High-resolution physicochemical dataset of atmospheric aerosols over the Tibetan Plateau and its surroundings

. Atmospheric aerosol in the Tibetan Plateau (TP) and its surroundings has attracted signiﬁcant scien-tiﬁc interest in recent decades due to its notable impacts on regional climatic and cryospheric changes, ecological and environmental security, and the hydrological cycle. However, our understanding of the atmospheric aerosol in this remote region is highly limited by the scarcity of available datasets owing to the extremely harsh natural conditions. This challenge has been mitigated in recent decades by establishing ﬁeld observatories at typical sites within the TP and its surroundings. A continuous project initiated in 2015 aims to explore the properties and sources of atmospheric aerosols, as well as their regional differences, through multiple short-term intensive observations obtained across this vast region utilizing a suite of high-time-resolution online instruments. This paper presents a systematic and hourly scale dataset of aerosol physicochemical and optical properties at eight sites across the TP and its surroundings that is derived from the project. It includes size-resolved chemical compositions of submicron aerosols, high-resolution mass spectra and sources of organic aerosols, size distributions of particle number concentrations, particle light-scattering and light-absorption coefﬁcients, particle light ab-sorptions attributed to different carbonaceous substances including black carbon and brown carbon, and number


S1. HR-ToF-AMS operation
The HR-ToF-AMS is one of the most advanced instruments that widely used for the study of atmospheric aerosol chemistry worldwide.The detailed principle of HR-ToF-AMS can be obtained elsewhere (DeCarlo et al., 2006).The HR-ToF-AMS is mainly composed by three different sections that separated by small apertures and differentially pumped, i.e., a particle beam generation section to form a concentrated and narrow particle beam through a critical orifice and a six-stage aerodynamic lens, a particle-sizing chamber to measure the particle aerodynamic sizes through a particle time-of-flight measurement (different velocities and arrival times for size-dependent particles in a known flight distance), and a particle chemical composition detection section to directly vaporize the particle beam at a ~600 ℃ resistively heated surface, ionize the particles into positively charged ion fragments by a 70 eV electron impact, and then detect their chemical composition by a high-resolution mass spectrometer (Jimenez et al., 2003).
There are two different operation modes in HR-ToF-AMS, i.e., V-mode (detection limits of about 10 ng m −3 ) and W-mode (~5000 m/Δm) with different signal-to-noise ratio (S/N).However, the HR-ToF-AMSs were only operated at the V-mode during almost all the seven field campaigns in consideration of the relatively low aerosol mass concentration level and low S/N ratio over the TP.The mass concentration and size distribution of non-refractory PM1 chemical species were obtained by further switching the instrument between mass spectrum (MS) mode and particle time-of-flight (PToF) mode every 15s under the V-mode operation.However, there are no observation of particle sizes during the NamCo and LHG measurements due to the malfunction of the chopper.In addition, the HR-ToF-AMS need to be calibrated for its flow, ionization efficiency (IE), and sizes at the beginning and end of each observation (Jayne et al., 2000).The relative IE (RIE) of ammonium and sulfate were calibrated using the monodispersed pure ammonium nitrate and ammonium sulfate particles, respectively, with the selected sizes of 200-300 nm, while the particle size was calibrated using the monodispersed ammonium nitrate particles with sizes varied from 60 to 600 nm.Finally, default RIE values were assumed to be 1.1, 1.3, and 1.4 for nitrate, chloride, and OA, respectively, during all the field campaigns, while different RIE values were set for ammonium and sulfate according to their calibration results during each campaign, e.g., 3.9 and 4.2 for ammonium and 1.6 and 1.4 for sulfate based on two calibrations in the QOMS measurement.

