Global-regional nested simulation of particle number concentration by combing microphysical processes with an evolving organic aerosol module

Aerosol microphysical processes are essential for the next generation of global and regional climate and air quality models to determine particle size distribution. The contribution of organic aerosol (OA) to particle formation, mass, and number 30 concentration is one of the major uncertainties in current models. A new global-regional nested aerosol model was developed to simulate detailed microphysical processes. The model combines an advanced particle microphysics (APM) module and a volatility basis-set (VBS) OA module to calculate the kinetic condensation of low-volatility organic compounds and equilibrium partitioning of 35 semi-volatile organic compounds in a 3-dimensional (3-D) framework using global-regional nested domain. In addition to the condensation of sulfuric acid, the equilibrium partitioning of nitrate and ammonium, and the coagulation process of particles, the microphysical processes of the OAs are realistically represented in our new model. The model uses high-resolution size-bins to calculate the size distribution 40 of new particles formed through nucleation and subsequent growth. The multi-scale nesting enables the model to perform high-resolution simulations of the particle formation processes in the urban atmosphere in the background of regional and global environments. By using the nested domains, the model reasonably reproduced the OA components obtained from the analysis of aerosol mass spectrometry measurements 45 through positive matrix factorization and the particle number size distribution in the megacity of Beijing during a period of approximately a month. Anthropogenic organic species accounted for 67 % of the OA of secondary particles formed by nucleation and subsequent growth, which is considerably larger than that of biogenic OA. On the global scale, the model well predicted the particle number concentration 50 in various environments. The microphysical module combined with the VBS simulated the universal distribution of organic components among the different aerosol populations. The model results strongly suggest the importance of anthropogenic organic species in aerosol particle formation and growth at polluted urban sites and over the whole globe under the influence of anthropogenic sources. 55

