Assessment of the Spatial Structure of Black Carbon Concentrations in the Near-Surface Arctic Atmosphere

Abstract

The results of the research are numerical estimates of the average fields of black carbon mass concentration in the surface layer of the atmosphere of the Arctic region obtained using the numeric technology referred to as fluid location of the atmosphere (FLA). The modelling has been based on measurements of the black carbon concentrations in the near-surface atmosphere obtained during the two cruises of the Professor Multanovskiy (28 July–7 September 2019) and Akademik Mstislav Keldysh (31 July–24 August 2020) research vessels. These measurements have been supplemented by measurements at stationary monitoring points located on the Spitsbergen and the Severnaya Zemlya archipelagoes. The simulation in the summertime demonstrates that areas of increased black carbon concentrations were observed over Northern Europe and, in 2019, also over the Laptev Sea basin. The obtained spatial distribution of mass concentrations of black carbon qualitatively agreed with the same data derived from the second Modern-Era Retrospective analysis for Research and Applications (MERRA-2) but showed quantitative differences. The average values of mass concentrations of black carbon in the modelling zones are as follows: 85.3 ng/m³ (2019) and 53.6 ng/m³ (2020) for fields reconstructed by the FLA technology; and 261.69 ng/m³ (2019) and 131.8 ng/m³ (2020) for the MERRA-2 data.

1. Introduction

While studying the processes related to changes in the Earth radiation balance and causing global warming of the planet, special attention is paid to the Arctic region. Though the Arctic is remote from large sources of atmospheric emissions, it is universally acknowledged that the Arctic is warming faster. This phenomenon has been called the Arctic or polar amplification [1,2]. By recent estimates, the Arctic is warming four times faster than the entire globe as a whole [3].

When solving the problems of forecasting climate changes in the Arctic region, the comprehensive impact of aerosol particles on climatic processes is exacerbated by their high spatial and temporal variability. Atmospheric aerosol can have both positive (heating) and negative (cooling) radiation effects on the lower and upper boundaries of the Arctic atmosphere, which significantly depends on the albedo of the underlying surface and the absorbing characteristics of atmospheric aerosol [4,5,6].

It has been established that black carbon (BC) is only a small part (just a few percent) of the aerosol mass in the atmosphere of the Arctic region, which mainly consists of sulfates, sea salt and organic aerosol [7]. Nevertheless, it is BC that is of great climatic interest, since it effectively interacts with solar short-wave radiation, as well as, in some cases, with long-wave radiation reflected from the Earth’s surface; it affects cloud formation and processes inside clouds and increases the melting of snow and ice cover after sedimentation on the underlying surface [5,8,9,10,11]. By some estimates, BC is the second most important anthropogenic emission in terms of its climate forcing and only carbon dioxide is estimated to have a greater forcing [11].

Many authors underline that the sources of BC in the Arctic are of anthropogenic origin (burning of fossil fuels and biofuels). Besides, in summer it is the smoke of forest fires that has the biggest impact. Thus, it is believed that BC mainly comes from the long-distance transport of air masses from the continents of Eurasia and North America [12,13,14].

There is a huge number of scientific publications on the problem of BC in the Arctic region. There is an excellent, almost exhaustive analysis of various aspects of this topic, including monitoring and modeling issues [11].

Since regular aerosol measurements in the Arctic face many technical difficulties, primarily due to the large territory of the region and inaccessible infrastructure, nowadays there is no well-connected network of points to monitor aerosol properties in the atmosphere of this region. Though it is of paramount importance to obtain up-to-date information on the spatial distribution of aerosols in the atmosphere to correctly predict climate changes. The density of the stationary measuring network [15,16], the number of ship and/or aviation research expeditions [17,18] as well as the complexity of space-borne remote sensing make it difficult or even almost impossible to use their data to directly build the spatial distributions of BC in this region [19]. The main methods for reconstructing the spatial distributions of BC in the Arctic atmosphere are simulation of climatic processes considering atmospheric chemistry [17,20,21,22,23,24]; reanalysis via the methods of assimilation of observations and space-borne remote sensing data [25,26,27,28]; and methods of back trajectories statistics and solving inverse problems of observational data interpretation [12,16,29].