S2. HR-ToF-AMS data processing
The HR-ToF-AMS data was processed using the standard data analysis software with SQUIRREL and PIKA toolkits written in Igor Pro (Wavemetrics Inc., Lake Oswego, OR, USA).The SQUIRREL used a fragmentation table to apportion the measured signals at each mass-to-charge ratio (m/z) into different species to quantify the chemical composition of non-refractory PM1 species, while the PIKA employed a modified Gaussian fitting algorithm to obtain the ion-speciated high-resolution mass spectra (HRMS) and elemental composition of OA (Allan et al., 2004;DeCarlo et al., 2006).The elemental ratios of OA, i.e., oxygen-to-carbon (O/C), hydrogen-to-carbon (H/C), organic matter-to-organic carbon (OM/OC), and nitrogen-to-carbon (N/C), were determined using the improved method (Canagaratna et al., 2015) during all the seven observation campaigns.
In addition, a collection efficiency (CE) was generally introduced to compensate for the incomplete transmission and detection of particles through the aerodynamic lens and bouncing at the vaporizer surface in most AMS studies.Previous study has revealed that the CE is significantly influenced by the relative humidity (RH) in sampling line and the acidity and ammonium nitrate mass fraction (ANMF) in the sampled aerosols, which has been concluded as a build-in composition-dependent CE (CDCE) algorithm in the standard data processing software (Middlebrook et al., 2012).Generally, a high RH, a high aerosol acidity, or a high ANMF often corresponds to a high CE value.
However, the RH in the sampling system is always maintained below 40% due to the professional deployments of dryers in the front of the sampling system and the ANMF is basically below 0.4 due to the low contributions of nitrate and ammonium during all the seven observation campaigns, which means the negligible effects of these two parameters on CE in our study.Therefore, default CE value of 0.5 were finally employed during the QOMS, NamCo, Ngari, Waliguan, and Lhasa campaigns in consideration of their overall neutralized or slightly acidic aerosols, whereas the CDCE values were adopted at Motuo, LHG, and Bayanbulak where bulk submicron aerosols were acidic.

S3. OA source apportionment using PMF analysis
Source apportionment of OA during each observation was conducted by the positive matrix factorization (PMF) analysis on organic matrix data using the PMF2.exealgorithm in robust mode (Paatero and Tapper, 1994) and the standard PMF Evaluation Tool (PET, Ulbrich et al., 2009) written in Igor Pro software.
The PMF analysis was evaluated thoroughly according to the standard procedures outlined in Zhang et al. (2011) by down-weighting, modifying, or removing some ion fragments in the data and error matrices.Firstly, those ions at m/z >120 and all the isotope ions were generally excluded because of the insufficient ability to resolve the deconvolution due to their low signals.Then, the signals of the four organic ions of O + , HO + , H2O + , and CO + were scaled to that of CO 2 + according to the suggested fragmentation table in Aiken et al. (2008) and further down-weighted in PMF analysis.
Thirdly, all those "bad" ions (S/N <0.2) were removed from the data matrices, while all the "weak" ions (0.2< S/N <2) were downweighted by increasing their errors.In addition, some runs and some ions which had obviously huge residual spikes were also removed in order to avoid their unnecessary interference.After the above preprocessing, the PMF solutions were investigated by selecting a certain variation range of factor number and rotational parameter (fPeak), e.g., 1−6 factors with fPeak varying from -1 to 1. Finally, the optimal solution of PMF analysis were determined after a comprehensive evaluation by examining the model residuals at each m/z and each time, The sample and sheath flow rates are 0.3 and 3.0 L min −1 , respectively, at both QOMS and Lhasa which measure particles between 14.6 and 661.2 nm in mobility diameter (Dm), whereas the sample and sheath flow rates are 0.5 and 5.0 L min −1 at LHG and Motuo and sample particles at a size range of 10.9−495.8nm in Dm.The number concentrations of submicron particles in 107 different size channels are firstly recorded at an initial time resolution of 5 min and then converted to the total number and volume concentrations according to the obtained size distribution of number concentration.

2) PAX
The PAX directly measures the Babs and Bscat of aerosol particles at 405 nm by using a modulated diode laser, namely measures the Babs by an in-situ photoacoustic technique while the Bscat using a wide-angle integrating reciprocal nephelometer.The Bext is the sum of Babs and Bscat while the single scattering albedo (SSA) is calculated as the ratio of Bscat to Bext.The BC mass concentration is calculated as the ratio of measured Babs to a fixed BC mass absorption cross-section (MAC) value of 10.19 m 2 g −1 at 405 nm.In addition, the Bscat is calibrated using the high-concentration ammonium sulfate particles generated by the aerosol generator, while the Babs is calibrated using the sufficient black smoke from a kerosene lamp before each field campaign according to the operator manual of this instrument.

3) Aethalometer
The Aethalometer (models AE31 or AE33) is used to measure the particle Babs at seven wavelengths, which firstly measures the light attenuation between particle-laden and particle-free sample spots on the filter and finally converts the attenuation to particle Babs in ambient air.Both AE31 and AE33 have seven bands, namely 370, 470, 520, 590, 660, 880 and 950 nm, and the concentration of black carbon is mainly measured according to the absorption coefficient at 880 nm.The filter-based loading effect and multiple scattering effect are corrected during all the three observations to eliminate the difference between the light attenuation measured at the filter and the ambient particle Babs.
The absorption Ångström exponents (AAE) value is acquired through a powerlaw fitting of Babs following the typical Beer-Lambert's law, i.e., AAE = ln (B abs,λ1 /B abs,λ2 ) /ln(λ 2 /λ 1 ) .Furthermore, a traditional AAE method was adopted to quantitatively apportion the total Babs into two parts from BC and BrC (Babs,BC and Babs,BrC) at 370−660 nm during each campaign.The contribution of BrC to total Babs (fBabs,BrC) at a short wavelength λ is calculated as fB abs,BrC,λ = 1 (B abs,880 /B abs,λ ) × (λ/880) AAE BC by assuming its negligible contribution at 880 nm.Detailed information about the data correction and calculation of this instrument can be found in our previous publication (Zhang et al., 2021b).