secondary formations are required to improve the ability of model in simulating 95 particle number size distribution (Mann et al., 2014). Spracklen et al. (2005) find that the assumption of the size distribution has a large impact on particle number concentrations in the boundary layer. The comparison between the simulation and the Single Particle Soot Photometer (SP2) measurements suggests that the model has large bias in simulating the number size distribution of black carbon particles 100 . Significant improvements in the simulation of the particle number concentration and aerosol optical properties were achieved by using an optimized size distribution of primary particles in polluted atmosphere over areas with large emissions (Zhou et al., 2012(Zhou et al., , 2018. Much work remains to reduce the uncertainty associated with primary emissions, especially over primary particles 105 dominated regions in terms of particle number concentration, like China. The main source of uncertainty in simulating new particle formation at regional and global scales can be attributed to the nucleation mechanism and particle growth rates unexplained. Although sulfuric acid has been identified as a major component and plays a central role in nucleation (Yu and Turco, 2001;Boy et al., 2005;Kirkby et 110 al., 2011), alone it could not explain the new particle formation rates (Wang et al., 2013;Kulmala et al., 2013). Recent studies revealed that certain organic vapors are involved in the particle nucleation (Metzger et al., 2010;Zhang et al., 2012;Yao et al., 2018) and contribute much to the particle growth (Kulmala and Kerminen, 2008;Tröstl et al., 2016). It is no doubt that reasonable representation of organic aerosol 115 (OA) is crucial for aerosol models to realistically simulate new particle formation and growth. However, it is still an open question which organic species are possibly involved in new particle formation process. Even the chemical composition and the sources of OA are still uncertain as they contain large number of compounds (Goldstein and Galbally, 2007). Up to now, OA is still the least understood one 120 among the components of aerosols (Kanakidou et al., 2005;Hallquist et al., 2009).
Clearly, the OA representation is the major uncertainty contributing to the huge gap in elucidate particle formation processes.
In recent years, much progress has been achieved in simulating the formation of by Odum et al. (1996) had been widely used in 3-D models to describe SOA formation process empirically. The volatility basis-set (VBS) approach was recently developed to represent the oxidation of primary OA (POA) and SOA and the partitioning of OA in different volatilities between gas phase and aerosol phase (Donahue et al., 2006). Many regional models have used VBS to simulated OA and 130 SOA (Shrivastava et al., 2008;Fountoukis et al., 2011;Ahmadov et al., 2012;Zhao et al., 2016;Han et al., 2016). However, application of VBS in global models is limited for the large number of tracers required and the uncertainty of the involved parameters (Farina et al., 2010;Hodzic et al., 2016). There are even fewer applications of this unified framework in 3-D global aerosol models to calculate the processes of particle 135 formation. Among the second phase AeroCom aerosol microphysical models, simplified parameterization and two-product method are the mostly used schemes to represent SOA (Mann et al., 2014). Recently，there has been some models with VBS incorporated in their microphysical module to simulate aerosol microphysical formation process. Patoulias et al. (2015) developed a new aerosol dynamics model 140 with VBS and explored the contribution of SOA with different volatility to particle growth in different stages, but the 3-D modeling was not presented. By assuming equilibrium partitioning for all volatility bins, Gao et al. (2017) implemented VBS in an aerosol microphysics model and examined the effect of semi-volatile SOA on the composition, growth, and mixing state of particles. Their simulation of box model 145 suggested that the volatility of organic compounds simulate rather different mixing states from those simulated by coagulation process alone in the scheme treating primary emission of organics as nonvolatile. Matsui (2017) represented aerosol size distribution with a two-dimensional sectional method in a global aerosol model coupled with the VBS scheme, but the size-bin resolution is not high enough to well 150 resolve the growth of new particles.
To our knowledge, there is currently scarce 3-D modeling study using VBS to account for both (1) the kinetic condensation of low-volatile organics and re-evaporation of semi-volatile organics and (2) the size-resolved kinetics of the mass 5 https://doi.org/10.5194/acp-2020-759 Preprint. Discussion started: 27 August 2020 c Author(s) 2020. CC BY 4.0 License.
transfer for new particles. In addition, the particle formation in the polluted 155 atmosphere was not well understood (Kulmala et al., 2016;Wang et al., 2017;Chu et al., 2019). Over the urban areas in northern China, observation and modeling studies indicate that anthropogenic SOA contributes a larger fraction to OA than that of biogenic one and play an significant role in particle formation (Yang et al., 2016;Guo et al., 2020;Han et al., 2016;Lin et al., 2016). Simultaneously calculating both 160 anthropogenic and biogenic SOA in microphysical models with high resolution is crucial to resolve the particle formation processes over the urban areas. Furthermore, the previous studies focusing on the sensitivity of particle number concentration to primary emission were based on models without considering the detailed microphysics of organic species (e.g., Chang et al., 2009;Chen 165 et al., 2018;Zhou et al., 2018). Therefore, it is urgently needed to establish a 3-D modeling framework of VBS with an aerosol microphysics module with high size-bin resolution to simulate the particle number size distribution and explore the uncertainties associated with the treatment of primary emission.
In our previous work, a regional model with detailed microphysical processes 170 has been developed to improve the new particle formation in summer in Beijing . In this study, we extend our work to the global scale and doing so to establish a new aerosol model by coupling a VBS organic aerosol scheme with a particle microphysics module in a global-regional nested model. The model performance was evaluated against the measurements at a tower and the collected 175 dataset from published papers. In addition, the model's sensitivity to the size distribution of primary emission and volatility distribution of POA are explored to understand and quantify the uncertainties associated. The new modeling framework can provide a useful tool to simulate aerosol microphysical process in both global and regional scales. The description of model and its development method are introduced

