Impacts of long-range transport of aerosols on marine-boundary-layer clouds in the eastern North Atlantic

Vertical profiles of aerosols are inadequately observed and poorly represented in climate models, contributing to the current large uncertainty associated with aerosol– cloud interactions. The US Department of Energy (DOE) Atmospheric Radiation Measurement (ARM) Aerosol and Cloud Experiments in the Eastern North Atlantic (ACEENA) aircraft field campaign near the Azores islands provided ample observations of vertical distributions of aerosol and cloud properties. Here we utilize the in situ aircraft measurements from the ACE-ENA and ground-based remotesensing data along with an aerosol-aware Weather Research and Forecast (WRF) model to characterize the aerosols due to long-range transport over a remote region and to assess their possible influence on marine-boundary-layer (MBL) clouds. The vertical profiles of aerosol and cloud properties measured via aircraft during the ACE-ENA campaign provide detailed information revealing the physical contact between transported aerosols and MBL clouds. The European Centre for Medium-Range Weather Forecasts Copernicus Atmosphere Monitoring Service (ECMWF-CAMS) aerosol reanalysis data can reproduce the key features of aerosol vertical profiles in the remote region. The cloud-resolving WRF sensitivity experiments with distinctive aerosol profiles suggest that the transported aerosols and MBL cloud interactions (ACIs) require not only aerosol plumes to get close to the marine-boundary-layer top but also large cloud top height variations. Based on those criteria, the observations show that the occurrence of ACIs involving the transport of aerosol over the eastern North Atlantic (ENA) is about 62 % in summer. For the case with noticeable long-rangetransport aerosol effects on MBL clouds, the susceptibilities of droplet effective radius and liquid water content are −0.11 and +0.14, respectively. When varying by a similar magnitude, aerosols originating from the boundary layer exert larger microphysical influence on MBL clouds than those entrained from the free troposphere. 1 Motivation and background It has been long hypothesized that increased high concentrations of aerosols serving as cloud condensation nuclei (CCNs) can reduce cloud droplet effective radius, enhance cloud albedo, suppress drizzle formation, and change cloud lifetime and fraction – the so-called aerosol indirect effects (AIEs; Twomey, 1977; Seinfeld et al., 2016). However, current radiative forcing stemming from cloud responses to anthropogenic aerosols remains highly uncertain in the climate system, representing the largest challenge in climate predictions (Fan et al., 2016). Note that the current IPCC assessment mainly considers the warm stratus and stratocumulus responses to aerosols, while aerosol-induced convective cloud response (Wang et al., 2014) and anthropogenic aerosol effects such as ice nuclei (Zhao et al., 2019) have not been fully accounted for yet. Even for warm clouds, the climate significance of whether liquid water content and cloud lifetime are enhanced or reduced by CCNs is still widely debated (Malavelle et al., 2017; Toll et al., 2019; Rosenfeld et al., 2019). Due to the nonlinear nature of cloud responses to CCN perturbations, the largest cloud susceptibility and AIEs Published by Copernicus Publications on behalf of the European Geosciences Union. 14742 Y. Wang et al.: Impacts of long-range transport of aerosols typically occur for the marine-boundary-layer (MBL) clouds over remote regions (Garrett and Hobbs, 1995; Carslaw et al., 2013; Dong et al., 2015). Under pristine conditions with extremely low background CCN concentration (Kristensen et al., 2016), any aerosol intrusion following long-range transport has great potential to alter the local aerosol and CCN budget (Roberts et al., 2006). Hence, in this study, we aim to characterize long-range transport of aerosols and to assess their impacts on MBL clouds by combining in situ aircraft measurements with cloud-resolving model simulations. For those aerosols resulting from long-range transport, one of the most important aspects pertinent to aerosol–cloud interactions (ACIs) is their vertical distribution, or in other words, their position relative to cloud layers. The vertical distribution of aerosols can be affected by a number of complex