Black carbon and particulate matter mass concentrations in the Metropolitan District of Caracas, Venezuela: An assessment of temporal variation and contributing sources

Atmospheric aerosols scatter and absorb solar and terrestrial radiation. They also participate in aerosol–cloud mechanisms because they act as cloud condensation nuclei. These interactions modify Earth’s radiative forcing and, hence, alter the climate (Haywood, 2021). Radiative forcing estimates from atmospheric aerosols still present high uncertainties (Intergovernmental Panel on Climate Change [IPCC], 2021). Furthermore, exposure to atmospheric aerosols with an aerodynamic diameter smaller than 2.5 µm (PM_2.5) is considered to impact human health (Landrigan et al., 2018).

Air pollution affects more than 80% of the population in densely populated urban areas (Zhu et al., 2013). High PM_2.5 mass concentrations are characteristic in these sites due to traffic emissions, among other sources (Chen et al., 2016). A significant portion of PM_2.5 from traffic emissions is composed of black carbon. This type of aerosol is produced and emitted from the incomplete combustion of fossil fuels, biofuels, and biomass burning, and it is considered a short-lived climate pollutant (Andreae, 2001; Bond et al., 2013). Additionally, black carbon aerosols have a fractal-like structure with a high surface area that adsorbs carcinogenic molecules of polycyclic aromatic hydrocarbons (Shrivastava et al., 2017).

Densely populated areas of developing countries experience rapid vehicle fleet growth (Wang et al., 2019). A significant portion of these vehicle fleets comprises old cars due to secondhand car importations from industrialized countries and inadequate air quality regulations (Wang and Yeo, 2016). Moreover, vehicles over 5 years in use are reported to have larger black carbon and PM_2.5 emission factors than newer vehicles. For instance, diesel and gasoline cars in use for less than 5 years have black carbon emission factors reduction by 60% and 47% from those in use for 5–10 years (Jezek et al., 2015). Therefore, old cars usage represents a technology that can significantly contribute to air pollution in low-income countries of South America. In particular, in Venezuela, most of the public transport (0.092 million vehicles) and cargo fleet (0.99 million vehicles) is made up of diesel-based engines (Instituto Nacional de Transporte Terrestre, 2013).

In the north of South America, aerosols derived from fossil fuel combustion, biomass burning, sea spray, and mineral dust are ubiquitous (Prospero, 1981; Sanhueza et al., 1987; Saturno, 2013; Morantes et al., 2021). Mineral dust from the Saharan desert in Africa can reach South America and intrude the mixing layers, contributing to significant effects on air quality, human health, and climate (Bozlaker et al., 2013; Prospero and Mayol, 2013; Kabatas et al., 2014). Mineral dust intrusions have been widely observed in Colombia (Ramirez, 2014), Puerto Rico (Colarco et al., 2003), Brazil (Gioda et al., 2011), Venezuela (Saturno, 2013; Morantes et al., 2021), and the Amazon Basin (Swap et al., 1992; Moran-Zuloaga et al., 2018), among other sites. Additionally, mineral dust from the Saharan desert can be enriched by biomass burning aerosols emitted and produced by the sub-Sahelian region in Africa (Salvador et al., 2016).

Forest fires are present throughout the north of South America due to land-use change and agricultural activities (Thornhill et al., 2018). The primary contributing source to biomass burning aerosol is the fires in Los Llanos region (Sanhueza et al., 1987). These events occur over the country and can experience long-range transport. A rise in biomass burning aerosols during the dry season has been reported at a high-altitude remote station (4,765 m.a.s.l.) in the lower free troposphere of the Andes region in Venezuela (Schmeissner et al., 2011; Hamburger et al., 2013).

Ground-based measurements of aerosol and short-lived climate pollutants are still scarce in the north of South America (Andrade-Flores et al., 2016). For the region, determining aerosol sources and their temporal variation is important because it can provide relevant information to improve air quality strategies, validate general circulation models’ outputs and reduce the regional radiative forcing uncertainties (Riojas-Rodriguez et al., 2016; UNEP and CCAC, 2018). In the northernmost area of South America, particularly in Venezuela, the only air quality regulation regarding aerosol is total suspended particles, which have a diameter range of <50–100 µm (Ministerio del Ambiente y Recursos Naturales, 1995). Studies done in the region include PM_2.5 and elemental carbon measurements, among other variables, with offline instruments from samples collected with high-volume samplers (Sanhueza et al., 1987; Saturno, 2013). However, continuous measurements are needed to lead to a better temporal characterization in concentration and chemical composition and increase the understanding of aerosol regional transport.

In this work, we aim to fill that gap of information by measuring PM_2.5 and equivalent black carbon (eBC) by online instruments in Caracas and a tropical montane cloud forest in the city suburbs, with the following main objectives: (1) To evaluate diurnal, weekly, and seasonal variability of PM_2.5 and eBC, (2) to identify potential sources of PM_2.5 and eBC, and (3) to infer the local and regional impacts with regard to air pollution and aerosol transport. This study is an essential contribution to reduce radiative forcing uncertainties and improving the existing air quality policies in the region, suggesting PM_2.5 and black carbon as better air quality suitable criteria than total suspended particles (Kutlar Joss et al., 2017).

We established 2 sampling sites: the city of Caracas (in the following referred to as the urban site) located on the roof (approximately 20 m a.g.l.) of the general command building Generalisimo Francisco de Miranda Air Base. The second site is a suburban area in a tropical montane cloud forest near Caracas (referred to as forest site). The site is located at a station belonging to the Venezuelan Institute for Scientific Research (IVIC) at 10 m a.g.l. (Figure 1, Table 1).