Host model
The host model employed in this study is the Atmospheric Aerosol and Chemistry Model developed by Institute of Atmospheric Physics, Chinese Academy of Sciences (IAP-AACM). The IAP-AACM is a 3-D atmospheric chemical transport model treating chemical and physical processes for gases and aerosols in global and 190 regional scales using multi-scale domain-nesting technique (Wang et al., 2001;Li et al., 2012;Chen et al., 2015). The model has been successfully used to explore mercury transport  and simulate the global and regional distribution of gaseous pollutants, aerosol components Wei et al., 2019). The calculation of some modules in the model has also been optimized recently (Wang et 195 al., 2017(Wang et 195 al., , 2019. The model calculates 3-D advection (Walcek, 1998), turbulent diffusion (Byun and Dennis, 1995), gas phase chemical reactions (Zaveri and Peters, 1999), dry deposition at the surface (Zhang et al., 2003), aqueous reactions in the cloud and wet scavenging (Stockwell et al., 1990), and heterogeneous chemical process (Li et al., 2012). The partition of nitric acid and ammonia into particle phase 200 to form nitrate and ammonium are simulated using a thermodynamic equilibrium model (Nenes et al., 1998). The model calculates the online emission of dimethyl sulfide (DMS) (Lana et al., 2011), sea salt (Athanasopoulou et al., 2008) and dust (Wang et al., 2000;Luo and Wang, 2006  .

APM module
Advanced Particle Microphysics (APM) module is an aerosol module using the sectional method to represent particle number size distribution. APM has been 210 coupled to several 3-D models, such as GEOS-Chem (Yu and Luo, 2009), WRF-Chem (Luo and Yu, 2011), and NAQPMS . In APM, there are two types of aerosol particles: one is secondary particles (SPs) and the other is primary particles (PPs) with a secondary species coating. The definitions of SPs and PPs in our model are different from secondary aerosol and primary aerosol commonly 215 used in the community. SPs indicate their source is the nucleation and the subsequent growth of newly nucleated particles whereas PPs are from the direct emission. PPs include dust particles, sea salt particles, organic carbon (OC) particles, and black carbon (BC) particles. The APM has a high size-bin resolution to accurately describe the formation and growth processes of SPs (composed of sulfate, nitrate, ammonium, 220 and organic compounds). SPs are represented by 40 size bins from 0.0012 μm to 12 μm in dry diameter. Among the PPs, representation of BC and OC are updated from a modal method in the original version (Yu and Luo, 2009) to a size-bin scheme in the revised version (Chen, 2015). Dust particles in 0.03-50 μm are represented by 4 size bins and sea salt particles in 0.0012-12 μm are represented by 20 size bins. SPs are 225 assumed to be internally mixed and PPs are assumed to be consisted of a primary core and coating species. SPs and PPs of different categories are externally mixed with each other. In addition to the primary core, the coated species are explicitly simulated in APM.
The basic microphysical processes in APM include nucleation, 230 condensation/evaporation, coagulation, and thermodynamic equilibrium partition. The nucleation scheme is the ion-mediated nucleation (IMN) (Yu, 2006(Yu, , 2010 physically-based and constrained by laboratory data, which have predicted reasonable distributions of global nucleation (Yu et al., 2008). Due to very low saturation vapor pressure, the condensation of H 2 SO 4 is explicitly calculated. The semi-volatile 235 inorganic species (nitrate and ammonium) and secondary organic species are simulated through equilibrium partitioning. The bulk mass concentrations of coating species are tracked to reduce the computational cost and the corresponding tracers used are defined as BC sulfate, OC sulfate, sea salt sulfate, and dust sulfate, respectively. For coagulation, APM not only calculates the self-coagulation of sea salt 240 particles, BC particles, OC particles, and SPs, and but also considers the coagulation scavenging of SPs by four types of PPs. Yu (2011) has further developed APM to explicitly calculate the co-condensation of sulfuric acid and low-volatility secondary organics gas (LV-SOG) on the secondary and primary particles. In the scheme, the 8 https://doi.org/10.5194/acp-2020-759 Preprint. Discussion started: 27 August 2020 c Author(s) 2020. CC BY 4.0 License. production rate of LV-SOG and the semi-volatile OA input to APM are simulated with 245 the extended two-product SOA formation model. For high calculation efficiency, a pre-calculated look-up table of coagulation kernels is used in the coagulation module.
The numerical scheme used is from Jacobson et al. (1994). More details on microphysical processes of APM can be found in the study of Yu and Luo (2009).

