The global simulations utilize version 12.8 of GEOS-Chem, with a horizontal resolution of 4° × 5° and 47 vertical levels, driven by meteorology fields from the Modern Era Retrospective-analysis for Research and Applications, version 2 (MERRA-2) (Gelaro et al., 2017) (https://gmao.gsfc.nasa.gov/reanalysis/MERRA-2/, last access: 20 February 2024). Global anthropogenic emissions are from EDGARv43 (Crippa et al., 2018), with region-specific updates, including APEI for Canada (Canada's Air Pollutant Emissions Inventory, 2016), NEI (2014) for the USA, MIX for Asia (Li et al., 2017), and DICE_Africa for Africa (Marais and Wiedinmyer, 2016). Biofuel emissions are from Bond et al. (2007), and daily resolved biomass burning emissions are from the Global Fire Emissions Database with small fires (GFED4s) (Giglio et al., 2013; van der Werf et al., 2017; Randerson et al., 2017). These emission inventories yield global organic carbon (OC) emissions of 42.74 Tg and BC emissions of 8.24 Tg in 2019. Of the total OC emissions, 21.17 Tg is from biomass burning, and 6.29 Tg is from biofuel burning. In the simulation, these sources are treated as BrC (see Sect. 2.2 for details), while the remaining OC emissions are modeled as purely scattering aerosols.

In the simulation, 80 % of BC, 50 % of anthropogenic OC, and 50 % of biomass burning OC is emitted as hydrophobic. These hydrophobic aerosols are converted into hydrophilic forms, which are subject to wet removal, with an e-folding timescale of 1.2 d (Park and Jacobb, 2003). BC and organic aerosol (OA) are treated as externally mixed, as described in Wang et al. (2014), with absorption enhancements of 1.1 for fossil BC and 1.5 for biofuel and biomass burning BC. Primary organic aerosol (POA) is diagnosed from simulated primary OC using an OA/OC mass ratio of 2.1 (Wang et al., 2018). Secondary organic aerosols (SOAs) are simulated using the simple scheme (Kim et al., 2015; Pye et al., 2010), with all SOAs being assumed to be hydrophilic. Inorganic sulfate–nitrate–ammonium aerosols are simulated following Shah et al. (2020) and Shao et al. (2019), with thermodynamic equilibrium computed using the ISORROPIA II (Fountoukis and Nenes, 2007). Dust and sea salt aerosols are simulated according to Duncan Fairlie et al. (2007) and Jaeglé et al. (2011), respectively.

To evaluate the radiative impact of BrC, we use the Rapid Radiative Transfer Model for Global Circulation Models (RRTMG), coupled with GEOS-Chem, to calculate both longwave and shortwave radiative fluxes and heating rates in the atmosphere. RRTMG is invoked every 3 h, and the computation of instantaneous atmospheric radiative fluxes is performed at 30 wavelengths between 0.23 to 56 µm in the clear sky (Heald et al., 2014).

Our simulation treats organic aerosols from biomass or biofuel burning as light-absorbing BrC and those from fossil fuel combustion as purely scattering. Although a few field studies, particularly in China, have reported weak light absorption by primary organic aerosols from fossil fuel sources (Yan et al., 2017; Chen et al., 2020), their overall contribution is minor on the global scale and thus is not considered in our simulation.

To capture the change in mass absorption efficiency (MAE) due to chemical processing, organic aerosols from biomass and biofuel burning are modeled as two distinct species, namely fresh BrC, which is strongly absorbing, and bleached BrC, which is weakly absorbing, following the approach by Wang et al. (2018). All freshly emitted OAs from biomass and biofuel burning are specified as fresh BrC. They are subsequently converted into bleached BrC in the atmosphere, with the rate being governed by bleaching lifetime τBrC.

The BrC optical properties are implemented following Wang et al. (2018). The mean absolute errors (MAEs) of fresh BrC at different wavelengths are estimated based on the Mie theory as a function of aerosol size, aerosol density, and the refractive index. The size of OA is assumed to follow a log-normal distribution, with a geometric median diameter of 180 nm and a standard deviation of 1.6 (Wang et al., 2018). The density of OA is set at 1.3 g cm⁻³. The real part of the refractive index is 1.5, while the imaginary part and its wavelength dependence are parameterized based on the mass emission ratios of BC/OA (June et al., 2020). The BC/OA ratios are specified as 0.05 for biomass burning and 0.12 for biofuel, representing average global burning conditions (Wang et al., 2018).