The metropolitan area of Caracas (urban site) has a high population density (approximately 1,223.41 inhabitants km^–2) (Instituto Nacional de Estadísticas, 2012). It is located in a relatively small, elongated, and narrow mountain valley (approximately 800 km²) in the central-north-coastal area of Venezuela. The Cordillera de la Costa mountain range (2,765 m.a.s.l., Naiguatá Peak) is located north and separates the city from the Caribbean Sea (about approximately 20 km). It experiences a well-defined dry (December–April) and wet (May–November) seasonality. Caracas encompasses the financial, commercial, and communications service center of the country. Urban land use has spread to the surrounding hills, with few industries in the valley (Sanhueza et al., 1982). It has a large vehicle fleet and, as commonly found in the developing world, it is very outdated. Most of the vehicles are gasoline-based fueled, while public transportation is diesel-based. Our sampling site in Caracas is located in the geographic center of Caracas on a flat plane. This location is representative of the background characteristics of the city. Large-scale meteorological conditions in the valley are dominated by trade winds coming from the NNE. Most of the day, Caracas has prevailing dispersion conditions, while temperature inversion forms at night, breaking up in the morning.

The valley has a tremendous dispersive capacity of gases and other pollutants, shown by a significant increase in the mixing layer height between 7:00 h and 9:00 h. (Sanhueza et al., 1982). Caracas mixing layer height average varies between 400 ± 230 m and 1190 ± 430 m, with an ascent rate of 395 ± 55 meters h^–1, equivalent to a 3.6 times dilution. The average mixing layer height from 9:00 h to 15:00 h ranges between 1927 and 2171 m. Later, after 16:00 h, the mixing layer’s height decreases proportionally to the solar radiation intensity reduction (Figure S1; Sena, 2003). At this site, we measured eBC from June 2018 to October 2019, while PM_2.5 from December 2018 to November 2019.

The forest site is about 10 km away from Caracas. It is located in Altos de Pipe, a mountainous area in Miranda State (approximately 8.4 km²), part of the Cordillera de la Costa mountain range. The main ecosystem is a tropical montane cloud forest with a well-defined dry (January–March) and wet (April–December) seasonality (Table 1). The annual temperature average is 17.6°C, the canopy of Altos de Pipe cloud forest ranges from 5 to 10 m., and the average relative humidity is 88%, which leads to the formation of mist with the rise of the moist air coming from the ocean (Cuenca, 1987; Sanhueza et al., 2000). Altos de Pipe is an excellent place to compare the atmospheric aerosols and trace gases concentrations with those found in Caracas due to the valley-mountain atmospheric dispersion effect related to its higher altitude (Sanhueza 2002). At this site, we measured PM_2.5 from June 2018 to June 2019.

The PM_2.5 mass concentration was measured at a 5-min time resolution using a continuous analyzer SHARP “Synchronized Hybrid Ambient Real-time Particulate” (Thermo Fisher Scientific, Waltham, MA, USA, model 5030i) equipped with a PM_2.5 cut-off size inlet, at a flow rate of 16.67 L min^–1. At both stations, the instruments were inspected weekly. The inspection included visual checks whether all the instrumental components were switched on and working correctly. The flow rate was checked and calibrated monthly with a reference flowmeter. The instrument consists of an array that combines a nephelometer detector and a radiometer. A detailed description of this instrument can be found in Hsu et al. (2016). PM_2.5 mass concentration is obtained from the linear relationship between the intensity of monochromatic light scattering by aerosols that pass through a light beam of 880 nm (nephelometer detector) and the beta radiation attenuation by particles that are deposited on a glass fiber filter (radiometer).

PM_2.5 concentration is calculated based on the PM_2.5 concentration average obtained by the nephelometer (N_average) and the ratio of PM_2.5 concentrations obtained between the radiometer and the nephelometer (B/N), which is considered a correction factor Equation 1,

The coefficient of the light scattering signal was a parameter set by the manufacturer, while the time constant in B/N was set for 20 min following the manufacturer’s recommendations (Thomas and Gebhart, 1994; Gebhart, 2001; Hsu et al., 2016). The concentration of particles is proportional to the intensity of light scattering by particles with the same size distribution and chemical composition (Gebhart, 2001; Wang et al., 2009). Therefore, PM_2.5 concentration from the nephelometer is calculated from the relationship between the scattered intensity detected by the detector (P), the incident intensity (I_o), the detection volume (V_m), the total concentration of the number of particles (C_n), the particle number size distribution (f(d_p)), with d_p (particle diameter), the integrated light scattering phase function for the corresponding solid angle (S), the wavelength (λ), and the refractive index (m) (Thomas and Gebhart, 1994; Gebhart, 2001; Hsu et al., 2016) (Equation 2),

In the radiometer, the attenuation of beta radiation (β^–) is analyzed when radiation (electrons) passes through a layer of particles suspended in the filter and is related to the concentration of PM_2.5 by Equation 3, where β is the electron flow transmitted, β_o is the electron flow incident in the sample, μ is the attenuation coefficient (dependent on the energy of the beta radiation source, ¹⁴C) (cm²g⁻¹), and x is the concentration of the particles (gcm⁻²). μ value is determined by comparing standards of known mass with β by the manufacturer (Kulkarni et al., 2011),

The instrument is calibrated with typical PM_2.5 aerosols standard compared to gravimetric reference methods (Jaklevic et al., 1981; Hsu et al., 2016). The sensitivity of the nephelometer depends on the particle size; therefore, the contributions of ultra-fine (Dp < 0.1 μm) and coarse (Dp > 2.5 μm) particles are underestimated. Particles larger than 2.5 µm size are underestimated since the mass scattering coefficients decrease with size, while ultra-fine particles are underestimated because they are within the Rayleigh scattering regime. The nephelometer automatically performs a zero in the measurement every 24 h to ensure the sensitivity of SHARP by circulating particle-free air after passing through a high-efficiency particulate air filter. On the other hand, the radiometer is automatically calibrated during the operation of the SHARP.

