Meteorol Atmos Phys (2015) 127:625–634 DOI 10.1007/s00703-015-0392-2 ORIGINAL PAPER Understanding the synoptic variability of stratocumulus cloud liquid water path over the Southeastern Pacific Lanxi Min1,3 • Wei Gong1 • Guangyi Liu2 • Qilong Min1,3 Received: 2 April 2015 / Accepted: 2 July 2015 / Published online: 17 July 2015 Springer-Verlag Wien 2015 Abstract The spatial and temporal variance of stratocumulus cloud liquid water path (CLWP) over Southeastern Pacific has been investigated by combining satellite moderate resolution imaging spectroradiometer cloud products, CLWP from advanced microwave scanning radiometerEOS observations and NCEP final analysis atmospheric products with empirical orthogonal function (EOF) Analysis. CLWP variance is the most complicated factor among three fundamental cloud quantities (the microphysical cloud droplet concentration, and the macrophysical CLWP and cloud fractional cover). The results show that EOF/ PC1 of CLWP represents the variation of domain-averaged CLWP, which is mainly controlled by surface meteorological factors. Sea surface temperature and cold advection drive the synoptic and seasonal scales of variance of CLWP, while surface wind speed plays a fundamental role in stratocumulus cloud formation and daily variance of CLWP. EOF/PC2 of CLWP describes the spatial variance of CLWP. This daily spatial variance of CLWP is controlled by the factors of lower tropospheric stability and cloud top relative humidity, which determine cloud thickness and, consequently, CLWP through thermodynamic and entrainment processes. Further study indicates a Responsible Editor: S. Hong. & Qilong Min qmin@albany.edu 1 State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan, China 2 Smart Grid Operation Research Center, China Electric Power Research Institute, Beijing, China 3 Atmospheric Science Research Center, State University of New York, Albany, NY 12203, USA twofold interaction of the surface wind speed on stratocumulus CLWP: (1) dynamically through modulation of surface latent heat and sensible heat fluxes and (2) microphysically through enhanced marine aerosol production. 1 Introduction Stratocumulus clouds play an important role in the global radiation budget system. The stratocumulus decks reflect solar radiation back to space, causing a cooling effect. Unique climatological conditions over the southeast pacific (SEP) support the formation of the world’s most persistent stratocumulus clouds (Bretherton et al. 2004). During Austral spring (August to October), the Andes forms a sharp barrier to the warm easterly wind from South American, helping to maintain strong divergence and temperature inversion in the southeast pacific region (Xu et al. 2004). Stronger winds are generated in this process, bringing cold water to the surface to sustain a lower sea surface temperature. The combination of cold SST and subsidence aloft is ideal for the persistence of large-scale stratocumulus clouds over the southeast pacific of South America (Wood et al. 2011). Furthermore, an anomalous shallow meridional circulation developed through the radiative cooling effect of these clouds enhance the subsidence over the SEP region (Wang et al. 2004a, b, 2005), helping to maintain the stratocumulus clouds there. However, the covarying impacts of meteorology and aerosols on those clouds and the complex interaction between clouds and the turbulent boundary layer (Lin et al. 2009; Stevens and Feingold 2009) make it hard to understand the SEP stratocumulus clouds, despite their importance (Klein 1997). These clouds are sensitive to their 123 626 environment, especially on timescales relevant to its physical evolution (Bony and Dufresne 2005). To understand the radiative cooling effect of stratocumulus clouds, it is important to study both cloud variability and associated microphysical and macrophysical mechanisms. A better understanding of stratocumulus cloud variability and associated microphysical and macrophysical mechanism will lead to improved representation of such clouds and may help provide constraints for model simulations (Wang et al. 2004a). In general, there are three fundamental cloud quantities that contribute to the cloud radiation effect: microphysical cloud droplet concentration (Nd), macrophysical cloud liquid water path (CLWP) and cloud fraction