.08).

When brown carbon is present in the particles of smoke aerosol (spectra 7‒10), the Angstrom exponents for absorption AOD increase as high as 1.43‒1.69, including for the spectra recorded at Kluane_Lake station and Mongu_Inn station (9 and 10).

The spectral dependencies of the absorption AOD during the anomalous absorption in the visible range in the near-infrared region of the spectrum (11‒15 in Fig. 2) differ significantly from the previous ones. In most similar cases (11‒14), the absorption AOD depends on the wavelength comparatively weakly \(({{\beta }_{{{\text{ab}}}}} = 0.24{\kern 1pt} - {\kern 1pt} 0.5)\). The exception is the spectral dependence recorded at Bonanza_Creek station on July 23, 2019, at 5:53 p.m. (15 in Table 1), when \({{\tau }_{{{\text{ab}}}}}\) grows rapidly at an increase in the wavelength from 0.022 at a wavelength of 440 nm to 0.046 at a wavelength of 1020 nm (the Angstrom exponent is 0.91). We mention the large values of absorption AOD observed at NEON DEJU station equal to 0.72 and 0.61 at wavelengths of 440 nm and 1020 nm, respectively, at 3:45 p.m. on July 10, 2019.

One more characteristic of the ACSA is a single-scattering albedo (Table 1). However, we did not record markedly expressed patterns of variability in \(\omega (\lambda )\) related to anomalous absorption of smoke aerosol. We mention unusually low values of \(\omega \) (from 0.62 to 0.32 at a change in the wavelength from 440 to 1020 nm) at NEON_DEJU station on July 10, 2019, at 3:45 p.m. (spectrum 11). Thus, we revealed a significant difference in the spectral dependences of the ACSA characteristics in the cases of anomalous selective absorption of smoke aerosol in the visible and near-infrared regions.

EXTINCTION AEROSOL OPTICAL DEPTH

The level of atmospheric contamination by smoke is determined by the extinction aerosol optical depth. In particular, the direct solar radiation flux on the underlying surface for the wavelength of 440 nm at \({{\tau }_{{{\text{ex}}}}} = 4.53\) for the solar zenith angle of 44° (2 in Table 1) decreased by approximately a factor of 500. Figure 3 presents the dependences of the extinction AOD on the wavelength. They are approximated by power functions with satisfactory accuracy (Angstrom approximation). For the cases presented in Table 1, when the absorption of direct solar and scattered radiation in the visible and near-infrared regions in aerosol particles is determined by black carbon (1‒6 in Table 1), including during the savanna fires (Table 1), the Angstrom exponent \({{\beta }_{{{\text{ex}}}}}\) varies from 1.44 to 1.80.

Fig. 3.
figure 3

Spectral dependences of the extinction aerosol optical depth of smoke aerosol in the presence of black carbon (curves 1‒6) and brown and black carbon (curves 7‒10) in the aerosol particles and for the cases of anomalous absorption in the spectral region of 440‒1020 nm (curves 11‒15).

Full size image

The appearance of brown carbon in the smoke aerosol particles (in addition to black carbon present in them) evidently leads to a small increase in \({{\beta }_{{{\text{ex}}}}}\), on average. At small contributions of brown carbon, this influence is comparatively low, which is typical of the cases we considered (7‒10 in Table 1), when \({{\beta }_{{{\text{ex}}}}}\) changed from 1.24 to 1.92. For large-scale smoke hazes, there are frequent deviations from the power approximation of \({{\beta }_{{{\text{ex}}}}}(\lambda )\), which are determined by the variations in the form of the particle size distribution for the fine mode of smoke aerosol particles [18].

In the cases of anomalous absorption, the Angstrom exponent noticeably decreases for the absorption AOD within 0.96‒1.39. The minimum value of \({{\beta }_{{{\text{ex}}}}}\) = 0.96 was recorded at NEON_DEJU station on July 10, 2019, at 3:45 p.m. (11 in Table 1), which cannot be explained by the contribution of the coarse aerosol mode to \({{\tau }_{{{\text{ex}}}}}(\lambda )\) (it does not exceed 1%).

Thus, in the case of anomalous absorption of smoke aerosol, significant changes occur in the spectral dependence of the AOD on extinction, which leads to significant decreases in \({{\beta }_{{{\text{ex}}}}}\) (by a factor of 1.5 and greater).

