1. Introduction
Climate models do not yet adequately parameterize the physical and dynamical processes associated with low-level marine stratiform clouds, in part because of inadequate observational records as well as insufficient knowledge of the relevant meteorological processes. These low-level clouds are particularly important, as they reflect a large amount of solar radiation back into space because they are often horizontally extensive and optically thick. However, they only reduce outgoing terrestrial radiation by a small amount because they are nearly as warm as the underlying sea surface. Previous studies (Randall et al. 1984; Slingo 1990; Wood 2012) pointed out that an 8%–12% relative increase in low-level cloud cover would be sufficient to offset the global warming induced by a doubling of CO2. A main contributor to low-level cloudiness is subtropical stratocumulus (Sc), which occurs predominantly in eastern ocean basins over relatively cold sea surface temperatures (SSTs), under strong subsidence, and in a boundary layer capped by a strong temperature inversion (Klein and Hartmann 1993). Stratocumulus clouds break up into shallow cumulus clouds as the trade winds advect the boundary layer over increasingly warmer SSTs, under weaker subsidence and a weaker temperature inversion (Albrecht et al. 1995; Norris 1998; Wood and Hartmann 2006).
Previous observational and modeling studies have examined the links between low-level cloud amount (LCA) and large-scale meteorological parameters on different time scales. Klein and Hartmann (1993) and Wood and Bretherton (2006) found that the inversion strength is the dominant factor that explains the geographical distribution and seasonal cycle of LCA. Stronger inversions favor a shallow, well-mixed marine boundary layer (MBL) with larger LCA by inhibiting cloud-top entrainment and trapping moisture within the MBL (Albrecht et al. 1988; Klein and Hartmann 1993; Stevens and Brenguier 2009). A weaker inversion favors a deep, decoupled MBL with reduced LCA by promoting enhanced cloud-top entrainment (Bretherton and Wyant 1997). A significant negative correlation between LCA and SST anomalies is observed on seasonal and interannual time scales (Hanson 1991; Mansbach and Norris 2007; Norris and Leovy 1994; Clement et al. 2009; Eastman et al. 2011; Eitzen et al. 2011). Cooler SST may increase the strength of the inversion and lead to a shallower and more well-mixed MBL, favoring increased LCA (Bretherton and Wyant 1997). Warmer SST may promote greater latent heat flux, favoring a weaker inversion and a deeper and less well-mixed MBL, and reduced LCA (Eastman et al. 2011; Norris and Leovy 1994). With no change in inversion strength, enhanced subsidence leads to a shallower MBL with reduced cloud thickness (Deardorff 1976; Schubert et al. 1979) and reduced LCA (Zhang et al. 2009; Sandu and Stevens 2011; Mauger and Norris 2010; Myers and Norris 2013). Climatologically, stronger subsidence typically occurs with a stronger inversion, however, leading instead to overall greater LCA. Increased sea level pressure (SLP) is associated with a shallower MBL with greater LCA (Klein et al. 1995). Enhanced cold-air advection amplifies evaporative flux from the sea surface and at the same time favors stronger inversions (Klein et al. 1995; Wood 2012), each of which favors larger LCA. The relationship between free-tropospheric moisture and LCA is unclear, as Klein et al. (1995) found that enhanced free-tropospheric moisture over the eastern subtropical Pacific was associated with smaller LCA, but Lacagnina and Selten (2013) found that it was associated with larger LCA. Clement et al. (2009) reported that an observed decadal reduction in LCA over the northeastern Pacific was linked to the local warming of SST, weaker estimated inversion strength, weaker trade winds, and weaker subsidence.
Most global climate models do not realistically simulate the observed relationships between subtropical low-level cloudiness and meteorological factors (Bony and Dufresne 2005; Forster et al. 2007; Clement et al. 2009; Myers and Norris 2015; Qu et al. 2015). This motivates the examination of observed, multidecadal trends in low-level cloud amount and related meteorological parameters as an alternative means of investigating climate change. Unfortunately, substantial discrepancies exist in global-scale cloud cover variability and decadal trends reported by all satellite and surface datasets (Evan et al. 2007; Arndt et al. 2010; Stubenrauch et al. 2013), and changes in MBL clouds are among the most uncertain (Boucher et al. 2013). For example, Eastman et al. (2011) report that surface-observed cloud cover decreased in all subtropical Sc regions from 1954 to 2008, whereas O’Dell et al. (2008) and Rausch et al. (2010) suggest that no long-term trends have occurred in satellite-observed cloud liquid water path or the optical properties of low-level marine clouds. Assessing trends in low-level clouds has been challenging because surface visual observations and satellite remote sensing of clouds are affected by various identified and unidentified retrieval artifacts that cause spurious trends in reported cloud properties, including instrumental calibration, drifts in satellite orbits, and other factors (Campbell 2004; Norris 2005; Evan et al. 2007). A new method of statistical correction, however, provides satellite cloud data from which spurious variability has largely been removed (Norris and Evan 2015).
The corrected satellite cloud data enable the present study, which examines multidecadal trends in LCA and related meteorological parameters over eastern subtropical Sc regions. Satellite observations provide the best source of quantitative information on changes in cloud properties, and reanalyses, an assimilation of measurements by numerical weather prediction models, provide the best source of large-scale meteorological information. We assess whether the observed changes in cloudiness in two independent satellite records are physically consistent with the observed changes in large-scale meteorology as well as what changes in meteorology independently contribute most to cloud changes. In particular, we construct multilinear regression models that predict long-term trends in LCA associated with long-term trends in SST, estimated inversion strength, and the strength of advection over the SST gradient. Agreement between observed and predicted trends will increase confidence in the datasets, and identification of meteorological parameters most responsible for cloud trends will provide insight into factors driving multidecadal cloud changes.
2. Datasets
a. Satellite observations of clouds
The International Satellite Cloud Climatology Project (ISCCP) provides cloud fraction, cloud-top pressure, and cloud optical thickness information retrieved from geostationary and polar-orbiting weather satellites from July 1983 through December 2009 (Rossow and Schiffer 1999). Low-level clouds are defined as those with tops below the 680-hPa level, midlevel clouds as those with tops between 680 and 440 hPa, and high-level clouds as those with tops above the 440-hPa level. The ISCCP data are available on an equal-angle grid with a latitude–longitude spacing of 2.5° × 2.5°. Another source of long-term cloud data is the Pathfinder Atmospheres–Extended (PATMOS-x) dataset, which provides fractional cloud cover, cloud type, and cloud height retrieved from the Advanced Very High Resolution Radiometer measurements beginning in October 1981 (Heidinger et al. 2014). We use data from January 1984 to December 2009 for consistency with ISCCP. The data are available in 1° × 1° equal-angle grid boxes, but we bilinearly interpolated them into 2.5° × 2.5° equal-angle grid boxes to be consistent with the ISCCP grid.