cover (Fc). These three fundamental cloud quantities are involved in cloud formation processes that are controlled by meteorological conditions and modulated microphysically by aerosols (George and Wood 2010). They also vary substantially at different spatial and temporal scales. Understanding meteorological controls on these fundamental cloud quantities may offer clues regarding the key physical processes. Due to the intrinsic coupling of dynamical and microphysical processes related to cloud formation, cloud cover, CLWP, and cloud droplet concentration are physically correlated. When clouds have greater spatial coverage, CLWP is usually greater (Considine et al. 1997). Also, cloud droplet growth processes link CLWP to cloud droplet concentration (Lohmann and Feichter 2005). Understanding the nature of the correlation of cloud microphysics with macrophysical variables is one simple constraint beyond the mean state. Cloud cover is the most widely studied cloud macrophysical variable from observations and from numerical modeling in the literature. On long time scales, the cloud fraction is strongly correlated with meteorological predictors, such as lower tropospheric stability (LTS), sea surface temperature (SST), and atmospheric circulation (Klein and Hartmann 1993; Norris and Leovy 1994; Stevens and Brenguier 2009; Weare 1994). The seasonal cycle of cloud cover controlled largely by the seasonal cooling of the free-tropospheric air temperature instead of the sea surface temperature. Seasonal variances depend largely on MBL depth, inversion strength, and cloud top entrainment (Lin et al. 2009; Wang et al. 2011). On short (subseasonal and synoptic) timescales, variations in low-cloud amount are significantly (albeit weakly) correlated with LTS, cold advection, and relative humidity of the cloud layer (George and Wood 2010; Klein 1997; Klein et al. 1995; Xu et al. 2005). On the diurnal scale, the decoupling effect plays an important role in the diurnal cycle of low clouds (Klein 1997) and is driven primarily by cloud solar absorption (Wood et al. 2002). However, the cloud cover is also a loosely defined variable, depending on not only the minimal detection limit of instruments and the minimal 123 L. Min et al. threshold of models but also the domain size. As the key cloud variable in cloud formation processes, CLWP is less studied than the cloud cover, partly because its temporal variance is much quicker and its spatial variance is of much more complex. Understanding the controls of the high spatial and temporal variability of cloud liquid path will be important for correctly representing stratocumulus clouds in the global climate system and identifying feedback processes associated with aerosol–cloud interaction. In this study, therefore, we focus on the variability of CLWP and investigate possible meteorological and microphysical controls at synoptic timescale in the Southeastern Pacific (SEP). 2 Data and methods 2.1 Data NASA Aqua satellite has visible–infrared moderate resolution imaging spectroradiometer (MODIS) and advanced microwave scanning radiometer-EOS (AMSR-E) sensors, providing cloud and aerosol information. We use spatially griddedLevel-3 Low-Resolution Daily (MYD08_D3) MODIS Images and AMSR-E Level 3 data over the oceanic part of a spatial domain of 0–30S and 70– 100W in the SEP at 1 9 1 resolution from September 1 to November 30, 2008. The selection of spatial domain and time span corresponds to the VAMOS Ocean-CloudAtmosphere-Land Study (VOCAL) field campaign. Not only are clouds prevalent during Austral spring in the region, but also satellite remote sensing can be compared with VOCALS in situ measurements (Min et al. 2012). Specifically, the daily variance of cloud properties, i.e., cloud fraction (CF), CLWP (CLWP), and cloud droplet concentration (CDN), derived from retrieved cloud effective radius and cloud optical depth (Szczodrak et al. 2001), are derived from MODIS; and seasonal sea surface temperature, column cloud water, column water vapor (CWV), and sea surface wind are retrieved from AMSRE. To explore the influence of meteorology factors, NCEP final analysis (FNL) operational global analysis data with 1 9 1 resolution is used in this study (Kalnay et al. 1996). Specifically, temperature, relative humidity, pressure, wind, and vertical velocity at multiple levels at 18:00UTC (close to Aqua overpass time) are taken directly from NCEP FNL. The LTS is calculated as the potential temperature difference between 850 and 1000 hPa (Klein and Hartmann 1993). As the boundary layer is lower in the SEP, we use the potential temperature at 850 hPa instead of at 700 hPa, as originally defined in (Klein et al. 1995). We also derive the temperature advection (TMPADV) using Understanding the synoptic variability… 627 ~sfc rSST from the measurements of SST the formula V and sea surface wind speed. 