Seasonal characteristics of emission , distribution and radiative effect of marine organic aerosols 1 over the western Pacific Ocean : an analysis combining observations with regional modeling 2 3

Abstract. Organic aerosols from marine sources over the western Pacific Ocean of East Asia were investigated by using an online-coupled regional chemistry-climate model RIEMS-Chem for the entire year 2014. Model evaluation against a wide variety of observations from research cruises and in-situ measurements demonstrated a good skill of the model in simulating temporal variation and spatial distribution of particulate matter with aerodynamic diameter less than 2.5 μm and 10 μm (PM2.5 and PM10), black carbon (BC), organic carbon (OC), and aerosol optical depth (AOD) in marine atmosphere. The inclusion of marine organic aerosols apparently improved model performance on OC aerosol concentration, reducing the normalized mean biases from −19 % to −13 % (KEXUE-1 cruise) and −21 % to −3 % (Huaniao Island) over the marginal seas of east China, and from 33 % to 5 % (Dongfanghong II cruise) and from −13 % to 3 % (Chichijima Island) over remote oceans of the western Pacific. It was found that marine primary organic aerosol (MPOA) accounted for majority of marine organic aerosol (MOA) mass in the western Pacific. High MPOA emission mainly occurred over the marginal seas of China and remote oceans of the western Pacific northeast of Japan. The seasonality of MPOA emission is determined by the combined effect of Chlorophyll-a (Chl-a) concentration and sea salt emission flux, exhibiting the maximum in autumn and the minimum in summer in terms of domain average over the western Pacific. The annual mean MPOA emission rate was estimated to be 0.16×10−2 μg m−2 s−1, yielding an annual MPOA emission of 0.78 Tg yr−1 over the western Pacific, which potentially accounted for approximately 8~12 % of global annual MPOA emission. The regional and annual mean near surface MOA concentration was estimated to be 0.27 μg m−3 over the western Pacific, with the maximum in spring and the minimum in winter, resulting from the combined effect of MPOA emission, dry and wet depositions. Marine secondary organic aerosol (MSOA) produced by marine biogenic VOCs (isoprene and monoterpene) was approximately 1~2 orders of magnitude lower than MPOA. The simulated annual and regional mean MSOA was 2.2 ng m−3, with the maximum daily mean value up to 28 ng m−3 over the western Pacific in summer. MSOA had a distinct summer maximum and winter minimum in the western Pacific, generally consistent with the seasonality of marine isoprene emission flux. In terms of annual mean, 26 % of the total organic aerosol concentration was contributed by MOA over the western Pacific, with an increasing importance of MOA from the marginal seas of China (13 %) to remote oceans of the western Pacific (42 %). MOA induced a minor direct radiative effect (DRE), with a domain and annual mean of −0.21 W m−2 at the top of the atmosphere (TOA) under all-sky condition over the western Pacific, whereas the mean indirect radiative effect (IRE) due to MOA at TOA (IREMOA) was estimated to be −4.2 W m−2. MSOA contributed approximately 6 % of the annual and regional mean IREMOA over the western Pacific, with the maximum seasonal mean contribution up to 14 % in summer, which meant MPOA dominated the IREMOA. It was noteworthy that the IREMOA accounted for approximately 32 % of that due to all aerosols over the western Pacific of East Asia, indicating an important role of MOA in perturbing cloud properties and shortwave radiation in this region.



