Enhanced heating rate of black carbon above the planetary boundary layer over megacities in summertime

The fast development of a secondary aerosol layer was observed over megacities in eastern Asia during summertime. Within three hours, from midday to early afternoon, the contribution of secondary aerosols above the planetary boundary layer (PBL) increased by a factor of three to five, and the coatings on black carbon (BC) also increased and enhanced its absorption efficiency by 50%. This tended to result from the intensive actinic flux received above the PBL which promoted photochemical reactions. The absorption of BC could be further amplified by the strong reflection of solar radiation over the cloud top across the PBL. This enhanced heating effect of BC introduced by combined processes (intensive solar radiation, secondary formation and cloud reflection) may considerably increase the temperature inversion above the PBL. This mechanism should be considered when evaluating the radiative impact of BC, especially for polluted regions receiving strong solar radiation.


Background
The absorption of shortwave radiation, and consequential atmospheric heating, by black carbon (BC) has an important impact on the atmospheric radiative balance (Ramanathan andCarmichael 2008, Ding et al 2019). In regions with high BC emissions, these climatic effects may be intensified by strong lower atmosphere heating and surface dimming, which could alter the thermodynamic structure of the planetary boundary layer (PBL) (Babu et al 2002, Ding et al 2016. The absorption efficiency of BC, described as the absorption coefficient per unit mass of refractory BC (rBC), will be enhanced if coated with non-BC materials, through the lensing effect (Liu et al 2017). In addition, the heating effect will depend on the actinic flux incidental on the BC particles, which could be significantly increased at higher altitudes, because less dimming will be caused by aerosol optical depth (AOD) (Norris and Wild 2009).
It has been demonstrated in modelling studies that the absorption capacity of BC depends considerably on the location of the BC layer relative to the cloud layer, e.g. absorption will be significantly enhanced if the BC layer is above the cloud layer due to strong reflection by the cloud top, whereas below the cloud layer the dimming effect will reduce the solar flux deposited on the BC (Nenes et al 2002, Jacobson 2012. The position of the BC layer relative to the cloud could be crucial in Original content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. determining its impact on cloud microphysics by heating at different levels (Johnson et al 2004, Koch andDel Genio 2010). This study presents aircraft in situ measurements, including the full aerosol size distribution and BC size-resolved mixing state, throughout the PBL (containing cloud layers) and lower atmosphere over a megacity in eastern China during summertime. The measurements of vertical profiles were conducted during different times of the day to reveal the diurnal evolution of the heating impact of BC.
2. Measurements and data analysis 2.1. Instrumentation and data processing The aircraft KingAir-350 was deployed (Liu et al 2018) to produce vertical profiles over three successive days in summertime (13-15 July 2018) over Xuzhou megacity in eastern China ( figure 1(a)). Each flight will be referred to by the date (0713, 0714, and 0715) from now on. The meteorological parameters, including ambient pressure, temperature, relative humidity and wind speed/direction, were characterized in situ by an aircraft integrated meteorological measurement system (AIMMS-20, Aventech Research Inc.), which was calibrated on an annual basis. The typical aircraft speed was about 250 km h −1 , and the ascent and descent rates during profiles were ∼2-5 ms −1 . As shown in figure 1(b), the morning (09:00) and midday (11:30) profiles were produced on 0713, while the midday (12:00) and early afternoon (14:00) profiles were on 0714 and 0715. Night flights were also performed on all three days. The profiles covered the time of day receiving the most intensive solar radiation across midday. HYSPLIT backtrajectory analysis (Draxler and Hess 1998) (figure 1(a)) using 1°×1°, three-hourly GDAS1 reanalysis meteorology, was performed to track the air mass histories for all profiles. Two  (Zhao et al 2019) only to support the phenomenon observed here, but the detailed radiative transfer calculation is not performed.