2.2 EOF and DINEOF The empirical orthogonal function (EOF) is a powerful tool to analyze the spatial and temporal variance of large-scale atmospheric fields (Wilks 2006). The EOF method, in which the original data are transformed into a set of orthogonal functions or EOFs, provides us an effective way to study cloud variability: (1) each EOF represents a spatial pattern, and (2) the corresponding principal component (PC) is the temporal variation that represents the magnitude of the EOF over time. We are able to project various meteorological and other fields on each of the EOFs and PCs. Due to a limited instrument swath, some data may be missing within the analysis domain. To apply the EOF method to a field with possible missing data, we use the Data INterpolating Empirical Orthogonal Function (DINEOF) (Beckers and Rixen 2003), in which the EOF modes and their relative temporal amplitudes are utilized to reconstruct the missing data. One advantage of this method compared to other interpolation methods is that the EOF analysis result will not be changed after using this DINEOF method. Also, DINEOF is self-consistent, parameter-free and does not need any a priori knowledge of the errors. The DINEOF method was widely used to reconstruct the missing data of the reduced MODIS data set (Mauri et al. 2007). 3 Results Many dynamical, thermodynamical, and microphysical processes are involved in the cloud formation, resulting in the complicated spatial and temporal variability of cloud quantities. EOF/PC analysis illustrates that liquid water path exhibits the most variable quantity among three key cloud quantities. As shown in Fig. 1, the first EOF/PC explains only 13 % of the variance of the CLWP, and the first 15 principal components explain 50 % of the variance CLWP. In contrast to the variance of CLWP, the first EOF/ PC of cloud droplet concentration and cloud fraction can explain 86 and 22 % variance, respectively. It indicates that CLWP variance is complicated and influenced by many factors at different scales. Although it is impossible to capture the variance of CLWP with a few EOF/PCs, we only focus on the first two leading EOF/PCs to illustrate possible controlling factors for simplicity. Various large-scale meteorological factors, such as sea surface temperature, surface wind speed, temperature advection, LTS, relative humidity, influence cloud Fig. 1 91 eigenvalues derived from 91 days of CLWP over spatial domain of 0–30S and 70–100W in the SEP from MODIS level 3 data. Value of each eigenvalue represents the weights of each CLWP EOF/PC component formation and CLWP variance (Ackerman et al. 2004; George and Wood 2010; Myers and Norris 2013; Xu et al. 2005). As listed in Table 1, surface level meteorology factors play a key role in the EOF/PC1 of CLWP variance, while the cloud layer meteorological factors control the EOF/PC2 of CLWP variance. 3.1 EOF/PC1 of CLWP The first component of EOF/PC explains the maximal amount of variance of CLWP. As shown in Fig. 2, the spatial pattern of EOF1 closely resembles the three-month averaged CLWP from AMSR-E, suggesting that the large variance is located where there is a larger liquid water path. The near coast region with high anthropogenic aerosol loading and the southwest corner with a lot of drizzle make a relatively small contribution to the CLWP EOF1 variance. The temporal variation of its corresponding principal component (PC1), representing the magnitude of EOF1 over time, is strongly correlated with the variation of the domain-averaged CLWP (Fig. 3). EOF/PC1 captures the variance of averaged CLWP in the domain, although it only explains 13 % of the overall variance of CLWP. Sea Surface Temperature (SST) is a central determinant of air–sea interactions and cloud variability. Evaporation from the oceans is the primary source of water vapor to fuel cloud growth in the atmosphere. Following the Clausius– Clapeyron relationship, the warmer the water, the greater is the evaporation. On the other hand, a decrease in SST increases the stability of the lower troposphere, leading to an increase in low-cloud amount. In turn, the increased cloud amount cools the ocean further to form a positive feedback (Xu et al. 2005). Certainly, such an effect is scale dependent. 