Strong Aerosol Absorption and Its Radiative Effects in Lhasa on the Tibetan Plateau

Knowledge of aerosol radiative effects in the Tibetan Plateau (TP) is limited due to the lack of reliable aerosol optical properties, especially the single scattering albedo (SSA). We firstly reported in situ measurement of SSA in Lhasa using a cavity enhanced albedometer (CEA) at λ = 532 nm from 22nd May to 11th June 2021. Unexpected strong aerosol absorbing ability was observed with an average SSA of 0.69. Based on spectral absorptions measured by Aethalometer (AE33), black carbon (BC) was found to be the dominated absorbing species, accounting for about 83% at λ = 370 nm, followed by primary and secondary brown carbon (BrCpri and BrCsec). The average direct aerosol radiative forcing at the top of atmosphere (DARFTOA) was 2.83 W/m2, indicating aerosol warming effect on the Earth‐atmosphere system. Even though aerosol loading is low, aerosol heating effect plays a significant role on TP warming due to strong absorbing ability.


Introduction
The Tibetan Plateau (TP) is the third pole with high altitude and strong radiation, located in the west of China.Over the past 50 years, there has been a growing concern that the warming rate over TP is twice more than the global average value (Kang et al., 2016;Mu et al., 2020;You et al., 2021).TP warming impacts not only local ecosystem but also regional or global atmospheric and climate changes (Huang et al., 2023;Wang et al., 2008).Moreover, there are large spatial heterogeneities in the air temperature increase in TP (Mu et al., 2020), adding complexity to figure out the driving mechanisms.Previous studies attributed accelerated TP warming to cloudradiation feedback (Duan & Xiao, 2015), stratospheric ozone depletion (Guo & Wang, 2012), CO 2 (Chen et al., 2002) or water vapor (Rangwala et al., 2009) increasing concentrations.However, the reasons of TP warming have not reached an agreement by now.
Aerosol has significant impacts on radiation balance of the Earth-atmosphere system by scattering and absorption of solar radiation (Chylek & Wong, 1995;Gui et al., 2022;Li et al., 2022;Ramanathan et al., 2005).The aerosol scattering or absorbing ability is represented by single scattering albedo (SSA).Stronger aerosol scattering ability (with larger SSA) can enhance reflected solar radiation from the Earth causing cooling effect, while aerosol absorbing (with lower SSA) leads to positive net radiation and contributes to warming effect.Among 31 sites connected to the Global Atmosphere Watch (GAW) network, the SSA mean values based on in situ measurement ranged from 0.8 to 1, with only six sites lower than 0.9 (Laj et al., 2020).SSA is the most sensitive factor to aerosol radiative forcing calculation compared with other optical parameters (Loeb & Su, 2010;Lu et al., 2022;McComiskey et al., 2008).Inorganic aerosols, like nitrate and sulfate, are treated as pure scattering aerosol.The absorbing aerosols include black carbon (BC), brown carbon (BrC) and dust.BC is the primary pollutant from incomplete combustion processes (Bond et al., 2013), while BrC can be produced from primary emission (BrC pri ) (Saleh et al., 2014) and secondary chemical reactions (BrC sec ) (Lambe et al., 2013).Deriving absorptions by different aerosols is helpful to identify the potential impactors, which could provide scientific basis for policy makers to balance different aerosol emissions.
Remote sensing observations showed that aerosol optical depth (AOD) in TP could be 10 times lower compared with sites in eastern China (de Leeuw et al., 2018;Li, Li, et al., 2020;Li, Xiao, et al., 2020;Lu et al., 2022).Previous studies evaluated that aerosol over TP had weak impact on radiative forcing based on ground based remote sensing observation, as it was less efficient to warm the atmosphere due to the low aerosol loading (Li, Li, et al., 2020;Li, Xiao, et al., 2020).However, aerosol radiative effects over TP still have large uncertainties because of inaccurate knowledge of aerosol optical properties, especially SSA.For example, quality-assured SSA retrieved from sun photometer is only available for AOD larger than 0.4 at 440 nm (Dubovik et al., 2000;Dubovik & King, 2000;Holben et al., 2006), while AOD is usually lower than 0.2 over TP (Zhu et al., 2019).Moreover, SSA retrieved from satellite observations are missing over TP (Ramachandran et al., 2020), and the retrieved SSA values depend highly on the retrieval algorithms (Devi & Satheesh, 2022).The lack of reliable SSA observed or retrieved over TP leads to large uncertainties to assess aerosol radiative forcing in this region.
Lhasa is the capital city of Tibet province, with the largest population (around 460,000 inhabitants).Zhu et al. (2019) reported that the yearly average AOD at 440 nm was lower than 0.2 in Lhasa.Apart from the low aerosol content, aerosol composition in Lhasa exhibits distinctive characteristics.Based on in situ measurements, Zhao et al. (2022) found lower concentration of secondary inorganic aerosols (by a factor of 2-5 times) in comparison to similar cities in eastern China (Huang et al., 2014;Sun et al., 2015), with nitrate and sulfate fractions of 5.4% and 3.4%, respectively.Zhu et al. (2017) revealed that Lhasa had low absorption Ångström exponent (AAE = 1.04) over the whole measured spectral range (370-950 nm), with large contribution of black carbon (>90%) and low contribution of brown carbon and dust.By now, the reliable SSA hasn't been reported in Lhasa, and absorption by BrC sec is still unknown.
In this work, in situ measurements were performed in Lhasa to obtain multiple aerosol optical parameters combining the cavity enhanced albedometer (CEA) and Aethalometer (AE33), including aerosol extinction, scattering, absorption coefficients and SSA at λ = 532 nm, as well as absorption coefficients at multiple wavelengths.The absorption contributions by different aerosols were distinguished to further identify the main compounds impacting on aerosol absorption in this city.Clear-sky direct aerosol radiative forcing (DARF) was then evaluated based on measured aerosol optical properties by using the Santa Barbara Discrete Ordinate Atmospheric Radiative Transfer (SBDART) model.Positive aerosol radiative forcing at the top of atmosphere was revealed due to strong absorption during the campaign, indicating that aerosol might play a significant role on TP warming.