The PM_2.5 measurements usually have deviations related to the condensation of the water vapor in the filter, commonly contained in the sample flow of air mass entering the instrument. To reduce this interference, SHARP is equipped with a heating system that turns on automatically when the relative humidity exceeds 35%. Therefore, when heating the sample, it is not possible to discard volatile organic compounds lost due to evaporation (GAW Report No. 227, 2016). Nevertheless, it was reported that the loss of volatile organic compounds in the SHARP is lower compared to other instruments that use reference techniques to determine PM_2.5, such as tapered element oscillating microbalance (Hsu et al., 2016). The SHARP fiberglass filter changed automatically every 24 h at midnight or when a saturation mass of 1500 µg was achieved in the filter.

The concentration of black carbon in PM_2.5 is derived from the measurement of the absorption coefficient (C_abs) at 1-min time resolution, obtained from the amount of light that is transmitted through a fiberglass filter loaded with aerosol particles using a multi-angle absorption photometer (MAAP) (Thermo Fisher Scientific, Waltham, MA, USA) (Petzold and Schönlinner, 2004). The instrument was equipped with a PM_2.5 cut-off size inlet and used at a flow rate of 16.67 l min^–1, checked, and calibrated monthly with a reference flowmeter. The MAAP automatically transforms the C_abs into black carbon mass concentration, from the relationship between the C_abs with the mass absorption cross-section of black carbon at 637 ± 1 nm, which is the operational wavelength of the MAAP. The instrument firmware uses a mass absorption cross-section value of 6.6 m² g^–1, the reference value used worldwide for rural and urban aerosols obtained during instrumental comparisons with a thermal reference method. For more information regarding the method, please visit Müller et al. (2011).

MAAP is considered the most sophisticated between the instruments that measure since it measures both the light transmittance in the orthogonal direction through the filter and the reflectance at 2 different angles, considering the effects of multiple light scattering and attenuation (Petzold and Schönlinner, 2004). This fact makes it more accurate than others and more consistent with the reference method (Petzold et al., 2005; Slowik et al., 2007; Moosmüller et al., 2009). In addition to this, the MAAP has a high time resolution (1 min), which allows daytime variability and rapid changes in environmental concentrations to be observed.

When using data derived from optical absorption methods, such as those obtained with the MAAP, to determine the concentration of black carbon, it is recommended to report these results as eBC concentration. In this way, the ambiguities in the literature are reduced, and it is also assumed that the concentration of eBC is equal to the concentration of elemental carbon measured by thermo-optical methods (Petzold et al., 2013). Therefore, we adopted the eBC terminology for reporting black carbon in this study.

The PM monitors methods calibrations (nephelometer and radiometer) were done at the laboratory prior to the campaign at IVIC institute following the procedure in the instruction manual for the instrument model 5030i SHARP. The radiometer was calibrated with a mass transfer standard foil from the Thermo Fisher Scientific. Checks and calibrations of volumetric flow, temperature, and RH were done once per month or when it was possible to visit the sampling station. The photometer MAAP was calibrated at the TROPOS institute in Germany prior to the campaign in Venezuela. The data obtained from the instruments were checked and flagged, consisting of relative humidity of the sample above the recommended 35%, data below the detection limit, flow calibration periods, electricity shortage, and instrument failures. The dates and intervals for the data flagged and instrument calibration are available online (https://owncloud.gwdg.de/index.php/s/cidoBjSXbN8u8xA).

Local meteorology parameters with 5 min resolution were monitored at both sites, approximately 3 m far from each sampling station, including wind speed and direction, temperature, pressure, and relative humidity. The meteorological station used in the urban site was a Campbell Scientific USA, model WxPRO Entry-Level Research-Grade Weather Station, equipped with a Wind Monitor model 05103. In contrast, in the forest site, the meteorological station was a Davis brand Vantage Pro2 model 6152. Back trajectories were performed by the Hybrid Single-Particle Lagrangian Integrated Trajectory model (HYSPLIT, http://www.hysplit.uhu.es/hysplitweb08/HYSPLIT_traj.php; Draxler and Hess, 1988). Trade winds air masses from the North Atlantic are predominant during the year at both sampling sites. During September–October–November, the stations frequently receive air masses from the Caribbean Sea and air below 1,000 m of altitude from the southeast, presumed to be mostly continental air. Active fire locations from the Fire Information for Resource Management System (FIRMS) MCD14ML fire database (available online at https://earthdata.nasa.gov/active-firedata) were used to determine biomass burning events around the study sites. This information was complemented using the statistics of fire events reported from the firefighter department of the National Parks Institute (INPARQUES, Instituto Nacional de Parques).