123 628 L. Min et al. Table 1 Correlation coefficients for liquid water path PC1 and PC2 with PC1 of selected cloud properties and macrophysical variables R CF PC1 CDN PC1 WIND10M PC1 SST PC1 CLWP_PC1 0.396 0.365 CLWP_PC2 0.327 -0.541 CF_PC1 1 0.141 -0.263 CDN_PC2 0.141 1 -0.305 0 TMPADV PC1 0.604 -0.709 0.658 0 -0.152 0.243 0.447 -0.148 0 LTS PC1 RH850 PC1 RH1000 PC1 VVEL850 PC1 -0.239 -0.148 -0.311 0.581 0.67 0.626 0.23 VVEL1000 PC1 0.616 0.379 0.352 -0.319 -0.221 -0.429 -0.239 -0.261 0.675 0.628 0.226 0.155 0.187 The italicized numbers indicate the confidence level over 99 % Fig. 2 The comparison between a EOF1 pattern of CLWP and b 3-month mean of CLWP averaged from 91 days MODIS level 3 daily mean data during austral spring Fig. 3 Comparison of 91 PC1 variables of CLWP and daily spatially averaged CLWP over the same subtropical southeast Pacific region (0–30S, 70–100W) Figure 4 shows the temporal variances of SST EOF/PC1 and CLWPEOF/PC1. The variance of SST EOF/PC1is much smoother than that of CLWPEOF/PC1. However, the long-term trend of CLWPEOF/PC1 is negatively correlated with the variance of SST EOF/PC1 (reversed plotted in 123 Fig. 4) with a correlation coefficient of 0.71. The strong negative correlation between the SST and CLWP leading modes over the 30 9 30 domain represents the SST control on large scale of both spatial and temporal domain. Although it is focused on the CLWP variability, the finding is consistent with previous research indicating that seasonal to inter-annual variations in low-cloud amount are closely related to variation in SST (Klein 1997; Klein and Hartmann 1993; Klein et al. 1995; Norris and Leovy 1994; Weare 1994; Xu et al. 2005). Temperature advection and stability of the lower atmosphere are two other large-scale factors that are significantly correlated with stratocumulus cloud cover (Klein 1997; Xu et al. 2005). However, temperature advection but not LTS (important in the CLWPEOF/PC2, more discussion later) plays an important role in the variance of CLWPEOF/PC1, as shown in Fig. 5 and listed in Table 1. The correlation coefficient between CLWPEOF/PC1 and TMPADVEOF/PC1 is 0.65. As discussed in Xu et al. (2005), the cold advection increases surface latent and sensible heat fluxes and destabilizes the low boundary layer. The stronger overturning and moisture transport into the cloud layer results in thicker clouds and larger CLWP. Further inspection into the temporal variation of EOF/PC1s indicates that unlike the SST, the temperature advection Understanding the synoptic variability… 629 Fig. 4 a Comparison between PC1 series of CLWP and reversed SST. Reversed SST PC1 trend is given to show the tendency that CLWP varies with the SST. It shows that when SST is lower, CLWP tends to be larger, and vice versa. b Scatterplot of CLWP PC1 vs. reversed SST PC1 Fig. 5 Relationship between PC1s of CLWP and temperature advection during the 91 days in austral spring over the southeast Pacific deck 0–30S, 70–100W captures some temporal variance at synoptic scale of a few days. Surface wind has long been thought as an important factor in controlling the variance of CLWP (Stevens and Brenguier 2009). Xu et al. (2005) found that SSM/I CLWP was correlated significantly with local wind velocity based on a buoy observation. As surface wind physically links to SST and TMPADV, the temporal variance of wind speed EOF/PC1 is significantly correlated with the temporal variance of CLWP EOF/PC1 with correlation coefficient of 0.60, shown in Fig. 6. We applied a Butterworth filter to both CLWP and wind EOF/PC1s with cutoff value of 30 days. The filtered CLWP and wind EOF/PC1s are correlated with correlation coefficient of 0.43, indicting the impact of surface wind on cloud formation and consequently CLWP. Notably, the temporal variation of wind speed EOF/PC1 is consistent with CLWP EOF/PC1, except for the period from Oct. 8 (day 38) to Oct. 15 (day 45). During that period, there was a strong cyclone passage through this region (Rahn and Garreaud 2010). The increase of CLWP on Oct. 8 is due to this cyclone, resulting in the inconsistency between the two. It is clear that SST, TMPADV, and wind speed are not mutually independent. Figures 4, 5 and 6 suggest that all three factors impact on cloud formation and CLWP at different temporal scales. To further illustrate this, we composite the daily difference of all four parameters [EOF/ PC1 (n ? 