Observation Site
Lhasa is located in the southern TP and characterized with its river valley terrain (Figure S1 in Supporting Information S1), surrounded by high mountains (over 5,500 m a.s.l), which prevents aerosol pollutants from dispersion.Aerosols in Lhasa are mainly issued from local anthropogenic activities, like traffic emissions, religious activities, coal combustion (Zhao et al., 2022), as well as long-range transport from South Asia (Chen et al., 2022).The field campaign was conducted in the Institute of Tibetan Plateau Research (29.63°N, 91.02°E, 3,640 m a.s.l.) from 22nd May to 11th June 2021.The observation site is located in the western part of urban Lhasa, and it is closed to Jinzhu Road, which is a major road in Lhasa with heavy traffic.

Aerosol Optical Properties and Meteorological Factors Measurements
CEA measures aerosol extinction and scattering coefficients at λ = 532 nm in the same sample volume simultaneously (Xu et al., 2018;Zhao et al., 2014;Zhou et al., 2020).Aerosol absorption coefficient and SSA can be Geophysical Research Letters 10.1029/2023GL107833 obtained from the difference and ratio of both measurements.The uncertainties of extinction, scattering and absorption coefficients and SSA are 3.3%, 3%, 4.5%, and 5.5%, respectively (Xu et al., 2018).AE33 measures aerosol absorptions at seven wavelengths (λ = 370, 470, 520, 590, 660, 880, and 950 nm) by using filter-based method (Drinovec et al., 2015).The AAE is expressed as the absorption dependence on wavelength, and it can be obtained with absorption coefficients at two wavelengths (λ and λ 0 ) as shown in Equation 1.Using spectral variation of absorption coefficients, AAE was calculated as the slope of least-square fit of the logarithm of Abs(λ) versus the logarithm of seven wavelengths in this study (Helin et al., 2021).During the campaign, ambient aerosol particles passed through a PM 2.5 cyclone inlet and a Nafion dryer upstream the CEA and AE33 to measure dry PM 2.5 optical properties.The flowrates were set at 1.3 and 3 L min 1 for CEA and AE33, respectively.To get the absorption coefficient correction factor measured by AE33 due to multiple scattering, aerosol absorption at λ = 532 nm determined by CEA was used as the reference (Li, Li, et al., 2020;Li, Xiao, et al., 2020).Since multiple scattering correction factor is wavelength-independent (Saleh et al., 2014;Zhao et al., 2020), the correction factor (2.62 for this study, Figure S2 in Supporting Information S1) could be also used for other wavelengths to obtain reasonable aerosol absorptions.The Cimel sun photometer (CE318) installed in Lhasa (29.6°N, 91.2°E) was used to obtain AOD at λ = 380, 440, 500, 670, 870, and 1,020 nm during the daytime.
Apart from aerosol optical properties, meteorological factors were measured during the campaign, including temperature and relative humility (RH) measured by meteorological sensor (HMP155A, Vaisala), wind speed and direction measured by a sensor (010C-1/020C-1, Metone), as well as upward and downward shortwave radiation observed by radiometer (300-2,800 nm for shortwave, CNR4, Kipp & Zonen).The date and time reported in this study are using China standard time (CST: UTC + 8 h).