Based on these parameters, Mie calculations yield an MAE of 1.3 m² g⁻¹ at the 365 nm wavelength for fresh BrC from biomass burning and 1.2 m² g⁻¹ for biofuel burning, with corresponding absorption Ångström exponents (AAEs) between 300 and 600 nm of 3.1 and 2.6, respectively. To account for weakened absorption by bleaching, we assume that bleached BrC absorbs only one-quarter as much as fresh BrC, resulting in MAEs of 0.33 and 0.30 m² g⁻¹ for biomass and biofuel burning aerosols, respectively, at 365 nm.

The conversion from fresh to bleached BrC is governed by the chemical lifetime τBrC (referred to as bleaching lifetime hereafter). Wang et al. (2018) parameterized this parameter as inversely proportional to the ambient hydroxyl radical (OH) concentration, specifying τBrC as 5×105OH d, where [OH] is the OH concentration in molec. cm⁻³. This formulation, mainly based on empirical evidence from field measurements in the lower troposphere (Forrister et al., 2015; Wang et al., 2018), yields a τBrC of roughly 1 d throughout the troposphere.

More recent laboratory experiments by Schnitzler et al. (2022) found that the chemical lifetime of water-soluble BrC, generated from smoldering pine wood in the presence of ozone, increases substantially with enhanced particle viscosity when temperature and RH get lower. Although the experimental setup (i.e., using ozone as the oxidant, generating aerosols from pine wood, and focusing on water-soluble BrC) does not fully represent atmospheric conditions, the finding that τBrC is highly sensitive to environmental parameters has critical implications for BrC's radiative and chemical effects.

To quantitatively evaluate these implications, we implement the parameterization of τBrC (in s) developed by Schnitzler et al. (2022) in GEOS-Chem. As shown in Eq. (1), τBrC depends on ambient temperature (T) and relative humidity (RH): 1τBrC=AD0Tη0TηBrCT,RHξ-12, where D0 is the diffusion rate of ozone in pure water, calculated with the Stokes–Einstein equation as a function of T. η0 is the viscosity of pure water as a function of T, and ηBrC is the viscosity of aqueous BrC aerosols, parameterized as a function of T and RH based on the experimental measurements (Schnitzler et al., 2022). A and ξ are treated as constants, where A=0.125 m s^−1∕2 and ξ=0.684. τBrC is calculated and updated online in the GEOS-Chem simulation, driven by T and RH data from MERRA-2.

We use observations from aircraft campaigns (ATom mission, Fig. S1 in the Supplement) and a ground-based remote sensing network (AERONET) to evaluate our simulations of BrC. The Atmospheric Tomography Mission (ATom) is a NASA-led aircraft campaign aimed at measuring and understanding the global distribution of greenhouse gases and reactive gases and aerosols (Thompson et al., 2022). The ATom flights, conducted between 2016 and 2018, sampled extensive latitudinal and longitudinal ranges, from the Arctic to the Antarctic and over both the Pacific and Atlantic oceans (Prather et al., 2017). During the ATom-2, ATom-3, and ATom-4, BrC was sampled at altitudes ranging from 0.2 to 12 km aboard the NASA DC-8 aircraft, alongside comprehensive measurements of various gases and aerosols. Water-soluble BrC (WS BrC) in aerosols was measured with two systems: a spray chamber (MC) coupled to a liquid waveguide capillary cell (LWCC) and a particle-into-liquid sampler (NOAA PILS-LWCC) coupled to an LWCC (Washenfelder et al., 2015). The MC-LWCC system collects aerosols through a spray chamber, trapping the particles in water; these are then analyzed using an LWCC and a spectrometer. The PILS-LWCC system collects aerosols in a liquid stream for similar analysis. The detection limits at 365 nm were 1.53 Mm⁻¹ (MC), 0.89 Mm⁻¹ (CSU), and 0.03 Mm⁻¹ (NOAA), respectively. A factor of 2 is used to convert the light absorption of BrC in the solution into the light absorption of atmospheric particles (Liu et al., 2013; Zhang et al., 2017). In this study, we evaluate our simulations against the measurements taken during the ATom-4, which occurred from 24 April to 21 May 2018 (https://espoarchive.nasa.gov/archive/browse/atom/id14/DC8, last access: 25 October 2024). Samples were collected every 5 min at altitudes below 3 km and every 15 min at higher altitudes, yielding a total of 1074 samples. The flight path of the ATom-4 is shown in Fig. S1.