Time series and univariate analysis (i.e., mean, median, and 15th, 25th, 75th, and 85th percentiles), were used to infer the temporal variability of eBC and PM_2.5. We performed analysis of variance analysis for mean comparison among groups, and statistical significance with a P value of <0.05 was determined. Data are presented as mean and standard deviation. The contribution of local sources to eBC and PM_2.5 was studied using the conditional bivariate probability function (CBPF) by percentiles. A summary of the measurement interval and the number of data points used to generate polar plots for the CBPF are described in Table S1. A complete description of the CBPF methodology can be found in Uria-Tellaetxe and Carslaw (2014). The function calculates the probability that a specific concentration of a particular pollutant is found in a specific wind vector. CBPF is plotted on bivariate polar graphs. In this way, clear contributions from pollutant sources based on wind vectors can be visualized.

CBPF uses wind speed as a third variable plotted on the radial axis and assigns percentile ranges of the pollutant concentration to wind speed and direction intervals. These CBPF ranges allow the identification of pollutant sources more rigorously because the sources tend to occupy precise concentration intervals. This characteristic of the sources is revealed because they tend to have a relatively constant emission rate. For example, high concentrations under low wind speeds (<3 m s^–1) indicate surface emissions released with little or no buoyancy. In contrast, high concentrations at high wind speeds (>6 m s^–1) are indicative of emissions at high velocity and a greater distance (Uria-Tellaetxe and Carslaw, 2014).

The function can be expressed by Equation 4, where $mΔθ,Δuy≥c≥x$ is the number of pollutant samples in a wind vector Δθ, with a speed interval of particular wind Δμ, at a concentration range c, established by a range of percentiles y – x, and n_{Δθ, Δμ} is the total number of samples present in a particular wind speed and direction interval,

In the daily analysis of the time series, we noted occasions when high eBC and PM_2.5 mass concentrations persisted for some days during the dry season. We, therefore, deemed exploration of the circumstances surrounding these events. We reviewed whether air masses arriving at the forest and urban site during these high concentration events crossed fire regions, using the FIRMS database combined with HYSPLIT back-trajectories. For these particular events, the backward trajectories were calculated for 24 h duration, arriving every 6 h at every site, with altitudes of 1000 and 500 m.a.g.l.

Main meteorological parameters at both sites are typical of tropical regions (Table S2). Wind roses show that the prevailing wind direction is from the southeast and northwest vectors for the urban and forest sites, respectively (Figure 2). PM_2.5 and eBC means (from daily data) for the urban site were 11.9 ± 8.5 µg m^–3, and 1.8 ± 0.9 µg m^–3, respectively; while PM_2.5 for the forest site was 10.2 ± 9.8 µg m^–3 (Figure 3, Table 2). In the urban site, eBC to PM_2.5 ratio is 1:6, and PM_2.5 values were higher for the period of measurements than at the forest site.

PM_2.5 and eBC for the urban site and PM_2.5 for the forest site exhibited defined daily and seasonal cycles, higher during the dry season (December–May) (Figures 3 and 4). In the urban area, eBC also displayed higher values during the dry season. Likewise, during the period of biomass burning in the dry season, we found higher eBC values in the urban site and PM_2.5 in both sites, in contrast to that of the rainy season (P < 0.05). For both sites, PM_2.5 mean values were larger during weekdays in comparison to weekends. The same is observed for eBC at the urban site (Table S3). During weekdays, the eBC diurnal cycle showed a maximum peak at 8:00 h., an absolute minimum at about 13:00 h, and a began rising again at 16:00 until about 21:00 h. Weekend patterns had the same trend but were flattened (Figure 4).

For the 7 days of the week, we determined daily hourly means of PM_2.5 and eBC concentrations in the urban site and PM_2.5 in the forest site during the biomass burning period in the dry and rainy seasons. A marked increment of eBC during the biomass burning period of approximately 30% higher is observed compared to the rainy season in the daily hour averages (Figure 5). The inter-hour variability of the eBC is bimodal, but with the highest concentration peak, between 17:00 h and 10:00 h, compared to what is observed for the rainy season and the entire sampling period.

The CBFP polar graphs for the concentrations of eBC and PM_2.5 for the urban site and PM_2.5 for the forest site showed seasonal patterns for different sources. For instance, during the wet season in the urban site, the CBPF for the percentiles 0–50th shows that the more probable eBC sources in the range of 0–1.1 µg m^–3 are located in the southeast, with wind speeds below 5 m s^–1. Percentiles 50–100th of the CBPF show that eBC sources (1.1–9.5 µg m^–3) are spread within the northwest, southwest, and south, with wind speeds above 3 m s^–1 (Figure 6). During the dry season, the results for the percentiles 0–50th indicate that the more probable sources of eBC (0–1.5 µg m^–3) are in the southeast, and southwest, with wind speeds between 0 and 6 m s^–1. For the percentiles 50–100th, eBC sources (1.5–11 µg m^–3) are spread between the northwest, southwest, south, and southeast, with wind speed above 2 m s^–1 (Figure 6c and d).

Probable sources for PM_2.5 (≤13 µg m^–3) in the urban site during the wet season are located in the north and west, with wind speed between 0 and 4 m s^–1. PM_2.5 sources for the percentiles 50–100th are found to the south and southeast, with wind speed >3 m s^–1 (Figure S2a and b). There is a specific contribution of PM_2.5 (0–9.7 µg m^–3) from the north and southeast during the dry season. For concentrations between 9.7 and 114 µg m^–3, the sources are spread around the west, southwest, and southeast (Figures S2c and d). In the forest site, during the wet season, probable sources of PM_2.5 (≤3.4 µg m^–3) are located at the north, with wind speed over 4 m s^–1. For concentrations between 3.4 and 63 µg m^–3, sources are mainly at the northwest (Figure S3a and b). During the dry season, probable PM_2.5 sources in the range of 0–8.5 µg m^–3 are shown in northwest and southwest directions, with wind speed (< 3 m s^–1). Sources of PM_2.5 with concentrations between 8.5 and 148 µg m^–3 have a very defined direction from the north, with wind speed over 3 m s^–1 (Figure S3c and d).