Identification of BC, BrC pri , and BrC sec
Previous study revealed Lhasa was rarely impacted by dust from May to September, as the observed PM 10 concentration and crustal elements (Na, Mg, Al, Ca, etc.) were significantly lower during this period (Cong et al., 2011).Ignoring the impact by dust, BC and BrC are treated as the only absorbers within the PM 2.5 fraction during the campaign.Primary BC and BrC are mainly emitted from combustion processes.The minimum Rsquared (MRS) method was used to differentiate BrC sec from primary BC and BrC (Wang et al., 2019;Wu et al., 2018;Zhu et al., 2021).First, the absorption coefficient by primary aerosols is estimated using the following equation: Where [BC] is the mass concentration of BC, and (Abs(λ)/BC) pri can be assumed to a constant value for a specific season at specific region (ranging from 0 to 100 with step of 0.01).BrC sec absorption is then calculated by the total absorption subtracting absorptions by primary BC and BrC (Equation 3) assuming different values of (Abs(λ)/ BC) pri .The optimal (Abs(λ)/BC) pri is determined when absorptions of calculated BrC sec and [BC] have minimum correlation coefficient (R 2 ).
Considering that the AAE of BC (AAE BC ) differs with BC size distribution (Li, Liu, et al., 2016;Li, Chen, et al., 2016;Wang et al., 2016), assuming AAE BC of 1 may bring larger uncertainties.In this study, AAE BC was assumed to be equal to AAE calculated at λ = 880 and 950 nm (AAE 880-950, Equation 4).Because BC is usually treated as the only absorbing aerosol at both wavelengths and previous experiment proved that AAE 880-950 values were almost the same before and after removing other absorbing aerosols and the coating (Zhang et al., 2019).Therefore, absorption of BrC pri can be derived by subtracting BC absorption with previously calculated absorptions of primary aerosol (Equation 5).

Calculation of Aerosol Radiative Effects
The net radiation flux is the difference between downward (F ↓ ) and upward (F ↑ ) radiation flux (Equation 6).Direct aerosol radiative forcing at the top of atmosphere (DARF TOA ) or the bottom (DARF BOT ) is defined as the net change in radiative flux caused by aerosols in the atmosphere (Equations 7 and 8).
The difference between DARF TOA and DARF BOT is calculated as the direct aerosol radiative forcing in the atmosphere (DARF ATM ).Radiation flux was calculated with SBDART model (Ricchiazzi et al., 1998), which was widely used to evaluate aerosol radiative effects using column integrated (Lu et al., 2022) or vertically resolved optical properties (Su et al., 2020) or a combination of those with surface measurements (Kuang et al., 2015).The input optical parameters to SBDART included AOD and Extinction Ångstrom Exponent (EAE) from CE318, SSA measured by CEA, and surface albedo calculated as the ratio of measured upward to downward shortwave radiation.Previous study (Lu et al., 2022) reported that DARF was less sensitive to asymmetric factor compared with AOD and SSA (Loeb & Su, 2010).This study used the average asymmetric factor of 0.71 reported at Namco site, which is close to Lhasa (Lu et al., 2022).