atmospheric processes, such as emission, transport, and deposition as well as microphysical and chemical processes. Previous studies suggest that aerosols can alter MBL cloud microphysical properties and enhance indirect effects through entrainment into the cloud top either when aerosol particles settle or the cloud deck deepens (Painemal et al., 2014; Lu et al., 2018). In the boundary layer of remote regions like the equatorial Pacific, the majority of CCNs were found to be supplied by long-range transport instead of local emission or formation (Clarke et al., 2013). In the northeast Pacific, where aerosol types are similar to the Azores, biomass-burning aerosols from the episodic wildfire events are found to be less efficient in altering cloud microphysics than the nonbiomass-burning aerosols (Hossein Mardi et al., 2019). Recent aircraft observations from the NASA’s ORACLES (ObseRvations of Aerosols above CLouds and their intEractionS) campaign showed distinctive MBL cloud responses to aerosols above and below clouds depending on the history of smoke entrainment (Diamond et al., 2018). Therefore, it is critical to understand aerosol variability as a function of height and its influence on the aerosol indirect forcing assessment over the regions where MBL clouds are abundant. Spaceborne active sensors that possess vertically profiling capabilities have been widely used to characterize aerosol and cloud spatial variations and to detect the aerosol above clouds (Painemal et al., 2014; Jiang et al., 2018). However, satellites likely miss the thin aerosol layers with relatively low concentration (but still higher than maritime background values), and thus overestimate the distance between the aerosol plume base and the cloud top. Also, when plumes are too thick near the aerosol source regions, satellite signals will be saturated, and the retrievals may underestimate the extent of thick layers (Rajapakshe et al., 2017). Therefore, aircraft observations with continuous vertical sampling are the most reliable source that can accurately characterize the vertical relationship between aerosol and cloud. The DOE ARM Aerosol and Cloud Experiments in the Eastern North Atlantic (ACE-ENA) aircraft field campaign near the Azores islands provided a unique opportunity to study aerosols from different sources and their impacts on MBL clouds (Wang et al., 2019). The eastern North Atlantic (ENA) site is located in the remote northeastern Atlantic Ocean, where MBL clouds are prevalent throughout the year due to the warm sea surface temperature and prevailing subsidence near the edge of the Hadley cell (Wood et al., 2015; Dong et al., 2015). The site also receives complex air mass dictated by different wind patterns. In addition to the local maritime air, the airflows originating from either the North American or the Saharan region complicate the local aerosol types and sources (Logan et al., 2014). This study leverages the airborne measurements of aerosol vertical profiles for different chemical species to understand aerosols and their influence on MBL cloud microphysical properties over the Azores, with the ultimate goal of providing observational constraints on the global climate model simulations. An aerosol reanalysis product is evaluated in the present study as well. Even with the aircraft-measured vertical relationship between aerosol and cloud, it is difficult to estimate whether the aerosol aloft can impact the cloud beneath as the microphysical processes such as entrainment into the cloud top cannot be directly measured. Hence, we employ an aerosolaware cloud-resolving model to simulate the MBL cloud development and aerosol transport in the free troposphere and to quantify the AIEs. Through the sensitivity experiment by imposing different aerosol vertical profiles, we can disentangle aerosol and other confounding meteorological factors in ACIs, which is challenging to do using only short-term observations. Section 2 describes the main observational data and introduces the numerical modeling tools. Section 3 reports the observed aerosols and clouds based on aircraft measurements and reanalysis products. Section 4 presents the analyses of cloud-resolving simulations using the Weather Research and Forecast (WRF) model. Section 5 summarizes the key findings in this study and provides additional discussions for the study’s caveats and future work.