Mean daily values of eBC in the urban site was 1.75 ± 0.89 µg m^–3, while for PM_2.5 was 11.91 ± 8.54 µg m^–3, and 10.23 ± 9.78 µg m^–3 for the urban and forest site, respectively, for the entire period of measurements (Table 2). The eBC results of the present study are the first to be obtained with a high temporal resolution for the urban site (Caracas) and the country. eBC concentration in the urban site and PM_2.5 for both sites are low compared to typical average values from other urban areas (Figure 3).

High eBC and PM_2.5 concentrations would be expected given the large population density in Caracas metropolitan area (approximately 1223.41 inhabitants km^–2) (Instituto Nacional de Estadísticas, 2012) and its high vehicle fleet. Surprisingly, eBC and PM_2.5 concentrations values were low at both sites compared to the average from typical urban areas due to this region’s particular meteorology, which prevents long residence time of the air masses. Every morning the planetary mixture layer breaks due to solar radiation, warming up the city surface. The warm and lower air masses expand as the atmosphere is heated from below and progressively warms the air above by convective heat transfer. This daily process releases the atmospheric aerosols and other trace gases air masses, resulting in a cleaner planetary mixture layer during the day, despite the large anthropic emissions from the surface (Sanhueza et al., 1982).

The inter-hourly variability of eBC concentration is pronounced and bimodal. A high concentration peak occurs between 6:00 h and 8:00 h and one of lesser magnitude between 18:00 h and 22:00 h (Figure 4). The diurnal cycle of eBC during working days and weekends may be associated with fossil fuel combustion emissions from the highways traffic, which are characteristic of the urban site. Higher eBC concentrations in the early morning, reaching its peak at around 8:00 h, are caused by a large automobile traffic starting at 5:00 h (particularly public transportation, which is mostly diesel-based) and the lower mixing layer height. Later, at around 13:00 h, the observed decrease of eBC is caused by the fast dispersion of the pollutants and the dilution effect due to the maximum mixing layer height. Based on the wind roses (Figure 2), we can expect that these pollutants are transported at this time of the day, mainly to the forest site, located in the southwest of the urban site.

Previous studies have suggested that the valley might experience an increment in aerosol mass concentration and other pollutants every day after 16:00 h because of the reduction of the mixing layer height due to the temperature inversion (Sena, 2003). This effect encompasses the elevation in air temperature as altitude in the atmosphere increases, which prevents lower air masses from rising, then the air density is higher at lower altitudes (Seinfeld and Pandis, 1998; Finlayson-Pitts, 2000). Therefore, in this study, the high eBC concentration peak at 17:00 h, which kept increasing until about 20:00 h, can be explained by the temperature inversion in the Caracas valley at night. In the urban site, PM_2.5 mass concentration on weekdays and weekends exhibited minimal inter-hour changes, with slightly similar behavior to the observed for eBC. This behavior is mainly due to the difference in sources of PM_2.5 and eBC. While eBC comes mainly from fossil fuel combustion, PM_2.5 can be comprised of secondary organic aerosols derived from heterogeneous nighttime reactions (Liu et al., 2020; Liu et al., 2021). Therefore, this can be the most relevant PM_2.5 source at early hours, limiting us to distinguish the relative contribution of fossil fuel combustion.

In the forest site, there was a decrease in PM_2.5 from midnight until 7:00 h. Since the forest site is in a high mountainous area, the PM_2.5 concentrations found early in the morning represent the concentrations of the free troposphere, when the air mass of the city is confined in the valley due to the nocturnal inversion layer at a lower altitude. Therefore, the values measured in early hours at the forest site are considered the regional PM_2.5 background concentrations of the area. By that time, the median concentration was around 3 µg m^–3 for the entire period of measurements. Later, from 7:00 h to 20:00 h, the PM_2.5 concentration showed a progressive increment due to the expansion of the planetary boundary layer and possible transport of aerosol particles from other areas, including the urban site. This behavior is observed during weekdays and weekends, with the difference that the PM_2.5 mean is slightly lower on weekends.

Average concentrations of eBC in the urban site and PM_2.5 in both sites are higher during the dry season (Figure 3). An increase in aerosol mass concentration has been observed before within the mixing layer in different sites of Venezuela during the dry season (December–April) (Guajardo et al., 2010; Hamburger et al., 2013). Also, during the dry season, there is a pronounced reduction of the mixing layer height in the north of South America caused by the North Atlantic Subtropical High-Pressure System, which minimizes the updrafts motion and hence vertical dispersion (Sanhueza et al., 1982). Additionally, biomass burning is higher during the dry season in Venezuela, mainly related to land-use change and agricultural activities, especially in the Venezuelan savannah region (Sanhueza et al., 1987). The effects of these events during the dry season are nationwide, with reported large aerosol concentrations at a high-altitude remote station (4,765 m.a.s.l.) in the lower free troposphere of the Andes region in Venezuela (Hamburger et al., 2013).

In Caracas, during the dry season, mineral dust intrusions are typical (Saturno, 2013), as well as numerous forest fires in the mountains surrounding the valley, for example, Waraira Repano National Park (Morales et al., 1979). These events, together with the orography of the site and the reduction of the mixing layer height, can make the city experience significant pollution events during that period of the year (Guajardo et al., 2010). Therefore, we propose that the increase in aerosol concentration during the dry season results from different factors, including a reduction in the height of the mixing layer, the occurrence of mineral dust plumes, and the rise of biomass burning events.