1) - EOF/PC1 (n), where n is the index of the day]. As shown in Fig. 7, the daily difference of CLWP EOF/PC1 is significantly correlated with that of wind speed with correlation coefficient of 0.57. The correlation coefficients are reduced to 0.22 and 0.14 for TMPADV and SST, respectively. It suggests that although the correlation coefficient of wind speed EOF/PC1 with CLWP EOF/PC1 is the lowest among the three factors, the wind speed plays a more fundamental role in stratocumulus cloud formation and CLWP. SST and cold advection mainly drive the synoptic or longer-term scales of variance of CLWP, while surface wind mainly controls the daily variance of CLWP. 3.2 EOF/PC2 of CLWP The CLWP EOF2 provides detailed spatial pattern of partial variability of CLWP, shown in Fig. 8a, which is orthogonal from the CLWP EOF1. The vertical velocity at 850 hPa is a good indicator of the boundary layer turbulent structure and strength. The spatial pattern of CLWP EOF2 is similar to the pattern of three-month averaged vertical velocity at 850 hPa (Fig. 8b). Strong upward velocity results in large variability. Strong subsidence, particularly at the southeast corner of the domain (the blue region in Fig. 8b), corresponds to a large positive variance of CLWP EOF2. On temporal variance of CLWP EOF2, LTS has the highest correlation coefficient of 0.67 among all other large-scale meteorological factors with CLWP EOF/PC2, 123 630 L. Min et al. Fig. 6 Relationship between PC1sof CLWP and wind speed at 10 m level; the upper panel for unfiltered data and the bottom panel for the filtered data Fig. 7 Comparison between CLWP PC1 daily difference (EOF/PC1 (n ? 1) - EOF/PC1 (n), where n is the index of the day) and a PC1 of wind at 10 m daily difference, b PC1 of SST daily difference and c PC1 of cold advection daily difference as listed in Table 1 and shown in Fig. 9. The temporal variation of EOF/PC1 of LTS captures the variance of CLWP EOF/PC2, except for the days when the cyclone came by. The second highest correlation is between the EOF/PC1 of relative humidity in the cloud layer (e.g., RH at 850 hPa) and CLWP EOF/PC2. The correlation coefficient of 0.63 is also above the 99 % confidence level. In Fig. 10, close inspection of relationship between LTS and RH at 850 hPa indicates that the two fields are highly anticorrelated on both observed fields and their EOF/PC1s. Both fields in concert control cloud thickness and consequently CLWP through thermodynamic and entrainment processes. 123 3.3 Overall CLWP variability Although the first two EOFs/PCs of CLWP explain small portion of CLWP variance (less than 20 %), the above analysis does provide the basic understanding of meteorological controls on CLWP variability. Previous studies have shown that on sub-seasonal timescales, LTS and relative humidity of the cloud layer (cold advection as well) are significantly correlated with variations in low-cloud amount (Klein 1997; Klein et al. 1995). However, the correlations of low cloud amount with meteorological predictors are substantially weaker than on longer timescales (George and Wood 2010). As for CLWP, Understanding the synoptic variability… 631 Fig. 8 The comparison between a spatial EOF2 pattern of CLWP and b 3 months mean of vertical velocity averaged from 91 days NCEP re-analysis during austral spring Fig. 9 Relationship between CLWP PC2 and LTS PC1 during the 91 days in austral spring over the southeast Pacific deck 0–30S, 70–100W Fig. 10 Relationship between CLWP PC2 and RH PC1 at 850 hPa during the 91 days in austral spring over the southeast Pacific deck 0–30S, 70– 100W meteorological conditions at the low boundary of clouds, such as SST, TMPADV, and wind, control overall CLWP (EOF1) of stratocumulus clouds. The meteorological factors at the cloud