A wing-mounted passive cavity aerosol spectrometer probe (PCASP-100X, DMT Inc., USA) was used to measure the particle size distribution at diameter0.12-2.5 μm, at a time resolution of 1 s. A wired heater on top of the inlet, and the dry sheath flow, ensured that the particles measured by the PCASP were in a dry state, with relative humidity (RH)<40%. In addition, the aerosol inlet of the aircraft included a silicate dryer, so aerosol measurements inside the cabin were also dry. The particulate matter (PM1) used in this study is derived from the PCASP optically measured size distribution by assuming an average density of 1.5 g m −3 (Cross et al 2007). The Aitken and accumulation mode particles at diameter6-520 nm were measured by an engine exhaust particle sizer (TSI Inc.) with a time resolution of 1 s. The aerosol scattering cross section (C sca ) and asymmetry parameter (g) for each size bin is calculated based on the PCASP-measured size distribution assuming a refractive index of 1.50+0i. A fast cloud droplet probe (FCDP, SPEC Inc.) (O'Connor et al 2008) was used to measure the droplet size distribution at ambient RH. All of the aerosol data in cloud were screened out, based on the FCDP-measured liquid water content (LWC) >0.001 g m −3 , but the LWC is used to indicate the location of cloud layers.
The physical properties of BC were characterized by a single particle soot photometer (SP2, DMT Inc.) (Schwarz et al 2006, Zhao et al 2015. The SP2 is able to measure the rBC mass and associated coating for each rBC-containing particle. The BC core size is measured at 0.05-0.45 μm and the remaining mass outside of the detectable range is obtained by a lognormal extrapolation (supplementary figure S4, available online at stacks.iop.org/ERL/14/124003/mmedia). As the actual coating thickness depends on both the rBC core and the coated BC size, a metric of coating information in bulk, described as a mass ratio of coating/rBC, is used to represent the overall coating status of the particle ensemble during a given time period (Liu et al 2014). Recent work shows this metric is able to represent the total mass of coatings associated with BC (Ting et al 2018). The absorption cross section (C abs , in m 2 ) or C abs normalized by rBC mass (MAC, in m 2 g −1 ) could be calculated based on the measured rBC core sizeresolved mixing state (an example is given in figure S10 (b)) via the core-shell mixing rule using a BC refractive index of 1.95-0.79i (Bond and Bergstrom 2006) and a coating refractive index of 1.50-0i (Liu et al 2015). Figure S5 shows an example of all size distributions measured on 0714. The scattering coefficient ( sca s ) is obtained by integrating the number concentration (N(D)) and C sca for all PCASP bins; and the absorption coefficient ( abs s ) is the integration of C abs and BC number concentration N(D c ) for all SP2 BC core size bins (D c up to 0.6 μm will include >95% of the total rBC mass in this study). The sum of both gives the extinction coefficient, as expressed in equation (1): This calculation is performed for every 200 m altitude bin using the mean PCASP size distribution. The single scattering albedo (SSA= sca ext s s / ) and asymmetry parameter (g) are also obtained for each altitude bin. The AOD for each altitude bin (h) is obtained from the altitude-integrated h , ext s ( ) as expressed in equation (2): The AOD, SSA and g as a function of altitude (figure S8) serve as inputs for the radiative transfer calculation given below.
Micro-pulse lidars at 532 nm (MPL-4B, Sigmaspace Co., USA), were located at Huaian and Hefei (marked by black dots in figure 1(a)) to monitor the temporal evolution of the aerosol layer. A wind profile radar (Airda-3000, Airda Co., China) was located close to Xuzhou (34.402°N, 118.017°E) to measure wind profiles.

Calculation of BC absorption and heating rate
The actinic flux spectrum (λ=250-2550 nm) was calculated using the pseudo-spherical version of the Discrete Ordinates Radiative Transfer Code (DIS-ORT), as implemented in the libRadtran software package (Emde et al 2016). In this study, the AOD, SSA and g used are the in situ-measured parameters based on the PCASP and SP2 measurements (see above) and calculated at each λ. The λ-dependent AOD is derived from the calculation based on the PCASP measurement, which is used as an input, and is expressed as (870) is the ratio of the AOD at specified λ over that at λ=870 nm.
The parametrization of the effect of cloud on actinic flux is, according to Hu and Stamnes (1993), to convert the cloud's microphysical properties to optical properties. The inputs used are the in situ-measured vertical profiles of LWC, with cloud cover set as 0.15 according to the aircraft camera (figure S5). For details of the settings for the radiative transfer calculation, refer to table S1.