FUNCTION OF THE SMOKE AEROSOL SIZE DISTRIBUTION

The AERONET website presents the functions of the aerosol size distribution averaged over the atmospheric column by radii r in the range of changes in r from 0.05 to 15 µm. Figure 4 shows the volume size distributions \(dV(r){\text{/}}d\ln r\) (in arbitrary units) for smoke aerosol, reconstructed from the data of monitoring at AERONET stations (the level L1.5) on the territory of Alaska, including Bonanza_Сreek, Kluane_Lake, and NEON_DEJU during large-scale smoke hazes in July 2019, as well as the distributions (for comparison) recorded during the mass fires in the savanna in August 2022. The recorded distributions (Fig. 4) clearly display a fine (submicron) and coarse modes of aerosol. It is not too difficult to see that smoke aerosol is dominated by the fine mode with modal radii of approximately 0.15–0.25 µm, which agrees with the results published earlier [11]. At low contents of smoke aerosol in the atmospheric column, a noticeable contribution to the total volume of aerosol particles can be made by the coarse aerosol mode.

Fig. 4.
figure 4

Functions of distribution of smoke aerosol particles by volume in the presence of black carbon (curves 1‒6) and brown and black carbon (curves 7‒10) in the aerosol particles and for the cases of anomalous absorption in the spectral region of 440‒1020 nm (curves 11‒15).

Full size image

The nature of anomalous absorption of smoke aerosol in the visible and near-infrared region is of interest.

It was stated in the monograph [19] (p. 365) that “smoke released during the combustion of fuels consists primarily of soot, resin, and ash. The particles of soot or smut form black smoke. Fine particles of liquid or semi-liquid substances have yellow or brown shades.” The indicated shade of smoke points to selective absorption of the smoke aerosol. In the summer of 2002, we collected aerosol samples during the forest-and-grass fires in Shatura district, Moscow oblast (in order to quantify the soot aerosol concentration in the smoky air); they were yellow in color, which can be explained by the presence of resin in the smoke aerosol particles. The experiment showed that subliming of resin results in the formation of aerosol, which creates a yellow film when the aerosol samples are collected on a quartz-fiber filter.

An important role in the formation of smoke aerosol during the forest fires on the territory of Alaska in the summer of 2019 could have been played by the meteorological conditions during the fires and in the time preceding the fires. In the summer of 2019, the US National Weather Service, as well as the FOBOS Center, reported that the air temperature broke the records in Alaska in July 2019. Moreover, July 2019 was the 12th month in a row when the air temperature in Alaska was above the norm almost every day (http://www.rp5.ru; http://www.ncdc.noaa.gov/). We note that in the case of forest fires in Alaska in 2004 and 2005, long-term increases in the air temperature were not recorded.

The question of the anomalous absorption nature requires further study.

CONCLUSIONS

According to the data of monitoring of the optical and microphysical characteristics of smoke aerosol at AERONET stations during mass fires in the boreal forests in Alaska in July 2019, anomalous absorption spectra of smoke aerosol in the visible and near-infrared regions were detected (range of wavelengths of 440‒1020 nm). It was established that, in the presence of the indicated anomalous absorption, the imaginary part of the refractive index of the smoke aerosol substance reached 0.315 at a wavelength of 1020 nm. A power approximation of spectral dependences of the imaginary part of the refractive index was proposed. Based on the data of monitoring, the index of power characterizing the growth rate of the imaginary part of the refractive index varies from 0.7 to 2.3 as the wavelength increases.

The variations in the spectral dependences of aerosol optical depths (extinction and absorption) were analyzed by using power approximations for which the indices of power are the known Angstrom exponents for the extinction and absorption spectra. In contrast to the Angstrom exponent values (1.08‒1.69) observed frequently for the smoke aerosol absorption spectrum, in Table 1 the indicated Angstrom exponents decrease as low as 0.16‒0.5 and in one case it was negative (‒0.91) when anomalous absorption appeared. The Angstrom exponent decreased also for the extinction aerosol optical depth (on average, from 1.65 to 1.18). The dominant mode of smoke aerosol in the presence and in the absence of anomalous absorption is the mode of fine aerosol.