clouds. The ECMWF-CAMS aerosol reanalysis data can reproduce the key features of aerosol 23 vertical profiles in the remote region. The cloud-resolving WRF sensitivity experiments with 24 distinctive aerosol profiles suggest that the transported aerosols and MBL cloud interactions (ACI) 25 require not only low-altitude aerosol preferably getting close to the marine boundary layer top, but 26 also large cloud top height variations. Based on those criteria, the observations show the 27 occurrence of ACI involving the transport of aerosol over the Eastern North Atlantic is about 62% 28 in summer. For the case with noticeable long-range transport aerosol effect on MBL cloud, the 29 susceptibilities of droplet effective radius and liquid water content are −0.11 and +0.14, 30 respectively. When varying on the similar magnitude, aerosols originating from the boundary layer 31 exert larger microphysical influence on MBL clouds than those entrained from free troposphere. 32

Motivation and Background 33
It has been long hypothesized that increased high concentrations of aerosols serving as 34 cloud condensation nuclei (CCN) can reduce cloud droplet effective radius, enhance cloud albedo, 35 suppress drizzle formation, and change cloud lifetime and fraction, the so-called aerosol indirect 36 effects (AIE) (Twomey, 1977;Seinfeld et al., 2016). However, current radiative forcing stemming 37 from cloud responses to anthropogenic aerosols remains highly uncertain in the climate system, 38 representing the largest challenge in climate predictions (Fan et al., 2016). Note that the current 39 IPCC assessment mainly considers the warm stratus and stratocumulus response to aerosols 40 (Myhre et al., IPCC, 2013), while aerosol induced convective cloud response (Wang et al., 2014) 41 as well as with anthropogenic aerosol effect as ice nuclei (Zhao et al., 2019) have not been fully 42 accounted for yet. Even for warm clouds, the climate significance of whether liquid water content 43 and cloud lifetime are enhanced or reduced by CCN is still widely debated (Malavelle et al., 2017;44 Toll et al., 2019;Rosenfeld et al., 2019). Due to the nonlinear nature of cloud responses to CCN 45 perturbations, the largest cloud susceptibility and AIE typically occurs for marine boundary layer 46 (MBL) clouds over remote regions (Garrett and Hobbs, 1995;Carslaw et al., 2014;Dong et al., 47 18 and 12 presents the typical high-and low-plume cases, respectively, so they will be investigated 188 thoroughly in the later aircraft data analyses and model simulations. The concentrations of OC, 189 BC, and sulfate are generally low in the MBL, so aerosol penetration from the free troposphere 190 into the lower MBL may be not significant during this month. One exception is sulfate during 18-191 21 July. Sulfate concentration experienced an increase in the MBL followed by a lag increase in 192 the free troposphere. Since there is no significant transport signal before and during that time 193 period, the elevated sulfate concentration within the boundary layer is due likely to some local 194 sources such as oxidation of marine dimethyl sulfate (DMS). 195 The aerosols of natural sources, namely sea salt and dust, show different vertical 196 distributions . Sea salt aerosols mainly reside near the surface and are rarely found 197 above 1000 m. Dust particles are mainly found at high altitudes, typically above 3 km, during  14 July, indicating their long-range transport. However, the dust spatiotemporal pattern in the free 199 troposphere are quite distinctive from sulfate and smoke, implying the different sources of long-200 range transport. Previous studies suggest the possible dust transport from the Saharan Desert to 201 the northeast Atlantic region (Logan et al., 2014;Weinzierl et al., 2015). To address those issues, 202 back-trajectory analyses were conducted, and the results will be discussed later. During 15-19 July, 203 dust particles are found within the boundary layer and even near the surface following the presence 204 of dust plume in the free troposphere earlier. Such a downward propagation does not occur for 205 anthropogenic aerosols either, likely explained by the fact that dust particles are bigger in size with 206 larger settling velocity. 207

Identification of source regions using back-trajectory analysis 208