The AERONET (AErosol RObotic NETwork) program is a federation of ground-based remote sensing aerosol networks established by NASA and LOA-PHOTONS (CNRS) (Holben et al., 2006). The program provides a long-term, continuous, and readily accessible public domain database of aerosol optical, microphysical, and radiative properties. AERONET's aerosol optical depth (AOD) observations are derived from direct solar radiation in multiple wavelength bands, mainly ranging from 340 to 1640 nm, while other aerosol properties (including single-scattering albedo or SSA and absorbing aerosol optical depth or AAOD) are derived from diffuse sky radiation at four wavelengths: 440, 675, 870, and 1020 nm (Dubovik and King, 2000). This study extracted AOD and AAOD at 440 nm from AERONET data in 2019 to compare with the model simulation results. We use high-quality level-2.0 AOD data (uncertainties of 0.01 in the visible range and 0.02 in the ultraviolet range) (Giles et al., 2019) and almucantar inversion AAOD data. To ensure the consistency in the comparison between model simulations and observations, we sample the simulation on days and at locations when and where AOD and AAOD measurements are available.

We perform a series of global 4° × 5° GEOS-Chem simulations for 2019 to examine the impact of environmental-condition-dependent BrC chemical lifetime (τBrC) on the abundance and distribution of BrC globally (Table 1). The baseline simulations (Base and BaFt) use the Wang et al. (2018) τBrC parameterization (bleaching lifetime of roughly 1 d, largely invariant with environmental conditions). The updated simulations (Upd and UpdFt) incorporate the Schnitzler et al. (2022) τBrC parameterization as a function of temperature and RH (Eq. 1). To test the impact of the vertical partitioning of fire emissions and its interactions with τBrC parameterizations, we perform GEOS-Chem simulations that assign 100 % (0 %) (Base and Upd) and 65 % (35 %) (BaFt and UpdFt) of wildfire emissions (all fire-emitted species including BC and BrC) in the boundary layer (free troposphere). The 65 % : 35 % partitioning is based on the averages of aerosol smoke plume heights observed by the Multi-angle Imaging SpectroRadiometer (MISR) (Val Martin et al., 2010) and has been previously applied in GEOS-Chem modeling studies to assess the impact of fire plume heights (Fischer et al., 2014; Jin et al., 2023). In addition to 2019, we also run the April and May 2018 simulation to evaluate against the ATom-4 observations. We also use the GFED5 beta version to test the sensitivity to biomass emission inventories.

Figure 1 shows the spatial distribution of τBrC as parameterized based on Schnitzler et al. (2022). In the planetary boundary layer (<1.5 km), τBrC is roughly 1–10 h, indicating that BrC's light-absorbing capacity rapidly degrades on the timescale of 1 d, generally consistent with ground or lower-troposphere observations (Laskin et al., 2015; Forrister et al., 2015). τBrC increases dramatically with altitude, driven primarily by decreased temperature. τBrC reaches 10^2.5 h after ≈2 weeks in the mid-troposphere at ∼4 km, comparable to the lifetime of aerosols against wet and dry deposition. At higher altitudes, τBrC increases further by several orders of magnitude, rendering photo-bleaching effectively negligible in the upper troposphere, which is supported by a few analyses of high-altitude measurements (Liu et al., 2014; Zhang et al., 2017).

In addition to vertical variations, τBrC also varies considerably across different regions of the globe (Fig. S2a). The horizontal variations at the same altitude appear to be controlled mainly by relative humidity (Fig. S2). At the surface and in the lower troposphere, τBrC tends to be several days longer in dry regions such as northern Africa, Middle Asia, the western US, northern Australia, southern Africa, and southern America. In the free troposphere, τBrC is shorter in the Intertropical Convergence Zone because of the higher humidity due to convective updrafts.

In contrast, the [OH]-dependent parameterization of τBrC implemented in Wang et al. (2018) exhibits much smaller spatial variations, predicting τBrC to be on the order of 1 d throughout the troposphere. The [OH]-dependent parameterization was constructed based on very few field studies, mostly in lower-troposphere conditions (Forrister et al., 2015). The two parameterizations predict similar τBrC in the planetary boundary layer but differ greatly in the middle and upper troposphere. These differences would have important implications for the global distribution and the radiative impact of primary organic aerosols from biomass and biofuel burning.