Variation of Meteorological Factors and Aerosol Optical Properties
Variations of shortwave radiation and meteorological factors are given in Figures 1a-1c.The peak value of downward shortwave radiation during daytime was around 1400 W/m 2 , which is twice higher than sites in Yangtze River Delta (YRD) of China (Duan et al., 2022).The average temperature was 17.6 ± 4.2°C during the campaign.RH had an average value of 40.1 ± 17.5% and was usually below 50% (taking about 79%), so aerosol hygroscopic growth effect can be ignored for most cases and the measured dry aerosol optical properties can represent the result in the real atmosphere (Tao et al., 2014;Yan et al., 2009).Episodically, RH could reach values larger than 80%, but it only occurred during nighttime.Wind speed was usually lower than 3 m/s with an average value of 1.1 ± 0.8 m/s, and the most frequent (46%) wind direction was southwest with high wind speed (around 2 m/s).
Aerosol optical properties variations during the campaign are shown in Figures 1d and 1e.It could be found that aerosol loading was generally low in Lhasa with average extinction, scattering and absorption coefficients of 13.6 ± 10.7, 9.3 ± 7.2 and 4.3 ± 4.0 Mm 1 , respectively.Extinction coefficient was usually lower than 20 Mm 1 (for about 86% of the period).There was no obvious regional pollution episode during the campaign except for some peaks (extinction coefficient peaked up to 60 Mm 1 ) caused by local emission starting from 21:00 and lasting for 1-2 hr (Figure S3 in Supporting Information S1).The larger values of aerosol extinction coefficients correspond to the lower values of SSA, meaning large influence of local emission on SSA.These late emissions might be related to traffic emissions due to heavy trucks circulation forbidden during daytime in Lhasa.
Furthermore, AOD at 440 nm was lower than 0.2 during the observation, with an average value of 0.08 ± 0.02.SSA ranged from 0.44 to 0.9 with an average value of 0.69 ± 0.09, which is lower than SSA values highlighted by the GAW network (Laj et al., 2020).The low SSA over Lhasa might be due to extremely low contribution from secondary inorganic aerosols (Zhao et al., 2022).In addition, regional transport from South Asia with absorbing aerosols might contribute to low SSA values.As shown in Figure 1 and Figure S4 in Supporting Information S1, lower SSA was observed with an average value of 0.61 ± 0.07 from 20:00 on 25th to 1:00 on 30th May 2021, and air masses during this period were coming from South Asia passing over regions with dense fire spots due to biomass burning (Liu et al., 2021).During observation period, the average AAE was 1.07 ± 0.08, which is close to the AAE of aerosol emissions from fossil fuel (Sandradewi et al., 2008).Table 1 is the list of aerosol extinction coefficients and SSA values among sites in TP, eastern China and Europe.Observation sites located in TP are distinguished with lower extinction coefficients and lower SSA values.