VBS module 250
To reproduce the formation and evolution of OA, a 1.5-D VBS approach (Koo et al., 2014) based on 1-D VBS framework but accounting for changes in the oxidation state and volatility of OA in the 2-D VBS space is coupled to the model. Both secondary and primary organic aerosols are distributed in five volatility bins ranging from 10 -1 to 10 3 μg/m 3 in saturation concentration (C*) at 298 K, and temperature 255 dependence of C* is calculated by the Clausiuse-Clapeyron equation (Sheehan and Bowman, 2001). The compounds distributed in the lowest bin with C* less than 10 -1 µg/m 3 represents the effectively nonvolatile OAs and they are considered as low-volatile organic compounds almost partitioned to the particulate phase in our model. The compounds in other four bins, i.e., C* = {10 0 , 10 1 , 10 2 , 10 3 } µg/m 3 , are 260 defined as semi-volatile organic compounds that can be partitioned between the gas and particulate phase by equilibrium assumption (Donahue et al., 2009). To track the oxidation state of OA, four basis sets are used in the scheme: two-basis sets for chemically aged OA from anthropogenic and biogenic sources, and two-basis sets for freshly emitted OA from anthropogenic sources and biomass burning. The molecular 265 properties for primary OA (POA) and SOA in each volatility bins are provided by the parameters calculated by 2-D volatility scheme (Donahue et al., , 2012. In this VBS module, gas phase organic compounds can be aged by extremely reactive hydroxyl radicals (OH) and other oxidants. Volatile organic precursors of SOA in this study include compounds with terminal olefin carbon bond (R−C = C), 270 internal olefin carbon bond (R−C = C−R). The associated species in the model are terpenes, isoprene, and aromatics. Aging of POA by OH is at a reaction rate of 4×10 -11 cm 3 ·molecule -1 ·s -1 (Robinsonet al., 2007). In the calculation, a conception of "partial conversion" is used, i.e., the oxidation products are a mixture of POA and 9 https://doi.org/10.5194/acp-2020-759 Preprint. Discussion started: 27 August 2020 c Author(s) 2020. CC BY 4.0 License.
oxidized POA (OPOA) in the adjacent lower volatility bins (Koo et al., 2014). In 275 addition, the multigenerational oxidation processes of intermediate VOCs (IVOCs) with OH radicals at a rate constant of 4×10 -11 cm 3 ·molecule -1 ·s -1 are taken into account in SOA formation. IVOCs emission was put into the bin of 10 4 µg/m 3 saturation concentration. The VBS module in this study does not consider OA formation through aqueous-phase/heterogeneous reactions although their importance is suggested in 280 some studies (e.g., Liu et al., 2012;Ervens et al., 2014;Lin et al., 2014). SOA generated from VOCs and IVOCs and anthropogenic OPOA are assumed to be further oxidized by OH radical at an aging rate of 2×10 -11 cm 3 ·molecule -1 ·s -1 based on the work in Koo et al. (2014). The volatilities of multi-generation oxidation products decrease and move down to the adjacent bin with an order of magnitude lower

Model development
In our previous work, the VBS module has been combined with APM to improve the simulation on new particle formation process in our regional model (NAQPMS+APM, Chen et al., 2019). Here, we use the similar method to couple the VBS and APM into the global model, i.e., IAP-AACM. The newly developed model 295 is named IAP-AACM+APM. In the model, not only the basic microphysical processes aforementioned but also the condensation of LV-SOG and equilibrium partition of semi-volatile OA (SV-OA) are calculated following the approaches described in Luo (2009) andYu (2011). In addition to the tracers of OAs and OGs mentioned above, a new tracer for LV-SOG is tracked in IAP-AACM+APM. The OAs in VBS module. The partition of this part of OA is similar with that of 315 equilibrium partition theory (Pankow, 1994a,b;Odum et al., 1996). By using the treatments above, the different microphysical behaviors of OAs with different volatilities are reasonably simulated. The dry deposition at the surface level and wet deposition by precipitation of LV-SOG are modeled using same scheme as H 2 SO 4 .
The dry deposition and wet scavenging of the coated LV-OA associated with SPs and 320 PPs are calculated using the same scheme as the sulfate coated on PPs (Yu, 2011).
The tracers associated aerosol microphysical processes in IAP-AACM+APM are listed in Table 1