The absorption power of BC is then calculated as the actinic flux multiplied by absorption coefficient integrated over all λ and BC core sizes, expressed in equation (4): where σ abs,rBC is the BC mass absorption cross section (in m 2 g −1 ), which is a function of incident λ and BC core size (D c ), M rBC is the rBC mass concentration at each D c bin (in μg m −3 ), multiplying both to obtain the absorption coefficient of rBC (in M m −1 ), and the actinic flux (F ac , in mW m −2 ) is calculated from the DISORT radiative transfer module. Integrating over all wavelengths (λ=250-2550 nm) and D c range (50-800 nm) gives the BC absorption power per unit volume of air (in mW m −3 ). The absorption power deposition efficiency (P eff ) is P abs normalized by rBC mass, in mW μg -1 .

Results
Figure 2 schematically shows the mechanism this study will illustrate, which is the enhanced heating rate of BC above the PBL, resulting from the combined effects of enhanced secondary formation, BC coatings and cloud reflection on actinic flux at this layer. These are discussed in detail in the following sections. For guidance, besides the main figures, supplementary figures S1 and S2 show the temporal evolution of wind profiles and aerosol extinction respectively; figure S3 shows medium resolution imaging spectrometer (MODIS) cloud and AOD images; figure S4 shows the measured typical size distribution; and figures S5 and S6 show the vertical profiles of the meteorological parameters particle number concentrations and BCrelated properties, respectively.

Meteorology
The flights from 13-15 July followed very nearly the same route ( figure 1(a)). The flight region was about 200 km away from the coast of the East China Sea and is influenced by sea-land breezes in summertime. The top of the PBL could be determined by the aircraft in situ-measured temperature inversion and stable potential temperature (figure S5), with the dashed lines showing the height of the PBL (PBLH). Radar wind profiles (figure S1) showed diurnal variation of wind shear which also reflected the PBLH. In the PBL, oceanic easterly air flow dominated, and above the PBL, continental southwesterly air mass. The PBLH increased from ∼0.5 km to ∼1.1-1.5 km from morning to early afternoon due to stronger convective mixing through daytime surface heating. At night the height of the wind shear top was significantly lowered, to ∼200 m, consistent with the aircraft in situ-measured shallow temperature inversion for the postsunset flight ( figure S5). The high pressure centred over the East China Sea, evident in the 700 hPa geopotential height (figure 1(c)), and led to southwesterly continental transport to the flight location. During all three flight days the synoptic conditions maintained a similar pattern. Backtrajectory analysis ( figure 1(a)) showed that the measured air masses were transported about 50 km (0.2°in latitude) from the south to the flight region in 3 h. The region within this distance was controlled by a similar synoptic system ( figure 1(c)). This means that the air masses observed at different times of the day (in 3 h transport) could be generally deemed to have similar air mass origin and regional influence. The Beijing winter and summer campaigns also chose flights without important regional transport or shift of sources, e.g. the variation of rBC mass loading was less than 20% in the lower free troposphere (FT) (figure S7), and the variation of rBC mass in the PBL was due to daytime boundary layer development when some rBC mass from ground sources could be transported upwards to higher levels. The atmospheric processing is thus considered to be mainly at the local scale for the results here. Persistent cloud layers were observed on 0714 and 0715 during the Xuzhou campaign, principally thin layers of cumulus humilis, with cloud coverage of about 15%-25% according to the aircraft camera and MODIS visible cloud images (figure S3). The FCDPmeasured LWC (figure S5) indicates the location of cloud layers. The presence of these layers may partly reduce the visibility of MODIS AOD data on 0714 and 0715 (figure S3), whereas on 0713 the cloud was not as intense and thus AOD data are fully visible.