Figure 2a shows the global distribution of fresh BrC column density in 2019, which reveals elevated concentrations in regions characterized by substantial biomass burning and biofuel emissions (Fig. S3). These hotspots include tropical Africa (biomass), tropical South America (biomass), East Siberia (biomass), South Asia (biofuel), and East China (biofuel). Compared to the baseline simulation (BaFt), the updated simulation (UpdFt) with temperature- and RH-dependent τBrC yields higher fresh BrC concentrations across most regions. The largest increase is observed in southern Africa, which is due to long τBrC near the surface (Fig. 1) resulting from the region's strong biomass burning emissions (Fig. S3) and dry environmental conditions (Fig. S2). Conversely, despite intensive wildfire emissions in East Siberia, the updated simulation shows minimal differences in terms of fresh BrC concentrations because of comparable τBrC values between the two parameterizations (Fig. S2a).

Figure 3 compares the vertical profiles and average column densities of fresh BrC concentrations across simulations. The updated simulation (UpdFt), with revised τBrC and fire injection parameterizations, yields the highest fresh BrC concentrations, with average column densities of 235 µgm-2 in the tropics, 172 µgm-2 in the northern extratropics, and 24 µgm-2 in the southern extratropics. These values represent increases of factors of 4.7, 1.6, and 8.6, respectively, compared to the base simulation. Globally, the updated simulation (UpdFt) produces an average column density of 144.4 µgm-2, about 3 times that of the baseline simulation (base, 53.0 µgm-2). Given the global average emission fluxes of BrC, this result translates into an effective τBrC of 1.45 d in the UpdFt simulation as compared to 0.45 d in the base simulation. Comparison with BaFt and Upd simulations (average column density of 52.9 and 120.7 µgm-2) suggests that the increase in BrC abundance in UpdFt relative to the base is mainly due to the bleaching lifetime update.

Vertical profiles of fresh BrC demonstrate significant differences between the UpdFt and base simulations. The absolute differences are particularly large in the lower troposphere of the tropics, reaching up to 0.05 µgm-3, while relative differences are most pronounced in the upper troposphere, exceeding a factor of 10 despite lower absolute concentrations. Meanwhile, the updated τBrC parameterization does not change the total BrC concentrations (fresh + bleached BrC) (Fig. S4) as the abundance of total BrC is controlled by the rate of wet and dry deposition.

The influence of τBrC parameterization on fresh BrC concentrations is modulated by fire emission injection height. Direct injection of emissions into the free troposphere results in slower BrC bleaching and higher concentrations at elevated altitudes. Comparing results from UpdFt and Upd simulations, we find that, under environmental-condition-dependent τBrC parameterization, injecting 35 % of fire emissions into the free troposphere (UpdFt) (Thapa et al., 2022) leads to 30 %–50 % greater fresh BrC burdens compared to those with emissions confined to the surface layer (Upd) (Fig. 3). In contrast, under the original uniform τBrC, different treatments of fire emission injections (BaFt and Base) affect only the vertical distribution of BrC without influencing its overall atmospheric burden (Fig. 3). These results underscore the interactions between τBrC and fire emission injection heights.

We compare our simulations with the aircraft campaign of the ATom-4, which measured the global distribution of BrC over remote oceans (Fig. S1) (Zeng et al., 2020). The baseline simulation severely underestimates BrC absorption at 365 nm (Abs₃₆₅) (Fig. 4a). The updated simulation with longer τBrC increases fresh BrC concentrations and, hence, their Abs₃₆₅ by about an order of magnitude, bringing the simulation results closer to but still lower than the observations (Fig. 4a). Meanwhile, the model generally captures observed levels of OA and CO during the ATom-4 but with an underestimation of OA in the upper troposphere and an overestimation of CO in the lower troposphere (Fig. 4).

Figures 5 and 6 show the comparison of the annual means of AOD and AAOD simulation results against the AERONET observations. The model, in general, reproduces both the magnitude and the distribution of observed AOD well. The model also captures the high AAOD observed in southern and eastern Asia, where the absorption is mainly due to BC from anthropogenic sources, but underestimates the high AAOD observed in regions with a strong influence of biomass burning, such as Africa, South America, and Siberia (Fig. 5). Both the baseline and updated simulation show good agreement with observed AOD (R=0.88 and slope = 0.79 in BaFt and R=0.89 and slope = 0.80 in UpdFt). The updated UpdFt simulation (R=0.56 and slope = 0.42) improves agreement with AAOD observations relative to the BaFt simulation (R=0.47 and slope = 0.30) but still has notable underestimation (Fig. 6). Meanwhile, at some stations with low SSA values, the simulated SSA is overestimated (Fig. S5).