4) CCN-100
The CCN-100 measures aerosol particles called cloud condensation nuclei that can form into cloud droplets.The instrument supersaturates the sampled aerosol particles in a 50-cm-high column with continuously wetted walls and a longitudinal thermal gradient, so that those particles grow into detectable CCN particles and are measured using an optical particle counter among 20 different size bins.The principle of CCN counters is that diffusion of heat in ambient air is slower than that of water vapor, which diffuses from the warm, moist column walls to the centerline faster than heat in the column.Detail information may refer to Roberts and Nenes (2005).The number concentrations of CCN are measured consecutively at five different SS values of 0.2%, 0.4%, 0.6%, 0.8%, and 1.0%.The CCN data is recorded every 5 minutes at each SS and finally has a time resolution of 30 minutes during a complete measurement cycle.

S5. Calculation and evaluation of the bulk acidity of submicron aerosols
Bulk acidity of submicron aerosols from AMS measurement was generally evaluated following the methods in Zhang et al. (2007) and Schueneman et al. (2021).
The mass concentration of ammonium was firstly predicted by assuming to fully neutralize these AMS measured sulfate, nitrate, and chloride using the following

⁄
) was further calculated to be a good indicator to evaluate the bulk acidity of submicron aerosols.In this study, linear regression analysis between the mass concentrations of measured and predicted ammonium was performed to evaluate the bulk acidity of submicron aerosols in the different TP regions (Fig. 3).
Aerosol particles are generally considered to be "acidic" if the calculated ratio is obviously lower than 1 and to be "more acidic" if the ratio is lower than 0.75, whereas a ratio that roughly near to 1 or larger than 1 indicates the particles are "bulk neutralized" and even there are more excess ammonium that needed to be neutralized.Note that the validity of using this method is based on the assumption that the influence from nitrogen-or sulfur-containing organic ions (e.g., organic acids and organic nitrogen compounds) as well as the mineral and metal ions are negligible (Zhang et al., 2007).