Both ISCCP cloud amount data and PATMOS-x cloud amount data suffer from inhomogeneities associated with satellite changes over time (Norris and Evan 2015). These include spurious variations resulting from systematic changes in satellite view angle (ISCCP), spurious variations resulting from drift in equatorial crossing time (PATMOS-x), and spurious variations resulting from changes in effective calibration and ancillary data for a single satellite or between satellites (ISCCP and PATMOS-x). Following Norris and Evan (2015), we use the cosine of the satellite zenith angle, the cosine of the solar zenith angle, and the standardized cloud anomaly spatially averaged over every grid box viewed by a satellite to represent factors associated with artifacts. We then calculate for each grid box the least squares best-fit line between cloud anomalies and artifact factor anomalies, and let the residuals from the best-fit line be the newly corrected data. Unfortunately, the correction method removes any real global-scale cloud variability from the datasets, so we are unable to assess how tropical average LCA has changed in recent decades. Fortunately, the correction method has very little impact on regional trends such as those examined in this study (Norris and Evan 2015).
b. Meteorology from reanalyses
The ERA-Interim, MERRA, and JRA-55 reanalyses are generated by assimilating various surface, upper-air, and satellite observations into atmosphere general circulation models (AGCMs) forced by observed SST, while CFSR is a partially coupled ocean–atmosphere reanalysis system (Onogi et al. 2007; Saha et al. 2006; Saha et al. 2010). ERA-Interim and JRA-55 use a four-dimensional variational data assimilation (4DVAR) system based on a spectral GCM (e.g., Simmons et al. 2005), while MERRA and CFSR use a 3DVAR algorithm based on the Gridpoint Statistical Interpolation scheme (Wu et al. 2002; Derber et al. 2003; Purser et al. 2003a,b). Although the monthly mean atmospheric datasets are available at a higher spatial resolution, we have bilinearly interpolated them onto a 2.5° × 2.5° grid to be consistent with the satellite cloud observations.
Several studies have analyzed the major differences in various reanalysis products (Bengtsson et al. 2004; Stachnik and Schumacher 2011; Simmons et al. 2014). These differences can result from changes in observations, biases in assimilated radiances due to instrument and calibration errors, and biases in models (Saha et al. 2010). Dee et al. (2011b) and Kumar and Hu (2012) claim that the long-term trends observed in new-generation reanalysis datasets are reliable, but Stachnik and Schumacher (2011) argue that the lack of consistency among reanalyses may skew trends in meteorological fields. The influence of unexpected drifts in tropospheric AMSU-A data (noted in Mears and Wentz 2009) is expected to be further exacerbated in ERA-Interim as a result of the assimilation of increasing numbers of biased aircraft reports in the upper troposphere (Dee and Uppala 2009). For long-term climate variability measures, Chelliah et al. (2011) noted that the CFSR was generally an outlier in comparison to the 40-yr ECMWF Re-Analysis (ERA-40), JRA, and MERRA; specifically, CFSR exhibited much stronger easterly trade winds, cooler tropospheric temperatures, and lower geopotential heights during the period of 1979–98. Willett et al. (2010) and Simmons et al. (2014) revealed consistency in trend estimates for temperature, both near the surface and at higher levels in the atmosphere among ERA-Interim, MERRA, and JRA. Assuming that biases in different reanalyses are independent to the extent that they average out, the ensemble mean values across the four reanalyses will be treated in this study as the best estimates of multidecadal trends in meteorological fields.
3. Results
a. Observed multidecadal trends in low-level cloud amount and meteorology
Figure 1 displays multidecadal trends in ISCCP and PATMOS-x LCA and ensemble mean large-scale meteorology over the ocean. We focus on the subtropical subsidence regions of the northeastern and southeastern Pacific, the northeastern and southeastern Atlantic, and the southeastern Indian Oceans, where large and persistent decks of Sc clouds occur. The regions are similar to those examined by Klein and Hartmann (1993) but are slightly modified to accommodate the spatially coherent trend patterns. In particular, we analyze trends in five regions denoted by the white, rectangular boxes in Fig. 1, covering the northeastern Pacific (NEP; 10°–25°N, 115°–145°W), southeastern Pacific (SEP; 5°–25°S, 70°–100°W), northeastern Atlantic (NEA; 10°–30°N, 20°–40°W), southeastern Atlantic (SEA; 12.5°–27.5°S, 10°E–15°W), and southeastern Indian Ocean (SEI; 20°–40°S, 95°–110°E). Multidecadal trends in LCA and meteorological variables are computed during 1984–2009 from the monthly anomalies by least squares linear regression. Meteorological trends for individual reanalyses are shown in Figs. S1–S4 of the supplementary material.
The 1984–2009 linear trends in (a) ISCCP LCA, (b) PATMOS-x LCA, and ensemble mean (c) SST, (d) SLP, (e) EIS, (f) ω700, (g) SST advection, and (h) q700. White rectangular boxes indicate the five marine Sc domains chosen for the analysis. In (a) and (b), grid boxes with trends significant at the 90% confidence level are shaded. In (c)–(h), grid boxes with all four reanalyses sharing the same trend sign are shaded.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
The 1984–2009 linear trends in (a) ISCCP LCA, (b) PATMOS-x LCA, and ensemble mean (c) SST, (d) SLP, (e) EIS, (f) ω700, (g) SST advection, and (h) q700. White rectangular boxes indicate the five marine Sc domains chosen for the analysis. In (a) and (b), grid boxes with trends significant at the 90% confidence level are shaded. In (c)–(h), grid boxes with all four reanalyses sharing the same trend sign are shaded.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
The 1984–2009 linear trends in (a) ISCCP LCA, (b) PATMOS-x LCA, and ensemble mean (c) SST, (d) SLP, (e) EIS, (f) ω700, (g) SST advection, and (h) q700. White rectangular boxes indicate the five marine Sc domains chosen for the analysis. In (a) and (b), grid boxes with trends significant at the 90% confidence level are shaded. In (c)–(h), grid boxes with all four reanalyses sharing the same trend sign are shaded.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
As given in Table 1, during 1984–2009, ISCCP LCA significantly increased by 4% and 3%, respectively, over the northeastern Pacific and southeastern Pacific; significantly decreased by 5% over the northeastern Atlantic; and weakly increased by 1% and 0.3%, respectively, over the southeastern Atlantic and southeastern Indian Ocean. For the same time period, PATMOS-x LCA significantly increased by 4%, 2%, and 3.5%, respectively, over the northeastern Pacific, southeastern Pacific, and southeastern Atlantic; significantly decreased by 2% over the northeastern Atlantic; and weakly increased by 1% over the southeastern Indian Ocean. The impact of correcting the satellite cloud datasets is apparent when the LCA trends in Table 1 are compared to LCA trends for the original datasets in Table S2 in the supplementary material. In many cases trends for the corrected and original cloud data have opposite sign, and trends are generally more negative for the original cloud data. Figure S5 in the supplementary material shows maps of LCA grid-box trends for the original data.