Experiments setting
One base experiment and four sensitivity experiments are used in our study. The sensitivity experiments involving size distribution of primarily emitted particles, including BC and POC, and the volatility distribution of POA, are designed to investigate the impact of these factors on the particle number concentration. Table 2   355 lists the experiments used in this study. In the BASE experiment, the volatility distributions of POA from vehicles and biomass burning are based on the chamber studies (May et al., 2013a,b,c); the factors of other POA emissions are from the estimation of Robinson et al. (2007). In the LV_POA and HV_POA experiment, quartiles of the above mentioned distribution factors are used. In the OCD0.5 and 360 PPD0.5 experiment, the geometric mean diameter is set half as the ones used in BASE experiment for POC, both BC and POC, respectively.

Observation data
The hourly observation of OA and particle number size distribution (PNSD) in Beijing is used to evaluate the model performance in the typical urban environment.  Table S1 gives the compiled mean concentrations of condensation nuclei larger than 10 nm (CN10) and the corresponding station information from published papers.

390
The simulated OA concentration is compared with the results of the PMF analysis of the AMS measurements before evaluating the simulated PNSD in Beijing.
Here, HOA and OOA components by the PMF analysis have been compared with the simulation assuming they are primary and secondary components of OA, i.e., POA contributor to OA (Zhao et al., 2017). Moreover, the nearby traffic emissions would have large influences on the observed OA concentrations at the measurement site (Sun et al., 2015). Same as BC, the temporal variation of POA was mainly influenced by emissions, transport and deposition, the disagreement between the simulated POA and the observed HOA can largely be attributed to the emissions. In additions, the 415 PMF analysis has its own uncertainties and deficiencies (Ulbrich et al., 2009 During the past decades, many field observations have been conducted to study the characteristics of PNSD in Beijing (Wehner et al., 2004;Wu et al., 2007Wu et al., , 2008 PNSD in winter in Beijing (Chen et al., 2017). However, 3-D modeling study on these issues are still limited (Kulmala et al., 2016;Wang et al., 2016). Here, the observed PNSD at 260m height is used to evaluate the model performance. Fig.2 shows the comparison of simulated PNSD with the observations. In Fig.2, the model well reproduced the evolution of PNSD at the height of 260m at the measurement site. In 430 the observation, there are five cycles of conversion from clean days to pollution days.
Once the pollution episode was over, an obvious new particle formation event occurred, such as the events in September 3, 12, 19 and 25. When the pollution level increased, the PNSD shifted to end of large diameter. The model well captured the new particle formation events and the growth of particles in the pollution episode 435 mentioned above. Because the atmosphere at higher level was not susceptible to local sources, the observation was more representative than that at the ground level. The number concentration of particles from 100 nm to 1000 nm was nicely reproduced, with normalized bias less than 40% and correlation coefficient being 0.70. The consistency between simulation and observation suggests the good performance of 440 model in producing reasonable number concentration of regional aerosol particles, especially in the climate-relevant size range. However, the number concentration of particles from 15 nm to 25 nm was overestimated. On one hand, the measurements have analytical errors (Du et al., 2017). On the other hand, the model has several uncertainties. First, the model used the monthly mean emissions and therefore could 445 not simulate the diurnal variation of traffic emission. In addition, the size distribution of primary emissions does not meet the assumed lognormal distribution. For example, traffic sources emit smaller particles than industrial sources (Paasonen et al., 2013;Kumar et al., 2014). Second, the nucleation scheme also has uncertainties (Zhang et al., 2010;Yu et al., 2018). For all this, the main features of new particle formation atmosphere and its large contribution to OA suggested the influence of regional transport of OA and precursors of OA from surrounding areas to Beijing. The aging and growth during the lifetime of SPs in the atmosphere could greatly enhance their 470 regional impact. In addition to the local emissions of OA precursors (Guo et al., 2014), our results also highlight the importance of regional sources of OA precursors in the growth of new particles.