49
Atmospheric aerosol is one of the most important and uncertain factors in climate change issues 50 (IPCC, 2013). Aerosols can alter radiation balance by scattering/absorbing solar/infrared radiation, and 51 affect cloud microphysics and lifetime by activating as cloud condensation nuclei (CCN), exerting 52 significant effects on climate system directly and indirectly. Aerosols are originated from anthropogenic 53 and natural sources and of high spatial and temporal variability and short atmospheric lifetime relative 54 to greenhouse gases. Consequently, aerosol radiative and climatic effects often have strong regional 55 characteristics. 56 The western Pacific Ocean is frequently influenced by continental outflow of both anthropogenic 57 and natural aerosols. Due to continuous growth of economy and energy consumption in the past decades, 58 the aerosol level in China has been enhanced (Smith et al., 2011;Li et al., 2017) and may have 59 potentially significant effects on radiation and cloud over not only the East Asian continent but also the 60 https://doi.org/10.5194/acp-2020-1016 Preprint. Discussion started: 27 October 2020 c Author(s) 2020. CC BY 4.0 License. capture processes of particle by falling hydrometeor through Brownian and turbulent shear diffusion, 151 interception and inertial impaction, and is parameterized by a scavenging rate, which is a function of 152 precipitation rate and collision efficiency of particle by hydrometeor (Slinn, 1984). 153 Totally 10 aerosol types are simulated in RIEMS-Chem, which are sulfate (SO4 2-), nitrate (NO3 -), 154 ammonium (NH4 + ), black carbon (BC), primary organic aerosol (POA), secondary organic aerosol 155 (SOA), anthropogenic primary PMs (PM2.5 and PM10), dust, and sea salt. Sulfate is mainly produced 156 from the oxidation of SO2 by OH radical in gas phase and the oxidation of dissolved SO2 by H2O2, O3, 157 and metal catalysis in aqueous phase (Chang et al., 1987). Nitrate and ammonium are produced through 158 thermodynamic processes represented by the ISORROPIA II model (Fountoukis and Nenes, 2007). BC, 159 POA, and anthropogenic primary PMs are considered chemically inert. SOA formation from 160 anthropogenic and biogenic VOC precursors is treated by a bulk yield scheme from Lack et al. (2004), 161 with SOA yield of 424 μg m -3 ppm -1 for toluene, 342 μg m -3 ppm -1 for xylene, and 762 μg m -3 ppm -1 for 162 monoterpene. For irreversible conversion of marine VOCs to SOA, a 28.6% mass yield is assumed for 163 isoprene (Surratt et al., 2010, Meskhidze et al., 2011 and 30% for monoterpene (Lee et al., 2006). 164 Based on the observational analysis of aerosol mixing state in eastern China (Wu et al., 2017), an 165 internal mixing assumption is adopted for anthropogenic aerosols and they are externally mixed with   The size-resolved marine primary organic aerosol (MPOA) emission is parameterized based on the 207 method of Gantt et al. (2011;2012a). A briefly introduction is provided below. 208 The emission rate of MPOA is the product of sea salt emission rate (Ess) and organic matter fraction    sources.
357 Table 1 shows that for all the 9 EANET sites, the overall mean PM10 concentration was 30.0 g m -   Kanghwa, and Imsil) and lower concentrations at the remote island site (Ogasawara) over the western 370 Pacific, which were also reasonably reproduced by RIEMS-Chem.

371
Seasonal mean statistics of PM10 and PM2.5 concentrations at the EANET and CNEMC sites were 372 also listed in Table 1 and Table 2  In all, RIEMS-Chem was able to reproduce the spatial distribution and seasonal variation of PM10 381 and PM2.5 concentrations over the western Pacific. The good performances of PM10 and PM2.5 in the 382 marine environment of less anthropogenic source influence also imply that the model may be able to 383 reproduce sea salt reasonably well, which is essential to the estimation of marine MPOA emission.

384
In addition to the validation for PM concentrations, the model performances for gas precursors (O3, 385 SO2, and NOx/NO2) were also evaluated against hourly observations at the EANET sites (Table S1).   (Table 3).  (Table 3). The inclusion of marine-OC 458 (including both primary and secondary OC) reduced the model bias from -33% to -5% along the cruise.

459
The average contribution of marine-OC to the total OC mass in the marine atmosphere was     (Table 4).  (Table 4).  (Table 5). The largest 640 contribution in SON was associated with the relatively lower total OC concentration as shown in Figure   641 5c. The relative contribution from marine-OC to total OC at Chichijima Island resembled that at 642 Okinawa in terms of annual and season averages.  Table 6 shows the performance 708 statistics for hourly AOD at these AERONET sites. The overall annual mean AOD for the 6 sites was   (Table 7). It is interesting to note that the seasonal variation of 732 MPOA emission was not consistent with that of Chl-a concentration, which exhibited higher values in 733 SON and JJA and the lowest one in DJF (Table 7). This is because MPOA emission rate is determined  (Table 7). However, although 738 Chl-a concentration was also high in JJA (1.07 mg m -3 , Table 7), the sea salt flux was the minimum in 739 JJA (0.14 μg m -2 s -1 , Table 7) due to the weakest wind speed (3.0 m s -1 , Table 9), resulting in the lowest 740 MPOA emission in summer (Table 7). Although the sea salt emission flux reached the maximum in DJF 741 (Table 7) due to the largest wind speed in this season (Table 9), the winter Chl-a concentration was 742 lowest, leading to a moderate MPOA emission in winter (Table 7), in a similar magnitude to that in For the EYB region, the maximum MPOA emission occurred in winter (DJF) (Figure 7b and Table   747 7) with a seasonal and domain average of 1.2×10 -2 μg m -2 s -1 , which was 10 times larger than the 748 minimum of 0.12×10 -2 μg m -2 s -1 in summer (JJA) (Figure 7d and Table 7). Although Chl-a 749 concentrations were similar between DJF and JJA, the sea salt flux in DJF was approximately 9 times 750 that in JJA (Table 7). So, the seasonality of MPOA emission in the EYB region was mainly determined  (Table 7). It is interesting to note that although both the Chl-a  (Table 7). It is interesting to note that the August was mainly resulted from the strongest solar radiation in summer, although the Chl-a 796 concentration was not highest in this season in the EYB region (Table 7). Table S2 also  dominant fraction of MOA, which will be discussed below. Figure 8a shows that high MOA 814 concentrations mainly occurred over the EYB and NWP regions, with the annual and regional averages 815 being 0.48 μg m -3 and 0.59 μg m -3 , respectively (Table 8), accounting for 13% (6~30%) and 42% 816 (30~60%) of total OA mass in these two regions, respectively ( Figure 8k and Table 8 (Table 7), MOA concentration peaked in MAM (Table 8). It is noticed that precipitation 829 was lowest and wind speed was low in MAM (Figure 9c and 9h, Table 9), leading to a smaller dry speed and relatively more precipitation in DJF (Figure 9b and 9g, Table 9), the mean MOA concentration 833 was lowest in winter.