To assess the spatial distribution of BC in the Arctic region the authors of this paper rely on the numeric technology of fluid-location of the atmosphere (FLA) [30,31,32,33]. This approach combines the statistical analysis and the general idea of the methods of back trajectory statistics [34,35,36] with solving the equations of impurity transfer along the air parcels trajectories (for Lagrangian particles). Moreover, it is not necessary to know the location and properties of the sources of impurity emissions into the atmosphere. The information on the amount of a pollutant in the atmosphere is transferred to the measurement devices from remote areas because of the airflows, which allows us to speak about a new kind of ground-based remote sensing—the passive wind location of the atmosphere.

2. Materials and Methods

The initial data for the FLA simulation are the results of continuous measurements of the impurity content in the atmosphere $\phi \left({\mathit{r}}_m\left(t_k\right),t_k\right)$. In case of moving measuring platforms (research vessels), the coordinates of the monitoring station ${\mathit{r}}_m$ vary depending on time moments of measurements $t_k$. For each measurement there is only one back trajectory of air (Lagrangian) particle movement, which is designated by the radius-vector ${\mathit{r}}_k\left(t\right)$.

Bellow we consider a special case of the general formulation of the problem where the following simplifying assumptions are used:

• Quasi two-dimensional approximation: back trajectories of Lagrangian particles are used in real 3D space, but the solution is sought on the surface of a ‘flat’ 2D computational grid;
• Long-range transport of the pollutant is considered when advection predominates over small-scale turbulent diffusion;
• The passive impurity, i.e., the contribution to the change in the admixture concentration from various physicochemical processes, is negligibly small compared with the contribution from the action external sources and sinks;
• The trend in the average concentration fields is absent.

In that case the average effective field of concentration ${\tilde{\phi }}_{V,T}$ obeys the following integral equations on the set of grid cells V:

$${\tilde{\phi }}_{V,T}\left(\mathit{r}\right)=\frac{1}{N_{V,T}}{\displaystyle \sum }_{k=1}^{N_{V,T}}\left\{\phi \left({\mathit{r}}_m\left(t_k\right),t_k\right)+\frac{1}{{\tau }_{k,V}}{{\displaystyle \int }}_{t_{k,V}}^{t_{k,V}+{\tau }_{k,V}}\left[{{\displaystyle \int }}_{t_k}^t{\tilde{J}}_{V,T}\left({\tilde{\phi }}_{V,T},{\mathit{r}}_k\left(t^{\prime }\right)\right)d{t}^{\prime }\right]dt\right\},$$

(1)

where $N_{V,T}$ is the total number of air particles passing through the grid cell V over the time period T; k is the index denoting of the air particle back trajectory (which coming from various monitoring points) passing through the grid cell V over the time period T; ${\tau }_{k,V}$ is the residence time of k-th back trajectory within the grid cell V; $t_{k,V}$ is the entry time of k-th back trajectory into the grid cell V; t and t′ are the variables of integration over the trajectories which have a sense of time; and t represents the time moments when the Lagrangian particle is within the grid cell V. The integration in Equation (1) is performed along the trajectories of the Lagrangian particles ${\mathit{r}}_k\left(t\right)$.

The average effective field of sources ${\tilde{J}}_{V,T}$ is a function that depends on the average effective field of concentration ${\tilde{\phi }}_{V,T}$:

$${\tilde{J}}_{V,T}\left(\tilde{\phi },\mathit{r}\right)\stackrel{\mathrm{def}}{=}\frac{1}{VT}{{\displaystyle \int }}_{t_0}^{t_0+T}\left[{{\displaystyle ∯}}_{S_V}^{}\left(\tilde{\phi }\mathit{v}\left(\mathit{r}, t\right)\right)·d\mathit{\sigma }\right]dt,$$

(2)

where $t_0$ is the beginning of the period of averaging; $S_V$ is the area of surface covering the grid cell V; $d\mathit{\sigma }$ is the element of surface oriented “outward” the cell; $\mathit{v}\left(\mathit{r}, t\right)$ is the wind velocity field (considered known).