The 1984–2009 trends for satellite cloud and ensemble mean reanalysis meteorology. Parentheses denote regression model LCA trends predicted from ensemble mean meteorology. The values in boldface are significant at the 90% confidence interval, using a t test.
Multidecadal trends in LCA are for the most part physically consistent with trends in ensemble mean large-scale meteorological conditions in all five subtropical Sc regions. Figure 1 and Table 1 indicate that the NEP, SEP, and SEA regions exhibit increasing LCA, decreasing SST, increasing SLP, increasing EIS, and strengthening cold advection over recent decades. The ensemble mean, and most individual reanalyses, agree on the signs of the changes in SST, SLP, and EIS (Table S1 in the supplementary material). Ensemble mean subsidence is also increasing over the NEP, SEP, and SEA regions, although not every individual reanalysis exhibits the same trend sign. Contrastingly, the NEA region exhibits trends of opposite sign to the other subtropical stratocumulus regions. Here, both of the satellite cloud datasets, the ensemble mean, and all four individual reanalyses report decreasing LCA, increasing SST, decreasing subsidence, decreasing SLP, and weakening cold advection over recent decades. The positive trend in EIS appears to be associated with the excessive warming trend in T700 in particular and free-tropospheric temperature in general in the reanalyses (Watanabe et al. 2011).
Trends in LCA and q700 have opposite signs in the NEP, SEP, and NEA regions but a similar sign in the SEA region. The SEI region exhibits little change in LCA over recent decades, and meteorological trends are generally weaker than is the case for the other regions.
b. Correlation between trends in low-level cloud amount and meteorology
In this section, we aggregate the respective grid-box meteorological trends and LCA trends from all five Sc regions. For a particular interval of meteorological trends, grid-box LCA trends are collected from all five subtropical Sc regions and the mean value is computed for each interval. Only intervals with at least 2% of the total number of data points are considered to avoid noise resulting from small sample size. Figures 2 and 3, respectively, display composite average grid-box trends in ISCCP and PATMOS-x LCA as a function of trends in SST, EIS, T700, ω700, SSTadv, and q700. Composite LCA trends are significant at the 90% confidence level in most of the intervals. Table 2 provides the spatial correlation between actual grid-box trends in LCA and ensemble mean meteorological parameters (Table S3 in the supplementary material lists correlation values for individual reanalyses).
Variability in 1984–2009 trends of ISCCP LCA with respect to variability in trends of (a) SST, (b) EIS, (c) T700, (d) ω700, (e) SSTadv, and (f) q700, for the ensemble mean and four reanalyses.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
Variability in 1984–2009 trends of ISCCP LCA with respect to variability in trends of (a) SST, (b) EIS, (c) T700, (d) ω700, (e) SSTadv, and (f) q700, for the ensemble mean and four reanalyses.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
Variability in 1984–2009 trends of ISCCP LCA with respect to variability in trends of (a) SST, (b) EIS, (c) T700, (d) ω700, (e) SSTadv, and (f) q700, for the ensemble mean and four reanalyses.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
As in Fig. 2, but for 1984–2009 trends in PATMOS-x LCA.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
As in Fig. 2, but for 1984–2009 trends in PATMOS-x LCA.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
As in Fig. 2, but for 1984–2009 trends in PATMOS-x LCA.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
Spatial correlation between 1984–2009 grid-box trends in satellite cloud and ensemble mean reanalysis meteorology calculated over all five selected Sc regions. The values in boldface are significant at the 90% confidence interval, using a critical value table.
Consistent with previous observational studies (e.g., Norris and Leovy 1994; Clement et al. 2009), trends in ISCCP and PATMOS-x LCA are negatively correlated with trends in SST across all reanalyses and their ensemble mean (Figs. 2a and 3a). Areas with warming SSTs are associated with diminishing LCA and vice versa. Spatial Pearson correlation values of about 0.8 imply that changes in SST alone can statistically explain about 64% of the variance of the LCA trend pattern. Also consistent with previous observational studies (e.g., Klein and Hartmann 1993; Wood and Bretherton 2006), trends in ISCCP and PATMOS-x LCA are positively correlated to trends in EIS across all reanalyses and their ensemble mean (Figs. 2b and 3b). Areas with increasing stratification of the lower troposphere are associated with increasing LCA and vice versa. Unlike the case for SST, the EIS correlation values substantially differ between reanalyses. This is because there are large disagreements among reanalyses over how trends in T700 are related to trends in LCA (Figs. 2c and 3c). The ERA-Interim and JRA-55 reanalyses and the ensemble mean exhibit negative and statistically insignificant correlations between trends in T700 and LCA, which is opposite what observational studies at synoptic to interannual time scales report (e.g., Watanabe et al. 2011). The fact that T700 trends are positive in almost all grid boxes suggests the presence of excessive free-tropospheric warming of varying magnitude in different reanalyses (Watanabe et al. 2011).
Trends in SSTadv exhibit a negative correlation with trends in LCA in all reanalyses (Figs. 2d and 3d), indicating that the strengthening of cold advection over the SST gradient is associated with increasing LCA, as reported for shorter time scales in previous studies (e.g., Klein et al. 1995). Although not displayed in Figs. 2 and 3, Table 2 shows that there is a strong positive spatial correlation between trends in LCA and trends in SLP. A stronger subtropical anticyclone is associated with stronger trade winds and thereby greater cold advection. Figures 2e and 3e show that trends in LCA are positively correlated with trends in ω700 for the ensemble mean as well as the individual reanalyses. This is consistent with the positive correlation between subsidence and LCA previously reported for the climatology (e.g., Myers and Norris 2013) and on decadal time scales (e.g., Clement et al. 2009). Trends in q700 have a negative correlation with trends in LCA for the ensemble mean and three reanalyses (Figs. 2f and 3f), indicating that drying of the free troposphere is associated with increasing LCA (e.g., Klein et al. 1995). For unidentified reasons, the JRA-55 reanalysis shows a slight positive relationship between the q700 trends and LCA trends. It is likely that there is not a direct physical mechanism connecting enhanced LCA with stronger subsidence and reduced free-tropospheric humidity. Rather, these meteorological conditions co-occur with stronger EIS and cold advection, which instead dominate the cloud response.