The impact of traffic emissions on the diurnal cycle of eBC is present all year around. Nevertheless, eBC is also emitted during biomass burning episodes. Considering that the number of forest fires is commonly higher in Venezuela during the dry season, we decided to calculate the relative contribution of biomass burning to atmospheric eBC concentration. Different cases of eBC daily pattern concentrations in the urban site were compared, being these: The biomass burning period, the entire dry season, the dry season without biomass burning, and the rainy season (Figure 7a–d).

The eBC highest concentration peak was observed in the afternoons during the months with more biomass burning events in the dry season (Figure 7a). The median of eBC from 17:00 h to 22:00 h during the biomass burning period (Figure 7a) is higher (3.12 µg m^–3) compared with that found during the dry season without biomass burning (Figure 7c) (2.18 µg m^–3). The difference is directly attributed to the increase in biomass burning events. For the urban site, we estimated that biomass burning increases the eBC mass concentration by 30% during the dry season, indicating the importance of this source during this short period of the year. Factors contributing to this effect are less atmospheric washout and pronounced reduction in the mixed layer height in northern South America accentuated at the end of the day, increasing the air masses’ residence time in the lower atmosphere (Sanhueza et al., 1982).

On the other hand, the difference between the median of eBC from 17:00 h to 22:00 h for the rainy season (1.38 µg m^–3) with the one found during the dry season (2.70 µg m^–3) evidence that atmospheric washout reduces eBC concentration by 49%. Similar results are found when comparing the median eBC during the rainy season with the one observed during the biomass burning period. The first one is lower throughout the day, which shows the efficiency of atmospheric washing during this climatic period (Figure 7b and d). Assuming that the eBC concentration is exclusively from the burning of fossil fuels during the rainy season, the burning of biomass and the lack of atmospheric washout doubles the eBC concentration by 45% during the dry season. Such increment in eBC concentrations attributed to biomass burning could be translated into increments in the radiative forcing and impacts in the local climate, given the long residence time of eBC in the atmosphere.

To determine the influence of biomass burning on the eBC and PM_2.5 concentrations, we looked for correlations with specific fire events during the dry season (shown by the fire pixel counts for main biomass burning events in the north of Venezuela, Figure S4). We then explored 6 specific days (January 05, 2019, February 14, 2019, March 15, 2019, March 25, 2019, April 11, 2019, April 28, 2019), when daily mean concentrations were the largest observed ones in the entire campaign (PM_2.5 > 25 µg m^–3, and eBC > 7 µg m^–3) (Figure 3). Five out of 6 of those specific high concentration events selected showed back-trajectories crossing biomass burning events that featured paths over the urban and forest site. This result indicates that aerosols formed and emitted during such events could be transported to the monitoring sites by N–NE winds, increasing the mass concentration of eBC and PM_2.5 at the measurement sites (Figures 8 and S5).

We found 2 possible sources of eBC for the urban site during the wet season: (1) Nearby sources with relatively low eBC (0–1.1 µg m^–3), such as airplanes leaving and arriving at the Air Base derived from the wind speed (<5 m s^–1) that accompany them (CBPF percentiles 0–50th, Figure 6), and (2) sources from the northwest, southwest, and south, with relatively high eBC (1.1–9.5 µg m^–3) such as the Francisco Fajardo major freeway (CBPF percentiles 50–100^th Figure 6a and b). On the other hand, for low concentrations CBPF percentiles 0–50th during the dry season, the sources of eBC are similar to those observed during the wet season, while for high eBC concentrations (1.5–11 µg m^–3), they are spread between the northwest, southwest, south, and southeast for wind speed above 2 m s^–1 (Figure 6c and d). We analyze the occurrence of these concentrations more in detail and conclude that most of them happen during biomass burning events, discussed previously in Section 4.4.

Regarding PM_2.5 in the urban site during the wet season, the most likely nearby sources are the freeway mentioned above and residential sites for low concentrations (≤13 µg m^–3) (Figure S2a and b). For higher PM_2.5 concentrations (13–80 µg m^–3) and wind speed >3 m s^–1 residential sites located in the south and southeast (Figure S2b). During the dry season, there is a specific contribution of PM_2.5 (0–9.7 µg m^–3) from the north and southeast. Still, for the highest concentrations range (9.7–114 µg m^–3), the sources are spread around the west, southwest, and southeast, where most of the biomass burning events take place (Figure S2c and d, see Section 4.4) and small industries in “La California” and “Petare” sectors are located. As seen for eBC, this is attributed to biomass burning events and is analyzed in Section 4.4.

For the forest site, which is located southwest from the urban site, the prevailing wind directions from north and northwest for all concentration ranges and CBPF percentiles indicate that the major contributor of PM_2.5 is the urban site during both the wet and dry seasons (Figure S3). PM_2.5 transport from the urban site may occur precisely during the first hours of the day when the mixing layer of the valley expands; then, air masses reach the forest site. The moderate contribution of PM_2.5 from sources located in the south of the forest site, with wind speeds in the range of 1–3 m s^–1, would indicate the influence of air masses from closer residential areas such as San Antonio de Los Altos and Los Teques. The highest concentrations observed during the dry season in the forest site, coming from the northeast and south sectors, are attributed to biomass burning events occurring throughout the mountainous area where the forest site is located at and the contribution of biomass burning events around the urban site.