Integral Equations (1) and (2) form a closed system of equations. The number of equations is equal to the total number of grid cells V. The average effective field of concentration and the average effective field of sources are statistical estimates of the true averages for the considered time period of the concentration fields and capacities of sources of atmospheric impurities. The functions of the average effective field of concentration of impurities and the average effective field of sources are defined in such a way as to obey the equations describing the processes of transport and change of impurities in the atmosphere, while ensuring the realization of a set of known (measured) concentrations at specified times and in specified coordinates. In this situation it is not necessary to know the location and properties of the sources of impurity emissions into the atmosphere. Additional information on the FLA problem (the definition of basic concepts, recording of differential and integral equations) and about the methods of its solution in differential and integral formulations, as well as errors and verification of the solutions can be found in the following works [30,31,32,33].

For numerical calculations, a triangular grid has been used. The grid is generated under the following rules: the initial triangles are the main spherical triangles whose vertices coincide with the vertices of an icosahedron put in a sphere whose radius is equal to the average radius of the Earth. The triangular grid is constructed by dividing the sides of the main triangle in half, and as a result there are four new triangles formed inside each cell. The authors of this work have used a sixth-generation triangular grid (obtained by six consecutive divisions of the base triangle). The relative difference in the areas of the triangular grid cells around the globe does not exceed 25% [37] and does not depend on latitude, which makes it convenient for calculations in the circumpolar regions.

In our previous works, the FLA problem was solved solely on the basis of measurement results in one or more stationary monitoring points (see [30,32]). In this work, to estimate the average fields of BC mass concentration, we have used not only the measurement results received at stationary monitoring points, but also from mobile measuring platforms located on research vessels (RV).

Our research is based on the input information—the results of measurements of aerosol characteristics received during the cruise of the Professor Multanovskiy RV (28 July–7 September 2019) from Vladivostok to Murmansk [38] and during the 80th cruise of Akademik Mstislav Keldysh (31 July–24 August 2020) from Kaliningrad to Arkhangelsk [39]. We have also used data obtained in similar periods in 2019 and 2020 at two polar stations: “Cape Baranov” and in the settlement of Barentsburg.

The “Cape Baranov” station (79°16′ N, 101°45′ E) is located in the northern part of the Bolshevik Island, which is part of the Severnaya Zemlya archipelago [40]. The Russian scientific center on the Spitsbergen archipelago is located in the southwestern part of Barentsburg (78°04′ N, 14°13′ E) [41].

To obtain aerosol characteristics in ship expeditions and at polar stations in 2019 and 2020, some sets of instruments were used, including the MDA-02 aetalometer [42]—which helped to carry out the concentration of the light-absorbing substance in the equivalent of BC—soot (M_bc) [43].

Results of local measurements of BC concentrations aboard the ships and in Barentsburg occasionally experience technogenic effects. Aboard a ship, it is smoke from the ship’s funnel or dust from ventilation shafts. The aerosol characteristics in Barentsburg (with the population of about 500 people) are affected by local sources of aerosol pollution: products of the coal mining industry and a thermal power plant located at the distance of about 0.5 km. In addition to technogenic effects, there may be omissions and false measurements in the data obtained due improper operation of devices. Therefore, the initial series of observations were filtered using a special algorithm [44]: by detecting defective values (lasting up to 3 h) and restructuring distorted data.

Massive calculations of back trajectories have been performed using the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT [45]) model software. To calculate back trajectories, we have relied on three-dimensional fields of meteorological quantities of the Northern Hemisphere as the input information received from the archive of the Global Data Assimilation System (GDAS1) database from the server: ftp://arlftp.arlhq.noaa.gov/archives/gdas1/ (accessed on 1 February 2022). The spatial resolution of the initial meteorological information was 1° × 1° with 23 pressure levels in height, and the time resolution was 3 h. For all monitoring points, back trajectories were “launched” every hour from the height of 10 m above the ground level (approximate height of measurements). All the back trajectories were calculated with a 15 min time step. The duration of Lagrangian particle movement along the trajectories was 96 h.

The balances of BC flows at the boundaries of fixed “Eulerian” calculation cells were calculated with the wind speed at the altitude of 100 m from the ERA5 reanalysis database of the European Center for Medium-Range Weather Forecasts (http://www.ecmwf.int, accessed on 1 February 2022) with the spatial resolution of 0.25° × 0.25° and a one-hour time step.