Spurious variability in the original cloud datasets tends to be spatially uniform within each subtropical Sc region and therefore has only a minor impact on the spatial correlation between grid-box trends in LCA and ensemble mean meteorological parameters. Values of the spatial correlation between grid-box meteorological trends and grid-box LCA trends from the original cloud datasets (Table S4 in the supplementary material) are generally weaker than those for the corrected cloud datasets (Table 2), but otherwise they are similar.
c. Estimated low-level cloud amount trends from meteorology
The above results indicate that some combination of changes in SST, EIS, ω700, SSTadv, q700, and SLP could explain the majority of the variability in observed LCA trends. We accordingly estimate multidecadal LCA trends by applying a multilinear regression model to the above-discussed meteorological fields. One advantage of a multilinear regression model is that it will distinguish the impacts of each of the predictor (meteorological) variables in predicting the LCA trends. We choose SST, EIS, and SSTadv as predictors because their trends are highly spatially correlated with LCA trends, as assessed in the previous sections. Moreover, Myers and Norris (2015) also found that LCA anomalies are most closely related to SST, EIS, and SSTadv anomalies. Including ω700, SLP, T700, or q700 as predictors does not improve the performance of the regression model, so we leave them out. Presumably, ω700, SLP, T700, and q700 provide little independent information beyond that delivered by SST, EIS, and SSTadv.
Table 3 lists the coefficients for the multilinear regression model based on ensemble mean meteorological anomalies. The good agreement between coefficients derived from ISCCP and PATMOS-x LCA anomalies provides confidence in the model and the data. Moreover, quantitatively similar coefficients are obtained from a partial-derivative compositing method, following Myers and Norris (2013). Coefficients calculated for individual reanalyses have the same sign but slightly vary in magnitude, providing additional confidence in the relationship between LCA and meteorology. The coefficients for all predictors are significant at the 90% confidence level using a t test, assuming an effective sample size that takes spatial and temporal autocorrelation into account. Coefficients obtained using the original cloud datasets, listed in Table S5 of the supplementary material, have the same sign but somewhat different magnitude than those obtained from the corrected datasets for SST and EIS. Spurious variability in the original datasets primarily occurs at time scales longer than interannual and thus has only a minor impact on regression coefficients calculated from detrended time series.
Regression coefficients for the multilinear regression model constructed from ensemble mean meteorology.
The coefficients from the regression model reveal that cooler SST, a stronger inversion, and enhanced cold-air advection are each independently associated with increased LCA. Covariability among these three meteorological factors does not affect this interpretation (the data pass a variance inflation factor test for multicollinearity commonly used by statisticians). The regression coefficients obtained from the standard deviation normalized anomalies of the ensemble mean as well as most of the individual reanalyses suggest that anomalies in SST advection are the most dominant factor in independently determining anomalies in LCA. Anomalies in the strength of the inversion layer are the second most important factor, and anomalies in SST are the least important factor.
Figure 4 presents regional mean ISCCP and PATMOS-x LCA trends predicted by the multilinear regression model from the ensemble mean and individual reanalyses for each of the five Sc regions. Regional mean LCA trends were obtained from regional mean trends in SST, EIS, and SSTadv via multiplication by the multilinear regression model coefficients. The separate contributions of the SST, EIS, and SSTadv terms are also displayed, along with the observed regional mean LCA trends. The regression model captures the sign of the observed trends in most of the Sc regions for the ensemble mean, as well as for the individual reanalyses. In most cases, the predicted trends are also within the observational uncertainty range, though the observed trends have relatively large uncertainty since the time record is only 25 yr. In all five Sc regions, the term (∂LCA/∂EIS) × ΔEIS contributes the most to the predicted LCA trends. Although SSTadv has the dominant influence on monthly LCA anomalies in the model, EIS trends are larger than SSTadv trends and therefore have the dominant influence on predicted LCA trends. Trends in SST generally make the smallest independent contribution to trends in LCA and are nonnegligible only for the NEP and NEA regions. Table 1 lists predicted trends based on the ensemble mean meteorology for each Sc region, and Table S1 lists predicted trends for individual reanalyses.
The 1984–2009 trends in (left) ISCCP and (right) PATMOS-x LCA as observed (gray) and as predicted by the multilinear regression model for ensemble mean meteorology (orange) along with the independent contributing terms due to SST (blue), EIS (yellow), and SSTadv (green) for each of five marine Sc regions. Vertical error bars indicate the 90% confidence interval of the observed LCA trends as determined by a t test. Predicted trends for individual reanalyses are indicated by symbols according to the legend.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
The 1984–2009 trends in (left) ISCCP and (right) PATMOS-x LCA as observed (gray) and as predicted by the multilinear regression model for ensemble mean meteorology (orange) along with the independent contributing terms due to SST (blue), EIS (yellow), and SSTadv (green) for each of five marine Sc regions. Vertical error bars indicate the 90% confidence interval of the observed LCA trends as determined by a t test. Predicted trends for individual reanalyses are indicated by symbols according to the legend.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
The 1984–2009 trends in (left) ISCCP and (right) PATMOS-x LCA as observed (gray) and as predicted by the multilinear regression model for ensemble mean meteorology (orange) along with the independent contributing terms due to SST (blue), EIS (yellow), and SSTadv (green) for each of five marine Sc regions. Vertical error bars indicate the 90% confidence interval of the observed LCA trends as determined by a t test. Predicted trends for individual reanalyses are indicated by symbols according to the legend.
Citation: Journal of Climate 28, 21; 10.1175/JCLI-D-15-0120.1
ISCCP and PATMOS-x report similar observed LCA trends for the NEP and SEP regions. The multilinear regression model based on ensemble mean meteorology overpredicts LCA trends in the NEP (Figs. 4a and 4b) but is closer in the SEP (Figs. 4c and 4d). Trends predicted from ERA-Interim and JRA-55 meteorology, however, are very close to the observed trends in the NEP. Observed ISCCP and PATMOS-x LCA trends exhibit greater differences for the SEA region, but model-predicted trends derived from ensemble mean meteorology are nearly the same and halfway between the observed ISCCP and PATMOS-x trends (Figs. 4g and 4h). ERA-Interim and CFSR underpredict the observed LCA trends and MERRA and JRA-55 overpredict the LCA trends relative to ISCCP in the SEA region. Over the SEI region, the multilinear regression model overpredicts LCA trends for the ensemble mean and individual reanalyses, though the observed trends are insignificant and exhibit a very large uncertainty range (Figs. 4i and 4j). In general, CFSR predicts the smallest LCA trends and underperforms over all domains. The fact that model-predicted LCA trends are nearly the same for ISCCP and PATMOS-x for every reanalysis and the ensemble mean suggests that monthly LCA anomalies are nearly the same for ISCCP and PATMOS-x.