Global and regional distribution of OA
There are two important characteristics of OA that influence particle growth and The condensation behavior of OA is closely related with the separation of POA and SOA and their volatility distribution. Therefore, these properties of OA are given as the background to discuss the global and regional particle number concentration.  (Spracklen et al. 2011). In the second domain simulation (shown in Fig.5c and 5d), it is more clearly seen that ASOA has the higher concentrations than BSOA over China.
In North China Plain, concentrations of ASOA were above 3 μg/m 3 while 520 concentrations of BSOA were below 1 μg/m 3 . Previous modeling studies using VBS suggested that roughly half of the condensing mass needs to be distributed proportional to the aerosol surface area to explain the observed aerosol particle growth. The condensation of this part of OA is governed by gas-phase concentration rather than the equilibrium vapour pressure, which is the way our model calculates the growth of LV-SOA to particles. The volatility distribution of SOA is an important 540 factor impacting the global and regional distribution of particle number concentration.  Table 3 are also shown in Fig.6a and environments. By a more specific comparison in Fig.8, where the values of simulation 575 are compared by a scatter plot with corresponding observations at 34 sites given in Table S3, the simulations of annual mean concentration of CN10 agree quite well with the observations, within a factor of two for most of the sites. The spatial pattern of CN10 over the second domain (in Fig.7c) is similar with that of the corresponding region in the first domain (in Fig.7a), but the gradients of CN10 is characterized more Both secondary particles formed through nucleation and subsequent growth and direct emission of primary particles can contribute to atmospheric particle number concentration. It is important to quantify the contribution of these two sources in different parts of the globe. In Fig.6b, it can be seen that secondary particles are 590 dominant in most parts of the globe except for the regions with large primary emissions, e.g., eastern China, India, and southern Africa. The low contribution of secondary particles in these regions is due to the strong scavenging of secondary particles by primary particles and the low nucleation rate caused by competing of primary particles for condensable gases. This spatial pattern is similar with the results 595 of previous studies (Yu and Luo, 2009). However, the fractions of secondary particles in CN10 are lower than those in CN3 showed in Yu and Luo (2009) due to the dominant contribution of secondary nucleation to particles in 3-10 nm. In Fig.7d, a boundary from northeast to southwest can be seen to separate the areas dominated by secondary particles from that by primary particles over China. This phenomenon is 600 also caused by the large difference of emissions between western region and eastern region.

The mixing state of organic aerosols and their growth to new particles
Besides particle number concentration, mixing state of aerosols is necessary to 20 https://doi.org/10.5194/acp-2020-759 Preprint. condensed on the particles. Fig.9 shows the fraction of organic species reside in aerosols of different types (i.e., SPs, sea salt, dust, BC, and OC) defined in our model.
In Fig.9, most of the organic species reside in OC, SPs, and BC particles, suggesting the intense mixing of anthropogenic aerosol species. In the southern hemisphere, the fractions of organic species residing in SPs are above 30%, larger than that of OC In China, significant difference also exists between the western and eastern region.
The dominant contribution of semi-volatile species to OA (shown in Fig.6) and their partition proportional to the low-volatility OA lead to a higher fraction of organic 625 species residing in OC particles over eastern China. The mixing of natural aerosols and organic species were also demonstrated in Fig.9. Over the most areas of the globe, 15% of organic species are distributed in dust particles, which could greatly modify the properties of dust particles and thus their climate forcing over the regions influenced by dust particles .

630
Previous study indicate that organic species are the major components of aerosols (e.g., Zhang et al., 2007;Jimenez et al., 2009) and low-volatility organic species can greatly enhance the growth of new particles (e.g., Yu, 2011;Tröstl et al., 2016). Our results presented above also indicated the substantial distribution of SPs is analyzed. Fig.10 shows the ratio of LV-SOG to H 2 SO 4 and the ratio of low-volatility organic species to sulfate that reside in SPs. The concentration of LV-SOG is a factor of ~1.5-10 higher than that of H 2 SO 4 over many parts of the continents and the adjacent oceans but is lower in East Asia, eastern United States, southern Europe, and northern Africa where emissions of SO 2 are high. Especially,

