Environmental Effects on Aerosol-Cloud Interaction in non-precipitating MBL 1 Clouds over the Eastern North Atlantic 2 3

Abstract. Over the eastern north Atlantic (ENA) ocean, a total of 20 non-precipitating single-layer marine boundary layer (MBL) stratus and stratocumulus cloud cases are selected in order to investigate the impacts of the environmental variables on the aerosol-cloud interaction (ACIr) using the ground-based measurements from the Department of Energy Atmospheric Radiation Measurement (ARM) facility at the ENA site during the period 2016–2018. The ACIr represents the relative change of cloud-droplet effective radius re with respect to the relative change of cloud condensation nuclei (CCN) number concentration at 0.2 % supersaturation (NCCN,0.2 %) in the water vapor stratified environment. The ACIr values vary from −0.004 to 0.207 with increasing precipitable water vapor (PWV) conditions, indicating that re is more sensitive to the CCN loading under sufficient water vapor supply, owing to the combined effect of enhanced condensational growth and coalescence processes associated with higher NC and PWV. The environmental effects on ACIr are examined by stratifying the data into different lower tropospheric stability (LTS) and vertical component of turbulence kinetic energy (TKEw) regimes. The higher LTS normally associates with a more adiabatic cloud layer and a lower boundary layer and thus results in higher CCN to cloud droplet conversion and ACIr. The ACIr values under a range of PWV double from low TKEw to high TKEw regime, indicating a strong impact of turbulence on the ACIr. The stronger boundary layer turbulence represented by higher TKEw strengthens the connection and interaction between cloud microphysical properties and the underneath CCN and moisture sources. With sufficient water vapor and low CCN loading, the active coalescence process broadens the cloud droplet size distribution spectra, and consequently results in an enlargement of re. The enhanced NC conversion and condensational growth induced by more intrusions of CCN effectively decrease re, which jointly presents as the increased ACIr. The TKEw median value of 0.08 m2 s−2 suggests a feasible way in distinguishing the turbulence-enhanced aerosol-cloud interaction in non-precipitating MBL clouds.


In this study, the non-precipitating cloud periods are determined when the KAZR reflectivity at the 163 ceilometer-detected cloud base height range does not exceed -37 dBZ (Wu et al., 2015(Wu et al., , 2020b, which 164 extensively rules out the wet-scavenging depletion on below-cloud CCN (Wood, 2006)

Aerosol, cloud, and meteorological properties of selected cloud cases 170
A total of 20 non-precipitating cloud cases are selected to conduct this study, with the detailed time 171 periods listed in Table 1, including 1143 samples in temporal resolution of 5-min, which corresponds to 172 ~95 hours. Among the selected cases, there are three, eight, five, and four cases for Spring, Summer, 173 Fall, and Winter seasons, respectively. MBL clouds often produce precipitation in the form of drizzle 174 (Wood 2012, Wu et al., 2015, 2020b. A recent study of the seasonal variation of the drizzling frequencies 175 (Wu et al., 2020b) showed that the MBL clouds in cold months (Oct-Mar) have the highest drizzling 176 frequency of the year (~70%), while the clouds in warm months (Apr-Sept) are found to have a lower 177 chance of drizzling (~45%). Therefore, the selection of a non-precipitating single-layer low cloud case 178 that lasts at least 2 hours is limited, with only 6 cases found in the cold months and 14 cases found during 179 the warm months. 180 The distributions of the aerosol and cloud properties, and the environmental conditions for the 181 selected cases are shown in Fig. 1. The ,0.2% presents a normal distribution with a mean value of 182 215 cm −3 and median value of 217 cm −3 . About 97% of the ,0.2% samples lay below 350 cm −3 and 183 represents a relatively clean environment (Logan et al., 2014(Logan et al., , 2018  It is noteworthy that PWV values exhibit a bimodal distribution with a median value of 2.4 cm (Fig.  206 1f). About 43% of the samples have their PWV values in the range of 1.0 -2.0 cm with the first peak in 207 1.2 -1.6 cm, and 56% of the samples have PWV values higher than 2.2 cm with a second peak in 2.4 -208 2.8 cm, which may be due to the seasonal difference of the selected cases. Fig. S1 shows the seasonal 209 variation of the PWV from 2016 to 2018 when single-layer low clouds present. The monthly PWV values 210 are as low as ~ 1.7 mm and remain nearly invariant from January through March, then monotonically 211 increase up to ~ 3.4 cm (doubled) in August, and finally decrease dramatically to December. The selected 212 cloud cases are distributed across the seasons with ~34% of the samples occurring during the months 213 with the lowest mean PWV (Jan-Mar), while ~43% of the samples fall in the highest PWV months (Jun-214 Sept). These two obvious PWV regions will provide a great opportunity for us to further examine the 215 ACI under different water vapor conditions. 