The largest disagreement between the observations and predictions by the multilinear regression model occurs for the NEA region (Figs. 4e and 4f). Here, the observed LCA trends are negative, especially for ISCCP, but the predicted LCA trends are near zero for the ensemble mean as well as for most of the individual reanalyses. It may be the case that the observed negative LCA trend over the NEA is unrealistic. The NEA is a major dust region, mainly transported from the Sahara Desert (Carlson and Prospero 1972; Kaufman et al. 2005). The dust layer lies above or overlaps the Sc layer depending on the respective season (summer or winter) (Pradelle et al. 2002). If the dust resides above the Sc clouds, this could introduce local heating at 700 hPa and enhance the inversion strength, leading to a larger cloud amount in the MBL (Johnson et al. 2004; Wilcox 2010). Alternatively, if it intersects with the cloud layer, it could interact with Sc microphysics and lead to longer cloud lifetimes through an aerosol indirect effect, though local dust heating could lead to thinner clouds (Levin et al. 1996; Hansen et al. 1997; Feingold et al. 1999; Wurzler et al. 2000; Johnson et al. 2004). Taken together, the recent reduction in dust over the NEA due to increased rain over the Sahel (Brooks and Legrand 2000; Prospero and Lamb 2003; Foltz and McPhaden 2008) may have caused a reduction in LCA over the NEA that is not related to any changes in SST, EIS, or SSTadv. Another likely possibility is that ISCCP has misclassified thick dust layers as clouds such that a negative dust trend appears as a negative LCA trend. However, additional studies are required to determine with certainty the sign of the LCA trend.
Figure S6 in the supplementary material presents regional mean ISCCP and PATMOS-x LCA trends predicted by the multilinear regression model using the original cloud datasets. The predicted LCA trends have opposite sign from the observed original ISCCP trends in four Sc regions and are outside the uncertainty range of the observed original PATMOS-x trends in two Sc regions. The disagreement between predicted and observed trends for the original cloud data is consistent with the presence of spurious trends in the original cloud data. It is interesting to note that the predicted LCA trends derived from the original cloud datasets are within the uncertainty range of the observed LCA trends from the corrected cloud datasets for every Sc region except the NEA region for ISCCP. These results suggest that the correction procedure applied to the satellite datasets removed spurious variability but not real variability at regional scales.
4. Summary and conclusions
Cloud cover changes modulate surface solar radiation on an interannual basis, but their contribution to longer-term trends has been uncertain because of questions about the reliability of cloud cover trends reported by satellite datasets and surface observations. For this reason, it is essential to examine the consistency of trends in clouds with trends in physically related meteorological parameters. In this study, we investigated multidecadal trends of low-level cloud amount in five marine subtropical stratocumulus regions using two independent satellite records corrected for inhomogeneities and meteorological parameters from four different reanalyses. We employed a multilinear regression model to predict trends in low-level cloud amount from trends in SST, EIS, and advection over the SST gradient on the basis of observed interannual relationships between low-level cloud amount and the meteorological parameters. Consistent with previous observational studies, we also found that trends in low-level cloud amount are positively correlated to trends in SLP and subsidence and are negatively correlated to the trends in free-tropospheric humidity, but these parameters do not add any more skill beyond that already included in our multilinear regression model.
During 1984–2009, low-level cloud amount increased over the northeastern Pacific, southeastern Pacific, and southeastern Atlantic according to both ISCCP and PATMOS-x. These regions experienced decreasing SST, increasing EIS, and strengthening cold advection during the same time period. The close agreement between the observed cloud trends and those predicted by the multilinear regression model demonstrates that observed trends in SST, EIS, and cold advection were key to generating the observed cloud trends. Contrastingly, the northeastern Atlantic region exhibits decreasing low-level cloud amount, increasing SST, stronger EIS, and weakening cold advection. These trends are generally of opposite sign to the other subtropical stratocumulus regions. The observed ISCCP cloud trend over the northeastern Atlantic is large and negative whereas the model-predicted trend is near zero. It may be the case that the decrease in cloud reported by ISCCP actually represents a decrease in dust or a response to this decrease in dust. The southeastern subtropical Indian Ocean region exhibits little change in low-level cloud amount over recent decades, and the meteorological trends are generally weaker than is the case for the other regions. Although regional mean cloud trends predicted by the model from the ensemble mean are within the range of observational uncertainty, this is not always the case for cloud trends predicted from individual reanalyses. The spread of the cloud trends predicted by the multilinear regression model among the four reanalyses can be quite substantial, indicating that uncertainties in multidecadal meteorological changes are comparable to uncertainties in satellite-observed cloud changes. Nevertheless, the predicted trends almost always have the same sign as the observed trends.
In our multilinear regression model, we normalize coefficients by dividing each by the standard deviation of monthly anomalies in their respective predictor variables. This enables us to compare the cloud response to a typical size anomaly in each predictor. Within this framework, anomalies in the strength of cold advection have the largest impact on predicted monthly low-level cloud anomalies, but EIS trends are the most important independent contributors to predicted low-level cloud trends. Out of the three parameters, SST trends have the smallest independent effect on predicted cloud trends and are nonnegligible only for the northeastern Pacific and Atlantic regions. The relative contributions of trends in SST, EIS, and the strength of cold advection to the total predicted trend in low-level cloud amount vary from one region to another. Trends in EIS almost entirely drive cloud trends over the southeastern Atlantic and southeastern Indian Oceans, whereas trends in the strength of cold advection substantially contribute to cloud trends over the northeastern and southeastern Pacific. In four of the regions, trends in the meteorological parameters independently produce cloud trends with the same sign, but over the northeastern Atlantic, trends in SST and the strength of cold advection produce cloud trends with signs opposing and canceling the cloud trend produced by the EIS trend.
Are the low-level cloud amount changes observed during 1984–2009 in subtropical stratocumulus regions the result of natural variability, or are they rather a systematic change associated with global warming? One argument for attribution to natural causes is that the sign and magnitude of cloud changes and the relative contributions of meteorological factors are dissimilar between regions. Indeed, Qu et al. (2014) find that climate models generally simulate spatially uniform low cloud trends over the five main stratocumulus regions of the globe in simulations of twenty-first century climate change. We find that the largest cloud increases occur over the northeastern and southeastern Pacific and appear to be associated with coupled atmosphere–ocean decadal variability (e.g., Burgman et al. 2008; Clement et al. 2009). Cloud trends in the northeastern Atlantic and southeastern Atlantic have opposite sign, and little cloud change occurs in the southeastern Indian Ocean. One direction for future work would be to examine subtropical stratocumulus trends in global climate models. If the pattern of cloud change in the last 25 years of a historical simulation of the twentieth century resembled the pattern of cloud change associated with long-term greenhouse warming, it would suggest that the observed cloud trends described in this study may provide insight into long-term cloud feedbacks.
Acknowledgments
This study was funded by NSF Award AGS-0946094. ISCCP data were downloaded from the Atmospheric Science Data Center located at NASA Langley Research Center. PATMOS-x data were obtained from the Cooperative Institute for Meteorological Satellite Studies located at the University of Wisconsin–Madison. The CFSR and JRA-55 data were provided by the National Center for Atmospheric Research. ERA-Interim data were downloaded from the ECMWF data server (apps.ecmwf.int). The Global Modeling and Assimilation Office and the Goddard Earth Sciences Data and Information Services Center provided the MERRA data. We also thank three anonymous reviewers for their constructive comments on the original manuscript.