In South America, biomass burning events (forest fires and burning for agricultural purposes) contribute about 20% of global emissions from this source (e.g., Koch et al., 2009; van der Werf et al., 2010; Lee et al., 2018). However, limited information is available regarding the relative contribution of biomass burning during the dry period (November–April) from north of South America, but it is suggested to be large (Poveda et al., 2006; van der Werf et al., 2010; Thornhill et al., 2018).

The large increase of PM_2.5 and eBC average concentration during the dry season in Caracas (7% and 29%, respectively, compared to the annual average) is likely to have local and regional implications on air quality, as found elsewhere (Tsimpidi et al., 2016; Yadav and Devi, 2019). The impact on regional air quality occurs because the aerosols emitted during biomass burning are mainly fine particles that can be transported long distances, causing significant air pollution episodes far from the fires that generated them (Dejchanchaiwong et al., 2020).

A recent study suggests that the deterioration in the air quality during the dry season in densely populated areas of Bogota, Colombia occurs from a rise by up to 80% in aerosol concentration in the fine mode (Rincón-Riveros et al., 2020). Such particles are transported more than 1,000 km away after being emitted during biomass burning events in the Los Llanos region, the largest savannah area across northern South America (occupying western Venezuela and northeastern Colombia). Our study coincides with the time of measurements of Rincón-Riveros et al. (2020) during the dry season (January to April 2019). In their study, the concentrations of biomass burning tracers such as brown carbon, levoglucosan, K⁺, and soluble organic carbon allow them to indicate that medium-range transport of biomass burning events during January 2019 causes an additional aerosol load of 15% ± 6%. Also, the monthly averages of PM_2.5, eBC, and brown carbon during February were 2, 5, and 8 times higher than that found in the month with the lowest concentration (July), respectively.

When comparing these results with ours, we infer that long-range transport of the biomass burning events that occur in the north of South America are likely to impact the air quality on a regional scale. The air masses charged with aerosols from biomass burning events reach the boundary layer, are transported long distances by the northern trade winds and can contribute to the aerosol load in long distance downwind areas. Proofs of this are as follows: (1) Enhancement of PM_2.5 concentrations in the planetary boundary layer during the dry season in the forest site (located at 1,400 m.a.s.l. and influenced by air masses from Caracas), as has been found in Guajardo et al. (2010); (2) eBC concentration nearly doubled during biomass burning events; and (3) biomass burning is responsible for 40% of the PM_2.5 increase (derived by eBC increase in Caracas during the same period).

The long-range transport of the fine aerosols from biomass burning in the north of South America can also increment the regional climate change. These aerosols can reach the Andean region and deposit on the snow surface, contributing to the melting of glaciers (Kang et al., 2020). For the first time for the north of South America, our results support that there is an influence of aerosol transport toward the Andean region through the movement of continental air masses, as has been previously inferred (Thornhill et al., 2018). For now, it is impossible to quantify this effect in the north of the Andean Cordillera. It would require black carbon measurements, snow cover, and runoff monitoring in the Andean region. Further continuous regional aerosol monitoring network and modeling efforts will allow assessing the relative contribution of different sources and the impact on regional downwind air quality.

It is expected an increase by 20%–50% in the number of days likely to cause extreme climate events, with sharper rises in the subtropical Southern Hemisphere (Bowman et al., 2017). Extreme climate events include heat waves, which create perfect conditions (reduction in rainfall and relative humidity, warmer air temperatures, vegetation drought) to increase the intensity and quantity of biomass burning events (Thornhill et al., 2018; Jones et al., 2020).

In recent decades, the proportion of extremely warm days doubled during December–January–February for Caracas (Feron et al., 2019). Therefore, positive climate feedback between heat waves and the occurrence of biomass burning might enhance their magnitude. In Caracas, for instance, an increase in the average temperature during the dry season would, at the same time, increase the probability of biomass burning events. These events emit and produce high concentrations of BC, which causes a positive radiative forcing 2,000 times greater than CO₂ on a molar basis (IPCC, 2021). Therefore, its positive feedback on the short-term regional climate will enhance the local warming, as observed in the Amazon (Liu, 2005).

By the middle of the 21st century, half or more of the days in December, January, and February could be scorching for northern South America if greenhouse gas and aerosol emissions continue increasing (Feron et al., 2019). The increase in the number of warm days would bring a higher number of biomass burning events, causing a rise of more than 2ºC by 2040. The temperature rise would severely impact the global climate and biological diversity. The projections for Caracas indicate an increase of around 46% in extreme events (Feron et al., 2019). Therefore, continuous measurements of black carbon in Caracas and their correlation with the number of fires during the dry season are needed to establish empirical relationships, which should be incorporated in models for regional radiative forcing estimations and future heat waves projection.

Coupled atmosphere-ocean interactions involve irregular, periodic variations in winds and sea surface temperatures during the ENSO, which affects the climate of much of the tropics and subtropics. ENSO modulates temperature and precipitation, impacting the PM variability at local and regional scales. Its effects in many sites are well documented, for example, in Asia and South America (Kim et al., 2013; Wie and Moon, 2017, Cai et al., 2020). Northeast South America and the Amazon are affected by severe drought during ENSO (Cai et al., 2020), which causes marked socio-economic, ecological, and environmental losses. For the period of measurements of this study, 2018–2019, the ENSO international dateline Index (ONI) was positive (0.5–0.9), corresponding to the El Niño phase (National Oceanic and Atmospheric Administration, n.d.). We only have 17 months of PM_2.5 and BC measurements for our study, and to capture a statistically significant ENSO effect in the regional atmospheric aerosol concentrations would require a more extended data set.