3.2. Enhanced heating rate of BC above the PBL in early afternoon Figure S6 shows the vertical profiles of aerosol number concentration for both Aitken and accumulation mode particles, PM1 mass concentration, and effective diameter (D eff ) at 120-800 nm. Note that on 0713 the daytime profiles were from morning (09:00) to midday (11:30), whereas on 0714 and 0715 they were from midday (12:00) to early afternoon (14:30). The development of the PBL led to the uplifting of aerosol from the surface, coming with some dilution effect for certain days, i.e. reduced surface concentration. The residue layer was observed at night above the shallow inversion layer (not on 0715 as no inversion was observed). On 0714 and 0715 there were notable increases in PM1 by a factor of 3-5 above the PBL ( figure S4(c)), which occurred in 2-3 h from midday to early afternoon. This came with a decrease of Aitken mode particle number concentration (figure S6(a)) and considerably increased particle size (figure S6(d)) by 20%. This increase of particulate mass was less pronounced on 0713 when profiles were conducted from morning to midday. This phenomenon was further validated by two lidar measurements away from the flight area with distance of about 170 km and 270 km respectively. As figure S2 showed, the extinction profile from lidar measurements at both locations featured a fast-developed PBL from 12:00 to 15:00, and in addition an aerosol layer formed above the PBL during this time. This wide spatial consistency confirmed the regional nature of this phenomenon. Given this growth occurred during the period when solar radiation is most intense, it is inferred that aerosol growth within this layer may be driven by photochemical processing of gaseous precursors. Figure 3(a) showed that there was no notable variation of rBC mass above the PBL from 12:00 to 14:30 on both 0714 and 0715. The change in total particulate mass (PM1) normalized by the rBC mass could broadly reflect the formation of secondary aerosol mass, because rBC is always primary while the addition of extra PM1 mass will be mainly controlled by secondary formation (assuming the variation in the relative emission factors of other species is not significant in this relatively short experimental period). The consistent PM1/BC ratio in the PBL (figure 3(b)) at different times of the day suggested well-mixed primary and secondary sources, whereas a remarkable increase of PM1/rBC by a factor of three to five occurred above the PBL from midday to early afternoon, and this was also consistent with the lidar-measured extinction across the region ( figure S2).
The coatings associated with rBC, indicated by the coating/rBC mass ratio (figure 3(c)), also increased by a factor of three to five, similar to PM1/rBC, from midday to early afternoon. The absorption efficiency of BC and the absorption enhancement relative to uncoated BC increased from 5% to 50% ( figure 3(d)). Both 0714 and 0715 showed consistent results, whereas on 0713 this enhancement was not observed (figure S7) as the measurements were from morning to midday. The PM1/rBC ratio and the coatings of BC above the PBL at night decreased compared to those in the early afternoon, consistent with the lidar extinction, and this in turn suggested the importance of solar radiation in the formation of secondary aerosol. There was an increase of RH from 60% to 70% on 0714 between the midday and early afternoon profiles, which was most likely to be from moisture uplift through convective mixing as there was no obvious wind profile (figure 1) or air mass shift (according to backtrajectory analysis). This may cause a more significant increase of PM1 and particle size (figure S6) compared to that on 0715, because more water molecules could also promote photochemical reactions and allow more semi-volatile species to condense (Donahue et al 2006). There was no obvious variation of RH on 0715 but there was still significant enhancement of secondary species, for which the sole photochemical reaction tended to dominate. There was no solar irradiance increase from morning to midday on 0713 and thus no obvious secondary species formed. The results over Beijing in summer ((Zhao et al 2019), figure 2) also confirmed the strong enhancement of BC coatings above the PBL in 2-3 h evolution time, by a factor of two to three, occurring in the early afternoon. However, this was at a lower scale compared to Xuzhou, and this may result from a drier air mass (RH<60%, figure S5) in the FT for northern cities. In the Beijing winter measurements this enhancement was significantly reduced, with coating enhancement increasing by less than a factor of 1.5, or even decreasing, from midday to early afternoon (figure 4), which may be due to the reduced solar radiation and enhanced AOD dimming effect in winter (given the winter flights were conducted during a heavily polluted period with surface rBC mass loading >4 μg m −3 ). Summer mountain valley breezes in Beijing can lift surface pollutants above the PBL (Chen et al 2009). A recent study conducted over Korea (Lamb et al 2018) also indicated some enhanced coating thickness for BC above the PBL (in figure 4(b) at about 800 hPa compared to 100 hPa), whereas the mixing state of BC at higher altitude was more influenced by synoptic conditions. The enhanced coating of BC above the boundary layer therefore tends to be a general phenomenon for sites where intensive solar radiation is received above the PBL.