216 217 3.2 Dependent of cloud microphysical properties on CCN and PWV 218 Figure 2 shows the cloud microphysical properties as a function of ,0.2% and PWV for the 219 samples from 20 selected cases. As illustrated in Fig. 2a, there is a statistically significant positive 220 correlation ( 2 =0.9) between ( ) and ln( ,0.2% ). The linear fit of ( ) to ln( ,0.2% ) is then 221 mathematically transformed to a power-law fitting function of to ,0.2% , and plotted as dash lines 222 in Fig. 2a The PWV values are represented as blue circles (larger one for higher PWV) in Fig. 2a in order to 233 study the role of water vapor availability on the CCN-conversion process. As demonstrated in Fig.  234 2a, the PWV values almost mimic the increasing ,0.2% trend, which is also governed by the seasonal 235 ,0.2% and the selected cloud cases. appears to be flattened for higher ,0.2% and PWV bins. Furthermore, the joint power-law fitting of 251 (to ,0.2% and PWV) appears to be constantly lower than the single power-law fitting of (to 252 ,0.2% solely) in each bin. The negative power of PWV in this relationship suggests that PWV might 253 play a stabilization role in the diffusional growth process, which will be further analyzed in the following 254 sections. 255 The relationship between and ,0.2% is shown in Fig. 2b where there is no significant 256 relationship between with ,0.2% solely, given a near-zero slope and the low correlation coefficient 257 (fitted line not plotted). However, after applying a multiple linear regression to the logarithmic form of terms of the increase. The result is as expected because the process of condensational growth is 300 predominant in the near-adiabatic cloud, that is, with increasing water vapor supply, the higher CCN 301 loading can effectively lead to more cloud droplets. However, in the sub-adiabatic cloud regime, 302 increases with increased ,0.2% but possesses a negative correlation with PWV, which results in a 303 slow increase of . The mean reduction of in the sub-adiabatic regime is computed to be ~33% 304 compared to that for the near-adiabatic cloud. As previously studied, the coalescence process contributes 305 significantly to depletion, even in a non-precipitating marine boundary cloud (Feingold et al., 1996;306 Wood, 2006). Thus, the lower in the sub-adiabatic regime may be partly due to the combined effect 307 of coalescence and entrainment (Wood, 2006 (2) 319 The ACI r represents the relative change of with respect to the relative change of  The increasing trends of large occurrences mimic the trend of ACI r and suggest that with 358 increased PWV, cloud droplets have a greater chance to grow via the effective coalescence process and 359 subsequently lead to an enlargement of ACI r . Although previous studies have brought up the potential 360 impacts of cloud droplet coalescence process on ACI, it is rarely seen that the relationship among them 361 has been discussed in detail. Here we provide possible explanations on how the enhanced coalescence 362 process can enlarge ACI r . Quantitatively, ACI r is described by the log partial derivative ratio of to 363 ,0.2% , thus a sharper decrease of with respect to a given ,0.2% range can result in a steeper 364 slope and in turn, larger ACI r (i.e., a ,0.2% increase of 49% leads to a decrease of 41% in the 2.8-365 3.0 cm bin in Fig. 5a). Physically, this relies on how the cloud droplet size distribution spectra would 366 change with different CCN loadings. Therefore, particularly in low CCN conditions, sufficient water 367 vapor availability allows cloud droplets to continuously grow via diffusion of water vapor (i.e., 368 condensational growth), and enter the active cloud-droplet coalescence regime. In contrast, the increase 369 in cloud droplets can effectively reduce via the process of large cloud droplets collecting small 370 droplets, and small droplets coalescing into large droplets. Consequently, the size distribution spectra are 371 effectively broadened toward the large tail by the coalescence, so that is enlarged. With more CCN 372 available, the size distribution spectra are narrowed by the enhanced condensational growth and regress 373 toward the small tail by increasing the amount of newly converted cloud droplets and result in decreased 374 . These interactions between CCNs and cloud droplets ultimately result in the broadened changeable 375 range of , and in turn, the enlarged ACI r . 376 377