layer, such as LTS and cloud top RH, influence CLWP spatial distribution (EOF2). Specifically, as shown in Fig. 11, the overall changes in CLWP from Sept. 20 to Sept. 27 as a cloud system correspond well with change in sea surface wind speed except for near coast regions. Temperature advection seems to play a minor role in controlling CLWP. The movement from southwest to northeast, shown in Fig. 11, corresponds well with changes in LTS and cloud top RH. The major cloud bands with large CLWP were also located in the regions with large gradients of LTS. The cloud seems to break into small parts where the LTS reduced substantially, as weaker LTS promotes decoupling which is favorable for stratocumulus breaking up into cumulus (Bretherton and Wyant 1997; Dal Gesso et al. 2011). In the meantime, RHs at the top of those cloud bands were very high (near 100 %). (Caldwell and Bretherton 2009) found that the cloud top height exhibits a larger variation than the cloud base and cloud layer thickening follows the entrainment pattern. High RH at the cloud top helps to slow the entrainment and to sustain the cloud longer with larger CLWP. 123 632 L. Min et al. Fig. 11 Corresponding meteorological fields from Sept. 20 to Sept. 27. a liquid water path (gm-2) from MODIS data, b surface wind speed (m/s) from NCEP data, c cold advection (C day-1) from NCEP data, d lower tropospheric stability (T) from NCEP data, e relative humidity at 850 hPa (%) from NCEP data As discussed previously, at short-time scales, surface wind speed plays more dominant role in determining CLWP variation than other factors. One reason is that the increased wind speed and associated temperature advection increase surface latent and sensible heat fluxes and destabilize the low boundary layer, resulting in stronger moisture transport into the cloud layer and thicker CLWP (Xu et al. 2005). However, it is demonstrated from Figs. 4, 5, 6 and 7 that surface wind speed captures much more daily variance than both temperature advection and sea surface temperature. (Stevens and Brenguier 2009) indicate that besides setting the rate of coastal ocean upwelling and hence surface temperatures, surface wind speed can also increase mechanical and bio-chemical production of marine aerosol as an compound effect. Former researches have also shown that the surface wind influences sea salt and Dimethyl sulfate (DMS) production in the atmosphere, which can be activated effectively into cloud droplets 123 Understanding the synoptic variability… 633 Fig. 12 Relationship between surface wind speed and liquid water path from AMSR-E measurements (Latham and Smith 1990; O’Dowd et al. 1999). It is highly possible that the daily correlationship between surface wind speed and CLWP results from wind-induced marine aerosol. Simultaneous retrievals of surface wind and CLWP from AMSR-E measurements allow us to investigate the relationship of wind and CLWP through the perspective of the aerosol indirect effect. Cloud formation is affected by both dynamical and microphysical processes. We segregate the observed data into three regions: drizzle influenced region, anthropogenic aerosol influenced region, and normal marine aerosol region (the rest of the remote ocean domain). The drizzle influenced region is mostly located at the southwest corner, and the anthropogenic aerosol influenced region is along the Chile coast, where the CDN is larger than 100 cm-3. To further limit other meteorological impacts, we only use the data in which SST is less than 20 C and the total water vapor path is within 20 mm. As shown in Fig. 12, the slope and correlation coefficient are substantially different among the three regions. The marine aerosol region without drizzle has the greatest slope of 0.011 with the highest coefficient of 0.72, exhibiting the strongest impact of surface wind on CLWP. Although there is tendency of increasing CLWP with surface wind in the anthropogenic aerosol region, the slope is about one-third of that in the marine aerosol region with the weakest coefficient of 0.23. It is understandable that drizzle removes liquid water from clouds, resulting in that the slope is only half of that in the marine aerosol region without drizzle. It is clear that the sea surface wind could impact cloud development micro-physically. The pattern of relationship between surface wind speed and CLWP corresponds very well with the first aerosol indirect effect and second aerosol indirect effect. More CDN due to enhanced marine aerosols would increase the CLWP. Most importantly, the anthropogenic aerosols due to the copper smelters near the Chile coast provide sufficient CCNs to be activated into cloud droplets, resulting in high CDN and inhibiting the precipitation initiation process (Albrecht 1989). Note that clouds with drizzle are formed mostly in the remote ocean with normal marine aerosol region. 