The backward ensemble trajectories were computed using the NOAA Hybrid Single-209 Particle Lagrangian Integrated Trajectory (HYSPLIT) (Stein et al., 2015) model, based on the 210 large-scale meteorological fields from Global Data Assimilation System (GDAS) with a spatial 211 resolution of 0.5°. We focus on three cases/days to examine the sources of typical high-and low-212 altitude plumes of anthropogenic aerosols and mineral dust. The model uses an end-point height 213 of 1.5, 2.4, and 3 km for three selected cases to represent the air parcels in the anthropogenic low-214 altitude, high-altitude, and dust plumes, respectively. To capture the different lengths of transport 215 procedure, the model was backward integrated for 7 days for the anthropogenic aerosols and 13 216 days for the mineral dust case. 20 ensemble members are employed for each case. They agree with 217 each other better on horizontal trajectory than vertical displacement. Larger differences are found among the ensemble members after three days for anthropogenic aerosols and after two days for 219 dust. 220 The back-trajectory analyses confirm that the source region of sulfate, BC, and OC in the 221 plumes is the North American continent (Fig. 2a,c), consistent with previous analyses of data from 222 the earlier field campaign over the ENA site (Logan et al., 2014). The westerly jet carries the 223 pollutants across the Atlantic Ocean, and it takes three to four days to arrive the Azores. Temporal 224 evolutions of trajectory vertical displacement reveal when aerosols are elevated from the PBL to 225 the free troposphere and such information can be used to pinpoint the aerosol source. Fig. 2b,d 226 suggests that aerosols are mainly from the central US in the high-plume case, and from eastern US 227 in the low-plume case. The curved trajectories in the low-plume case reflect the influence of the 228 Bermuda/Azores High located to the south. The dust transports exhibit a much different pathway. 229 Starting at 3km altitude, the back-trajectory develops westward initially, but sharply turn around 230 and point to the North Africa (Fig. 2e,f). It suggests that Sahara is the most likely source for the 231 dust particles observed over the Azores. 232 Note that back-trajectory analysis of air mass has its own limitations. For example, 233 shipping emissions over Northern Atlantic Ocean are not considered in the present analysis. Also, 234 the source attribution based on episodic events may be not representative for the climatological 235 mean scenario. Therefore, the source attribution results here need to be further evaluated in future 236 studies which can utilize more sophisticated approach such as source tagging in the GCM nudged 237 by the reanalysis data (Wang et al., 2014). 238

Vertical distributions of different aerosols in aircraft observations 239
Aircraft observations during the ACE-ENA provide more accurate depictions of aerosol 240 vertical distribution and aerosol layer heights relative to cloud layer heights, with differentiation 241 of aerosols type and hygroscopicity. During the summer IOP, quite diverse aerosol vertical profiles 242 are found. Here we focus on those with noticeable aerosol plumes in the free troposphere. Fig. 3  243 shows two representative vertical distributions of aerosol mass concentrations averaged over the 244 flights on July 18 and 12, corresponding to the high-and low-altitude aerosol plume, respectively. 245 In the high-altitude plume case, BC, OC, and sulfate concentrations all increase with height above 246 clouds, indicating downward propagation of aerosol plumes and possible interaction with MBL 247 clouds. BC and OC concentrations are even higher than that of sulfate in the free troposphere, 248 suggesting the biomass burning signature of the plume on that day. Conversely, within MBL, much higher concentration of sulfate in the MBL than those of BC and OC. This phenomenon is also 250 captured by the CAMS aerosol reanalysis (Fig. 1a), lending support to the fidelity of the reanalysis 251 dataset. For the low altitude plume (Fig. 3b), the vertical gradients of aerosol concentrations are 252 not clear above clouds, but aerosol concentrations within 500 m right above clouds are higher than 253 those near the cloud base ( Fig. 3b), corroborating the physical contact between aerosol plumes and 254 MBL clouds. Comparing Fig. 3 and 1, the CAMS reanalysis data generally agree with aircraft 255 observed aerosol profiles on the selected days, but the predicted aerosol mass mixing ratios are an 256 order of magnitude higher in the reanalysis data. Those discrepancies point out that any 257 quantitative usage of aerosol reanalysis product should be cautious. 258 Aerosol and CCN concentration vertical profiles are also available from the aircraft 259 observations. For the high-altitude plume, NCN reaches a peak of ~ 600 cm -3 at 2.5 km, and then 260 decreases dramatically downwards to ~180 cm -3 near cloud top (~ 1.1 km), which is even lower 261 than NCN values within the boundary layer ranging from 200 to 300 cm -3 on that day (Figure 4a). 