Although we cannot analyze the ENSO effect statistically, we infer that the large PM2.5 mean concentrations found in our study for both sites and BC in the urban site during the 2019 dry season related to biomass burning, could be in part caused by El Niño phase in Northeast South America. For instance, the 2015–2016 El Niño led to extreme drought conditions in northern South America, which, combined with elevated temperatures derived from global-warming trends, increased fire incidence by 36% compared with the preceding 12 years and pushed active fire detection beyond the agricultural transition zone (Cai et al., 2020). Consequently, PM_2.5 and BC concentrations are expected to increase proportionally compared to neutral conditions. For years when the ONI phase is neutral or negative, we could expect lower concentrations of PM_2.5 and BC, as has been observed in Colombia, for PM_2.5 concentrations (Arregocés et al., 2021).

Monitoring PM_2.5 can help policymakers reduce impacts on human health and air pollution and has been established as a mandatory air quality standard worldwide. Nevertheless, there are still no regulations for PM_2.5 in Venezuela, and total suspended particles is the only air quality standard in the country.

The PM_2.5 results obtained in both study sites were compared with the most recently updated limit values established by the World Health Organization (WHO) in 2021 (Figure 3). PM_2.5 annual averages from both measurement sites are above WHO’s annual limit value (5 µg m^–3). In addition, the limit value (15 µg m^–3) of the 24-h average for PM_2.5 is exceeded during biomass burning events of the dry season (WHO, 2021), being for urban and forest sites 27 ± 18 µg m^–3, and 28 ± 21 µg m^–3, respectively.

Air quality regulation for PM_2.5 in Venezuela should be established, and this is of great relevance considering that both study sites experience air pollution episodes during the burning season bringing risks to the human health of the population. The PM_2.5 regulation becomes crucial considering that out of the most common diseases and injuries globally, upper respiratory infections have the most significant prevalence, incidence, years of living with disabilities, and high socioeconomical costs (Vos et al., 2016). In the meantime, mitigation strategies to minimize the effects in upper respiratory diseases and the potential asthmatic sequelae of exposure to biomass burning should be implemented during extreme events (D’Amato et al., 2015).

This work shows continuous measurements of PM_2.5 and eBC for the first time in the northernmost area of South America. We assessed the temporal variation and contributing sources of BC and PM_2.5 in Caracas, Venezuela. Evaluation of fire pixel maps with backward air masses trajectories showed that biomass burning events in the dry season is the major source of PM_2.5 and eBC at the urban and forest sites. The transport of air masses with large PM_2.5 and eBC concentrations from biomass burning events contributes to local and regional pollution. It also suggests a positive radiative forcing enhancement and, therefore, contribution to regional warming. The determined eBC concentrations from this study can be used for more accurate radiative forcing estimates from the north of South America and regional climate models.

Since biomass burning impacts the climate and air quality and, hence, human health, mitigation strategies to reduce the number of biomass burning events that have been increasing in recent years are imperative for the region. Also, international PM_2.5 daily limits are exceeded in both study sites during biomass burning episodes in the dry season, highlighting the importance of including PM_2.5 as an air quality standard in the country. PM_2.5 and eBC continuous measurements in Caracas and Altos de Pipe are recommended to develop further studies assessing their interannual variability and provide early warning systems to reduce health impacts for the population. Likewise, measurements of additional pollutants, such as O₃, sulfur dioxide, and nitrogen gases, among other atmospheric variables, would allow investigate the possible effects of natural oscillations in secondary aerosol formation and forecast changes in the local and regional air quality.

All data used in this article are publicly available in (https://owncloud.gwdg.de/index.php/s/cidoBjSXbN8u8xA).

The supplemental files for this article can be found as follows:

Figures S1–S5. Tables S1–S3. docx

VE thanks all the Venezuelan Institute for Scientific Research (IVIC) technical staff and, in particular, Gregorio Maldonado for his assistance during the measurement campaign. We are deeply grateful to the Generalísimo Francisco de Miranda Air Base of the Bolivarian Military Aviation of Venezuela in Caracas for allowing us to place our urban site station in their base. Scientific advice, intellectual discussions, and lending of MAAP photometer from TROPOS institute are highly acknowledged. VE is grateful for the Ph.D. scholarship from IVIC.

This work was supported by the Fondo Nacional de Ciencia, Tecnología e Innovación de Venezuela (FONACIT), Project No 2012-001679, and the European Commission funded Prediction of Air Pollution in Latin America and the Caribbean (PAPILA) project (No 777544).

The authors declare that they have no conflict of interest.

Contributed to conception and design: VE, TP, LD.

Contributed to acquisition and analysis of the data: VE.

Contributed to interpretation of data: VE, TP, LD.

Drafted and revised the article: VE, TP, LD.

Approved the submitted version for publication: VE, TP, LD, TM, AW.

How to cite this article: Engelhardt, V, Pérez, T, Donoso, L, Müller, T, Wiedensohler, A. 2022. Black carbon and particulate matter mass concentrations in the Metropolitan District of Caracas, Venezuela: An assessment of temporal variation and contributing sources. Elementa: Science of the Anthropocene 10(1). DOI: https://doi.org/10.1525/elementa.2022.00024

Domain Editor-in-Chief: Detlev Helmig, Boulder AIR LLC, Boulder, CO, USA

Associate Editor: Balram Ambade, Department of Chemistry, National Institute of Technology, Jamshedpur, India

Knowledge Domain: Atmospheric Science