3. Results and Discussion

3.1. Results of the FLA Technology Modelling

The average effective concentration field is the average impurity concentration calculated in Lagrangian representation, i.e., based on the concentrations of impurity carried by moving air particles—over the period of measurement and by the volume of each calculated cell. Figure 1 shows the average effective fields of BC mass concentration reconstructed by the FLA technology based on the measurement results in the summer of 2019 (a) and 2020 (b). The calculation results are presented on a triangular calculation grid (shown below in the figures), with the cell areas being 6341.6 km² on average. Stationary monitoring points are marked with blue dots: “Barentsburg” (1) and “Cape Baranov” (2). The RVs routes are marked with blue lines. The background is the raster map from the Natural Earth public cartographic data set (https://www.naturalearthdata.com, accessed on 1 February 2022). The coastlines of continents and large islands are also mapped.

The shaded areas of the map represent estimates of concentration fields within the 96-h zone of impact, i.e., the area of space from which different devices receive information on BC concentration in the atmosphere by wind transport. Of course, BC is also present outside this area, but without involving data from other measurement points, it is impossible to estimate the values of its concentrations by the FLA technology, since not a single 4 day back trajectory was received from this territory during the analyzed period.

In 2019, the maximum concentrations of BC were observed over Eastern Siberia and the Scandinavian peninsula, with a trail of increased concentrations also recorded—which stretches from Canada across the Arctic Ocean. In 2020, the highest average concentrations in the modelling area were observed over Northern Europe, but their values were lower than the Siberian and Scandinavian maximum observations in 2019.

3.2. Comparison with MERRA-2

The Modern-Era Retrospective analysis for Research and Applications, version 2 (MERRA-2 [46]) contains an aerosol module based on the Goddard Earth Observing System, Version 5 (GEOS-5 [47]) and Goddard Chemistry Aerosol Radiation and Transport (GOCART [48]) models, which considers five aerosol components, including BC. For comparison, we have used surface mass concentrations of BC from an hourly time-averaged two-dimensional data collection (M2T1NXAER [49]). The spatial resolution of the MERRA-2 data is 0.5° × 0.625° (latitude × longitude).

Figure 2 shows the time series of BC mass concentrations obtained on the basis of in situ measurements and from the MERRA-2 reanalysis for two stationary monitoring points and along the RVs routes.

Table 1. Descriptive statistics of the data on surface concentrations of BC on the basis of in situ measurements and from the MERRA-2 reanalysis.

Characteristics	Barentsburg
	2019		2020
	In Situ	MERRA-2	In Situ	MERRA-2
Mean	30	118	40	24
Std	56	88	85	18
Min.	0	30	0	0
Q1	6	68	5	11
Median	15	101	10	20
Q3	30	142	38	33
Max.	704	789	1067	97
Characteristics	Cape Baranov
	2019		2020
	In Situ	MERRA-2	In Situ	MERRA-2
Mean	82	149	17	25
Std	253	168	26	17
Min.	2	0	3	1
Q1	9	50	7	13
Median	21	100	12	21
Q3	56	186	19	32
Max.	3678	1392	339	79
Characteristics	RV
	2019		2020
	In Situ	MERRA-2	In Situ	MERRA-2
Mean	85	202	76	80
Std	80	197	96	120
Min.	2	13	0	0
Q1	28	81	19	18
Median	63	123	40	38
Q3	118	254	75	85
Max.	494	1380	485	860

contains descriptive statistics of the data shown in Figure 1. The graphs and the table demonstrate that in the analyzed summer period of 2019, the reanalysis results generally exceed the results of surface observations. As noted in [50], the reason for such a discrepancy may be the high values of the aerosol optical depth of the atmosphere as a result of large fires in Eastern Siberia. It should be noted that the aerosol optical depth of the atmosphere is an integral characteristic throughout the entire column of the atmosphere and may weakly correlate with surface concentrations of BC [51]. At the same time, it is the aerosol optical depth obtained on the basis of the results of satellite and surface measurements that is the only aerosol characteristic involved in the MERRA-2 reanalysis data assimilation procedure. Thus, in certain situations the MERRA-2 reanalysis data can overestimate the BC mass concentration in the near-surface atmosphere. For this reason, before comparing the spatial distributions of surface mass concentrations of BC, the fields were normalized to their average values, i.e., there was a comparison of the spatial structure of the fields obtained.

Since the results of the FLA technology and the MERRA-2 reanalysis are presented on different spatial grids, all the data has been presented as an azimuthal orthographic projection with the point of contact of the geoid with the projection plane at the north pole and with a spatial resolution of 100 × 100 km. Then the procedure of normalization of all concentrations to average values was performed; the normalization constants were 85.3 ng/m³ (2019) and 53.6 ng/m³ (2020) for the fields obtained by the FLA technology, and 261.69 ng/m³ (2019) and 131.8 ng/m³ (2020) for the MERRA-2 data.