4 Conclusions It is important to understand the fundamental processes of stratocumulus clouds. The EOF method provides us an effective way to untangle the varieties of factors, and separates them into spatial and temporal aspects. In particular, stratocumulus CLWP, the most complex quantity among all key cloud properties, is influenced by both microphysical and macrophysical factors from all levels and each of these factors may be tangled with each other. The first principal component of CLWP can explain 13 %, while the first 15 principal components explain 50 % of the variance of CLWP. Although explained less than 20 % of CLWP variance, the EOF analysis over the first two principal components of CLWP provides insights into how key meteorological factors influence CLWP. The overall CLWP (EOF1) of stratocumulus clouds is controlled by the sea surface processes and associated factors, such as SST, TMPADV, and wind, while the meteorological factors at the top of the cloud, such as LTS and RH, influence the second principal of CLWP (EOF2) and control the daily spatial movement of CLWP. SST significantly correlates with the variance of CLWP at seasonal scale. The correlations of CLWP with LTS, RH, and TMPADV are significant at sub-seasonal or synaptic scales. The sea surface wind is highly correlated with CLWP at daily scale. The influence of surface wind on CLWP is twofold: (1) dynamically through modulation of surface latent heat and sensible heat fluxes and (2) microphysically through enhancement of marine aerosol production. 123 634 Acknowledgments This work was supported by US NSF under contract AGS-1138495, by the DOE’s Atmospheric System Research program (Office of Science, OBER) under contract DE-FG0203ER63531, and by the NOAA Educational Partnership Program with Minority Serving Institutions (EPP/MSI) under cooperative agreements NA17AE1625 and NA17AE1623, and supported by Science and Technology Research Foundation of SGCC contract DZB17201200260. References Ackerman AS, Kirkpatrick MP, Stevens DE, Toon OB (2004) The impact of humidity above stratiform clouds on indirect aerosol climate forcing. Nature 432:1014–1017 Albrecht BA (1989) Aerosols, cloud microphysics, and fractional cloudiness. Science 245:1227–1230 Beckers J, Rixen M (2003) EOF calculations and data filling from incomplete oceanographic datasets. J Atmos Ocean Technol 20:1839–1856 Bony S, Dufresne J-L (2005) Marine boundary layer clouds at the heart of tropical cloud feedback uncertainties in climate models. Geophys Res Lett 32:L20806. doi:10.1029/2005GL023851 Bretherton CS, Wyant MC (1997) Moisture transport, lower-tropospheric stability, and decoupling of cloud-topped boundary layers. J Atmos Sci 54:148–167 Bretherton CS, Uttal T, Fairall CW, Yuter SE, Weller RA, Baumgardner D, Comstock K, Wood R, Raga GB (2004) The EPIC 2001 stratocumulus study. Bull Am Meteorol Soc 85:976–977. doi:10.1175/bams-85-7-967 Caldwell P, Bretherton CS (2009) Response of a subtropical stratocumulus-capped mixed layer to climate and aerosol changes. J Clim 22:20–38 Considine G, Curry JA, Wielicki B (1997) Modeling cloud fraction and horizontal variability in marine boundary layer clouds. J Geophys Res Atmos 102:13517–13525. doi:10.1029/97JD00261 Dal Gesso S, Siebesma A, de Roode S, van Wessem J (2011) A mixed-layer model perspective on stratocumulus steady-states in a perturbed climate. Q J R Meteorol Soc. doi:10.1002/qj.2282 George R, Wood R (2010) Subseasonal variability of low cloud radiative properties over the southeast Pacific Ocean. Atmos Chem Phys 10:4047–4063. doi:10.5194/acp-10-4047-2010 Kalnay E et al (1996) The NCEP/NCAR 40-year reanalysis project. Bull Am Meteorol Soc 77:437–471 Klein SA (1997) Synoptic variability of low-cloud properties and meteorological parameters in the