Identification of BC, BrC pri , and BrC sec
In order to explain what causes the low SSA observed in Lhasa, aerosol absorptions of BC, BrC pri and BrC sec were distinguished with MRS and AAE methods.As shown in Figure 2a, BC was the most important absorbing compound with an average absorption coefficient of 5.6 ± 2.5 Mm 1 at λ = 370 nm, followed by BrC pri and BrC sec with average absorption coefficients of 0.71 and 0.45 Mm 1 , separately.The absorption contributions by BC, BrC pri and BrC sec were 83%, 11% and 6% at 370 nm, respectively.Previous study (Li, Liu, et al., 2016;Li, Chen, et al., 2016) performed measurement over the same site highlighted that PM 2.5 in Lhasa was mainly impacted by primary emission and less from secondary formation based on the low ratio of OC/EC (1.46).With the increasing wavelength, absorptions from BrC pri and BrC sec declined sharply.At λ = 520 nm, absorption contributions by BC, BrC pri and BrC sec were around 90%, 8% and 2%, respectively.
Absorptions of BC and BrC pri had similar daily variation (Figure 2b), with higher values in the morning and night due to traffic activities.The decreasing of primary aerosols after 8:00 could be explained by the development of planetary boundary layer (PBL), which diluted pollutants near surface.BrC sec peaked in the morning (8:00) indicating the formation of BrC sec due to photochemical reactions after sunrise.After 8:00 BrC sec tended to decrease, which is probably associated to the bleaching of light-absorbing chromophores via oxidative processes and PBL development (Wang et al., 2019).From 8:00 to 17:00, absorptions by BrC pri and BrC sec were comparable with average absorption coefficients at λ = 370 nm of 1.5 and 1.3 Mm 1 separately, indicating the nonnegligible role of BrC sec absorption during daytime.The large difference between BrC pri and BrC sec absorptions appeared during the nighttime (after 21:00), which is attributed to the emission of primary aerosols but weak formation of BrC sec without sunlight.

Implication of Aerosol Radiative Effects
The meteorological conditions in Lhasa were simulated during the observation period using the Weather Research & Forecasting Model (WRF, v4.4), and results showed good performance comparing with surfaced measured meteorological parameters (Figure S5 in Supporting Information S1).The daily averaged RH vertical distribution and PBL height (PBLH) are showed in Figure S6 in Supporting Information S1.It could be found that PBL was  well developed (PBLH > 1,400 m) from 11:00 to 19:00 in Lhasa, and RH in the PBL was usually lower than 50% except for layers close to the top of PBL.Since organic aerosols are the dominated species over Lhasa with an average contribution of 82.6% (Zhao et al., 2022), which has weaker hygroscopicity compared with inorganic aerosols, such as sulfate and nitrate (Kuang et al., 2020), aerosol hygroscopic growth could be ignored in the PBL over Lhasa during the observation period (Tao et al., 2014;Yan et al., 2009).Assuming that the PBL is well mixed and most aerosol particles are concentrated in the PBL, SSA can be treated as a constant value in the column when PBL is well developed (Li et al., 2017;Ran et al., 2022).To explore DARF sensitivity to aerosol optical properties and minimize the possible SSA uncertainties, DARF TOA , DARF BOT and DARF ATM calculated during 11:00-19:00 using SBDART model were set as the base case, and the average AOD and SSA values for this period were 0.06 and 0.70, respectively.The sensitivity experiments were designed via perturbing SSA and AOD values by 10%, 20%, 10% and 20%, respectively.
As shown in Figure 3a, aerosol exerted positive DARF TOA for the base case with an average value of 2.83 W/m 2 from 11:00 to 19:00, which is greater than the global average value of CO 2 effective radiative forcing (2.16 W/m 2 ) from IPCC Sixth Assessment Report (Masson-Delmotte et al., 2021), indicating strong aerosol heating effect on the Earth-atmosphere system.If SSA decreased by 20% (mean value of 0.56), DARF TOA increased about twice times (5.52 W/m 2 ).When SSA increased by 10% (SSA = 0.77), DARF TOA decreased to 1.45 W/m 2 , but still had warming effect.Only when SSA increased by 20% (SSA = 0.84), aerosol warming effect could be neglected with DARF TOA of 0.04 W/m 2 .Compared with SSA, DARF TOA was less sensitive to AOD with only 1.09 W/m 2 difference for AOD varying from 20% to 20%.As given in Table 1, the four sites with surfaced measured low SSA values (SSA ≤0.77) over TP are all locating in the middle and west of TP (Figure S7 in Supporting Information S1), where experienced obvious warming with air temperature increasing rate over 0.06°C per decade from 1901 to 2018 (Mu et al., 2020).This consistency might highlight the non-negligible direct radiative effect of aerosol on TP warming.
The average DARF BOT and DARF ATM were 0.78 and 3.61 W/m 2 for the base case, respectively (Figures 3b and  3c), and their absolute values were extremely low due to the low aerosol content in comparison to results obtained for sites locating in the eastern China (Che et al., 2018;Kuang et al., 2015;Lu et al., 2022;Zheng et al., 2019).For example, Che et al. (2018) reported that the average DARF BOT at different sites ranged from 93 to 79 W/m 2 .Similar to DARF TOA , DARF BOT and DARF ATM were more sensitive to SSA than AOD.Even though DARF BOT and DARF ATM over Lhasa are very weak compared with sites in eastern China (Che et al., 2018;Kuang et al., 2015;Lu et al., 2022;Zheng et al., 2019), DARF TOA is distinguished with positive value resulting from strong aerosol absorption, implying the vital role of aerosol heating on the Earth-atmosphere system.Considering that aerosol had increasing trends in most area of TP over the past decades (Rahman et al., 2023;Zhu et al., 2019), that is, the annual AOD trend was about 0.001 years 1 from 2007 to 2017 in Lhasa, aerosol warming effect would be strengthened without limiting aerosol emissions.The sensitivity experiments reveal that confining emission fractions of absorbing aerosols, especially BC, is an effective way to reduce DARF TOA and relieve TP warming caused by aerosol.