REFERENCES
Albrecht, B. A., D. A. Randall, and S. Nicholls, 1988: Observations of marine stratocumulus during FIRE. Bull. Amer. Meteor. Soc., 69, 618–626, doi:10.1175/1520-0477(1988)069<0618:OOMSCD>2.0.CO;2.
Albrecht, B. A., M. P. Jensen, and W. J. Syrett, 1995: Marine boundary layer structure and fractional cloudiness. J. Geophys. Res., 100, 14 209–14 222, doi:10.1029/95JD00827.
Arndt, D. S., M. O. Baringer, and M. R. Johnson, 2010: State of the climate in 2009. Bull. Amer. Meteor. Soc., 91 (7), S1–S224, doi:10.1175/BAMS-91-7-StateoftheClimate.
Bengtsson, L., S. Hagemann, and K. I. Hodges, 2004: Can climate trends be calculated from reanalysis data? J. Geophys. Res., 109, D11111, doi:10.1029/2004JD004536.
Berrisford, P., D. P. Dee, K. Fielding, M. Fuentes, P. Kallberg, S. Kobayashi, and S. M. Uppala, 2009: The ERA-Interim archive. ERA Rep. Series 1, ECMWF, 16 pp.
Bony, S., and J.-L. Dufresne, 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.
Boucher, O., and Coauthors, 2013: Clouds and aerosols. Climate Change 2013: The Physical Science Basis, J. T. Stocker et al., Eds., Cambridge University Press, 571–657.
Bretherton, C. S., and W. C. Wyant, 1997: Moisture transport, lower-tropospheric stability, and decoupling of cloud-topped boundary layers. J. Atmos. Sci., 54, 148–167, doi:10.1175/1520-0469(1997)054<0148:MTLTSA>2.0.CO;2.
Brooks, N., and M. Legrand, 2000: Dust variability over northern Africa and rainfall in the Sahel. Linking Climate Change to Land Surface Change, S. J. McLaren and D. R. Kniveton, Eds., Kluwer Academic, 1–25.
Burgman, R. J., A. C. Clement, C. M. Mitas, J. Chen, and K. Esslinger, 2008: Evidence for atmospheric variability over the Pacific on decadal timescales. Geophys. Res. Lett., 35, L01704, doi:10.1029/2007GL031830.
Campbell, G. G., 2004: View angle dependence of cloudiness and the trend in ISCCP cloudiness. Preprints, 13th Conf. on Satellite Meteorology and Oceanography, Norfolk, VA, Amer. Meteor. Soc., P6.7. [Available online at https://ams.confex.com/ams/pdfpapers/79041.pdf.]
Carlson, T. N., and J. M. Prospero, 1972: The large-scale movement of Saharan air outbreaks over the northern equatorial Atlantic. J. Appl. Meteor., 11, 283–297, doi:10.1175/1520-0450(1972)011<0283:TLSMOS>2.0.CO;2.
Chelliah, M., W. Ebisuzaki, S. Weaver, and A. Kumar, 2011: Evaluating the tropospheric variability in National Centers for Environmental Prediction’s climate forecast system reanalysis. J. Geophys. Res., 116, D17107, doi:10.1029/2011JD015707.
Clement, A. C., R. Burgman, and J. R. Norris, 2009: Observational and model evidence for positive low-level cloud feedback. Science, 325, 460–464, doi:10.1126/science.1171255.
Deardorff, J., 1976: Entrainment rate of stratocumulus-topped mixed layer. Quart. J. Roy. Meteor. Soc., 102, 563–582, doi:10.1002/qj.49710243306.
Dee, D. P., and S. Uppala, 2009: Variational bias correction of satellite radiance data in the ERA-Interim reanalysis. Quart. J. Roy. Meteor. Soc., 135, 1830–1841, doi:10.1002/qj.493.
Dee, D. P., and Coauthors, 2011a: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, doi:10.1002/qj.828.
Dee, D. P., E. Källén, A. J. Simmons, and L. Haimberger, 2011b: Comments on “Reanalyses suitable for characterizing long-term trends.” Bull. Amer. Meteor. Soc., 92, 65–70, doi:10.1175/2010BAMS3070.1.
Derber, J. C., R. J. Purser, W.-S. Wu, R. Treadon, M. Pondeca, D. Parrish, and D. Kleist, 2003: Flow-dependent Jb in a global grid-point 3DVAR. Proc. Annual Seminar on Recent Developments in Data Assimilation for Atmosphere and Ocean, Reading, United Kingdom, ECMWF, 125–134.
Eastman, R., S. G. Warren, and C. J. Hahn, 2011: Variations in cloud cover and cloud types over the ocean from surface observations, 1954–2008. J. Climate, 24, 5914–5934, doi:10.1175/2011JCLI3972.1.
Ebita, A., and Coauthors, 2011: The Japanese 55-year Reanalysis “JRA-55”: An interim report. SOLA, 7, 149–152, doi:10.2151/sola.2011-038.
Eitzen, Z. A., K.-M. Xu, and T. Wong, 2011: An estimate of low-cloud feedbacks from variations of cloud radiative and physical properties with sea surface temperature on interannual time scales. J. Climate, 24, 1106–1121, doi:10.1175/2010JCLI3670.1.
Evan, A. T., A. K. Heidinger, and D. J. Vimont, 2007: Arguments against a physical long-term trend in global ISCCP cloud amounts. Geophys. Res. Lett., 34, L04701, doi:10.1029/2006GL028083.
Feingold, G., W. R. Cotton, S. M. Kreidenweiss, and J. T. Davis, 1999: The impact of giant cloud condensation nuclei on drizzle formation in stratocumulus: Implications for cloud radiative properties. J. Atmos. Sci., 56, 4100–4117, doi:10.1175/1520-0469(1999)056<4100:TIOGCC>2.0.CO;2.
Foltz, G. R., and M. J. McPhaden, 2008: Trends in Saharan dust and tropical Atlantic climate during 1980–2006. Geophys. Res. Lett., 35, L20706, doi:10.1029/2008GL035042.
Forster, P., and Coauthors, 2007: Changes in atmospheric constituents and in radiative forcing. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 130–234.
Garay, M. J., S. P. de Szoeke, and C. M. Moroney, 2008: Comparison of marine stratocumulus cloud top heights in the southeastern Pacific retrieved from satellites with coincident ship-based observations. J. Geophys. Res., 113, D18204, doi:10.1029/2008JD009975.
Georgakakos, K., and R. Bras, 1984: A hydrologically useful station precipitation model. 1. Formulation. Water Resour. Res., 20, 1585–1596, doi:10.1029/WR020i011p01585.
Hansen, J., M. Sato, and R. Ruedy, 1997: Radiative forcing and climate response. J. Geophys. Res., 102, 6831–6864, doi:10.1029/96JD03436.