S6. Estimation of aerosol DRFs using the SBDART and OPAC models.
The aerosol direct radiative forcing (DRF) was modelled by the widely used Santa Barbara DISORT (Discrete Ordinate Radiative Transfer) Atmospheric Radiative Transfer (SBDART) model in the shortwave spectral range of 0.25−4.0μm.SBDART is a software tool that computes the plane-parallel radiative transfer under both clear and cloudy conditions (Ricchiazzi et al., 1998).Specific simulation of aerosol DRF in the atmosphere (ATM) can be described by the following equations ( 2) and (3).In brief, the net fluxes (∆F, difference between the downward and upward radiation fluxes) with and without the investigated variable were calculated twice in this model under cloudfree conditions at both the earth's surface (SUR) and the top of the atmosphere (TOA).
The differences of net fluxes between the two simulations were then considered as the DRFs of the specific investigated variable at the SUR and TOA.Finally, the DRF in the ATM (DRF ATM ) was obtained using the DRF at TOA (DRF TOA ) minus the DRF at SUR (DRF SUR ).The details of the model description can be found in previous studies (Xin et al., 2016;Gong et al., 2017).
The principle and specific steps for the estimation of aerosol DRFs using the SBDART mode can be described briefly as follows: Firstly, the mass concentrations of organic carbon (OC), BC, and water-soluble ions (WSIs), which were measured from the corresponding off-line filter samples with one-or two-days time resolutions during each campaign, were the initial measured parameters used for the estimation of aerosol DRFs.
Then, the particle numbers of each species in per cubic meter air (denoted as ρ ) were estimated using the above measured mass concentrations of each species divided by the referred M values in the OPAC model (Hess et al., 1998), which represented the aerosol mass in per cubic meter air and integrated over the size distribution and normalized to 1 particle per cubic centimeter of air.Specifically, values of 5.99E-5 was used for soot/BC particles and 1.34E-3 were used for OC and WSIs.
comparing the factor mass spectrum with corresponding reference spectrum, comparing the temporal variation of individual factor with external tracers, and analyzing the diurnal variation pattern of each factor.Totally, 2-, 3-, or 4-factor solution were selected during the different field campaigns in this study.The specific high-resolution mass spectrum of each OA factor identified among the eight different field campaigns are shown in Figure S4.particle sizer (SMPS) developed by the TSI Inc. is composed by an electrostatic classifier (EC, model 3080) equipped with a longdifferential mobility analyzer (long-DMA, model 3081) and a condensed particle counter (CPC, model 3772).Ambient particles are first screened by a particle impactor installed at the front of the DMA and large particles are removed.Ambient particles are measured through an electrical mobility detection technique in this instrument, e.g., a bipolar charger in the EC is utilized to charge the particles to a known charge distribution, then classify them according to their ability to traverse an electrical field in the long-DMA, and finally count those screened monodisperse particles by the CPC.
Next, four crucial input optical parameters of aerosol optical depth (AOD), single scattering albedo (SSA), Ångström exponent (AE), and asymmetry factor (ASY), and the light absorption and scattering coefficients were estimated for OC, BC, WSIs and total particles, respectively, using the Optical Properties of Aerosol and Cloud (OPAC) model.Detailed introduction of OPAC model can be found inHess et al. (1998).Noting that the above estimation in OPAC model were performed six times during each campaign by setting the ρ as the above calculated value multiplier 1 to 6, respectively.Then, the entire six datasets of modelled total light absorption and scattering coefficients were further used to evaluate the performance of OPAC model by comparing them with those corresponding measured light absorption and scattering coefficients from PAX and Aethalometer (FiguresS6-10).Finally, the optimal ρ value was selected when the modelled total light absorption and scattering coefficients were comparable.The performance of the OPAC model was then evaluated and tuned by comparing the modelled total light scattering and absorption coefficients with those correspondingly measured values from online Aethalometer and PAX measurements during the three campaigns, as shown in FiguresS5-10.The consistent variation trends were found between them with the coefficient of determination (R 2 ) varying between 0.69 and 0.99.The slightly lower modelled values compared with those measured values was mainly attributed to their inconsistent wavelengths, i.e., the modelled light scattering and absorption coefficient at 550 nm in the OPAC model, whereas the measured light scattering coefficients at 405 nm for PAX and the light absorption coefficients at 520 nm for Aethalometer.Overall, the small difference between the modelled and measured values indicated the reasonable simulation of aerosol optical parameters (e.g., AOD, AE, SSA, and ASY) in the OPAC model in this study.Finally, these four optical parameters belonging to each species were all used in the SBDART model for the simulation of DRFs caused by OC, BC, and WSIs.After the above evaluation, the finally obtained four input parameters of AOD, SSA, AE and ASY belong to each species (OC, BC, WSIs, and total particles) were inputted into the SBDART model to simulation the DRFs of each species, respectively.

FiguresFig
Figures Fig. S1 High-time-resolution temporal variations of the mass concentrations of PM 1 chemical compositions during the eight online aerosol field measurement campaigns over the Tibetan Plateau and its surroundings.

Fig. S6
Fig. S6 Time series of the particle light scattering coefficients at 405 nm measured by the PAX and those modelled light scattering coefficients at 550 nm under different particle numbers in per cubic meter air (the calculated value multiplier 1 to 6, respectively) from the OPAC model during the QOMS campaign.

Fig. S7
Fig. S7 Time series of the particle light absorption coefficients at 520 nm measured by the Aethalometer and those modelled light absorption coefficients at 550 nm under different particle numbers in per cubic meter air (the calculated value multiplier 1 to 6, respectively) from the OPAC model during the QOMS campaign.

Fig. S8
Fig. S8 Time series of the particle light scattering coefficients at 405 nm measured by the PAX and those modelled light scattering coefficients at 550 nm under different particle numbers in per cubic meter air (the calculated value multiplier 1 to 6, respectively) from the OPAC model during the Waliguan campaign.

Fig. S9
Fig. S9 Time series of the particle light absorption coefficients at 520 nm measured by the Aethalometer and those modelled light absorption coefficients at 550 nm under different particle numbers in per cubic meter air (the calculated value multiplier 1 to 6, respectively) from the OPAC model during the Waliguan campaign.

Fig. S10
Fig. S10 Time series of the particle light absorption coefficients at 520 nm measured by the Aethalometer and those modelled light absorption coefficients at 550 nm under different particle numbers in per cubic meter air (the calculated value multiplier 1 to 6, respectively) from the OPAC model during the NamCo campaign.