The role of lower tropospheric thermodynamics 379
The LTS parameter is used to infer the large-scale thermodynamic structures for the selected 380 cases in order to examine their impacts on ACI. The samples are separated into two regimes: high LTS 381 and low LTS using the median LTS value (19.1K) as a threshold. The values for the high LTS regime 382 are generally higher than those in the low LTS region (Fig. S3), though their difference is only 4.7%. 383 Since LTS is calculated by the difference between free tropospheric and surface potential temperatures, thin entrainment zone that restricts the effectiveness of the cloud-top entrainment. Moreover, a more 386 stable lower troposphere is prone to boundary layer cloud formation with a lower cloud base height and 387 accompanies a well-mixed boundary layer that couple the surface moisture and aloft (Klein and 388 Hartmann, 1993; Wood and Bretherton, 2006; Wood, 2012). Thus, the high LTS values are often found 389 to be associated with clouds that more close to adiabatic (Kim et al., 2008), which results in more 390 with less depletion. 391 To examine the impact of LTS on the water constrained ACI r , the samples are further separated by 392 the median PWV (2.4 cm) and median LTS, so each regime has ~25% of the total samples. As shown in 393 Fig. 6, where the regression lines for the four regimes are fitted to the 95% confidence interval, the ACI r 394 differences between low and high PWV regimes are still retained. In the low PWV regime, the ACI r 395 values are limited to 0.03 and 0.053 for low and high LTS regimes, respectively. In the high PWV regime, 396 the ACI r values are 0.154 and 0.171 for low and high LTS regimes, respectively, which is about 3-5 times 397 greater than those in low PWV regime. It appears that PWV plays a more important role in ACI r than 398 LTS since the LTS is mostly capturing the large-scale thermodynamical structures, and is obtained from 399 a coarser temporal resolution. Thus, the LTS does not essentially have strict correspondence to the 400 strength of boundary layer turbulence (which more directly interferes with the cloud processes). LTS 401 may not effectively represent the connection between cloud layer and boundary layer CCN and moisture 402 in terms of both spatial and temporal scales, and thus induces limitations in assessing the role of 403 thermodynamics on the ACI r . 404 405

The role of boundary layer turbulence 406
To examine the role of the dynamical factors on ACI, the TKE w parameter is used to represent 407 the intensity of below-cloud boundary layer turbulence. The median TKE w (0.08 m 2 s −2 ) is used to 408 separate the variation with ,0.2% for low and high TKE w regimes (Fig. S4). The values are 409 higher (with a mean increase of 20%) under high TKE w environments than those under lower TKE w , 410 across all CCN bins. The higher logarithmic ratios of to PWV for high TKE w regime suggest a more 411 sensitive response to CCN with an increased water vapor supply. This is mainly due to a closer 412 connection between the CCN below and the cloud layer loft, accompanied by stronger boundary layer 413 turbulence, so that more CCN can be converted into cloud droplets. When using the mean values of 215 414 cm −3 for ,0.2% and 2.2 cm for PWV as an example, the calculated values from the multiple 415 regressions are 83 and 68 cm −3 for high and low TKE w regimes, respectively. Thus, under the condition at given ,0.2% and PWV, the boundary layer with strengthened turbulence would be favorable for 417 more cloud droplets to be converted from CCN by water vapor condensation. 418 Similar to the aforementioned data separation method, the samples are further separated into four 419 regimes demarcated by PWV and TKE w in Fig. 7. Similar to the results in Fig. 6, the ACI r values in the 420 higher PWV regime are also much higher than those in the lower PWV no matter low or high TKE w 421 regimes, whereas TKE w plays a more important role in ACI r than LTS because the ACI r values in the 422 high TKE w regime are double those in the low TKE w regime. In the regimes of higher TKE w and PWV, 423 is highly sensitive to the CCN loading with the highest ACI r of 0.252. The sufficient water vapor 424 availability allows CCN to be converted into cloud droplets more effectively, while the relatively higher 425 Under high PWV but low TKE w conditions, the mean ACI r reduces to 0.125 (~ 50% of that under 436 high TKE w ). The weaker turbulence loosens