Hanson, H. P., 1991: Marine stratocumulus climatologies. Int. J. Climatol., 11, 147–164, doi:10.1002/joc.3370110204.
Heidinger, A. K., M. J. Foster, A. Walther, and X. Zhao, 2014: The Pathfinder Atmospheres–Extended AVHRR climate dataset. Bull. Amer. Meteor. Soc., 95, 909–922, doi:10.1175/BAMS-D-12-00246.
Johnson, B. T., K. P. Shine, and P. M. Forster, 2004: The semi-direct aerosol effect: Impact of absorbing aerosols on marine stratocumulus. Quart. J. Roy. Meteor. Soc., 130, 1407–1422, doi:10.1256/qj.03.61.
Kaufman, Y. J., I. Koren, L. A. Remer, D. Tanre, P. Ginoux, and S. Fan, 2005: Dust transport and deposition observed from the Terra-Moderate Resolution Imaging Spectroradiometer (MODIS) spacecraft over the Atlantic Ocean. J. Geophys. Res., 110, D10S12, doi:10.1029/2003JD004436.
Klein, S. A., and D. L. Hartmann, 1993: The seasonal cycle of low stratiform clouds. J. Climate, 6, 1587–1606, doi:10.1175/1520-0442(1993)006<1587:TSCOLS>2.0.CO;2.
Klein, S. A., D. Hartmann, and J. Norris, 1995: On the relationships among low-cloud structure, sea surface temperature, and atmospheric circulation in the summertime northeast Pacific. J. Climate, 8, 1140–1155, doi:10.1175/1520-0442(1995)008<1140:OTRALC>2.0.CO;2.
Kumar, A., and Z. Z. Hu, 2012: Uncertainty in the ocean–atmosphere feedbacks associated with ENSO in the reanalysis products. Climate Dyn., 39, 575–588, doi:10.1007/s00382-011-1104-3.
Lacagnina, C., and F. Selten, 2013: A novel diagnostic technique to investigate cloud-controlling factors. J. Geophys. Res. Atmos., 118, 5979–5991, doi:10.1002/jgrd.50511.
Levin, Z., E. Ganor, and V. Gladstein, 1996: The effects of desert particles coated with sulfate on rain formation in the eastern Mediterranean. J. Appl. Meteor., 35, 1511–1523, doi:10.1175/1520-0450(1996)035<1511:TEODPC>2.0.CO;2.
Mace, G., and Coauthors, 2006: Cloud radiative forcing at the Atmospheric Radiation Measurement Program Climate Research Facility: 1. Technique, validation, and comparison to satellite-derived diagnostic quantities. J. Geophys. Res., 111, D11S90, doi:10.1029/2005JD005921.
Mansbach, D. K., and J. R. Norris, 2007: Low-level cloud variability over the equatorial cold tongue in observations and models. J. Climate, 20, 1555–1570, doi:10.1175/JCLI4073.1.
Mauger, G., and J. Norris, 2010: Assessing the impact of meteorological history on subtropical cloud fraction. J. Climate, 23, 2926–2940, doi:10.1175/2010JCLI3272.1.
Mears, C. A., and F. J. Wentz, 2009: Construction of the Remote Sensing Systems V3.2 lower-tropospheric temperature datasets from the MSU and AMSU microwave sounders. J. Atmos. Oceanic Technol., 26, 1040–1056, doi:10.1175/2008JTECHA1176.1.
Minnis, P., P. Heck, D. Young, C. Fairall, and J. Snider, 1992: Stratocumulus cloud properties derived from simultaneous satellite and island-based instrumentation during FIRE. J. Appl. Meteor., 31, 317–339, doi:10.1175/1520-0450(1992)031<0317:SCPDFS>2.0.CO;2.
Myers, T. A., and J. R. Norris, 2013: Observational evidence that enhanced subsidence reduces subtropical marine boundary layer cloudiness. J. Climate, 26, 7507–7524, doi:10.1175/JCLI-D-12-00736.1.
Myers, T. A., and J. R. Norris, 2015: On the relationships between subtropical clouds and meteorology in observations and CMIP3 and CMIP5 models. J. Climate, 8, 2945–2967, doi:10.1175/JCLI-D-14-00475.1.
Norris, J. R., 1998: Low cloud type over the ocean from surface observations. Part I: Relationship to surface meteorology and the vertical distribution of temperature and moisture. J. Climate, 11, 369–382, doi:10.1175/1520-0442(1998)011<0369:LCTOTO>2.0.CO;2.
Norris, J. R., 2005: Trends in upper-level cloud cover and surface divergence over the tropical Indo-Pacific Ocean between 1952 and 1997. J. Geophys. Res., 110, D21110, doi:10.1029/2005JD006183.
Norris, J. R., and C. B. Leovy, 1994: Interannual variability in stratiform cloudiness and sea surface temperature. J. Climate, 7, 1915–1925, doi:10.1175/1520-0442(1994)007<1915:IVISCA>2.0.CO;2.
Norris, J. R., and A. T. Evan, 2015: Empirical removal of artifacts from the ISCCP and PATMOS-x satellite cloud records. J. Atmos. Oceanic Technol., 32, 691–702, doi:10.1175/JTECH-D-14-00058.1.
O’Dell, C., F. Wentz, and R. Bennartz, 2008: Cloud liquid water path from satellite-based passive microwave observations: A new climatology over the global oceans. J. Climate, 21, 1721–1739, doi:10.1175/2007JCLI1958.1.
Onogi, K., and Coauthors, 2007: The JRA-25 Reanalysis. J. Meteor. Soc. Japan, 85, 369–432, doi:10.2151/jmsj.85.369.
Pradelle, F., G. Cautenet, and I. Jankowiak, 2002: Radiative and microphysical interactions between marine stratocumulus clouds and Saharan dust 1. Remote sensing observations. J. Geophys. Res., 107, 4412, doi:10.1029/2000JD000155.
Prospero, J. M., and P. J. Lamb, 2003: African droughts and dust transport to the Caribbean: Climate change implications. Science, 302, 1024–1027, doi:10.1126/science.1089915.
Purser, R. J., W.-S. Wu, D. F. Parrish, and N. M. Roberts, 2003a: Numerical aspects of the application of recursive filters to variational statistical analysis. Part I: Spatially homogeneous and isotropic Gaussian covariances. Mon. Wea. Rev., 131, 1524–1535, doi:10.1175//1520-0493(2003)131<1524:NAOTAO>2.0.CO;2.
Purser, R. J., W.-S. Wu, D. F. Parrish, and N. M. Roberts, 2003b: Numerical aspects of the application of recursive filters to variational statistical analysis. Part II: Spatially inhomogeneous and anisotropic general covariances. Mon. Wea. Rev., 131, 1536–1548, doi:10.1175/2543.1.