subtropical trade wind boundary layer. J Clim 10:2018–2039 Klein SA, Hartmann DL (1993) The seasonal cycle of low stratiform clouds. J Clim 6:1587–1606 Klein SA, Hartmann DL, Norris JR (1995) On the relationships among low-cloud structure, sea surface temperature, and atmospheric circulation in the summertime northeast Pacific. J Clim 8:1140–1155 Latham J, Smith M (1990) Effect on global warming of wind-dependent aerosol generation at the ocean surface. Nature 347:372–373 Lin W, Zhang M, Loeb NG (2009) Seasonal variation of the physical properties of marine boundary layer clouds off the California coast. J Clim 22:2624–2638 Lohmann U, Feichter J (2005) Global indirect aerosol effects: a review. Atmos Chem Phys 5:715–737. doi:10.5194/acp-5-7152005 Mauri E, Poulain PM, Južnič-Zonta Ž (2007) MODIS chlorophyll variability in the northern Adriatic Sea and relationship with forcing parameters. J Geophys Res Oceans 112:C03S11. doi:10. 1029/2006JC003545 123 L. Min et al. Min Q et al (2012) Comparison of MODIS cloud microphysical properties with in situ measurements over the Southeast Pacific. Atmos Chem Phys 12:11261–11273. doi:10.5194/acp-12-112612012 Myers TA, Norris JR (2013) Observational evidence that enhanced subsidence reduces subtropical marine boundary layer cloudiness. J Clim 26:7507–7524 Norris JR, Leovy CB (1994) Interannual variability in stratiform cloudiness and sea surface temperature. J Clim 7:1915–1925 O’Dowd CD, Lowe JA, Smith MH, Kaye AD (1999) The relative importance of non-sea-salt sulphate and sea-salt aerosol to the marine cloud condensation nuclei population: an improved multi-component aerosol-cloud droplet parametrization. Quarterly Journal of the Royal Meteorological Society 125:1295–1313 Rahn D, Garreaud R (2010) Marine boundary layer over the subtropical southeast Pacific during VOCALS-REx–Part 1: mean structure and diurnal cycle. Atmos Chem Phys 10:4491–4506. doi:10.5194/acp-10-4491-2010 Stevens B, Brenguier J-L (2009) Cloud controlling factors—low clouds. In: Heintzenberg J, Charlson RJ (eds) Clouds in the perturbed climate system, 1st edn. MIT Press, Cambridge, pp 173–196 Stevens B, Feingold G (2009) Untangling aerosol effects on clouds and precipitation in a buffered system. Nature 461:607–613 Szczodrak M, Austin PH, Krummel P (2001) Variability of optical depth and effective radius in marine stratocumulus clouds. J Atmos Sci 58:2912–2926 Wang Y, Xie S-P, Xu H, Wang B (2004a) Regional model simulations of marine boundary layer clouds over the Southeast Pacific off South America. Part I: control experiment. Mon. Wea. Rev. 132:274–296 Wang Y, Xu H, Xie S-P (2004b) Regional model simulations of marine boundary layer clouds over the Southeast Pacific off South America. Part II: sensitivity experiments. Mon. Wea. Rev. 132:2650–2668 Wang Y, Xie S-P, Wang B, Xu H (2005) Large-scale atmospheric forcing by Southeast Pacific boundary-layer clouds: a regional model study. J Clim 18:934–951 Wang L, Wang Y, Lauer A, Xie S-P (2011) Simulation of seasonal variation of marine boundary layer clouds over the eastern Pacific with a regional climate model. J Clim 24:3190–3210 Weare BC (1994) Interrelationships between cloud properties and sea surface temperatures on seasonal and interannual time scales. J Clim 7:248–260 Wilks DS (2006) Statistical methods in the atmospheric sciences. Elsevier, Amsterdam Wood R, Bretherton C, Hartmann D (2002) Diurnal cycle of liquid water path over the subtropical and tropical oceans. Geophys Res Lett 29:2092. doi:10.1029/2002GL015371 Wood R, Mechoso CR, Bretherton CS, Weller RA, Huebert B, Straneo F, Albrecht BA, Coe H, Allen G, Vaughan G, Daum P, Fairall C, Chand D, Gallardo Klen-ner L, Garreaud R, Grados C, Covert DS, Bates TS, Krejci R, Russell LM, de Szoeke S, Brewer A, Yuter SE, Springston SR, Chaigneau A, Toniazzo T, Min-nis P, Palikonda R, Abel SJ, Brown WOJ, Williams S, Fochesatto J, Brioude J, Bower KN (2011) The VAMOS oceancloud-atmosphere-land study regional experiment (VOCALSREx): goals, platforms, and field operations. Atmos Chem Phys 11:627–654. doi:10.5194/acp-11-627-2011 Xu H, Wang Y, Xie S-P (2004) Effects of the Andes on eastern Pacific climate: a regional atmospheric model study. J Climate 17:589–602 Xu H, Xie S-P, Wang Y (2005) Subseasonal variability of the Southeast Pacific Stratus Cloud Deck. J Clim 18:131–142