834
For the EYB region, northwesterly winds prevailed In DJF and SON and turned to northeasterly 835 winds over marginal seas of southeast China (Figure 9b and 9e), which transported MOA from the major MPOA source region (EYB) to the northern part of the South China Sea (Figure 8b and 8e). As wind 837 speed over the EYB was low in MAM and JJA (Figure 9c and 9d, Table 9), MOA was mainly restricted 838 within this region (Figure 8c and 8d). In terms of seasonal average, MOA concentration experienced its  (Table 8). The different seasonality 842 between MOA concentration (Table 8) and MPOA emission (Table 7) in the EYB region could also be 843 mainly attributed to meteorological conditions. The MPOA emission was relatively low in MAM (Table   844 7), but the second lowest wind speed and less precipitation (Table 9) (Table 7), the maximum wind speeds (Table 9) led to stronger 848 dry deposition of aerosols and thus a moderate MOA concentration in the two seasons (Table 8).  Table 7), which was just smaller than that in SON, and partly due to the relatively weak dry deposition 859 and wet scavenging caused by moderate wind speed and precipitation in this season (Table 9). In JJA, 860 although the MPOA emission was small, the lowest wind speed and precipitation in JJA over the NWP

861
(2.5 m s -1 and 3.7 cm grid -1 month -1 , Table 9) led to the weakest dry deposition and wet scavenging of Furthermore, the secondly largest precipitation over the NWP in SON (Table 9) caused strong wet 867 scavenging of particles, also contributed to the relatively low MOA level. In DJF, the wind speed was 868 largest, about 2 times those in the other seasons, and the precipitation was also the maximum (Table 9,  MOA concentration due to large marine emissions there; whereas, the large MOA/OA ratios over the 875 subtropical oceans of low latitude were mainly due to the low total OA level (small denominator).

876
Averaged over the NWP region, the annual mean MOA/OA ratio was 42%, with higher contributions 877 in MAM (52%) and SON (48%) and lower ones in DJF (36%) and JJA (32%) ( Table 8). Although MOA 878 concentration over the NWP was secondly highest in JJA, its contribution was small because OA 879 transported from land sources also subject to weak dry deposition and wet scavenging, which led to   (Table 8). The regional and seasonal averages of DREMOA over the NWP  (Table 10). The weaker DREMOA over the EYB (-0.24 W m -2 in terms of annual mean) than that over the NWP (-0.41 W m -2 ) could be attributed to both the lower MOA concentration 957 (Table 8) and lower relative humidity (73% vs 83%, Table 9).

958
It is of interest to estimate the relative importance of MOA in directly perturbing solar radiation 959 compared with that of total aerosols over the western Pacific.   (Table 10), which could be mainly due to the largest cloud fraction 997 in DJF ( Figure S1g), although MOA concentration was lower in winter (Table 8). On the contrary,  (Table 8, Figure S1i). In springtime when MOA concentration was highest over the western 1002 Pacific (Table 8), the domain and seasonal mean IREMOA can be as high as -14.8 W m -2 (Table 10).  (Table 10)

1122
The observational data can be accessed through contacting the corresponding author.