Qu, X., A. Hall, S. A. Klein, and P. M. Caldwell, 2014: On the spread of changes in marine low cloud cover in climate model simulations of the 21st century. Climate Dyn., 42, 2603–2626, doi:10.1007/s00382-013-1945-z.
Qu, X., A. Hall, S. A. Klein, and P. M. Caldwell, 2015: The strength of the tropical inversion and its response to climate change in 18 CMIP5 models. Climate Dyn., 45, 375–396, doi:10.1007/s00382-014-2441-9.
Randall, D. A., J. A. Coakley, C. W. Fairall, R. A. Knopfli, and D. H. Lenschow, 1984: Outlook for research on marine subtropical stratocumulus clouds. Bull. Amer. Meteor. Soc., 65, 1290–1301, doi:10.1175/1520-0477(1984)065<1290:OFROSM>2.0.CO;2.
Rausch, J., A. Heidinger, and R. Bennartz, 2010: Regional assessment of microphysical properties of marine boundary layer cloud using the PATMOS-x dataset. J. Geophys. Res., 115, D23212, doi:10.1029/2010JD014468.
Rienecker, M. M., and Coauthors, 2008: The GEOS-5 Data Assimilation System—Documentation of versions 5.0.1 and 5.1.0, and 5.2.0. NASA/TM-2008-104606, NASA Tech. Rep. Series on Global Modeling and Data Assimilation, Vol. 27, 92 pp.
Rienecker, M. M., and Coauthors, 2011: MERRA: NASA’s Modern-Era Retrospective Analysis for Research and Application. J. Climate, 24, 3624–3648, doi:10.1175/JCLI-D-11-00015.1.
Rossow, W., and R. Schiffer, 1999: Advances in understanding clouds from ISCCP. Bull. Amer. Meteor. Soc., 80, 2261–2287, doi:10.1175/1520-0477(1999)080<2261:AIUCFI>2.0.CO;2.
Rozendaal, M., C. Leovy, and S. Klein, 1995: An observational study of diurnal variations of marine stratiform cloud. J. Climate, 8, 1795–1809, doi:10.1175/1520-0442(1995)008<1795:AOSODV>2.0.CO;2.
Saha, S., and Coauthors, 2006: The NCEP Climate Forecast System. J. Climate, 19, 3483–3517, doi:10.1175/JCLI3812.1.
Saha, S., and Coauthors, 2010: The NCEP Climate Forecast System Reanalysis. Bull. Amer. Meteor. Soc., 91, 1015–1057, doi:10.1175/2010BAMS3001.1.
Sandu, I., and B. Stevens, 2011: On the factors modulating the stratocumulus to cumulus transitions. J. Atmos. Sci., 68, 1865–1881, doi:10.1175/2011JAS3614.1.
Schubert, W. H., J. S. Wakefield, E. J. Steiner, and S. K. Cox, 1979: Marine stratocumulus convection. Part I: Governing equations and horizontally homogeneous solutions. J. Atmos. Sci., 36, 1286–1307, doi:10.1175/1520-0469(1979)036<1286:MSCPIG>2.0.CO;2.
Simmons, A. J., M. Hortal, G. Kelly, A. McNally, A. Untach, and S. Uppala, 2005: ECMWF analyses and forecasts of stratospheric winter polar vortex breakup: September 2002 in the Southern Hemisphere and related events. J. Atmos. Sci., 62, 668–689, doi:10.1175/JAS-3322.1.
Simmons, A. J., P. Poli, D. P. Dee, P. Berrisford, H. Hersbach, S. Kobayashi, and C. Peubey, 2014: Estimating low-frequency variability and trends in atmospheric temperature using ERA-Interim. Quart. J. Roy. Meteor. Soc., 140, 329–353, doi:10.1002/qj.2317.
Slingo, A., 1990: Sensitivity of the Earth’s radiation budget to changes in low clouds. Nature, 343, 49–51, doi:10.1038/343049a0.
Stachnik, J. P., and C. Schumacher, 2011: A comparison of the Hadley circulation in modern reanalyses. J. Geophys. Res., 116, D22102, doi:10.1029/2011JD016677.
Stevens, B., and J.-L. Brenguier, 2009: Cloud controlling factors—Low clouds. Perturbed Clouds in the Climate System: Their Relationship to Energy Balance, Atmospheric Dynamics and Precipitation, J. Heintzenberg and R. J. Charlson, Eds., Strungmann Forum Reports, Vol. 2, The MIT Press, 173–196.
Stubenrauch, C. J., and Coauthors, 2013: Assessment of global cloud datasets from satellite: Project and database initiated by the GEWEX radiation panel. Bull. Amer. Meteor. Soc., 94, 1031–1049, doi:10.1175/BAMS-D-12-00117.1.
Watanabe, M., H. Shiogama, T. Yokohata, T. Ogura, M. Yoshimori, S. Emori, and M. Kimoto, 2011: Constraints to the tropical low-cloud trends in historical climate simulations. Atmos. Sci. Lett., 12, 288–293, doi:10.1002/asl.337.
Wilcox, E. M., 2010: Stratocumulus cloud thickening beneath layers of absorbing smoke aerosol. Atmos. Chem. Phys., 10, 11 769–11 777, doi:10.5194/acp-10-11769-2010.
Willett, K. M., P. D. Jones, P. W. Thorne, and N. P. Gillett, 2010: A comparison of large scale changes in surface humidity over land in observations and CMIP3 general circulation models. Environ. Res. Lett., 5, 025210, doi:10.1088/1748-9326/5/2/025210.
Wood, R., 2012: Stratocumulus clouds. Mon. Wea. Rev., 140, 2373–2423, doi:10.1175/MWR-D-11-00121.1.
Wood, R., and C. S. Bretherton, 2006: On the relationship between stratiform low cloud cover and lower-tropospheric stability. J. Climate, 19, 6425–6432, doi:10.1175/JCLI3988.1.
Wood, R., and D. L. Hartmann, 2006: Spatial variability of liquid water path in marine low cloud: The importance of mesoscale cellular convection. J. Climate, 19, 1748–1764, doi:10.1175/JCLI3702.1.
Wu, W.-S., R. J. Purser, and D. F. Parrish, 2002: Three-dimensional variational analysis with spatially inhomogeneous covariances. Mon. Wea. Rev., 130, 2905–2916, doi:10.1175/1520-0493(2002)130<2905:TDVAWS>2.0.CO;2.
Wurzler, S., T. G. Reisin, and Z. Levin, 2000: Modification of mineral dust particles by cloud processing and subsequent effects on drop size distributions. J. Geophys. Res., 105, 4501–4512, doi:10.1029/1999JD900980.
Zhang, Y., B. Stevens, B. Medeiros, and M. Ghil, 2009: Low-cloud fraction, lower-tropospheric stability, and large-scale divergence. J. Climate, 22, 4827–4844, doi:10.1175/2009JCLI2891.1.