the connection between cloud layer and the underlying 437 boundary layer, results in a less effective conversion of CCN into cloud droplets, and then diminishes 438 the sensitivity to CCN. Although the constraints of insufficient water vapor on the ACI r are still 439 evident, the ACI r value is doubled from 0.017 in low TKE w regime to 0.035 in high TKE w regime. The 440 ACI r differences between the two TKE w regimes attest that ACI r strongly depends on the connection 441 between the cloud layer and the boundary layer CCN and moisture, that is, strong turbulence can enhance 442 the susceptibility of to CCN. Given the significant increase of ACI r , the TKE w demarcation line of 443 0.08 m 2 s −2 , which corresponds to the mean vertical velocity variation of 0.16 m 2 s −2 , may be a feasible 444 way to distinguish the impact of turbulence effect on the cloud microphysical responses to the change in 445 CCN loadings. 446 In this study, the relationship between turbulence and ACI is found to be valid in non-precipitating 447 marine boundary-layer clouds. Theoretically, the effect of turbulence on ACI r would appear to be artificially amplified, if in the presence of precipitation. The intensive turbulence can enhance the 449 coalescence process and accelerate the CCN-cloud cycling, and subsequently, the CCN depletion due to 450 precipitation and coalescence scavenging would result in quantitatively enlarged ACI r (Feingold et al., 451 1996(Feingold et al., 451 , 1999

When
,0.2% is greater than 250 cm −3 and PWV values are also relatively high, appears to 466 decrease with increasing ,0.2% under similar water vapor conditions. In a more adiabatic cloud 467 vertical structure, the cloud droplet is dominated by condensational growth, so responses to increased 468 ,0.2% and PWV are strengthened. When the cloud layer become more sub-adiabatic, the effect of 469 coalescence leads to the depletion of and thus, the competition between the condensational growth 470 and coalescence processes has a strong impact on the variations of cloud microphysics to CCN loading. 471 The ACI r values vary from -0.004 to 0.207 for different PWV conditions where the ACI r appears to 472 be diminished under limited water vapor availability due to the limited droplet activation and 473 condensational growth process. While under relatively sufficient water supply condition, shows more 474 sensitive responses to the changes of ,0.2% , due to the combined effect of condensational growth and 475 coalescence processes accompanying the higher and PWV. The coalescence process further enlarges 476 , particularly in low CCN loading, while the enhanced condensational growth narrows the cloud DSD 477 and decreases , so that a broader variable range of with respect to ,0.2% change results in a 478 higher ACI r value.
To investigate the impacts of environmental effect on the ACI r , the LTS parameter is used as a proxy 480 of the thermodynamic structure. A higher LTS regime is favorable to the adiabatic cloud with lower 481 cloud base height, accompanied by a well-mixed boundary layer, which likely enhances the cloud 482 microphysical responses to CCN loadings. However, the ACI r in different LTS regimes cannot be 483 distinctly differentiated, partly due to the competing effect of adiabaticities and turbulence characteristics 484 on the cloud droplet development processes. 485 In contrast, the intensity of boundary layer turbulence, which is represented by TKE w , plays a more 486 important role in ACI r than LTS. The shows more sensitive response to CCN with increased water 487 vapor supply for the higher TKE w regime, which may be due to enhanced CCN to cloud droplet 488 conversion induced by intensive boundary layer turbulence. As for ACI r , assessments in different TKE w 489 and PWV regimes, the constraints of insufficient water vapor on the ACI r are still evident, but in both 490 PWV regimes the ACI r values increase more than double when going from low TKE w to high TKE w 491 regimes. Noticeably, the ACI r increases from 0.125 in low TKE w regime to 0.252 in high TKE w regime, 492 under high PWV conditions. The intensive below-cloud boundary layer turbulence strengthens the 493 connection between the cloud layer and below-cloud CCN and moisture. So that with sufficient water 494 vapor, an active coalescence leads to further enlarged , particularly for low CCN loading condition, 495 while the enhanced from condensational growth induced by increased ,0.2% can effectively 496 decrease . Combining these processes together, the enlarged ACI r is presented. 497 In this study, the non-precipitating MBL clouds are found to be most susceptible to the below-cloud 498