262 The measured 200-m average of NCN above cloud top is 185 cm -3 , smaller than that below cloud 263 base 290 cm -3 (Table 1). From the surface to the 2.5 km height, the minimum NCN occurs near 264 cloud top, reflecting the disconnection between MBL aerosols and those from long-range transport 265 aloft. The characteristics of NCCN profile are similar with those of NCN. In the low-altitude plume, 266 both NCN and NCCN show a slower decline of above the cloud layer ( Fig. 4c,d). Also, the right-267 above-cloud-top NCN and NCCN at 1 km are higher than those below the cloud layer, indicating the 268 physical contact of the aerosol plume with the cloud deck. 269 During the summer IOP, the aircraft was deployed in twenty days to collect data. Among 270 those days, only eight of them have stable MBL clouds during the flight hours, according to the 271 ground-based cloud radar. We summarize the aircraft observed aerosol and cloud vertical 272 distribution characterizations for those eight days/cases in Table 1 The large-scale wind pattern and boundary layer structure from the model control run are 294 compared against the interpolated soundings over the ARM ENA site. Fig. 6 shows that the model 295 exhibits good agreement with the observed air temperature, moisture content, and relative 296 humidity. The model captures the cold/dry air advection at 1 km height in the morning followed 297 by the warm/moist air in the afternoon. The persistent supersaturation between 500 and 1000 m 298 and associated cloud deck are also reproduced in the simulation. We find that the key model 299 configuration to reproduce the main features of meteorological variability is to have appropriate 300 domain nesting and dynamical downscaling. Particularly, the outmost domain with 19.2 km grid 301 spacing is crucial and necessary for this mid-latitude region. The region is featured by frequent 302 mesoscale weather systems, and local wind and moisture fields vary drastically even within a day. 303 The model setup with only three domains of 4.8 km, 1.2 km, and 300 m horizontal resolution 304 induce large errors in the vertical profiles of moisture and temperature. A persistent dry bias occurs 305 near the MBL top when the outmost domain with 19.2 km grid spacing is absent. Such 306 meteorological biases further influence cloud simulation and result in discontinuous cloud layer in 307 its temporal evolution. 308 MBL cloud properties simulated by WRF are evaluated against the retrievals from a 309 combination of ground-based observations. The simulation captures the cloud top height at 1km 310 and cloud bottom height at 500 m during the day (Fig. 7a,b). Therefore, the cloud physical thickness is comparable between model and observation. LWC is generally smaller in the model 312 than that in the observation. Meanwhile, the simulation captures the larger LWC near the top of 313 the cloud, reflecting the adiabatic growth of cloud droplet starting from the cloud bottom. The 314 temporal evolution of simulated LWCs does not match well with retrievals, partly due to the spatial 315 sampling bias. Cloud droplet effective radius (Re) in the model is calculated as a function of 316 volume-mean droplet radius as well as relative dispersion (a ratio between standard deviation and 317 mean radius in a size distribution) (Liu and Daum, 2002). The model shows the comparable 318 vertical distribution of Re with cloud radar retrievals, e.g. the larger Re near the cloud top, but with 319 larger variability in the size range than observations (Fig. 7c,d). 320 To explore the sensitivity of MBL cloud microphysical properties to the long-range aerosol 321 transport, we contrast the simulations with and without observed long-range aerosol plumes in the 322 free troposphere. For the high-altitude plume (July 18) case, the comparisons of model run with 323 different aerosol vertical profiles show that both LWC and cloud fraction remain largely 324 unchanged, whether the aerosol plume above 1.5 km exists or not. In fact, the cloud top height on 325 that day experienced some temporal variations near the Azores, as it extended to 1.5 km during 326 the night due to strong radiative cooling and reduced to 1 km during the most of daytime. As a 327 result, the distance between the aerosol plume and cloud deck varied from 500 m to less than 100 328 m. Fig. 8a-f show that the long-range transported aerosols have no significant impacts on the MBL 329 cloud properties underneath when the physical distance between aerosol plume and cloud layer is 330 greater than 100 m. This finding does not support the previous study based on satellite products 331 arguing that aerosol-cloud interactions are still discernable with aerosol plumes 1 km above the 332 cloud deck (Painemal et al. 2014). 333 To answer the question at what height aerosol plume starts to influence MBL cloud 334 microphysical properties, we perform an additional simulation by lowering the aerosol plume 335 bottom from 1.5 km to 1.1 km which is considered as the height of MBL and cloud tops during 336 the daytime. In this sensitivity run, the aerosol indirect effect remains largely muted during the 337 daytime. It suggests that when boundary layers and cloud decks are relatively stable, long-range 338 transport aerosols have a low chance of being entrained into the cloud top and being activated to 339 cloud droplets. However, when the cloud deck becomes deeper at night, particularly after 2200 340 UTC when a significant part of the cloud extends into the aerosol layer above 1.1 km, an increase 341 in LWC by up to 0.1 g m -3 is observed (Fig. 8g-h).