Figure 3 and Figure 4 show the average normalized fields of BC mass concentration in the near-surface atmosphere, reconstructed on the basis of the FLA technology (a) and the MERRA-2 reanalysis (b) based on the results of measurements in 2019 and 2020, respectively. The red lines indicate the boundaries of the FLA modelling zones with the number of trajectories in the calculation cells more than 10. The modelling results outside these areas cannot be considered reliable due to the scarce statistics for estimating the average value. It can be seen that the average fields of BC mass concentrations reconstructed using the FLA technology are mosaic and contrast, while the fields constructed on the MERRA-2 data are smoother. The background values (first quartile, Q₁) of the average normalized BC concentrations in the zone of reliable modelling were 0.41 in 2019 and 0.32 in 2020 according to the results of the FLA technology; and 0.46 in 2019 and 0.22 in 2020 according to the MERRA-2 data (

Table 2. Descriptive statistics of normalized concentrations of BC (Figure 3 and Figure 4) in the zones of reliable modelling.

Characteristics	2019		2020
	FLA	Merra-2	FLA	Merra-2
Mean	0.97	0.71	0.75	0.58
Std	0.89	0.40	0.71	0.84
Min.	0.03	0.35	0.03	0.10
Q1	0.41	0.46	0.32	0.22
Median	0.77	0.56	0.55	0.31
Q3	1.18	0.84	0.91	0.49
Max.	7.32	3.56	4.66	5.65

).

From Figure 3 and Figure 4 it is clearly observed that there is a satisfactory correlation between the quality of the average fields of BC concentrations by the FLA results and that of the MERRA-2 results. For the fields constructed on the basis of measurement results in 2019, a vast area of increased values is observed over the north of Eastern Siberia and the Laptev Sea. The reason for the increased surface concentrations of BC in this area is most likely due to large forest fires in the Eastern Siberia region in August 2019 [38]. Also, the modelling shows areas of increased concentrations over Europe. According to the FLA results, this area covers almost the entire Scandinavian peninsula and stretches along the northern coast of the Eurasian continent. The MERRA-2 results show that the area of increased concentrations is located over the Baltic Sea and some part of Scandinavia.

As for the average fields of BC concentrations, based on the results of measurements in 2020, areas of increased concentrations are also observed over Northern Europe, partially covering the Baltic and North Seas. The FLA results show that increased concentrations of BC are also observed over the Norwegian and Greenland Seas. At the same time, the waters of the northern seas, which are located to the east of the island of Spitsbergen, have normalized concentration levels of less than one both by the results of the FLA technology and the MERRA-2 reanalysis.

4. Conclusions

The authors present the estimates of the average fields of surface mass concentrations of BC in the Arctic region, obtained on the basis of measurement data during the cruises of Professor Multanovskiy (28 July–7 September 2019) and Akademik Mstislav Keldysh (31 July–24 August 2020) research vessels and supplemented by measurements at stationary monitoring points located on the Spitsbergen and Severnaya Zemlya archipelagos.

The spatial distribution of the fields of BC concentrations over the waters of the Arctic seas is significantly heterogeneous. In addition, we can see a significant temporal variability recorded: the location of areas of increased concentrations of BC and their absolute values in 2019 and 2020 are different. In 2019, the maximum concentrations of BC were observed over Eastern Siberia and the Laptev Sea. The reason for the increased surface concentrations of BC in this area is most likely due to large forest fires in the Eastern Siberia region in August 2019. In 2020, the highest average concentrations in the modelling area were observed over Northern Europe.

The spatial structure of the obtained mass concentrations of surface BC is quite consistent with the results of estimates based on the MERRA-2 reanalysis data. At the same time, the absolute values of BC concentrations by the MERRA-2 reanalysis exceed the results of the FLA technology. The average values of mass concentrations of BC in the modelling zone are as follows: 85.3 ng/m³ (2019) and 53.6 ng/m³ (2020) for fields reconstructed by the FLA technology; and 261.69 ng/m³ (2019) and 131.8 ng/m³ (2020) for the MERRA-2 data.