The observations here showed strong enhancements for secondary formation, BC coatings and absorption above the PBL in the hours with most intense solar radiation, however these enhancements were less pronounced in the PBL. This could be caused  by strong photochemical activities above the PBL, while the increased AOD may have a significant optical shielding effect in the PBL (Streets et al 2006, Prabha andHoogenboom 2009). Radiative transfer calculations (figure S9) show that direct solar irradiance was reduced, especially within the PBL, by adding the measured aerosol loadings, whereas the downward diffuse irradiance was enhanced above the PBL due to increased particle size. The overall actinic flux thus showed significant enhancement above the PBL, as compared to within the PBL, due to the aerosol loading (especially for the early afternoon on 0714 and 0715). As figure 3(b) shows, the absolute absorbing power of BC was largely determined by the rBC mass loading, with heating rate 0.3-0.5 K d −1 in the PBL, while 0.1-0.18 K d −1 above the PBL. The power deposition efficiency (P eff , as normalized by rBC mass) depended on the absorption efficiency of BC (MAC). In the PBL, the clear-sky P eff of 6-8 mW μg −1 was broadly within the range of that measured in a North American city (7±2.5 mW μg −1 ) (Schwarz et al 2009). Corresponding with the increase of BC coatings from midday to early afternoon, the P eff showed an enhancement of up to 30% from ∼7.0 to 9.5 mW μg −1 above the PBL from midday to early afternoon.
The presence of cloud layers above the PBL further altered the irradiance, i.e. enhancing the dimming at lower level but increasing the reflectance above the cloud layer ( figure 5(a)). The thin cloud layer in this study was mainly cumulus humilis with 15%-25% cloud cover and LWC 0.1-0.3 g m −3 . The cloud layer enhanced the F ac above the PBL by 30% but weakened the F ac below the PBL by 15%. This study finds the P eff above the PBL will be further enhanced by 10% if cloud coverage of 15% is considered ( figure 5(c)). Note that the cloud fraction is only an approximate estimate here but the F ac will be further amplified if there is more cloud coverage. The absorption enhancement due to cloud reflection was previously studied for the BC above oceanic stratocumulus (Johnson et al 2004) but this study provides the direct evidence. It should also be noted that the patchy nature of the cloud layer (figures S3 and S5) may have allowed aerosols or precursors to penetrate the cloud layer, forming a BC layer with significant secondary coatings above the cloud layer, which may explain the results of this study.

Discussion and conclusions
In this study, the fast formation of a secondary aerosol layer was observed during summertime: within the 3 h from midday to early afternoon, the contribution of secondary aerosols above the planetary boundary layer (PBL) increased by a factor of three to five. This is likely to be due to the higher rates of photochemical processing at these altitudes, which is suppressed in the PBL due to dimming caused by the high AOD. The secondary species formed by this processing will condense on the BC and increase its coating content, leading to an enhancement of absorption efficiency by 50%. Consequently, the absorbing power deposited on the BC will be enhanced by the combined effects of increased coatings and solar flux. These processes are schematically illustrated in figure 2. The results here are consistent with the chamber simulation study by (Peng et al 2016) that a BC E abs of ∼50% occurred in 2-3 h ageing time under pollution conditions. In addition, the solar flux received above the PBL as in this study may be more intense than that on the ground because of the lower AOD dimming effect in addition to the cloud reflection above the PBL.
Cloud layers regularly form on top of the PBL in this region, and strong solar reflection by the cloud top will significantly increase the actinic flux received by the BC above the cloud layer, further amplifying the amount of solar radiation absorbed by the BC. Given the strong solar radiation in summertime, all of these processes will occur in a short time scale. This strong heating effect of BC introduced by combining processes (intense solar radiation, secondary coatings and cloud reflection) would considerably increase the temperature above the PBL, which may introduce feedback effects and accumulate more pollutants across this layer, further promoting the secondary formation. A previous study found the absorbing aerosols above the cloud may stabilize the underlying layer and tend to enhance the cloud coverage below (Brioude et al 2009), which may in turn enhance this feedback. The mechanism raised in this study should be considered when evaluating the BC heating effect in polluted regions rich in BC and precursors, especially in summertime when solar radiation is strong. Further chemical measurements in gas and aerosol phases are also needed to elucidate the complex interactions above the PBL.