In contrast, the simulated clouds in the low-altitude plume (July 12) case exhibit large 343 variations in the vertical (Fig. 9), and consequently the aerosol plume just above the cloud top 344 imposes significant influence on the MBL cloud micro-and macro-physical properties. The mean 345 LWC is increased by 5.7%, and cloud fraction is increased by 5.4%, due to a 48.0% increase in 346 CCN under the influence of the long-range aerosol transport. The distinctive responses of MBL 347 clouds to aerosol plumes at different heights reinforce the notion that the vertical overlap between 348 aerosol and cloud layers is crucial for ACI pertinent to the long-range aerosol transport. Moreover, 349 the extent of overlap is jointly controlled by aerosol plume height and cloud top variation. The 350 latter is particularly important, when the boundary layer is relatively stable, and the aerosol vertical 351 mixing is rather weak for most marine stratus. 352 It is a nontrivial task to identify the physical contact between an aerosol plume and a cloud 353 deck based on the aircraft measurements. Especially when the center of an aerosol plume is 354 hundreds of meters above cloud top and aerosol concentration right above the cloud is lower than 355 that within PBL, it is difficult to estimate whether aerosols can be entrained into the cloud layer. 356 As the above model results suggested, ACI requires critical mass of aerosols immersed into the 357 cloud layers. Here we define a "critical altitude" at which above-cloud NCN is equal to the below-358 cloud NCN. With such a concept, we can compare this altitude to the cloud top variation during a 359 period of interest. Take the July 18 case for example, according to the airborne measurements, the 360 critical altitude is 1674 m, well beyond the range of cloud top variation (880 -1300 m) on that 361 day (Table 1). Thus, we can reach a conclusion that, even though long transport of aerosols was 362 found in the free troposphere on that day, they were unlikely to interfere with MBL clouds below. 363 Here we take all the airborne measured vertical information into account, including aerosol 364 changes above clouds, comparison of above-and below-cloud NCN, as well as cloud top height 365 variations, and We revisit the eight observed cases in Table 1. We find that five days (0628, 0630, 366 0706, 0712, and 0715) out of eight during the summer phase of the ACE-ENA field campaign 367 clearly show the interactions between aerosols from long-range transport and local MBL clouds, 368 corresponding to a 62.5% occurrence frequency. 369 The previous cloud-resolving modeling studies of aerosol effects on MBL cloud properties 370 either used a constant CCN concentration throughout the whole domain (Yamaguchi et al., 2019) 371 or the CCN profiles in MBL were prescribed with an exponential decrease in the free troposphere 372 by perturbing CCN at different heights with the same scaling factor, without differentiating the 374 aerosols from different sources. Therefore, those studies share a common assumption that the 375 CCNs are solely from a local source impacted by local boundary layer processes. Here we repeat 376 this type of CCN perturbation experiment and compare the resultant aerosol effects with our 377 current assessment for the effects of long-rang transported aerosols only. Three idealized CCN 378 profiles are used for the July 18 cases. The cloud susceptibility (ratio between logarithmic cloud 379 property change and logarithmic CCN change) derived from the comparison of those three 380 idealized runs are found to range from −0.22 to −0.25 for Re and from +0.18 to +0.30 for LWC 381 ( Fig. 10a-b). Both Re and LWC susceptibility values are close to the high ends of the most of 382 current AIE assessments (Sato and Suzuki, 2018;Zheng et al., 2020). For the noticeable long-383 range transport effect in the July 12 case, the Re and LWC susceptibilities are −0.11 and +0.14, 384 respectively. They are smaller than those from the idealized MBL aerosol perturbation experiments. 385 Hence, this suggests that the aerosols of long-range transport are less efficient in altering MBL 386 cloud properties than those originating from local sources. It can be attributed to the fact that dry 387 air likely enters cloud layer along with CCN, resulting in less supersaturation and reduced 388 activation rate. 389

Conclusion and Discussion 390
Located in the remote eastern North Atlantic, the Azores islands experience frequent long-391 range transport of smoke and anthropogenic aerosols from continental U.S. A recent DOE ARM 392 ACE-ENA aircraft field campaign near the Azores in the summer of 2017 provides ample 393 observations of aerosols and clouds with detailed vertical information. In this study, we combine 394 the aircraft measurements, CAMS aerosol reanalysis, and an aerosol-aware and cloud-resolving 395 WRF model to characterize spatial variations of aerosols from long-range transport over the 396 Azores islands and assess their possible influence on the marine boundary layer clouds. The 397 reanalysis data show high frequency of occurrence of long-range transport over this area. 398 Evaluated by airborne aerosol measurement, the CAMS reanalysis data generally reproduce 399 observed aerosol profiles over this remote region, but the predicted aerosol mass mixing ratios are 400 still significantly biased. Our back-trajectory analyses confirm that anthropogenic and/or biomass 401 burning aerosols were mainly from the U.S. continent during the summer phase of ACE-ENA, 402 while the dust plumes are mainly originated from Sahara. importance of long-range transport aerosols on MBL clouds. Note that, due to the limited sample 430 size, the frequency may not be accurate to represent the true value on the daily basis. To our 431 knowledge, our study represents the first effort to utilize the ACE-ENA aircraft campaign data to 432 study the impacts of long-range transported aerosols on MBL clouds. Future study will focus on 433 the comparison of AIE involving long-range transport aerosols between different ARM sites and 434 field campaigns.