The underestimation of Abs₃₆₅ against ATom-4 data (Fig. 4) and AAOD against AERONET data (Figs. 5 and 6) may be partly explained by underestimation of biomass burning emissions. We perform additional simulations with the newly released GFED5 fire emission inventory (https://globalfiredata.org/, last access: 10 April 2025), which generally predicts higher biomass burning emissions than the GFED4s inventory. This simulation leads to better alignment with both ATOM-4 Abs₃₆₅ (Fig. 4) and AERONET AAOD observations (Fig. 6). However, the GFED5 simulation also leads to an overestimation of OA in the lower troposphere against ATom-4 data and of AOD against AERONET data (Figs. 4 and 6).

Additionally, there are also considerable uncertainties in the BrC simulation associated with its optical properties. The MAE values applied in different modeling studies vary considerably (Zhang et al., 2020; Jo et al., 2016), and laboratory measurements have also demonstrated source- and season-dependent differences in MAE (Chen et al., 2018; Xie et al., 2020). Moreover, assumptions about particle size distribution and refractive index also contribute to the uncertainties and need to be further constrained using observational and experimental data (Wu et al., 2020; Shamjad et al., 2018). Future improvement of BrC simulations may include refined treatment of these factors.

The presence of BrC enhances the overall absorptivity of aerosols, leading to positive direct radiative effects (DREs) at the top of the atmosphere. Incorporating an improved τBrC parameterization into the model increases the DRE of BrC absorption from +0.059 W m⁻² in the baseline simulation to +0.088 W m⁻² in the UpdFT simulation (Fig. 7). This 48 % enhancement in DRE highlights the sensitivity of BrC's radiative effects to the rate of bleaching. Our estimate of the BrC DRE is within the estimates by previous studies (0.04–0.11 W m⁻²) (Feng et al., 2013; Zhang et al., 2020; Wang et al., 2018). The value is higher than the estimate by Wang et al. (2018) (+0.048 W m⁻²), which did not account for longer τBrC in the cold and dry environment, but is lower than that reported by Zhang et al. (2020) (+0.10 W m⁻²), who made an assumption that convection-lofted BrC does not photo-bleach. The light absorption of BrC reduces the negative DRE of OA by 5.7 % in the base simulation and by 8.4 % in the UpdFT simulation (Fig. 7).

BrC reduces the ultraviolet light available for photochemical reactions, affecting atmospheric chemistry. Figure 8 evaluates the effect of BrC light absorption on the photolysis rate of NO₂ (JNO₂) and concentrations of O₃ and OH radicals. The effects are most pronounced in regions with substantial biomass burning emissions. The annual average of JNO₂ decreases by 0 %–7.4 %, while surface ozone levels decline by 0 %–2.5 %. These results are consistent with but slightly lower than the results of Jo et al. (2016). The light absorption of BrC also results in a 0 %–6.9 % reduction in tropospheric OH concentration, which is also lower than the values reported for the Northern Hemisphere in Jo et al. (2016) (0 %–10 %) and Jiang et al. (2012) (up to 15 %).

The effects of BrC on atmospheric chemistry are more substantial in the season and region with intensive biomass burning. For instance, the reductions in JNO₂ and the surface ozone concentration due to BrC reached 36.3 % and 4 ppb (17.5 %), respectively, in high-latitude Siberia, Russia, in August 2019 (Fig. S6), when a large wildfire event occurred (Konovalov et al., 2021a, b). Another example is the large wildfire over tropical Malaysia in September 2019, which accounted for approximately 40 % of the annual total wildfire emissions. In this case, JNO₂ decreased by about 27.6 %, and surface O₃ concentrations decreased by as much as 6.4 ppb (10.8 %) (Fig. S7).

Figure S8 shows the overall responses of atmospheric chemistry to wildfire emissions, including reactive gases and aerosols. Wildfire emissions increase tropospheric ozone concentrations across the globe while increasing OH concentrations near fire sources and suppressing OH concentrations away from the source. Comparison of Figs. S8 and 8 shows that including BrC light absorption partially offsets the ozone enhancement from wildfire emissions. Moreover, BrC absorption reduces near-source OH enhancement caused by fires while reinforcing OH reduction globally. These results demonstrate that neglecting BrC light absorption in models leads to moderate biases in simulated chemical responses to wildfire emissions.