Sensitivity of particle number concentration to volatility of POA
In the VBS, POA is treated as volatile species and allowed to be aged by oxidation in the atmosphere, it is necessary to explore the uncertainties associated with this treatment of volatility distribution. In addition, the size distribution of POA 660 and the associated microphysical processes are also modified due to this treatment.
For this reason, the sensitivity of particle number concentration to the volatility of POA and the assumed size distribution of PPs are discussed here. Fig.11  Overall, concentration of CN10 changed a little when POA volatilities were in the inter-quartile range of measurements (shown in Fig.11a and b). When using the low/high volatility distribution of POA, PPs number concentrations were increased/decreased by 5-10% over the most areas in the northern hemisphere. By Compared with the most models in the second phase AeroCom (Tsigaridis et al., 2014;710 Mann et al., 2014) and the recently developed new models (e.g., Yu, 2011;Patoulias et al., 2015;Gao et al., 2017), our model includes the more comprehensive sources of SOA by using the VBS framework, especially the anthropogenic SOA. In addition, allowing POA to evaporate and re-condense onto the particles make its microphysical behavior more like SOA and therefore give new meaning to the POA-SOA split 715 which significantly affects the global CCN formation (Trivitayanurak and Adams, 2014). The flexible framework of APM combined with VBS produces the different distribution of organic species in aerosols, i.e., the mixing state of OA, which has been found to cause substantial difference in radiative effects of aerosols (Zhu et al., 2017). Box model analyses showed that the low-volatility SOA has a large fraction in 720 the growing nucleation mode particles . The comprehensive thermodynamic-kinetic approach treating the condensation and the partitioning of organic species originated from biogenic and anthropogenic sources allows us to investigate the full role of organic species in the growth of new particles, which is 24 https://doi.org/10.5194/acp-2020-759 Preprint. Discussion started: 27 August 2020 c Author(s) 2020. CC BY 4.0 License.
important for understanding the formation processes of particles relevant for radiative 725 forcing and clouds (Shrivastava et al., 2017).
The model with three nested domains was applied to simulate the aerosol components and PNSD in Megacity Beijing during a period of about a month. The simulation results were evaluated by the observations at the high level of IAP tower, which is more representative than the ground level in regional scale. The simulated 730 BC and OA components agreed well with the PMF analysis of AMS measurements.
The evolution of PNSD and NPF events were also nicely reproduced by the model.
Our modeling analyses showed that AOA accounts for the larger part of OA of SPs and thus significantly contributed to the growth of SPs in Beijing. Molteni et al. (2018) indicated highly oxygenated organic compounds formed from anthropogenic VOCs anthropogenic sources (Matsui et al., 2014). Together with these studies, our modeling Although the size distribution of primary emitted particles has large impact on the simulation of CN10 as suggested by other studies (e.g., Chang et al., 2009;Zhou et al., 2018), the simulation of the base experiment gave the better agreement with the observations than the sensitivity experiments and the conclusions 765 will not be changed. Even so, the importance of the size distribution of primary emitted particles should be emphasized. The global model results have suggested the high sensitivity of CCN to to the assumed emission size distribution (Lee et al., 2013).
Recently, Xausa et al. (2018) found that using the size-segregated primary particle evident influence on the particle properties and total SOA mass (Ervens et al., 2011) 785 and these processes can close the gap between the simulation and observation (Lin et al., 2014). It is necessary to refine the description of aerosol microphysical processes by including aqueous formation of SOA in our model.    Table 3 are also overlapped with shaded circles on the plots for comparison. 825 30 https://doi.org/10.5194/acp-2020-759 Preprint. Discussion started: 27 August 2020 c Author(s) 2020. CC BY 4.0 License. Fig. 8. Comparison of simulated and observed annual mean number concentrations of particles condensation larger than 10 nm at 34 sites listed in Table 3. The solid carmine line shows a 1:1 ratio and the dashed turquoise lines show ratios of 3:1, 2:1, 1:2, and 1:3.     Data availability. All of the observation in this paper are provided in the manuscript.