Conclusions
In situ measurement of aerosol optical properties was performed over Lhasa to explore the role of aerosol in TP warming.Unexpected strong absorption was observed in this city, with average SSA of 0.69.This value is significantly smaller than SSA observed over sites in eastern China, and even lower than most of GAW sites in the globe, implying the potential heating effect on the Earth-atmosphere system by aerosol and non-negligible contribution to TP warming.BC, BrC pri and BrC sec contributions to absorption were distinguished combining MRS with AAE method to identify the key compounds influencing the strong absorption.BC was the main absorption contributor, accounting for 83% and 90% at λ = 370 and 520 nm, respectively.Despite the low aerosol content in Lhasa (average AOD of 0.08 at 440 nm and extinction coefficient of 13.6 Mm 1 at 532 nm), aerosol had positive DARF TOA (average value of 2.83 W/m 2 from 11:00 to 19:00) due to its strong absorbing property.As other three sites over TP were also reported with low SSA values (≤0.77) and with obvious warming trend over the past decades (>0.06°C per decade), aerosol might exert a significant role on accelerating TP warming.Under the increasing trend of aerosol content, the fractions of absorption species should be controlled to weaken aerosol warming effect in TP.
It should be noted that the observation over Lhasa lasted for a short time in this study.To get comprehensive knowledge about aerosol impacts on TP warming and the spatial heterogeneities, it is necessary to measure multiple and long-term aerosol optical and physical parameters at more typical sites in the future.Meanwhile, vertical aerosol optical property observations based on in situ techniques are necessary to evaluate vertical aerosol effects, as remote sensing means have their limitations under extremely low aerosol content.

Figure 1 .
Figure 1.Time series of meteorological factors (downward and upward shortwave radiation: SW_dn and SW_up; temperature: T; relative humility: RH; wind speed and direction: WS and WD) and aerosol optical properties (extinction and absorption coefficients and SSA at 532 nm: Ext. and Abs.; AOD at 440 nm; fitted AAE from 370 to 950 nm) in Lhasa.

Figure 2 .
Figure 2. The average spectral absorption coefficients (a) and daily variation of aerosol absorption coefficients (b) by BC, BrC pri and BrC sec at 370 nm.

Table 1
Aerosol Extinction Coefficients and SSA Observed at Different Sites Note.The rows with italics correspond to sites located in TP. a Neph, b PAX, c CAPS, d PAS and e MAAP are the abbreviations of integrating Nephelometer, Photoacoustic extinctionmeter, Cavity Attenuated Phase Shift, Photoacoustic Spectrometer and Multi Angle Absorption Photometer, respectively.